NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Gartlehner G, Hansen RA, Morgan LC, et al. Second-Generation Antidepressants in the Pharmacologic Treatment of Adult Depression: An Update of the 2007 Comparative Effectiveness Review [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 Dec. (Comparative Effectiveness Reviews, No. 46.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Second-Generation Antidepressants in the Pharmacologic Treatment of Adult Depression

Second-Generation Antidepressants in the Pharmacologic Treatment of Adult Depression: An Update of the 2007 Comparative Effectiveness Review [Internet].

Show details

Results

This chapter is organized as follows: first by Key Question (KQ), second by subquestion or subpopulation, and third by intervention comparison. In addition, according to the specifications from the Agency for Healthcare Research and Quality (AHRQ) for comparative effectiveness reviews (CER), within each KQ section we present an overview, then key points, and finally detailed analyses. Finally, as explained in Methods, we graded the strength of evidence for all major comparisons and outcomes in the key points. Table 7 summarizes the main issues that we address here.

Table 7. Key Questions about the comparative efficacy and safety of second-generation antidepressants.

Table 7

Key Questions about the comparative efficacy and safety of second-generation antidepressants.

We focus on randomized controlled trials (RCTs) for all questions; for KQ 2 on maintaining remission and treating unresponsive or recurrent disease, and KQ 4 on harms, we also include observational studies. Evidence tables for all included studies, by Key Question, are presented in Appendix C.

Reasons for exclusion were based on eligibility criteria or methodological criteria. We excluded 77 studies that originally met eligibility criteria but were later rated as poor quality for internal validity (Appendix D). The two main reasons for rating RCTs as poor were high loss to followup (more than 40 percent overall) and lack of intention-to-treat (ITT) analysis. Among meta-analyses, lack of a systematic literature search was the main reason for exclusion; this problem leads to a selected spectrum of trials and subsequently to biased results.

Studies reviewed for this report employed a notable array of diagnostic scales and health status or quality-of-life instruments. Most were pertinent to depressive and other disorders considered in this report, but some are considered more generic instruments that assess, for example, health-related quality of life. Table 8 lists abbreviations of diagnostic scales and health status or quality-of-life instruments encountered in this literature.

Table 8. Abbreviations and full names of diagnostic scales and other instruments.

Table 8

Abbreviations and full names of diagnostic scales and other instruments.

Because this report is an update of the original CER on second-generation antidepressants,12 we identify all new studies in the summary tables of included studies in each detailed analysis section.

Overview of all Key Questions

We identified 3,722 citations from searches and reviews of reference lists. Figure 3 documents the disposition of the 267 included articles in this review, working from 1,457 articles retrieved for full text review and 1,190 excluded at this stage.

Figure 3 is a PRISMA tree that documents the disposition of the articles reviewed for this review. There were 6,186 citations identified from searches across databases. Additionally, 167 articles were detected from manually reviewing the reference lists of pertinent review articles. Of the total 3,722 abstracts screened, 2,265 citations were excluded. Working from 1,457 articles retrieved for full review, 1,190 were excluded at this stage. Of the studies excluded at the full review 197 were excluded for wrong outcome, 84 were excluded for including the wrong population, 260 were excluded for wrong publication type, 279 were excluded due to the analysis of outcomes not of interest, 464 were excluded because the study was the wrong study design, 7 for foreign language, 10 for too short of a duration, 142 for wrong drug and 79 for poor quality. Multiple exclusion reasons are possible for each article. We included 267 published articles reporting on 228 studies in the qualitative synthesis and 92 studies were included in the quantitative synthesis mixed treatment comparisons and meta-analyses).

Figure 3

Results of literature search (PRISMA diagram).

We included 264 articles reporting on 248 studies of good or fair quality: 104 head-to-head randomized controlled trials (RCTs), 84 placebo-controlled RCTs, 46 meta-analyses or systematic reviews, observational studies, and studies of other design. We incorporated data from 14 additional placebo-controlled studies for indirect comparisons only. We attempted to contact 26 authors. We sent emails soliciting HAM-D response rates to 21 authors (current contact information for 5 authors could not be found). Fourteen authors responded to our query, but most could not provide data as it is no longer available. In the end, only two authors were able to provide us with HAM-D response rates from their studies. We were able to use the HAM-D data provided by Boulenger, 200631 in our mixed-treatment comparison; Blumenthal, 200732 HAM-D data is used in our sensitivity analysis.

Key Question 1a. Efficacy or Effectiveness in Treating Depressive Disorders and Symptoms

Major Depressive Disorder: Overview

In all, 91 RCTs (reported in 93 articles) compared the efficacy or effectiveness of one second-generation antidepressant with that for another for treating patients with MDD. Details can be found in the evidence tables in Appendix C.

Tables 9 through 14 provide selected information on all these studies. Studies are grouped according to the main drug classes compared—SSRIs versus SSRIs (Table 9); SSRIs versus SSNRIs and SNRIs (Table 10); and SSRIs versus other second-generation antidepressants (Table 11); SNRIs versus SSNRIs and SNRIs (Table 12); SNRIs versus other second-generation antidepressants (Table 13); and other second-generation antidepressants versus other second-generation antidepressants (Table 14). They are then listed alphabetically by the specific drugs compared.

Table 9. SSRIs versus SSRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 9

SSRIs versus SSRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 10. SSRIs versus SSNRIs and SNRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 10

SSRIs versus SSNRIs and SNRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 11. SSRIs versus other second-generation antidepressants study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 11

SSRIs versus other second-generation antidepressants study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 12. SNRIs versus SSNRIs and SNRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 12

SNRIs versus SSNRIs and SNRIs study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 13. SNRIs versus other second-generation antidepressants study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 13

SNRIs versus other second-generation antidepressants study characteristics, response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 14. Response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Table 14

Response and remission rates, and quality ratings of studies in adults with major depressive disorder.

Most subjects were younger than 60 years; 11 trials were conducted in populations of 55 years or older. We discuss these 11 studies in more detail in KQ 5 on subgroups. In the text below, studies are of fair quality unless otherwise specified.

In general, studies enrolled patients according to a criteria-based diagnosis of MDD relating to the Diagnostic and Statistical Manual of Mental Disorders (DSM, either revised third edition or fourth edition [DSM-III-R, DSM-IV]) and a predefined cutoff point of a widely used depression scale (i.e., Hamilton Rating Scale for Depression [HAM-D]=18 or Montgomery-Asberg Depression Rating Scale [MADRS]=19). Most patients had moderate to severe depression as measured by a variety of scales. Most studies excluded patients who had additional Axis I disorders, high suicidal risk, or progressive medical diseases or who used psychotherapy, electroconvulsive therapy, or psychotropic medications.

Of 78 possible comparisons of included second-generation antidepressants, we found direct head-to-head evidence for only 40 comparisons. Table 9 and Table 10 depict possible comparisons and the numbers of available head-to-head trials for each comparison. For those with fewer than three head-to-head trials, we conducted indirect comparisons. Appendix E presents studies included in our mixed-treatment comparisons.

Study investigators rarely assessed quality of life and functional capacity; if they did, they typically considered these as only secondary outcomes. Most studies employed both physician-rated scales; these included, for instance, HAM-D, MADRS, Clinical Global Impressions Scale (CGI) and patient-rated scales (e.g., Hospital Anxiety and Depression Rating Scale [HAD-A], Battelle Quality of Life Scale [BQOLS]).

In the majority of studies, the primary endpoints were either changes from baseline or rates of response or remission on investigator-rated diagnostic depression scales such as the HAM-D or MADRS. Changes on such diagnostic depression scales are generally viewed as intermediate outcomes rather than health outcomes, and they are not always reliably related to changes in health outcomes. Response or remission, even when deducted from such a scale (e.g., response is defined as a 50 percent improvement of scores on HAM-D or MADRS), can be seen as proxies to health outcomes. Therefore, we focused on differences in response or remission rates rather than differences in changes of scores.

We rated the quality of most studies as fair for internal validity. Most trials (68 percent) were of either short (6 weeks to 8 weeks) or medium (9 weeks to 11 weeks) duration; 32 percent reported followup of 12 weeks or more. Short-term studies may be limited in their ability to account appropriately for response rates and long-term adverse events. In addition, reviewed studies were conducted over a time span of more than 2 decades. Therefore, study populations differ with respect to cotreatment, prior exposures to other second-generation antidepressants, and other factors.

Trial reporting was often incomplete. Most articles did not report the method of randomization or allocation concealment. Last-observation-carried-forward methods (or LOCF analysis, which means that the last observed measurement serves as the substitute for missing values because patients drop out at different time points), were a frequent approach to ITT analysis. Few authors, however, reported the overall number of patients lost to followup from the point of randomization to the end of the trial.

Loss to followup (number of patients randomized who did not proceed to endpoint), a potential source of bias, was a frequent problem for internal validity. The high rates of loss to followup for many studies may be attributable to specific characteristics of a psychiatric outpatient population and a high rate of adverse events in the examined drug classes.

Major Depressive Disorder: Key Points

Ninety-one head-to-head studies (Tables 9 to 14) were included for a total of 40 comparisons (Tables 15 to 17) between the 13 second-generation antidepressants addressed in this report. Of these, only nine trials116-124 directly compared any non-SSRI second-generation antidepressant with any other non-SSRI agent (Table 18); of these, only three comparisons were evaluated in more than one trial.

Table 15. Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus SSRIs.

Table 15

Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus SSRIs.

Table 16. Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus SNRIs.

Table 16

Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus SNRIs.

Table 17. Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus other second-generation antidepressants.

Table 17

Number of head-to-head trials of selective serotonin reuptake inhibitors for treating major depressive disorders: SSRIs versus other second-generation antidepressants.

Table 18. Number of head-to-head trials of selective serotonin norepinephrine reuptake inhibitors, serotonin norepinephrine reuptake inhibitors, and other antidepressants for treating major depressive disorders.

Table 18

Number of head-to-head trials of selective serotonin norepinephrine reuptake inhibitors, serotonin norepinephrine reuptake inhibitors, and other antidepressants for treating major depressive disorders.

Overall, 37 percent of patients did not achieve a treatment response during 6 weeks to 12 weeks of treatment with second-generation antidepressants; 53 percent did not achieve remission.

Based on our meta-analyses of head-to-head trials and our mixed treatment comparisons, second-generation antidepressants had similar efficacy. Statistically significant differences for some comparisons are likely not to be clinically relevant. The overall strength of evidence for the comparative efficacy was rated moderate.

Direct evidence was considered sufficient to conduct meta-analyses for six drug-drug comparisons:

  • Citalopram versus escitalopram (5 published studies33-36, 38 and 1 FDA review;37 1,802 patients): For patients on escitalopram the odds ratio (OR) of response was statistically significantly higher than for patients on citalopram (OR, 1.47; 95% CI, 1.07 to 2.01). The number needed to treat (NNT) to gain 1 additional responder at week 8 with escitalopram compared with citalopram was 13 (95% CI, 8 to 39). These results are based on metaanalyses of head-to-head trials. Results of mixed-treatment comparisons, taking the entire evidence base on second-generation antidepressants into consideration, did not confirm these findings. (OR, 0.51; 95% credible interval [CrI], 0.13 to 4.14).
  • Fluoxetine versus paroxetine (5 studies;49-52, 55, 82 690 patients): Pooled response rates between fluoxetine and paroxetine were similar (OR, 1.08; 95% CI, 0.79 to 1.47).
  • Fluoxetine versus sertraline (4 studies;55, 56, 58, 60 940 patients): The odds ratio of response was statistically significantly higher for sertraline than for fluoxetine (OR, 1.42; 95% CI, 1.08 to 1.85). The NNT to gain 1 additional responder at 6 to 12 weeks with sertraline was 13 (95% CI, 8 to 58).
  • Fluoxetine versus venlafaxine (six studies;78, 80-84 1,197 patients): The odds ratio of response was statistically significantly higher for patients on venlafaxine than on fluoxetine (OR, 1.47; 95% CI, 1.16 to 1.86).
  • Paroxetine versus duloxetine (three studies;87-89 849 patients). Pooled response rates were similar between patients on paroxetine and duloxetine (OR, 0.84; 95% CI, 0.63 to 1.12).
  • Sertraline versus venlafaxine (three studies;97-99 470 patients). Pooled response rates were similar between patients on sertraline or venlafaxine (OR, 1.18; 95% CI, 0.81 to 1.72).

Seventeen studies (n=3,960) comparing one second-generation antidepressant with another indicated no differences in health-related quality of life.33, 57, 58, 60, 66, 67, 72, 76, 77, 82, 94, 99, 103, 106, 118, 125, 126 Quality of life, however, was rarely assessed as a primary outcome measure. The strength of evidence is moderate.

Seven studies, all funded by the maker of mirtazapine, reported that mirtazapine has a statistically significantly faster onset of action than citalopram, fluoxetine, paroxetine, and sertraline (Table 19).67, 75-77, 90, 92, 96 The pooled NNT to yield one additional responder after 1 or 2 weeks of treatment is seven (95% CI, 5 to 12). This treatment effect was consistent across all studies. The strength of evidence is moderate.

Table 19. Characteristics of trials comparing mirtazapine to SSRIs on onset of action (response rate).

Table 19

Characteristics of trials comparing mirtazapine to SSRIs on onset of action (response rate).

Major Depressive Disorder: Detailed Analysis

Head-to-Head Evidence: SSRIs Versus SSRIs

Citalopram Versus Escitalopram

Citalopram and escitalopram are produced by the same manufacturer, which funded all available studies. Generic brands of citalopram are available in the United States; escitalopram is still under patent protection.

Five published trials33-36, 38 and one unpublished trial37 compared the efficacy of citalopram and escitalopram (Table 20). Five studies were conducted over 6 to 8 weeks33, 35-38 and one over 24 weeks.34 One study was a flexible dose trial.35 Table 20 summarizes study characteristics and differences in effect sizes of studies comparing citalopram with escitalopram.

Table 20. Characteristics and effect sizes of studies comparing citalopram with escitalopram.

Table 20

Characteristics and effect sizes of studies comparing citalopram with escitalopram.

Overall, results of individual studies favored escitalopram over citalopram. In four studies, differences in response rates reached statistical significance at 8 weeks.34-36, 38 The flexible dose trial was a European-Canadian study that compared efficacy and harms of citalopram (20-40 mg/day), escitalopram (10-20 mg/day) in 315 depressed outpatients attending primary care centers.35 ITT results showed that the escitalopram group had significantly more patients responding (63.7 percent vs. 52.6 percent; P=0.021) and achieving remission (52.1 percent vs. 42.8 percent; P=0.036) than the citalopram group. Escitalopram was numerically better at all time points on three scales (MADRS, Clinical Global Impressions Improvement Scale [CGI-I], Clinical Global Impressions Severity Scale [CGI-S]). The study did not assess health outcomes.

The 24-week study was a fixed-dose trial (escitalopram 10 mg/day, citalopram 20 mg/day) of 357 European primary care patients over 24 weeks.34 Escitalopram patients had significantly higher response rates at week 8 (63 percent vs. 55 percent; P<0.05) but not at week 24 (80 percent vs. 78 percent; P=NR). Escitalopram had significantly lower CGI-S scores (1.75 vs. 2.00) and significantly fewer withdrawals (12.7 percent vs. 22.4 percent) than citalopram at week 24.

We conducted two meta-analyses of these studies comparing the effects of citalopram with those of escitalopram on MADRS scores at weeks 6-8. The outcome of the first meta-analysis was the odds ratio of being a responder on the MADRS scale after 6–8 weeks of treatment (Figure 4). In addition to the five published trials, we included data from one unpublished study from the FDA Center for Drug Evaluation and Research (CDER) database.37 A “response” was defined as an improvement of 50 percent or more on the MADRS. Pooled results included 1,802 patients and yielded a statistically significant additional treatment effect for escitalopram. The odds ratio that a patient would respond was 1.47 (95% CI, 1.07 to 2.01) for escitalopram relative to citalopram. The NNT to gain one additional responder based on the pooled risk difference is 13 (95% CI, 8 to 39). As mentioned above, all available head-to-head trials have been funded by the manufacturer of citalopram and escitalopram. Publication bias, therefore, is conceivable.

Figure 4 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio of MADRS response rates comparing citalopram with escitalopram; odds ratios larger than 1 favor escitalopram, odds ratios smaller than 1 favor citalopram. Burke et al., 2002 report an odds ratio of 1.19 (95% CI, 0.76 to 1.88). Colonna et al., 2005 report an odds ratio of 1.31 (95% CI, 0.84 to 2.04). Lepola et al., 2003 report an odds ratio of 1.60 (95% CI, 0.99 to 2.58). Moore et al., 2005 report an odds ratio of 2.12 (95% CI, 1.26 to 3.59). An unpublished study (SCT-MD-02) reports an odds ratio of 0.85 (95% CI, 0.50 to 1.45). Yevtushenko et al., 2007 report an odds ratio of 3.81 (95% CI, 1.39 to 12.06). Overall, the meta-analysis showed a pooled odds ratio of 1.47 (95%CI, 1.07 to 2.01; I2 = 17%)

Figure 4

Odds ratio meta-analysis of MADRS response rates comparing citalopram with escitalopram.

Results of mixed treatment comparisons of good or fair studies, taking comparisons of each drug with other second-generation antidepressants into consideration, revealed no statistically significant difference of response rates on HAM-D between the two medications (OR, 0.51; 95% CrI, 0.13 to 4.14). Although not statistically significant, the point estimate of MTC results was in favor of citalopram over escitalopram.

In a sensitivity analysis we extended the evidence base to all available studies (i.e. including studies that were rated poor because of high risk of bias). Results increased the precision of estimates and yielded similar response rates between citalopram and escitalopram (OR, 1.04; 95% CrI, 0.57 to 2.12).

The second meta-analysis was an effect size meta-analysis of all six studies (1,802 patients) assessing the pooled difference of points on the MADRS (Figure 5). The weighted mean difference (WMD) presented an additional treatment effect of a 1.52 point reduction (95% CI, 0.59 to 2.45) for escitalopram compared with citalopram.

Figure 5 shows the study statistics and forest plot for the random effects meta-analysis of the weighted mean differences between citalopram and escitalopram for decreases in depressive severity (on the MADRS). Differences greater than 0 favor escitalopram, differences smaller than 0 favor citalopram. The meta-analysis included the following studies: Burke et al., 2002, Colonna et al., 2005, Lepola et al., 2003, Moore et al., 2005, an unpublished study (SCT-MD-02) and Yevtushenko et al., 2007. Differences of individual studies are not reported in the forest plot. Overall, the meta-analysis showed a weighted mean difference of 1.52 (95%CI, 0.59 to 2.45; I2 = 42%)

Figure 5

Effect size meta-analysis comparing citalopram with escitalopram on the MADRS.

Citalopram Versus Fluoxetine

In a French trial, 357 outpatients with MDD attending general practices were randomly assigned to citalopram (20 mg/day) or fluoxetine (20 mg/day) over 8 weeks.39 Citalopram had a faster onset of efficacy than fluoxetine; significantly more patients were rated as responding (35 percent vs. 24 percent; P=0.048) or completely recovered (27 percent vs. 16 percent; P=0.034) on the MADRS after 2 weeks. At 8 weeks, however, response rates for the citalopram and the fluoxetine group were similar (78 percent vs. 76 percent; P=NR).

Citalopram Versus Fluvoxamine

A Dutch study (n=217) did not find any differences in efficacy (HAM-D, CGI, Zung self-rating depression scale at 6 weeks) between citalopram (20-40 mg/day) and fluvoxamine (100–200 mg/day).40 Remission rates did not differ significantly between citalopram and fluvoxamine treatments (14 percent vs. 8 percent; P=NR).

Citalopram Versus Sertraline

A Swedish study rated good quality assessed the effectiveness of citalopram (20-60 mg/day) and sertraline (50-150 mg/day) in 400 patients in general practice during 24 weeks of treatment.41 The majority of patients suffered recurrent depression (citalopram, 65 percent; sertraline, 56 percent) and used other medications for medical illnesses (citalopram, 44.5 percent; sertraline, 55 percent). The investigators found no significant differences between treatment groups in any outcome measures at any point in time (MADRS, CGI-S, CGI-I). Also, subgroup analyses of patients with recurrent depression or single episode depression did not report any differences in effectiveness between drugs. Response rates (defined as a 50 percent or greater in MADRS from baseline, CGI-S score of 1-3 and CGI-I score rated “much” or “very much” improved) were similar at week 24 (citalopram, 81.0 percent; sertraline, 75.5 percent; P=NR). This study was one of only a few trials not funded by the pharmaceutical industry; it can be considered an effectiveness trial.

Escitalopram Versus Fluoxetine

Two RCTs assessed the comparative efficacy of escitalopram and fluoxetine.42, 43 One study (n=240) was conducted in a Chinese population43 and the other (n=518) in European patients older than 65 years.42 Both trials had a fixed-dose design (escitalopram 10 mg/day, fluoxetine 20 mg/day) and lasted 8 weeks.

In both studies, patients showed similar treatment effects. The Chinese study found no significant difference between groups in HAM-D response (80 percent vs. 79 percent, P>0.05) or remission (46 percent vs. 55 percent, P=NR) rates at week 8. MADRS response and remission rates were similar.43

In the European trial, neither escitalopram nor fluoxetine achieved statistically significantly different MADRS response (46 percent vs. 37 percent) or remission rates (40 percent vs. 30 percent) compared with placebo (response: 47 percent, remission: 42 percent). We discuss this study in more detail for KQ 5 (subgroups).42

Escitalopram Versus Paroxetine

Two RCTs provided mixed results about the comparative efficacy of escitalopram and paroxetine.31, 44 Both studies were funded by the producer of escitalopram.

A double-blind, flexible-dose RCT compared the efficacy of escitalopram (10-20 mg/day) and paroxetine (20–40 mg/day) during the acute and maintenance phases of the treatment of 325 patients with MDD.44 After 8 weeks both groups achieved similar MADRS response (67.9 vs. 71.2 percent and remission [56 vs. 62 percent]) rates. Similarly, no differences in response and remission could be observed during the maintenance period (8–27 weeks).

The second study was a fixed-dose RCT of 459 patients undergoing treatment with escitalopram 20 mg/day or paroxetine 40 mg/day.31 After 24 weeks of treatment, patients on escitalopram achieved higher MADRS remission rates than patients treated with paroxetine (75 percent vs. 67 percent; P<0.05). No statistically significant differences in response rates could be detected (82.0 percent vs. 76.7 percent), however.

Escitalopram Versus Sertraline

An 8-week, multicenter study randomized 215 patients to fixed-dose escitalopram (10 mg/day) or flexible-dose sertraline (50-200 mg/day).45At study endpoint no substantial differences in efficacy between patients in both treatment arms could be detected. Overall, 72 percent of patients on escitalopram and 69 percent of patients treated with sertraline achieved a HAM-D response. Remission rates were also similar between treatment groups (49 percent vs. 53 percent; P=NR).

Fluoxetine Versus Fluvoxamine

Two studies evaluated the comparative efficacy and safety of fluoxetine and fluvoxamine in 284 outpatients with MDD.46, 47 A 7-week flexible-dose study (fluoxetine: 20-80 mg/day; fluvoxamine 100-150 mg/day) did not identify any statistically or clinically significant differences in efficacy between the two treatment groups (HAM-D, HAM-A, CGI-S, Raskin-Covi Scale, Hopkins Symptoms Checklist [HSCL-D20]).47 Both treatment regimens significantly improved scores on assessment scales over 7 weeks.

In a 6-week fixed-dose European trial (fluoxetine 20 mg/day; fluvoxamine 100 mg/day) in 184 outpatients with MDD,46 results are consistent with those of the flexible-dose study; scores on the primary outcome measure (HAM-D) were not significantly different at any time. At endpoint, the drugs were equally effective for secondary outcome measures such as suicidal ideation, sleep, anxiety, and severity of illness (CGI, Clinical Anxiety Scale [CAS], the Irritability, Depression, and Anxiety Scale [IDAS], Beck's Scale for Suicide Ideation [Beck's SSI]). Fluvoxamine had significantly more responders on the CGI-S (29 percent vs. 16 percent; P<0.05) and a greater reduction of CGI-S scores (P<0.05) at week 2 but not at weeks 4 or 6.

Fluoxetine Versus Paroxetine

Nine studies compared fluoxetine with paroxetine.48-55 Two trials were conducted in populations older than 60 years of age,48, 53 which we discuss for KQ 5 (subgroups).

Most studies lasted from 6 to 12 weeks. Efficacy measures included HAM-D, HAM-A, MADRS, CGI-S, CGI-I, Covi Anxiety Scale, and others. Overall, these studies did not indicate substantial differences in outcome measures between fluoxetine and paroxetine. The largest study was a Canadian RCT (n=203) with a study duration of 12 weeks.49 At study endpoint, fluoxetine (20-80 mg/day) and paroxetine (20-50 mg/day) presented similar response (68 percent vs. 67 percent; P=0.93) and remission rates (59 percent vs. 58 percent; P=0.84).

One study was conducted in an inpatient population.54 Results were consistent with findings of the other studies.

We conducted a meta-analysis of five studies using HAM-D scores at the end of followup;49-52, 55 i.e., we excluded the three studies that did not report data on HAM-D or had been conducted in elderly populations.48, 53, 54, 82 We defined “response” as an improvement of 50 percent or more on the HAM-D. The meta-analysis included 690 patients. The pooled estimate of the random effects model, presented in Figure 6, indicates that fluoxetine and paroxetine do not differ significantly in efficacy (OR, 1.08; 95% CI, 0.79 to 1.47). An effect size meta-analysis (Figure 7) also did not detect a statistically significant difference between fluoxetine and paroxetine (0.52; 95% CI, -0.42 to +1.47).

Figure 6 shows the study statistics and forest plots for the fixed effects meta-analysis of the odds ratio of response rates comparing fluoxetine with paroxetine on the HAM-D; odds ratios larger than 1 favor paroxetine, odds ratios smaller than 1 favor fluoxetine. Chouinard et al., 1999 report an odds ratio of 0.97 (95% CI, 0.52 to 1.81). DeWilde et al., 1993 report an odds ratio of 0.92 (95% CI, 0.39 to 2.17). Fava et al., 1998 report an odds ratio of 1.03 (95% CI, 0.45 to 2.37). Fava et al., 2002 report an odds ratio of 1.23 (95% CI, 0.65 to 2.33). Gagiano et al., 1993 report an odds ratio of 1.33 (95% CI, 0.52 to 3.45). Overall, the meta-analysis showed a pooled odds ratio of 1.08 (95%CI, 0.79 to 1.47; I2 = 0%)

Figure 6

Odds ratio meta-analysis of response rates comparing fluoxetine with paroxetine on the HAM-D.

Figure 7 shows the study statistics and forest plot for the random effects meta-analysis of the weighted mean differences between fluoxetine and paroxetine for decreases in depressive severity (on the HAM-D). Differences greater than 0 favor paroxetine, differences smaller than 0 favor fluoxetine. The meta-analysis included the following studies: Chouinard et al., 1999, Fava et al., 1998 and Fava et al., 2002. Differences of individual studies are not reported in the forest plot. Overall, the meta-analysis showed a weighted mean difference of 0.52 (95%CI, 0.42 to 1.47; I2 = 0%)

Figure 7

Effect size meta-analysis comparing fluoxetine with paroxetine on the HAM-D.

Four studies did not detect differences between fluoxetine and paroxetine in improvement of anxiety in patients with depression (HAM-A, Covi Anxiety Scale).49, 51, 52, 55

Fluoxetine Versus Sertraline

Seven studies compared fluoxetine with sertraline.55-61, 127 The best evidence consisted of one effectiveness60 and one efficacy trial57 with long periods of followup.

Two multicenter trials in France comparing fluoxetine (20-60 mg/day) and sertraline (50–150 mg/day) were conducted in office settings (private psychiatrists and general physicians [GPs]).57, 60 The psychiatrist study60 randomized 238 patients for 24 weeks; the GP study57 randomized 242 patients for nearly 26 weeks (180 days). The majority of patients had concomitant medical conditions. Both studies assessed quality of life as a secondary outcome measure (Sickness Impact Profile [SIP], Functional Status Questionnaire [FSQ]). Exclusion criteria were less stringent in the GP trial than the psychiatrist trial. Loss to followup was 4.5 percent in the GP trial and 29.8 percent in the psychiatrist trial. In the GP trial, researchers conducted outcome assessments only at day 120 and day 180, but patients could choose to consult the physician at any time. ITT analyses in both studies did not reveal any statistically significant differences in any primary (MADRS, HAM-D, CGI) or secondary (Covi Anxiety Scale, HAD, SIP, Leeds Sleep Evaluation) efficacy measures or in the incidence of adverse events.

The ARTIST (A Randomized Trial Investigating SSRI Treatment) trial was an open-label RCT designed as an effectiveness study and carried out in primary care physician settings over 9 months.128 This study did not meet our eligibility criteria because of lack of blinding; we present it because it is one of only a few effectiveness trials. This study enrolled 601 patients at 76 sites. Initial diagnosis for enrollment was not based on diagnostic criteria but rather on the judgment of the treating physician. Criteria-based evaluation classified 74 percent of patients as having MDD, 18 percent dysthymia, and 8 percent minor depression. Patients' treatments could be switched among study drugs or to other antidepressive medications as needed. ITT analysis maintained the original randomization. Outcome measures assessing changes in depression and health-related quality of life measures (work, social, and physical functioning, concentration and memory, and sexual functioning) were administered over the telephone by a blinded third party. Range of dosage and loss to followup were incompletely reported.

Results of the ARTIST trial did not reveal any significant differences among drugs in any outcome measures at either 3 or 9 months.128 Compared with baseline measures, all treatment groups significantly improved during the study. Subgroup analyses did not show different effectiveness for patients with MDD or for those older than 60 years.

Four additional trials did not find any significant differences in primary outcome measures (HAM-D, MADRS, CGI-S).55, 56, 58, 59, 61 Studies lasted from 6 weeks to 16 weeks.

One study was conducted in 236 participants older than 60 years.58 and will be discussed in more detail in KQ5 (subgroups). Briefly, in this RCT, outcome measures also included quality of life (Q-LES-Q) and cognitive assessments (Shopping List Task [SLT], MMSE, Digital Symbol Substitution Test). Results on these health outcome measures were similar for both drugs.

We conducted two meta-analyses of four studies55, 56, 58, 60 comparing the effects of fluoxetine and sertraline at study endpoint. The outcome of the first meta-analysis was the odds ratio of being a responder on the HAM-D (improvement of 50 percent or more) at study endpoint (Figure 8).

Figure 8 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio of response rates comparing fluoxetine with sertraline on the HAM-D; odds ratios larger than 1 favor sertraline, odds ratios smaller than 1 favor fluoxetine. Bennie et al., 1995 report an odds ratio of 1.36 (95% CI, 0.83 to 2.23). Fava et al., 2002 report an odds ratio of 1.65 (95% CI, 0.85 to 3.21). Newhouse et al., 2000 report an odds ratio of 1.11 (95% CI, 0.60 to 2.03). Sechter et al., 1999 report an odds ratio of 1.67 (95% CI, 0.94 to 2.96). Overall, the meta-analysis showed a pooled odds ratio of 1.42 (95%CI, 1.08 to 1.85; I2 = 0%)

Figure 8

Odds ratio meta-analysis of response rates comparing fluoxetine with sertraline on the HAM-D.

Pooled results including 940 patients yielded a statistically significant additional treatment effect for sertraline (OR, 1.42; 95% CI, 1.08 to 1.85). Both random effects and fixed effects models presented similar, statistically significant results. The NNT to gain one additional responder based on the pooled risk difference was 13 (95% CI, 8 to 58).

The second, effect size meta-analysis assessed the pooled difference of points on the HAM-D scale (Figure 9). Because of lack of reported data, we limited the analysis to three studies.55, 58, 60 We found no statistically significant difference in points on the HAM-D scale between fluoxetine and sertraline. Relative to fluoxetine, sertraline had an additional treatment effect of a 0.76 point reduction in HAM-D (95% CI, -0.44 to +1.95).

Figure 9 shows the study statistics and forest plot for the random effects meta-analysis of the weighted mean differences between fluoxetine and sertraline for decreases in depressive severity (on the HAM-D). Differences greater than 0 favor sertraline, differences smaller than 0 favor fluoxetine. The meta-analysis included the following studies: Fava et al., 2002, Newhouse et al., 2000 and Sechter et al., 1999. Differences of individual studies are not reported in the forest plot. Overall, the meta-analysis showed a weighted mean difference of 0.76 (95%CI, 0.44 to 1.95; I2 = 0%)

Figure 9

Effect size meta-analysis comparing fluoxetine with sertraline on the HAM-D.

Fluvoxamine Versus Paroxetine

Two RCTs, one flexible-dose62 and one fixed-dose,63 compared the efficacy and safety of fluvoxamine and paroxetine. The flexible-dose trial was a 7-week RCT comparing the efficacy and safety of fluvoxamine (50–150 mg/day) and paroxetine (20–50 mg/day) in 60 outpatients with MDD.62 Loss to followup was 30 percent. Results presented no statistically significant differences on HAM-D, HAM-A, CGI, and HSCL-56. The fixed-dose trial enrolled 105 perimenopausal women with MDD and provided consistent findings with the flexible-dose trial.63 Neither trial assessed response or remission rates.

Fluvoxamine Versus Sertraline

Two 7-week trials compared the depression scores and harms of fluvoxamine (50–150 mg/day) and sertraline (50–200 mg/day).64, 65 One trial was conducted in a mixed (84 percent unipolar, 16 percent bipolar depression) inpatient population.65 In both trials, efficacy did not differ significantly between treatment groups. Both regimens led to significant improvements in depression scores from baseline (HAM-D, CGI).

Paroxetine Versus Sertraline

Two studies assessed the comparative efficacy of paroxetine and sertraline.55, 66 A Swedish RCT compared paroxetine (20–40 mg/day) with sertraline (50–150 mg/day) in a 24-week study involving 353 patients.66 Outcome measures included MADRS, CGI, and Battelle Quality of Life Measure (BQOL). LOCF analysis yielded no significant differences in primary outcome measures (MADRS, CGI) at any point in time. Clinically significant improvement occurred over baseline among all quality of life factors. Treatment groups did not differ significantly on BQOL factors. Likewise, the second study yielded similar response rates between paroxetine and sertraline.55

Head-to-Head Evidence: SSRIs Versus SSNRIs and SNRIs

Citalopram Versus Mirtazapine

An 8-week European study (n=270) determined the comparative efficacy of citalopram (20-60 mg/day) and mirtazapine (15–60 mg/day) on depression and anxiety symptoms in a mixed inpatient and outpatient population.67 At study endpoint, results on efficacy measures (MADRS, HAM-A, CGI-S, Leeds Sleep Evaluation Questionnaire) and a quality of life measure (Q-LES-Q) were similar between treatment groups. Response rates on MADRS reached 88 percent in the citalopram and 85 percent in the mirtazapine group (P=0.54). Mirtazapine, however, had a faster onset of action with significantly greater response rates on MADRS, HAM-A, CGI-S, and Q-LES-Q at day 14. Overall discontinuation rates because of adverse events did not differ significantly between the two groups.

Citalopram Versus Venlafaxine

A 6-month European study compared citalopram (10–30 mg/day) with venlafaxine XR (75–150 mg/day) for the treatment of depression in elderly outpatients (mean age 73 years) found no statistical differences in any outcome measures (MADRS, CGI-S, CGI-I) at study endpoint.68 We discuss these results in more detail for KQ 5 (subgroups).

Escitalopram Versus Duloxetine

Three RCTs compared the efficacy and safety of escitalopram and duloxetine in 1,257 patients with MDD.69-71 Two of these trials were funded by the maker of escitalopram,69, 71 the third by the manufacturer of duloxetine.70 Two studies compared fixed-dose regimens of escitalopram (10 and 20 mg/day) and duloxetine (60 mg/day).70, 71 The third trial assessed the efficacy and safety of a flexible dose escitalopram (10–20 mg/day) treatment with a fixed dose regimen of duloxetine (60 mg/day).69 Overall, results rendered similar response and remission rates between patients on escitalopram and duloxetine.

Escitalopram Versus Venlafaxine

Two 8-week trials assessed the comparative effectiveness of escitalopram and venlafaxine XR.72, 73 One assigned 293 patients to escitalopram (10–20 mg/day) or venlafaxine XR (75–150 mg/day).73 The groups did not differ significantly in response (escitalopram, 77.4 percent; venlafaxine XR, 79.6 percent; P=NR) or remission (escitalopram, 69.9 percent; venlafaxine XR, 69.7 percent; P=NR). Survival analysis of the ITT population indicated that escitalopram-treated patients achieved sustained remission 6.6 days earlier than patients on venlafaxine XR (P<0.01).

The second trial also reported that no statistically significant differences were apparent between escitalopram (20 mg/day) and venlafaxine XR (225 mg/day) in response (61 percent vs. 48 percent; P=NR) and remission rates.72

Fluoxetine Versus Duloxetine

An 8-week RCT assigned 173 patients to duloxetine (40–120 mg/day), fluoxetine (20 mg/day), or placebo.74 Results revealed no statistically significant differences between fluoxetine and duloxetine in response rates (45 percent vs. 49 percent; P=0.39). Remission rates at study endpoint favored duloxetine but did not reach statistical significance (43 percent vs. 30 percent; P=0.82). However, the fixed-dose design for fluoxetine but not for duloxetine introduces equivalency issues and reduces the validity of this direct comparison.

Fluoxetine Versus Mirtazapine

Three trials compared the efficacy of fluoxetine and mirtazapine.75-77 Two studies enrolled either exclusively76 or a large percentage77 of inpatients and outpatients with severe depression (HAM-D>25). In both of these trials, treatments did not differ on any efficacy measures (MADRS, HAM-D, CGI) or quality of life measures (Q-LES-Q) at endpoint (6 and 8 weeks). Both trials reported a faster onset of mirtazapine but no differences in remission rates at endpoint. These findings are consistent with results from the third study, which was conducted in Taiwanese outpatients with moderate depression.75

In all three studies, patients treated with mirtazapine gained weight; by contrast, those treated with fluoxetine lost weight. In two studies, the differences reached statistical significance.76, 77 In one trial, 10.3 percent of patients in the mirtazapine group experienced an increase in body weight of more than 7 percent from baseline as did 0.9 percent of patients on fluoxetine.76

Fluoxetine Versus Venlafaxine

Nine studies compared the efficacy of fluoxetine to venlafaxine.78-86 One study was conducted in inpatient populations.85 One trial was conducted in outpatients with concomitant anxiety (minimum score of 8 on Covi Anxiety Scale).80 The studies lasted from 6 weeks to 12 weeks. Except in one study,86 results consistently presented greater efficacy of venlafaxine than fluoxetine; in three studies, this difference reached statistical significance.78, 80, 81

We conducted a meta-analysis of seven studies comparing fluoxetine with venlafaxine,78, 80-85 all supported by the manufacturer of venlafaxine. The main outcome measure was the odds ratio of being a responder on the HAM-D scale at study endpoint.

Results (Figure 10), based on 1,197 patients, reflect higher response rates of venlafaxine than fluoxetine (OR, 1.47; 95% CI, 1.16 to 1.86). A meta-analysis of changes on the HAM-D rendered a significantly greater reduction of points for venlafaxine than fluoxetine.

Figure 10 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio of response rates comparing fluoxetine with venlafaxine on the HAM-D; odds ratios larger than 1 favor venlafaxine, odds ratios smaller than 1 favor fluoxetine. Alves et al., 1999 report an odds ratio of 2.40 (95% CI, 0.69 to 9.55). De Nayer et al., 2002 report an odds ratio of 2.20 (95% CI, 1.07 to 4.54). Dierick et al., 1996 report an odds ratio of 1.62 (95% CI, 0.99 to 2.65). Nemeroff et al., 2005 report an odds ratio of 1.31 (95% CI, 0.73 to 2.36). Rudolph et al., 1999 report an odds ratio of 1.15 (95% CI, 0.64 to 2.07). Silverstone et al., 1999 report an odds ratio of 1.25 (95% CI, 0.71 to 2.19). Overall, the meta-analysis showed a pooled odds ratio of 1.47 (95%CI, 1.16 to 1.86; I2 = 0%)

Figure 10

Odds ratio meta-analysis of response rates comparing fluoxetine with venlafaxine on the HAM-D.

These findings are consistent with results of a meta-analysis reported by Smith et al.129 Compared with fluoxetine, venlafaxine yielded a modest but significantly greater standardized effect size (-0.14; 95% CI, -0.22 to -0.06) and a significantly greater odds ratio for remission (OR, 1.42; 95% CI, 1.17 to 1.73). The odds ratio for response was numerically greater for venlafaxine but not statistically significant (OR, 1.17; 95% CI, 0.99 to 1.38).

Paroxetine Versus Duloxetine

Three 8-week, fixed-dose trials assessed the comparative efficacy of duloxetine (60, 80, and 120 mg/day) and paroxetine (20 mg/day).87-89 In all three trials, efficacy outcomes were similar for duloxetine and paroxetine regimens, although dosages were not always equivalent. In the largest study (n=478), 60 percent of patients on duloxetine (60 mg/day) achieved response and 49 percent remission as did 65 percent and 50 percent, respectively, of patients on paroxetine.89

We pooled response rates on the HAM-D from low-dose paroxetine (20 mg/day) and low-dose duloxetine arms (60 and 80 mg/day) (Figure 11). Results indicate that the two drugs have similar efficacy (OR, 0.84; 95% CI, 0.63 to 1.12). Data were too heterogeneous to achieve a meaningful pooled estimate of the mean change of scores on the HAM-D (I2, 99 percent).

Figure 11 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio of response rates comparing paroxetine with duloxetine on the HAM-D; odds ratios larger than 1 favor duloxetine, odds ratios smaller than 1 favor paroxetine. Detke et al., 2004 report an odds ratio of 0.63 (95% CI, 0.31 to 1.23). Lee et al., 2007 report an odds ratio of 0.81 (95% CI, 0.55 to 1.19). Perahia et al., 2006 report an odds ratio of 1.17 (95% CI, 0.62 to 2.20). Overall, the meta-analysis showed a pooled odds ratio of 0.84 (95%CI, 0.63 to 1.12; I2 = 4%)

Figure 11

Odds ratio meta-analysis of response rates comparing paroxetine with duloxetine on the HAM-D.

Paroxetine Versus Mirtazapine

Three trials assessed the efficacy of paroxetine (20–40 mg/day) and mirtazapine (15–45 mg/day).90-92 One study among depressed patients 65 years or older92 is discussed in more detail for KQ 5.

In all three trials, paroxetine and mirtazapine were equally effective in reducing HAM-D and MADRS scores at the endpoint. Mirtazapine led to a faster response in two trials.90, 92 For example, in a German study, 23.2 percent of mirtazapine-treated patients and 8.9 percent of paroxetine-treated patients responded to the treatment at week 1 (P<0.002).90 A Kaplan-Meier analysis in the other trial also showed a significantly faster time to response for mirtazapine than for paroxetine (mean 26 days vs. mean 40 days; P=0.016).92 The NNT to yield one additional patient responding with mirtazapine at weeks 1 or 2 is seven. No significant difference in response rates on the CGI scale was noted. All three trials reported weight gain in significantly more patients treated with mirtazapine than with paroxetine (P<0.05).

Paroxetine Versus Venlafaxine

Three studies compared paroxetine with venlafaxine.93-95 A Spanish study compared the effects of paroxetine (20-40 mg/day) with venlafaxine (75–150 mg/day) in outpatients (n=84) with either MDD or dysthymia over 24 weeks.93 The majority of patients (88 percent) were female. The percentage of dysthymic patients was not reported, and the authors did not differentiate between dysthymia and mild or moderate depression. Loss to followup was 32 percent, with a substantially higher loss to followup in the venlafaxine group (39 percent vs. 26 percent). Response and remission rates favored venlafaxine at all time points. The difference in remission rates reached statistical significance at week 12 (57 percent vs. 33 percent; P=0.011). ITT analysis yielded no significant differences between treatment groups on any primary outcome measures (HAM-D, MADRS, CGI) at 24 weeks.

A British fixed-dose trial lasting 12 weeks randomized 361 mainly moderately ill patients (based on CGI severity score) treated in 43 general practices to either paroxetine (20 mg/day) or venlafaxine XR (75 mg/day).94 Study groups did not differ significantly in efficacy measures, quality of life scores, or adverse events.

Similarly, a trial comparing extended-release formulations of paroxetine and venlafaxine (paroxetine CR 75 mg/day; venlafaxine XR 375 mg/day) yielded similar treatment effects between the two medications.95

Sertraline Versus Mirtazapine

One European study examined the onset of efficacy of sertraline (50-150 mg/day) compared with that of mirtazapine (30-45 mg/day) in 346 outpatients.96 Onset of action was faster for the mirtazapine group than for the sertraline group on HAM-D and MADRS. Significantly more patients achieved response and remission on mirtazapine than on sertraline after the first 2 weeks (data not reported in the article; P<0.05) No significant difference could be detected at endpoint. Subgroup analysis in patients with severe depression (HAM-D>25) led to similar findings. A significantly higher number of patients withdrew because of adverse events in the mirtazapine group (12.5 percent vs. 3 percent; P=NR), and significantly more patients on mirtazapine than on sertraline had an increase in body weight of more than 7 percent (14.6 percent vs. 0 percent; P=0.01).

Sertraline Versus Venlafaxine

Three 8-week trials, two rated good97, 99 and one rated fair,98 compared sertraline with venlafaxine or venlafaxine XR;98 all three studies were funded by the makers of venlafaxine. In a Scandinavian study (n=147), venlafaxine (75–150 mg/day) was significantly more efficacious than sertraline (50–100 mg/day) with respect to remissions achieved on the HAM-D (68 percent vs. 45 percent; P=0.008).97 We pooled response rates of these three studies on the HAM-D rating scale for 470 patients (Figure 12); fluoxetine and venlafaxine had similar treatment effects (OR, 1.18; 95% CI, 0.81 to 1.72).

Figure 12 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio of response rates comparing sertraline with venlafaxine on the HAM-D; odds ratios larger than 1 favor venlafaxine, odds ratios smaller than 1 favor sertraline. Mehtonen et al., 2000 report an odds ratio of 1.42 (95% CI, 0.69 to 2.93). Shelton et al., 2006 report an odds ratio of 1.39 (95% CI, 0.70 to 2.75). Sir et al., 2005 report an odds ratio of 0.82 (95% CI, 0.40 to 1.68). Overall, the meta-analysis showed a pooled odds ratio of 1.18 (95%CI, 0.81 to 1.72; I2 = 0%)

Figure 12

Odds ratio meta-analysis of response rates comparing sertraline with venlafaxine on the HAM-D.

Head-to-Head Evidence: SSRIs Versus Other Second-Generation Antidepressants

Fluoxetine Versus Bupropion

Two trials compared the efficacy and harms of fluoxetine and bupropion.100, 101 Both trials reported similar response rates at endpoint; efficacy measures (changes of HAM-D, HAM-A, CGI-S, CGI-I scores) did not differ significantly. In the larger trial (n=456), bupropion SR (150-400 mg/day) treatment yielded a higher rate than fluoxetine (20-60 mg/day) of patients achieving remission, but this difference was not significant (47 percent vs. 40 percent; P=NR).100 From week 1 until endpoint (week 8), significantly more patients on fluoxetine than on bupropion SR were dissatisfied with their overall sexual function (data not reported; P<0.05).

Fluoxetine Versus Nefazodone

One trial (n=44) compared the efficacy of fluoxetine (20 mg/day) and nefazodone (400 mg/day) in patients with MDD and insomnia.102 After 8 weeks both groups had similar reductions in HAM-D scores. Authors did not report on response or remission rates. A pooled data analysis that did not meet our eligibility criteria combined results of this trial with two other trials with identical protocols.130 Fluoxetine and nefazodone were similarly efficacious in producing response on the HAM-D scale (45 percent vs. 47 percent; P=NR).

Fluoxetine Versus Trazodone

Two 6-week trials compared the efficacy and harms of fluoxetine (20-60 mg/day) and trazodone (50-400 mg/day).103, 104 The groups did not differ significantly in any outcome measures (HAM-D, CGI-I, CGI-S, PGI-I). Remission rates in the larger study (n=126), however, favored fluoxetine over trazodone at study endpoint (51 percent vs. 42 percent; P=NR).103 Moreover, significantly fewer patients on fluoxetine than on trazodone experienced sedation or adverse events associated with sedation (22 percent vs. 43 percent; P=0.11)

Paroxetine Versus Bupropion

A 6-week, flexible-dose RCT compared paroxetine (20–40 mg/day) with bupropion SR (150-300 mg/day).105 The main objectives of the study were to assess comparative efficacy and to evaluate sexual functioning. Response rates on HAM-D were similar for patients treated with paroxetine or with bupropion SR (52 percent vs. 56 percent; P=NR). Men treated with paroxetine reported a greater worsening of sexual functioning than men on bupropion SR. Sexual functioning did not appear to differ for women.

A second RCT examined the efficacy of paroxetine (10–40 mg/day) and bupropion SR (100-300 mg/day) in 100 outpatients ages 60 years or older (range 60–88 years) over 6 weeks;106 it is discussed in more detail in KQ5 (subgroups). Briefly, relative to baseline, both groups significantly improved in all outcome measures (HAM-D, HAM-A, CGI-I, CGI-S), but the treatment groups did not differ significantly. Response rates were similar in both groups (paroxetine, 77 percent; bupropion SR, 71 percent; P=NR).

Paroxetine Versus Nefazodone

Two studies determined the comparative efficacy of paroxetine and nefazodone on depression and sleep improvement.107, 108 The larger trial enrolled 206 moderately depressed patients to an 8-week, acute-phase trial comparing nefazodone (200–600 mg/day) with paroxetine (20–40 mg/day).107 Both groups showed significant improvements from baseline HAM-A, HAM-D, and MADRS scores. Response rates were similar for paroxetine and nefazodone (60 percent vs. 58 percent; P=NR). The second trial provided similar results for the comparative antidepressive efficacy.108 Nefazodone, however, led to significantly greater improvements than paroxetine in objective sleep measures.

Paroxetine Versus Trazodone

A European study compared paroxetine (20–40 mg/day) with trazodone (150–400 mg/day) in 108 outpatients with MDD.109 Study duration was 6 weeks. No differences in any efficacy outcome measures could be detected (HAM-D, CGI-S, CGI-I, MADRS). Response rates (91 percent vs. 87 percent; P=NR) and remission rates (68 percent vs. 69 percent; P=NR) did not differ significantly between paroxetine and trazodone.

Sertraline Versus Bupropion

Three studies compared the efficacy and harms of sertraline and bupropion.110-113 Studies lasted from 8 weeks to 16 weeks. All three studies reported no statistically significant differences in efficacy on any outcome measure (HAM-D, CGI-I, CGI-S, HAM-A). Response rates in the largest trial (n=364) were 61 percent for sertraline and 66 percent for bupropion SR (P=NR).110

In all three studies, patients on sertraline had statistically significantly higher rates of sexual dysfunction than patients on bupropion. Two RCTs assessed the incidence of sexual dysfunction during 8 weeks of treatment with sertraline (50–200 mg/day), bupropion SR (150–400 mg/day), or placebo as primary outcome measures using DSM-IV definitions for sexual dysfunction disorders.110, 111 In another study, discontinuation rates because of sexual adverse events were significantly higher in the sertraline group than the bupropion SR group (13.5 percent vs. 3.3 percent, P=0.004).112 In addition, in this study some adverse events (nausea, diarrhea, somnolence, sweating) were significantly more common among patients treated with sertraline than among those on bupropion SR (P<0.05).

Sertraline Versus Nefazodone

A multicenter European study assessed the efficacy and harms of sertraline (50–200 mg/day) and nefazodone (100–600 mg/day) among 160 outpatients with moderate to severe depression.114 ITT analysis in this 6-week trial did not yield significant differences in efficacy between treatment groups. Response rates were similar between patients treated with sertraline and those treated with nefazodone (57 percent vs. 59 percent; P=NR). Additional outcome measures assessed by questionnaire were sexual function and satisfaction under antidepressant treatment. Overall satisfaction with sexual function was significantly higher in the nefazodone group (P<0.01). Among men, 67 percent in the sertraline group and 19 percent in the nefazodone group reported difficulty with ejaculation (P<0.01). Other adverse events did not differ significantly between the two groups.

Sertraline Versus Trazodone

A 6-week Italian trial (n=122) randomized outpatients with MDD to sertraline (50-100 mg/day) or trazodone prolonged release (150–450 mg/day).115 At study endpoint sertraline and trazodone did not differ significantly in efficacy (HAM-D, MADRS, CGI-I, CGI-S). Overall, response rates were lower for sertraline than trazodone (HAM-D, 63 percent vs. 74 percent; MADRS: 66 percent vs. 78 percent). The mean changes of HAM-D and MADRS scores from baseline, however, were similar for sertraline- and trazodone- treated patients (-11.5 vs. -12.9 and -15.0 vs. -16.5, respectively).

Head-to-Head Evidence: SNRIs Versus SSNRIs or SNRIs

Desvenlafaxine Versus Duloxetine

An 8-week, fixed-dose RCT compared desvenlafaxine 50 and 100 mg/day with duloxetine 60 mg/day in 638 outpatients with MDD.116 At study endpoint no significant differences in efficacy could be detected among treatment arms (HAM-D, MADRS, CGI-I, CGI-S, HAM-A). Overall, response rates were numerically lower for patients on desvenlafaxine 50 mg/day than for patients on desvenlafaxine 100 mg/day or duloxetine 60 mg/day (39 percent vs. 49 percent vs. 47 percent; P=NR). Similarly, the percentage of patients on desvenlafaxine 50 mg/day who achieved remission was lower than the figure for patients in the other treatment arms (20 percent vs. 28 percent vs. 29 percent). The differences, however, did not reach statistical significance.

Mirtazapine Versus Venlafaxine

Two European trials compared the efficacy of mirtazapine and venlafaxine.117, 118 One 8-week trial evaluated efficacy and harms in hospitalized, severely depressed patients (mean HAM-D 29.3) with melancholic features.118 At study endpoint, no significant differences in any efficacy or quality of life measures were apparent (HAM-D, MADRS, CGI-S, Q-LES-Q, QLDS); however, response rates favored mirtazapine over venlafaxine (62 percent vs. 52 percent; P=NR). During the study, significantly fewer patients on mirtazapine than on venlafaxine dropped out because of adverse events (5.1 percent vs. 15.3 percent; P=0.037). Mirtazapine led to weight gain in significantly more patients than did venlafaxine (10.3 percent vs. 5.1 percent; P<0.05). Venlafaxine had significantly lower rates of constipation (17.1 percent vs. 31.1 percent; P=0.056) and sweating (15.8 percent vs. 35.1 percent; P≤0.05) than venlafaxine.

The other study enrolled 242 outpatients treated at private practices in Germany.117 Like the trial described above, mirtazapine ODT (orally disintegrated tablets; 45 mg/day) had a faster onset of action than venlafaxine XR (225mg/day). At day 8, 19.7 percent of patients on mirtazapine and 6.1 percent of patients on venlafaxine XR (P=0.002) had responded to treatment. At study endpoint, mirtazapine and venlafaxine XR did not differ significantly in efficacy measures (data not reported).

Venlafaxine Versus Duloxetine

A pooled data analysis of two RCTs that have not been published individually provides the only available head-to-head evidence comparing venlafaxine with duloxetine;131 both RCTs were funded by the makers of duloxetine This study did not meet our eligibility criteria; however, because it is the only available direct evidence on the comparative efficacy of venlafaxine and duloxetine, we briefly summarize its results.

The two RCTs used a 6-week fixed-dose period comparing venlafaxine XR (150 mg/day) with duloxetine (60 mg/day) followed by a 6-week flexible dose period in 667 patients with MDD. Overall, response rates (69.1 percent vs. 62.6 percent) and remission rates (50.3 vs. 48.1 percent) did not differ significantly between the two groups. Discontinuation rates, however, were significantly lower in the venlafaxine group than in the duloxetine group (25 percent vs. 35 percent; P=0.006).

Head-to-Head Evidence: SNRIs Versus Other Second-Generation Antidepressants

Mirtazapine Versus Trazodone

Two studies compared mirtazapine with trazodone in patients with MDD.119, 120 One trial was conducted in depressed patients 55 years of age and older;119 the other was done in hospitalized patients with MDD.120 Efficacy measures in both trials favored mirtazapine, but differences did not reach statistical significance. In the hospitalized patients, response rates at endpoint were 61 percent for mirtazapine and 51 percent for trazodone (P=NR).120

Venlafaxine Versus Bupropion

Two 8-week RCTs compared the efficacy and safety of venlafaxine XR and bupropion XR.121, 122 Both studies were flexible-dose trials treating patients with venlafaxine XR (75–150 mg/day), bupropion XR (150–300 mg/day), or placebo. After 8 weeks of treatment, response and remission rates for patients treated with venlafaxine XR or bupropion XR were similar. For example, in one study, MADRS response (65 percent vs. 57 percent; P=NR) and remission rates (51 percent vs. 47 percent; P=NR) did not differ significantly between patients on venlafaxine XR and bupropion XR. Likewise, no substantial differences in health outcomes (Q-LES-Q-SF, Shehan Disability Scale) were apparent at study endpoint.121

Venlafaxine Versus Trazodone

A 6-week study enrolled 225 patients to assess efficacy and harms of venlafaxine (150-400 mg/day), trazodone (75-200 mg/day), and placebo.123 Efficacy outcomes (HAM-D, MADRS, CGI-S) did not differ significantly between active treatment groups. Response rates at endpoint, however, favored venlafaxine over trazodone (72 percent vs. 60 percent; P=NR). Trazodone led to improvements in sleep disturbance that were statistically significantly superior to those with venlafaxine. Significantly more patients on venlafaxine than on trazodone suffered from nausea (44 percent vs. 19 percent; P<0.05); however, trazodone led to a significantly higher rate of dizziness than venlafaxine (36 percent vs. 17 percent; P<0.05).

Head-to-Head Evidence: Other Second-Generation Antidepressants Versus Other Second-Generation Antidepressants

Bupropion Versus Trazodone

In a two-center study, 124 outpatients were randomly assigned to bupropion (225–450 mg/day) or trazodone (150–450 mg/day).124 Because of a statistically significant treatment-by-center interaction, the article reported results separately for each center. Overall, in both centers, efficacy results did not differ significantly between the two treatment groups.

Mixed Treatment Comparisons

Of 78 possible comparisons, the evidence was sufficient to pool data in meta-analyses for only six comparisons for MDD (those documented in Figures 4 through Figure 12). For the remaining 72 MDD comparisons, we conducted mixed treatment comparisons, as outlined in the Methods chapter. Studies used for the mixed treatment comparisons can be found in Appendix E; those excluded are listed in Appendix B.

We assessed the odds ratio of response to treatment on the HAM-D scale. The majority of comparisons did not reflect statistically significant differences in response rates among compared antidepressants. For those comparisons that reached statistical significance in favor of one drug, differences in treatment effects were small and are likely not to be clinically significant.

In general, findings from mixed treatment comparisons were consistent with available head-to-head studies. Results of direct (denoted by an asterisk) and indirect comparisons are depicted in Figures 13 to 15.

Figure 13 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio meta-analysis of HAM-D response rates comparing head to head trials of SSRI's with SSRIs. Odds ratios smaller than 1 favor the first SSRI in the comparison, odds ratios larger than 1 favor the second SSRI. Comparisons of citalopram versus escitalopram show an odds ratio of 1.49 (95% CI, 1.07 to 2.01). For citalopram versus fluoxetine, the odds ratio is 0.34 (95% CI, 0.09 to 2.73). Citalopram versus fluvoxamine resulted in an odds ratio of 0.84 (95% CI, 0.84 to 1.90). Citalopram versus paroxetine shows an odds ratio of 0.41 (95% CI, 0.41 (0.11 to 3.27). Citalopram with sertraline resulted in an odds ratio of 0.42 (95% CI, 0.11 to 3.34). For escitalopram versus fluoxetine, an odds ratio of 0.65 (95% CI, 0.49 to 0.90) is shown. Escitalopram versus fluvoxamine shows an odds ratio of 1.63 (95% CI, 0.26 to 5.77). Escitalopram versus paroxetine, the odds ratio is 0.78 (95% CI, 0.58 to 1.08). Escitalopram versus sertraline resulted in an odds ratio of 0.80 (95% CI, 0.60 to 1.11). Comparisons of fluoxetine versus fluvoxamine show an odds ratio of 2.45 (95% CI, 0.39 to 8.56). Fluoxetine versus paroxetine, the odds ratio is 1.08 (95% CI, 0.79 to 1.47). Fluoxetine versus sertraline resulted in an odds ratio of 1.42 (95% CI, 1.08 to 1.85). Fluvoxamine versus paroxetine shows an odds ratio of 0.49 (95% CI, 0.14 to 3.07). Fluvoxamine versus sertraline, the odds ratio is 0.50 (95% CI, 0.14 to 3.10). Paroxetine versus sertraline resulted in an odds ratio of 1.02 (95% CI, 0.78 to 1.35)

Figure 13

Odds ratios of response rates comparing SSRIs with SSRIs. * Based on meta-analysis of head-to-head trials.

Figure 14 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio meta-analysis of HAM-D response rates comparing head to head trials of SSRI's , SNRI's with SNRI's and SSNRI's. Odds ratios smaller than 1 favor the first drug in the comparison, odds ratios larger than 1 favor the second drug. Comparisons for SSRI's versus SSNRI's: Citalopram versus duloxetine shows an odds ratio of 0.38 (95% CI, 0.10 to 3.07). Escitalopram versus duloxetine, the odds ratio is 0.74 (95% CI, 0.56 to 0.98). Fluoxetine versus duloxetine resulted in an odds ratio of 1.14 (95% CI, 0.87 to 1.46). Fluvoxamine versus duloxetine shows an odds ratio of 0.46 (95% CI, 0.13 to 2.87). Paroxetine versus duloxetine, the odds ratio is 0.93 (95% CI, 0.72 to 1.22). Sertraline versus duloxetine resulted in an odds ratio of 0.93 (95% CI, 0.70 to 1.20). Comparisons of SSRI's versus SNRI's: Citalopram versus desvenlafaxine shows an odds ratio of 0.43 (95% CI, 0.11 to 3.45). Escitalopram versus desvenlafaxine, the odds ratio is 0.81 (95% CI, 0.56 to 1.22). Fluoxetine versus desvenlafaxine resulted in an odds ratio of 1.27 (95% CI, 0.89 to 1.76). Fluvoxamine versus desvenlafaxine shows an odds ratio of 0.51 (95% CI, 0.14 to 3.25). Paroxetine versus desvenlafaxine, the odds ratio is 1.07 (95% CI, 0.73 to 1.51). Sertraline versus desvenlafaxine resulted in an odds ratio of 1.04 (95% CI, 0.72 to 1.44). Citalopram versus mirtazapine shows an odds ratio of 0.48 (95% CI, 0.12 to 4.14). Escitalopram versus mirtazapine, the odds ratio is 0.99 (95% CI, 0.57 to 1.59). Fluoxetine versus mirtazapine resulted in an odds ratio of 1.48 (95% CI, 0.91 to 2.27). Fluvoxamine versus mirtazapine shows an odds ratio of 0.57 (95% CI, 0.16, 3.85). Paroxetine versus mirtazapine, the odds ratio is 1.24 (95% CI, 0.76 to 1.89). Sertraline versus mirtazapine resulted in an odds ratio of 1.21 (95% CI, 0.73 to 1.89). Citalopram versus venlafaxine shows an odds ratio of 0.49 (95% CI, 0.13 to 3.91). Escitalopram versus venlafaxine, the odds ratio is 0.93 (95% CI, 0.67 to 1.33). Fluoxetine versus venlafaxine resulted in an odds ratio of 1.46 (95% CI, 1.16 to 1.83). Fluvoxamine versus venlafaxine shows an odds ratio of 0.58 (95% CI, 0.17 to 3.67). Paroxetine versus venlafaxine, the odds ratio is 1.18 (95% CI, 0.86 to 1.65). Sertraline versus venlafaxine resulted in an odds ratio of 1.17 (95% CI, 0.88 to 1.53). Comparisons of SSNRI's and SNRI's versus SNRI's: Duloxetine versus desvenlafaxine shows an odds ratio of 1.13 (95% CI, 0.80 to 1.55). Duloxetine versus venlafaxine, the odds ratio is 1.25 (95% CI, 0.92 to 1.75). Duloxetine versus mirtazapine resulted in an odds ratio of 1.32 (95% CI, 0.78 to 2.09). Desvenlafaxine versus venlafaxine shows an odds ratio of 1.12 (95% CI, 0.78 to 1.67). Mirtazapine versus desvenlafaxine, the odds ratio is 0.84 (95% CI, 0.50 to 1.50). Mirtazapine versus venlafaxine resulted in an odds ratio of 0.96 (95% CI, 0.59 to 1.65).

Figure 14

Odds ratios of response rates comparing SSRIs and SNRIs with SNRIs and SSNRIs. * Based on meta-analysis of head-to-head trials.

Figure 15 shows the study statistics and forest plots for the random effects meta-analysis of the odds ratio meta-analysis of HAM-D response rates comparing head to head trials of SSRI's , SNRI's, SSNRI's and other second-generation antidepressants with other second-generation antidepressants. Odds ratios smaller than 1 favor the first drug in the comparison, odds ratios larger than 1 favor the second drug. Comparisons for SSRIs versus other second-generation antidepressants are as follows: Citalopram versus bupropion shows an odds ratio of 0.37 (95% CI, 0.10 to 3.03). Citalopram versus nefazodone, the odds ratio is 0.42 (95% CI, 0.11 to 3.57). Citalopram versus trazodone resulted in an odds ratio of 0.33 (95% CI, 0.08 to 2.79). Escitalopram versus bupropion shows an odds ratio of 0.74 (95% CI, 0.50 to 1.06). Escitalopram versus nefazodone, the odds ratio is 0.87 (95% CI, 0.48 to1.44). Escitalopram versus trazodone resulted in an odds ratio of 0.67 (95% CI, 0.38 to 1.11). Fluoxetine versus bupropion shows an odds ratio of 1.11 (95% CI, 0.81 to 1.48). Fluoxetine versus nefazodone, the odds ratio is 1.30 (95% CI, 0.75 to 2.10). Fluoxetine versus trazodone resulted in an odds ratio of 1.01 (95% CI, 0.60 to 1.58). Fluvoxamine versus bupropion shows an odds ratio of 0.44 (95% CI, 0.13 to 2.82). Fluvoxamine versus nefazodone, the odds ratio is 0.50 (95% CI, 0.14 to 3.38). Fluvoxamine versus trazodone resulted in an odds ratio of 0.39 (95% CI, 0.11 to 2.62). Paroxetine versus bupropion shows an odds ratio of 0.93 (95% CI, 0.65 to 1.28). Paroxetine versus nefazodone, the odds ratio is 1.09 (95% CI, 0.62 to 1.78). Paroxetine versus trazodone resulted in an odds ratio of 0.84 (95% CI, 0.50, 1.32). Sertraline versus bupropion shows an odds ratio of 0.90 (95% CI, 0.66 to 1.20). Sertraline versus nefazodone, the odds ratio is 1.06 (95% CI, 0.62, 1.67). Sertraline versus trazodone resulted in an odds ratio of 0.82 (95% CI, 0.49 to 1.29). Comparisons for SNRIs versus other second-generation antidepressants are as follows: Desvenlafaxine versus bupropion show an odds ratio of 0.89 (95% CI, 0.59 to 1.31). Desvenlafaxine versus nefazodone, the odds ratio is 1.05 (95% CI, 0.57 to 1.76). Desvenlafaxine versus trazodone resulted in an odds ratio of 0.81 (95% CI, 0.45 to 1.36). Mirtazapine versus bupropion shows an odds ratio of 0.74 (95% CI, 0.46 to 1.26). Mirtazapine versus nefazodone, the odds ratio is 0.83 (95% CI, 0.45 to 1.69). Mirtazapine versus trazodone resulted in an odds ratio of 0.66 (95% CI, 0.43 to 1.08). Venlafaxine versus bupropion shows an odds ratio of 0.78 (95% CI, 0.53 to 1.10). Venlafaxine versus nefazodone, the odds ratio is 0.92 (0.51 to 1.51). Venlafaxine versus trazodone resulted in an odds ratio of 0.71 (95% CI, 0.41 to 1.16). Comparisons of SSNRI's versus other second-generation antidepressants: Duloxetine versus bupropion show an odds ratio of 0.99 (95% CI, 0.69 to 1.37). Duloxetine versus nefazodone, the odds ratio is 1.16 (0.66 to 1.89). Duloxetine versus trazodone resulted in an odds ratio of 0.90 (0.52 to 1.45). Comparisons of other second-generation antidepressants versus other second-generation antidepressants: Bupropion versus nefazodone show an odds ratio of 1.20 (0.66 to 2.00). Bupropion versus trazodone, the odds ratio is 0.92 (95% CI, 0.54 to 1.47). Nefazodone versus trazodone resulted in an odds ratio of 0.82 (95% CI, 0.40 to 1.50).

Figure 15

Odds ratios of response rates comparing SSRIs, SNRIs, SSNRIs and other second-generation antidepressants with other second-generation antidepressants. * Based on meta-analysis of head-to-head trials.

Sensitivity analyses including studies with high risk of bias increased the precision of the estimates and confirmed the overall conclusion that no substantial differences in response rates exist among second-generation antidepressants. In most cases, broadening the body of evidence to all available studies moved the point estimates towards the null.

Dysthymia: Overview

We did not find any head-to-head trials on patients with dysthymia. Five placebo-controlled trials (Table 21) assessed effectiveness, efficacy, and harms of fluoxetine, paroxetine, and sertraline in populations with dysthymia.125, 126, 132-136 Four studies were of fair quality; the fifth was of good quality. Details can be found in the evidence tables in Appendix C.

Table 21. Interventions, numbers of patients, results, and quality ratings of studies in adults with dysthymia.

Table 21

Interventions, numbers of patients, results, and quality ratings of studies in adults with dysthymia.

Dysthymia: Key Points

We identified no head-to-head trials in a population with dysthymia. The substantial differences in population characteristics in placebo-controlled trials make the evidence too inconsistent to identify differences between treatments. The strength of evidence is insufficient.

Five placebo-controlled trials (seven articles) provide conflicting evidence on the general efficacy and effectiveness of fluoxetine, paroxetine, and sertraline for the treatment of dysthymia.125, 126, 132-136 Specifically:

  • Two studies provide mixed evidence about the general efficacy of fluoxetine for the treatment of dysthymia. 126, 136
  • One effectiveness study did not detect any statistically significant difference between paroxetine and placebo. 134, 135
  • Two studies indicate that sertraline has a significantly greater efficacy in the treatment of dysthymia than placebo.125, 132, 133

Dysthymia: Detailed Analysis

Head-to-Head Evidence

We identified no head-to head trials.

Placebo-Controlled Evidence

Fluoxetine Versus Placebo

Two trials evaluated the efficacy of fluoxetine for treating patients with dysthymia over 12 weeks; the studies provide mixed results.126, 136 An RCT of good quality examined the efficacy and safety of fluoxetine (20–60 mg/day) in patients 60 years of age and older;136 we discuss this trial in more detail for KQ 5 (subgroups). Briefly, ITT analysis indicated that fluoxetine had limited efficacy. Response rates on the HAM-D favored fluoxetine over placebo, but the two groups did not differ significantly (27.3 percent vs. 19.6 percent; P=0.4). Likewise, the investigators found no difference in quality of life.

The other trial was conducted in patients 18 years of age and older (mean 43 years).126 Significantly more patients on fluoxetine than on placebo were rated as responders (58 percent vs. 36 percent; P=0.03). Remission rates favored fluoxetine but did not reach statistical significance (44.4 percent vs. 25.6 percent; P=0.07).

Paroxetine Versus Placebo Versus Behavioral Therapy

A large, primary-care-based effectiveness study randomized 656 patients with dysthymia or minor depression to 11 weeks of paroxetine (10–40 mg/day), placebo, or behavioral therapy.134, 135 Participants were stratified into patients 60 years of age and older (n=415) and patients younger than 60 years of age (n=241) for ITT analysis. We discuss the results of the subgroup analysis on older patients in more detail for KQ 5 (subgroups).

Briefly, in patients 60 years or older, paroxetine-treated patients showed a greater change in HSCL-D-20 scores than placebo-treated patients (P=0.004).134 Effects were similar for patients with dysthymia and minor depression. Among the younger patients, treatment groups did not differ significantly on the HSCL-D-20.135 For dysthymia only, the remission rate of patients with at least 4 weeks of treatment was significantly higher in the paroxetine group than in the placebo group (80 percent vs. 44 percent; P=0.008). Paroxetine was not more efficacious than placebo in patients with minor depression.

Sertraline Versus Placebo

Two RCTs that assessed the efficacy of sertraline (50–200 mg/day) for the treatment of dysthymia over 12 weeks provided similar results.125, 132, 133 In both studies, only patients who had had the diagnosis of dysthymia for more than 5 years were eligible; outcomes included quality of life and measures of functional capacity. Patients on sertraline had significantly greater antidepressant responses than those on placebo (64 percent vs. 44 percent; P<0.001133 and 52 percent vs. 34 percent; P=0.001125). In addition, sertraline was more efficacious than placebo on psychosocial and quality of life instruments (Global Assessment of Functioning Scale, Social Adjustment Scale [SAD], Quality of Life Enjoyment and Satisfaction Questionnaire [QLSQ], BQOLS).

Subsyndromal Depressive Disorders: Overview

We found no head-to-head RCTs on patients with subsyndromal depressive disorders. The only head-to-head evidence was a nonrandomized, single-blinded trial comparing citalopram with sertraline.137 Because of the lack of head-to-head evidence, we briefly summarize this study, although it did not meet eligibility criteria. In addition, two placebo-controlled studies assessed the efficacy and tolerability of fluoxetine138 and paroxetine134, 135 in patients with subsyndromal depression (Table 22). Details can be found in the evidence tables in Appendix C.

Table 22. Interventions, numbers of patients, results, and quality ratings of studies in adults with subsyndromal depressive disorders.

Table 22

Interventions, numbers of patients, results, and quality ratings of studies in adults with subsyndromal depressive disorders.

Subsyndromal Depressive Disorders: Key Points

We identified no head-to-head RCTs in a population with subsyndromal depression. A nonrandomized, open-label trial did not detect any differences in efficacy between citalopram and sertraline.137

In placebo-controlled trials, differences in population characteristics make the evidence insufficient to identify differences between treatments.134, 135, 138 In one effectiveness study in a primary care setting, effectiveness did not differ significantly between paroxetine and placebo for the treatment of minor depression.134, 135 The strength of evidence is insufficient.

Subsyndromal Depressive Disorders: Detailed Analysis

Head-to-Head Evidence

We did not find any head-to-head RCTs. A nonrandomized, single-blinded trial (n=138) lasting 1 year assessed the comparative efficacy and safety of citalopram and sertraline in patients with late-life minor depression or other subsyndromal depressive disorders.137 Overall, both treatments improved depressive symptoms but the groups did not differ significantly at any time point. At the end of the study, remission was achieved by 53 percent of patients on citalopram and 42 percent on sertraline (P=0.25). Likewise, no differences in psychosocial functioning emerged.

Placebo-Controlled Evidence

Two studies were conducted in populations with minor depression.

Fluoxetine Versus Placebo

A 12-week trial (n=162) evaluated the efficacy of fluoxetine in patients with minor depression.138 Improvements on depression scales (HAM-D, Beck Depression Inventory [BDI], IDS-C) were statistically significantly greater for patients receiving fluoxetine than for those receiving placebo. Likewise, the overall severity of illness (CGI-S) improved statistically significantly more in the fluoxetine than in the placebo group (P=0.002). No significant differences could be detected in psychosocial outcomes.

Paroxetine Versus Placebo

A large primary-care-based effectiveness study randomized 656 patients with dysthymia or minor depression to 11 weeks of paroxetine (10-40 mg/day), placebo, or behavioral therapy.134, 135 Participants were stratified into patients 60 years and older (n=415) and patients younger than 60 years (n=241) for ITT analysis.

In the 60 or older subgroup, patients receiving paroxetine showed a greater change in HSCL-D-20 scores than those receiving placebo (P=0.004), but those on paroxetine did not demonstrate more change than patients on behavioral therapy (P=0.17).134 Effects were similar for patients with dysthymia and minor depression. Paroxetine was not more efficacious than placebo in patients with minor depression in the younger subgroup.135

Key Question 1b. Response to Antidepressant Agents After Successful Response in the Past

We did not find any evidence that answered this Key Question.

Key Question 1c. Differences in Efficacy and Effectiveness between Immediate- and Extended-Release Formulations

Efficacy of Immediate- Versus Extended-Release Formulations: Overview

We found five head-to-head trials that investigated the comparative efficacy of daily versus weekly dosing (Table 23) and immediate- versus extended-release formulations (Table 24).139-143 Two of these trials compared fluoxetine daily with fluoxetine weekly;139, 140 two good-quality trials assessed paroxetine IR (immediate-release) versus paroxetine CR (controlled-release);141, 142 and one trial compared venlafaxine IR with venlafaxine XR (extended release).143 We could not find any studies on other medications, such as bupropion or fluvoxamine, that are available as both immediate- and extended-release formulations.

Table 23. Interventions, numbers of patients, results, and quality ratings of studies comparing daily with weekly fluoxetine regimens during continuation treatment.

Table 23

Interventions, numbers of patients, results, and quality ratings of studies comparing daily with weekly fluoxetine regimens during continuation treatment.

Table 24. Interventions, numbers of patients, results, and quality ratings of studies comparing immediate- with extended-release formulations.

Table 24

Interventions, numbers of patients, results, and quality ratings of studies comparing immediate- with extended-release formulations.

Efficacy of Immediate- Versus Extended-Release Formulations: Key Points

Five head-to-head trials investigated the comparative efficacy of daily versus weekly dosing and immediate- versus extended-release formulations.139-143

Two RCTs reported similar rates of maintenance of response and relapse for patients treated with fluoxetine daily or fluoxetine weekly during the continuation phase of MDD therapy.139, 140 The strength of evidence is moderate.

One RCT and a pooled analysis of two identical RCTs did not find any differences in response rates in patients treated with paroxetine IR or paroxetine XR for acute phase MDD.141, 142 The strength of evidence is moderate.

One RCT reported higher response rates for patients on venlafaxine XR than venlafaxine IR. The strength of evidence is low.

Efficacy of Immediate- Versus Extended-Release Formulations: Detailed Analysis

Head-to-Head Evidence

Fluoxetine Daily Versus Fluoxetine Weekly

No extended-release formulation of fluoxetine exists. Because of the long elimination half-lives of fluoxetine and its active metabolite norfluoxetine, investigators have explored different dosing regimens for fluoxetine during continuation treatment. Of particular interest has been weekly treatment regimens. Unlike daily treatments, the weekly treatment is administered with an enteric-coated formulation to reduce gastrointestinal adverse events.

Two double-blinded RCTs compared the efficacy of fluoxetine (20 mg/day) with fluoxetine (60 mg/week and 90 mg/week) during the continuation phase of patients with MDD who had responded to 20 mg/day of fluoxetine during the acute-treatment phases. The acute-treatment periods in both studies were open-label and lasted between 7 and 13 weeks.139, 140 Patients who achieved response were randomized to double-blinded continuation treatment with fluoxetine (20 mg/day) or fluoxetine (60 mg/week or 90 mg/week). Treatment durations during the continuation periods were 7 and 25 weeks, respectively.

The larger study randomized 501 patients to fluoxetine (20 mg/day), fluoxetine (90 mg/week), or placebo.140 After 25 weeks of continuation treatment, 37 percent of patients on weekly fluoxetine weekly and 26 percent of patients on daily fluoxetine experienced a relapse (P=NR). Both groups (weekly vs. daily) also exhibited similar changes in CGI-S (1.0 vs. 0.9) and HAM-D (6.6 vs. 6.4) scores. The smaller study also did not detect any statistically significant differences in the main outcome measures (MADRS, Hopkins Symptom Checklist).139

Paroxetine IR Versus Paroxetine CR

One double-blinded RCT142 and a pooled analysis of two identical RCTs141 compared the efficacy and safety of paroxetine IR with paroxetine CR. The RCT enrolled 319 elderly patients with acute MDD, randomizing them to paroxetine IR (up to 40 mg/day), paroxetine CR (up to 50 mg/day), or placebo.142 The primary outcome measure was the change of HAM-D scores after 12 weeks of treatment. Patients in both active treatment arms had similar changes on the HAM-D (paroxetine IR, -12.3; paroxetine CR, - 12.1). Likewise, response (65 percent vs. 72 percent) and remission rates (44 percent vs. 43 percent) were similar for the two groups.

The other study pooled data (n=820) of two identical RCTs conducted in adult outpatients between 18 and 65 years of age who had MDD.141 Patients received treatment with paroxetine IR (20 to 50 mg/day), paroxetine CR (25–62.5 mg/day), or placebo. After 12 weeks of treatment, patients on the IR and CR formulations exhibited similar response rates (73 percent vs. 74 percent) and remission rates (53 percent vs. 56 percent).

Venlafaxine IR Versus Venlafaxine XR

One flexible-dose, placebo-controlled RCT compared the efficacy and safety of twice-daily venlafaxine IR (37.5–150 mg/2× per day) with once-daily venlafaxine XR (75–150 mg/day) in 293 patients with acute-phase MDD.143 Primary outcome measures were the HAM-D, the MADRS, and the CGI scales. After 12 weeks of treatment, significantly more patients on venlafaxine XR experienced a response to treatment than patients treated with venlafaxine IR (data not reported; P<0.05 for response on HAM-D, MADRS, and CGI).

Key Question 2. Efficacy or Effectiveness for Maintaining Remission or for Treating Patients With Unresponsive or Recurrent Disease

This section deals with two key aspects of treating patients with major depressive disorder (MDD). KQ 2a addresses maintaining remissions and preventing relapses or recurrences for patients who have responded to antidepressant treatment; KQ 2b focuses on addressing ongoing depressive disease for those who have not responded to such therapy or who have experienced relapses or new episodes. For patients who have responded, two subquestions are important: the efficacy or effectiveness of (1) continuing the initial (existing) medication or (2) switching to a different one (KQ 2a). For patients who have not responded, the issues focus on using different antidepressants (KQ 2b).

For purposes of exposition in this section, we use the phrase “maintaining remission” to encompass preventing relapse or recurrence; we also use the phrase “achieving response” to encompass treating patients who have not responded in an acute phase of disease or who have experienced a relapse or recurrence. Detailed information on all trials reviewed for KQ 2 can be found in the evidence tables in Appendix C.

Maintaining Remission: Overview

Continuing Initial Medications

In all, we had 38 trials relating to KQ 2a about continuing existing medications (Table 25). We also identified two additional systematic reviews and meta-analyses, but we did not formally assess them because their component trials were already included in our work.144, 145 Seven head-to-head studies (eight articles) compared the efficacy of one second-generation antidepressant with another for preventing relapse or recurrence.44, 61, 123, 146-150 Comparisons included escitalopram versus desvenlafaxine,148 escitalopram versus paroxetine,44 fluoxetine versus sertraline,61 fluoxetine versus venlafaxine,149, 150 fluvoxamine versus sertraline,146, 147 and trazodone versus venlafaxine.123

Table 25. Number of head-to-head trials and placebo-controlled trials of second-generation antidepressants for preventing relapse, by comparison.

Table 25

Number of head-to-head trials and placebo-controlled trials of second-generation antidepressants for preventing relapse, by comparison.

Another 31 RCTs140, 149, 151-187 provide additional placebo-controlled evidence to support the general efficacy of bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, sertraline, trazodone, and venlafaxine for maintaining remission in patients with depressive disorders (Table 26).

Table 26. Number of head-to-head trials and placebo-controlled trials of second-generation antidepressants for recurrence of major depressive disorder, by comparison.

Table 26

Number of head-to-head trials and placebo-controlled trials of second-generation antidepressants for recurrence of major depressive disorder, by comparison.

Using the disease treatment framework depicted in Figure 1 of the Introduction chapter, we characterized studies that assessed continuation treatment of patients who had responded or remitted with acute-phase treatment as relapse-prevention studies. Relapse-prevention studies typically included an open-label, acute-phase treatment and a double-blind, randomized, placebo-controlled continuation-phase treatment. The duration of continuation treatment in these trials ranged from 14 weeks to 72 weeks.

We further denoted studies that assessed maintenance treatment among patients who had remained in remission following acute and continuation treatment as recurrence-prevention studies. These studies usually included an open-label acute phase, then an open-label continuation phase for acute-phase responders, followed by a randomized, double-blind, placebo-controlled maintenance phase for patients who had not relapsed. The maintenance phase in these trials lasted from 36 weeks to 100 weeks.

Investigators generally determined the initial inclusion of patients on a criteria-based diagnosis (e.g., DSM-III-R, DSM-IV) and a predefined cutoff point of a universally used depression scale (e.g., HAM-D≥18 or MADRS≥19). Subsequent inclusion criteria varied. Some trials randomized patients who had demonstrated a clinically significant response to open-label treatment (e.g., ≥50 percent improvement from baseline on the HAM-D or MADRS). Others used a predefined cutoff point on a depression scale to identify and randomize those who were in remission (e.g., HAM-D≤9, MADRS≤12, CGI-I≤2). Most studies assessed relapse or recurrence using a predefined cutoff point on a depression rating scale (e.g., HAM-D>18, MADRS>19, CGI-S≥4), but the specific cutoff point varied widely.

Because we rated most of these trials as fair quality (internal validity), we denote quality in this section only for those rated good. Poor-quality studies are not included here; a listing of these studies can be found in Appendix D. Trial reporting was often incomplete. Most articles did not report their methods of randomization or allocation concealment. Even though investigators frequently used intention-to-treat analysis, few authors reported the overall number of patients lost to followup from randomization to the end of the trial.

Because of heterogeneous study designs and the relatively small number of trials, we did not make indirect comparisons between drugs.

Switching Medications

No trial specifically addressed the efficacy or effectiveness of any second-generation antidepressant for preventing relapse (i.e., continuation phase) or recurrence (i.e., maintenance phase) when a patient had previously responded to one antidepressant and switched to an alternative antidepressant.

Maintaining Remission: Key Points

Continuing Initial Medications

In six head-to-head studies,44, 61, 123, 146-150 the overall efficacy for maintaining remission does not differ between escitalopram and desvenlafaxine,148 escitalopram and paroxetine,44 fluoxetine and sertraline,61 fluoxetine and venlafaxine,149 fluvoxamine and sertraline,146, 147 and trazodone and venlafaxine.123 One naturalistic study provided evidence that rehospitalization rates do not differ between patients continuing fluoxetine versus continuing venlafaxine.150 We rated the strength of head-to-head evidence as moderate.

We found 14 placebo-controlled relapse-prevention trials that provide consistent efficacy evidence favoring active treatment over placebo.140, 151-164, 176 Seventeen placebo-controlled recurrence-prevention trials provide consistent evidence for active treatment over placebo.149, 152, 165-175, 177-187 We rated the strength of this evidence as moderate.

Effect sizes generally were similar across drugs in placebo-controlled efficacy trials. This observation is consistent with effect sizes noted in two published meta-analyses of placebo-controlled trials: (1) relapse prevention with venlafaxine (OR, 0.37; 95% CI, 0.27 to 0.51);145 (2) relapse prevention with second-generation antidepressants (RR, 0.54; 95% CI, 0.46 to 0.62);144 and (3) recurrence prevention with second-generation antidepressants; (RR, 0.56; 95% CI, 0.48 to 0.66).144

Switching Medications

As noted, we identified no studies on this point. The strength of evidence in this case is graded insufficient. We do not comment further on this treatment option.

Maintaining Remission: Detailed Analysis

Head-to-Head Evidence on Continuing Initial Medications

Six head-to-head trials44, 61, 123, 146-149 and one naturalistic (nonrandomized) study150 compared one second-generation antidepressant with another for maintaining remission (Table 27). Findings for acute-phase treatment are reported in KQ 1 (above) and not replicated here, although we list acute-phase treatments and duration for context with other studies.

Table 27. Head-to-head studies of maintaining remission (preventing relapse or recurrence).

Table 27

Head-to-head studies of maintaining remission (preventing relapse or recurrence).

SSRIs Versus SSRIs

Escitalopram Versus Paroxetine

One trial compared the acute-phase and continuation-phase efficacy of escitalopram (10–20 mg/day) with paroxetine (20–40 mg/day).44 Although this study was designed primarily to assess discontinuation effects during treatment interruption and during tapered withdrawal, it provided data for the 19-week continuation period that followed on an 8-week acute phase. At the end of 27 weeks, response rates were similar for escitalopram and paroxetine (≥50 percent improvement in MADRS total score from baseline [85 percent vs. 79 percent; P=NR]). Relapse rates were not explicitly reported, but we calculated them from the sample flow data to be 10 percent and 9 percent, respectively, for escitalopram and paroxetine.

Fluoxetine Versus Sertraline

One trial compared the efficacy of fluoxetine and sertraline for preventing relapse during a 24-week continuation phase.61 A total of 165 patients with major depression were randomized to fluoxetine (20–40 mg/day) or sertraline (50–100 mg/day). At 8 weeks, 56 responders (≥50 percent reduction in HAM-D or MADRS) in the fluoxetine group and 49 responders in the sertraline group entered the continuation phase, continuing the same dose attained at the end of the acute phase. Relapse rates were similar in the two groups (13 percent and 10 percent, respectively; P=NR). This design may be prone to bias and confounding because patients had not been rerandomized at the start of the continuation phase.

Fluvoxamine Versus Sertraline

One Italian trial of 64 patients with recurrent depression compared the efficacy of fluvoxamine and sertraline for maintaining remission over 2 years146 and 4 years.147 After at least 4 months of remission with tricyclic antidepressants (n=49), SSRIs (n=4), monoamineoxidase inhibitors (n=2), or combination treatment (n=9), investigators randomized patients to fluvoxamine (200–300 mg/day) or sertraline (100–200 mg/day) and followed them for up to 4 years. Recurrence rates (HAM-D>15) for fluvoxamine and sertraline were similar at 2 years (19 percent vs. 22 percent, respectively; P=0.88) and 4 years (20 percent vs. 14 percent, respectively; P=0.92).

SSRIs Versus SNRIs

Escitalopram Versus Desvenlafaxine

One trial compared escitalopram (10–20 mg/day) with desvenlafaxine (100–200 mg/day) for relapse prevention during 6 months of continuation-phase treatment in postmenopausal women with MDD.148 At 8 weeks, 160 responders (≥50 percent reduction in HAM-D17 total score) in the escitalopram group and 137 responders in the desvenlafaxine group entered the continuation phase, continuing the same dose attained at the end of the acute phase. Relapse rates were similar in the two groups (20 percent and 18 percent, respectively; P=0.7).

Fluoxetine Versus Venlafaxine

One trial149 and a longitudinal naturalistic study150 assessed continuation-phase treatment comparing fluoxetine with venlafaxine. One trial, the Prevention of Recurrent Episodes of Depression with Venlafaxine for Two Years (PREVENT), randomized patients to double-blind treatment with fluoxetine 20-60 mg/day or venlafaxine ER 75–300 mg/day for 10 weeks; it then allowed patients achieving a response (≥50 percent reduction in HAM-D17 or total score≤12) or remission (HAM-D17≤7) to continue through 6 months of continuation treatment.149 Continuation-phase response rates (92 percent vs. 90 percent) and remission rates (69 percent vs. 72 percent) were similar for fluoxetine and venlafaxine, respectively. Only 3 fluoxetine-treated patients (2 percent) and 5 venlafaxine-treated patients (1 percent) relapsed during continuation-phase treatment (P=0.44).

A naturalistic study compared time to rehospitalization in Chinese patients with major depression who had received acute treatment in an inpatient setting.150 Patients were not randomly assigned to treatment, although patient characteristics at discharge were similar between the fluoxetine and venlafaxine groups. Patients continued the same antidepressant at the same dose as used at discharge; they were followed over 1 year to monitor clinical condition and rehospitalization status. Rehospitalization rates did not statistically significantly differ between fluoxetine and venlafaxine during this 1 year (46 percent vs. 43 percent of patients were rehospitalized, respectively; P=0.695).

SNRIs Versus Other Second-Generation Antidepressants

Venlafaxine Versus Trazodone

One trial of 225 patients with major depression compared the efficacy and safety of trazodone and venlafaxine over a 1-year continuation/maintenance phase.123 Investigators randomized patients for acute treatment with venlafaxine 75–200 mg/day (n=72), trazodone 150–400 mg/day (n=77), or placebo (n=76). After 6 weeks, 37 in the venlafaxine group and 30 responders in the trazodone group (CGI-I score of 1 or 2) were allowed to continue into the long-term phase. Relapse rates were similar in the three groups (8 percent, 13 percent, and 14 percent, respectively; P=NR). Fewer patients treated with venlafaxine than with either trazodone or placebo withdrew from treatment for any reason; the difference between venlafaxine and trazodone reached statistical significance (P≤0.05) during the long-term phase.

Placebo-Controlled Evidence on Continuing Initial Medications

Fourteen placebo-controlled trials (16 publications) assessed relapse prevention140, 151-164, 176 and 17 trials (24 publications) assessed recurrence prevention.149, 152, 165-175, 177-187 Because the duration of acute, continuation, and maintenance phase treatment is not consistent in all patients, and because the definition of these treatment phases is not universal, some studies described below (Table 28) can be categorized as addressing both relapse and recurrence prevention.

Table 28. Placebo-controlled studies of relapse prevention and recurrence prevention.

Table 28

Placebo-controlled studies of relapse prevention and recurrence prevention.

SSRI: Citalopram Versus Placebo

Two trials assessed relapse prevention;153, 188 two other trials assessed recurrence prevention.165, 166 Both relapse-prevention trials randomized patients who had responded in the acute phase (MADRS≤12) to placebo or continuation treatment with citalopram (20-60 mg/day). Statistically significantly fewer patients on citalopram than on placebo relapsed after 24 weeks in both trials. Relapse rates were 14 percent and 24 percent, respectively (P=0.04), in one trial, and 11 percent (pooled) and 31 percent, respectively (P<0.02), in the other trial.

Both recurrence-prevention trials included open-label, acute-phase treatment with citalopram (20-60 mg/day; 6 weeks to 9 weeks), followed by 16 weeks of open-label continuation treatment at the same dose for responders (MADRS≤11).165, 166 Patients who had not relapsed (MADRS≤22) during the continuation phase were randomized to 48 weeks of double-blind maintenance treatment with citalopram or placebo. Recurrence rates were lower for citalopram-treated patients than for placebo-treated patients in both trials (18 percent vs. 43 percent, respectively; P<0.001,165 and 32 percent vs. 67 percent, respectively; P=NR166).

SSRI: Escitalopram Versus Placebo

Three trials compared escitalopram with placebo; two assessed relapse prevention154, 163 and one recurrence prevention.178 The two trials on relapse prevention reported that patients continuing on escitalopram had statistically significantly lower relapse rates than patients on placebo.

One trial focused on 405 older patients (age≥65 years; mean age 73).163 Participants received open-label escitalopram (10–20 mg/day) for 12 weeks; responders (MADRS total score≤12) were eligible for randomization to 24 weeks of double-blinded treatment with escitalopram (10-20 mg/day; n=152) or placebo (n=153). Significantly fewer escitalopram-treated patients (MADRS≥22 or lack of efficacy as judged by the investigator) than placebo-treated patients experienced a relapse (9 percent vs. 33 percent; P<0.001). The risk of relapse was 4.4 times higher for placebo- than for escitalopram-treated patients (P<0.001), and the time to relapse was shorter for escitalipram- than for placebo-treated patients (P<0.001).

Another trial openly treated 502 MDD patients with escitalopram (10–20 mg/day) for 8 weeks.154 Patients who responded (MADRS≤12) were randomized to 36 weeks of double-blind continuation treatment with escitalopram (n=181) or placebo (n=93). Relapse rates (MADRS≥22) were statistically significantly lower for escitalopram-treated patients than for placebo-treated patients (26 percent vs. 40 percent, respectively; P=0.01), and the time to depressive relapse was significantly longer in patients who received escitalopram than in patients who received placebo (P=0.013).

One trial assessed recurrence prevention in 515 patients with recurrent depression (two or more previous episodes) who had responded (MADRS≤12) to 8 weeks of acute open-label treatment with citalopram, fluoxetine, paroxetine, or sertraline.178 The 234 responders were openly treated with escitalopram (10–20 mg/day) for 16 weeks. Patients who continued to respond (MADRS≤12) were randomized to 52 weeks of maintenance-phase treatment with escitalopram (n= 73) or placebo (n=65). Recurrence rates were lower for patients receiving escitalopram than for those receiving placebo (27 percent vs. 65 percent), and time to recurrence was significantly longer for patients receiving escitalopram than placebo (hazard ratio 0.26; 95% CI, 0.13 to 0.52; P<0.001).

SSRI: Fluoxetine Versus Placebo

Three trials (five publications) assessed relapse prevention.140, 155, 156, 179, 190 One of these trials,179 plus one additional trial,167 assessed recurrence prevention.

Of the relapse-prevention studies, one trial sought to determine the optimal length of continuation treatment by randomizing patients who were in remission (HAM-D<7 for 3 consecutive weeks) during 12 weeks to 14 weeks of acute-phase treatment with fluoxetine (20 mg/day) to 14 weeks, 38 weeks, or 50 weeks of continuation treatment with fluoxetine or placebo.155, 190 Relapse rates were significantly lower for fluoxetine-treated patients than for placebo-treated patients at 14 weeks (26 percent vs. 49 percent, respectively; P<0.001) and 38 weeks (9 percent vs. 23 percent, respectively; P=0.04), but not at 50 weeks (11 percent vs. 16 percent, respectively; P=0.54). The other trial openly treated 932 patients with MDD for 13 weeks with fluoxetine.140, 156 Responders (HAM-D≤9 and CGI-I≤2) were randomized to 25 weeks of continuation treatment with fluoxetine (20 mg/day; n=189), fluoxetine (90 mg/week; n=190), or placebo (n=122). Relapse rates were statistically significantly lower for both the daily and the weekly doses of fluoxetine than for placebo (26 percent and 37 percent vs. 50 percent, respectively; P<0.01 for placebo comparisons).

Another trial assessed both relapse and recurrence rates in patients who had responded (response criteria not reported) to 12 weeks of open-label treatment with fluoxetine (10-60 mg/day).179 Patients were randomized (n=131 fluoxetine and n=131 placebo) only at the beginning of the continuation phase, but the authors reported results for a conventional 6-month continuation phase and an additional 6-month maintenance phase; statistical tests reflected only aggregate 52-week data. After 6 months, relapse rates (relapse criteria not reported) were 35 percent for fluoxetine and 62 percent for placebo; after 1 year, relapse rates were 45.9 percent for fluoxetine and 72.0 percent for placebo (hazard ratio 1.73; 95% CI, 1.20 to 2.51; P=0.004).

A different recurrence-prevention trial randomized patients who continued to meet remission criteria (HAM-D≤8) during a 6-month continuation period to 1 year of double-blind maintenance treatment with either fluoxetine (20-40 mg/day; n=70) or placebo (n=70).167 Recurrence rates were statistically significantly lower for fluoxetine-treated patients than for placebo-treated patients (20 percent vs. 40 percent, respectively; P=0.01).

SSRI: Fluvoxamine Versus Placebo

One trial assessed recurrence prevention with fluvoxamine (100-300 mg/day).168 Of 436 patients with major depression treated openly with fluvoxamine for 6 weeks, 283 responders (MADRS<10 and CGI-I≤2) entered 18 weeks of continuation treatment with fluvoxamine 100 mg/day. Patients who sustained their response (MADRS<12 and no CGI-I score>2) were randomized to 1 year of double-blind treatment with fluvoxamine (n=110) or placebo (n=94). Recurrence rates were statistically significantly lower for fluvoxamine-treated patients than for placebo-treated patients (13 percent vs. 35 percent, respectively; P<0.001).

SSRI: Paroxetine Versus Placebo

Three trials compared paroxetine with placebo for relapse and recurrence prevention.152, 170, 180 One trial focused specifically on patients 70 years old and older (mean age 77.1 years), comparing recurrence rates among four groups: (1) paroxetine plus clinical management (n=35); (2) paroxetine plus psychotherapy (n=28); (3) placebo plus psychotherapy (n=35); and (4) placebo plus clinical management (n=18).180 We focused on the comparison of paroxetine with placebo for patients receiving clinical management services, which included monthly 30-minute visits to assess symptoms and possible adverse events. Major depression recurred (HAM-D17≥15) among 37 percent of the paroxetine (10-40 mg/day) group and 58 percent of the placebo group (P=0.06).

One U.K. trial152 and one U.S. trial170 assessed long-term treatment with paroxetine. Both trials randomized patients who had responded to acute-phase paroxetine therapy to 1 year of paroxetine or placebo.

The U.K. study assessed relapse prevention after 16 weeks of double-blind treatment and recurrence prevention after an additional 36 weeks of continued double-blind treatment with paroxetine 20-30 mg/day.152 After 16 weeks, significantly fewer paroxetine-treated patients had relapsed than placebo-treated patients (3 percent vs. 19 percent, respectively; P<0.01). Of the patients who maintained a response through the continuation phase and entered the maintenance phase, recurrence rates were lower for paroxetine-treated patients than for placebo-treated patients (14 percent vs. 30 percent, respectively; P<0.05).

The U.S. study was an extension of a 6-week acute-phase trial that compared paroxetine, imipramine, and placebo.170 Investigators invited patients who had responded in the 6-week trial to continue flexible-dose, double-blind treatment for up to 1 year. Treatment allocation in the long-term extension was not randomized; the authors reported only aggregated relapse rates. More placebo-treated patients withdrew from the long-term trial because of “lack of efficacy”170(n=10; 22 percent) than did patients treated with either paroxetine 10-50 mg/day (n=11; 12 percent) or imipramine 65-275 mg/day (n=3; 4 percent).

SSRI: Sertraline Versus Placebo

Two studies assessed relapse prevention;159, 164 four other studies171, 173, 174, 181 assessed recurrence prevention. In one relapse-prevention study, 295 patients who had responded in the acute phase were randomized to 44 weeks of double-blind treatment with sertraline (50-200 mg/day; n=185) or placebo (n=110).159 Statistically significantly fewer sertraline-treated patients than placebo-treated patients experienced a relapse (13 percent vs. 46 percent, respectively; P<0.001). In a Japanese relapse-prevention study, 235 patients who had responded to 8 weeks of open sertraline treatment were randomized to 16 weeks of double-blind sertraline (50-100 mg/day; n=117) or placebo (n=118).164 The relapse rate was significantly lower for sertraline patients than for placebo patients (9 percent vs. 20 percent; P=0.016). Time-to-relapse also was significantly longer for sertraline- than placebo-treated patients (P=0.026).

The good-quality relapse/recurrence-prevention trial addressed potential methodological biases by including patients with recurrent depression who had been successfully treated for at least 4 months with any antidepressant other than sertraline.173 Active treatment was replaced with placebo for 2 months to identify patients truly in remission; patients who continued to remain in remission were randomized to sertraline 50 mg/day; (n=95), sertraline 100 mg/day (n=94), or placebo (n=99) and followed for 18 months. Patients treated with sertraline were statistically significantly less likely to have a recurrent depressive episode than patients treated with placebo (17 percent vs. 33 percent, respectively, for the pooled comparison; P=0.002).

Two other recurrence-prevention studies found that patients treated with sertraline had fewer recurrences than did those on placebo.171, 174 In a 76-week maintenance phase, 6 percent of sertraline-treated and 23 percent of placebo-treated patients had a recurrent depressive episode (P=0.002).171 Differences did not reach statistical significance in a 100-week maintenance treatment of community residents 65 years of age and older with major depression; 45 percent of sertraline-treated patients and 54 percent of placebo-treated patients had a recurrent episode (P=0.21).174 This trial is described in further detail in KQ 5.

Another recurrence-prevention trial was conducted in patients with diabetes mellitus.181 Patients who recovered from depression (four consecutive BDI scores≤9) during 16 weeks of open-label treatment with sertraline (25–200 mg/day) were randomized to 52 weeks of maintenance sertraline (n=79) or placebo (n=73). Recurrence of major depression (defined by DSM-IV criteria) was more common among placebo- than sertraline-treated patients (52 percent vs. 34 percent; P=0.02). This trial is described in further detail in KQ 5.

SSNRI: Duloxetine Versus Placebo

One trial (two articles) compared duloxetine with placebo for preventing relapse;161, 162 one trial compared duloxetine with placebo for preventing recurrence.177 The relapse-prevention trial treated MDD patients (n=533) openly with duloxetine (60 mg/day) for 12 weeks and then randomized responders (HAM-D17≤9, CGI-S≤2, and did not meet DSM-IV criteria for a major depressive episode) to 26 weeks of double-blinded duloxetine (60 mg/day; n=136) or placebo (n=142).161, 162 Duloxetine-treated patients had significantly longer time to relapse (P=0.004); the estimated probability of relapse was 38.3 percent for duloxetine and 19.7 percent for placebo (P<0.05).

The recurrence-prevention trial treated MDD patients (n=514) openly with duloxetine 60-120 mg/day for 10 weeks, and then continued patients (n=413) meeting response criteria (HAM-D17≤9, CGI-S≤2, and did not meet DSM-IV criteria for a major depressive episode) openly on duloxetine 60-120 mg/day for 24 weeks.177 Patients continuing to meet response criteria were randomized to 52 weeks of maintenance treatment with duloxetine (n=146) or placebo (n=142). Time to depressive recurrence was significantly longer for duloxetine-treated patients than for placebo-treated patients (depressive recurrence of 14 percent vs. 33 percent, respectively; P<0.001).

SNRI: Desvenlafaxine Versus Placebo

One trial compared desvenlafaxine with placebo for preventing relapse.176 After 12 weeks of open-label treatment with desvenlafaxine 200-400 mg/day, 375 responders (HAM-D17 total score≤11 on day 84) were randomized to 6 months of double-blind treatment with desvenlafaxine (n=189) or placebo (n=185). Patients receiving desvenlafaxine had significantly longer times to relapse compared with patients receiving placebo (log-rank test, P<0.0001). The percentage of patients relapsing were 24 percent and 42 percent in the desvenlafaxine and placebo groups, respectively (P<0.001).

SNRI: Mirtazapine Versus Placebo

One trial of relapse prevention openly treated patients with recurrent or chronic major depression (n=410) with mirtazapine 15-45 mg/day for 8 weeks to 12 weeks.157 Those in remission (HAM-D≤7 and CGI-I≤2) were randomized to 40 weeks of continuation treatment with mirtazapine (n=76) or placebo (n=80). Relapse rates were statistically significantly lower for mirtazapine-treated patients than for placebo-treated patients (20 percent vs. 44 percent, respectively; P=0.001).

SNRI: Venlafaxine Versus Placebo

Three trials studied venlafaxine;160, 175, 182 one of these trials, the Prevention of Recurrent Episodes of Depression with Venlafaxine for Two Years (PREVENT) study, had multiple phases and was reported in multiple publications.149, 182-187 The PREVENT trial's head-to-head continuation-phase comparison of fluoxetine with venlafaxine has already been presented (Table 28).149 Among completers of this part of the trial, the venlafaxine responders (HAM-D17 total score≤12 and≥50 percent decrease from baseline) were randomized to 12 months of continued venlafaxine 75-300 mg/day (n=129) or placebo (n=129). At month 12, the recurrence probabilities were, respectively, 23.1 percent and 42.0 percent for venlafaxine- and placebo-treated patients (P=0.005).182 Patients taking venlafaxine who maintained their response through 12 months were then again randomly assigned to a second 12 months of venlafaxine (n=43) or placebo (n=40). At the end of this second 12 months of maintenance treatment, recurrence was more common among placebo-treated patients than venlafaxine-treated patients (45 percent vs. 8 percent; P<0.001).183 For the 2-year combined maintenance treatment, the recurrence probability was 47 percent for placebo- and 29 percent for venlafaxine-treated patients (P=0.005).183

One additional study assessed relapse prevention,160 and one study assessed recurrence prevention.175 The relapse-prevention study openly treated 490 patients with major depression with venlafaxine XR 75–225 mg/day for 8 weeks.160 Patients who responded (CGI-S≤3 and HAM-D≤10) were randomized to 26 weeks of double-blind treatment with venlafaxine (n=161) or placebo (n=157). Statistically significantly fewer venlafaxine-treated patients than placebo-treated patients experienced a relapse (28 percent vs. 52 percent, respectively; P<0.001).

The recurrence-prevention study openly treated 495 patients with recurrent major depression for 6 months with venlafaxine 100–200 mg/day.175 After 6 months, those who had responded (HAM-D≤12) were randomized to 12 months of venlafaxine (n=109) or placebo (n=116). The recurrence rate was statistically significantly lower for venlafaxine-treated patients than for placebo-treated patients (22 percent vs. 55 percent, respectively; P<0.001).

Other Second-Generation Antidepressants: Bupropion Versus Placebo

One trial assessed relapse prevention with bupropion.151 Patients with recurrent major depression (n=816) were treated openly for 8 weeks with bupropion SR 300 mg/day. Those who responded (CGI-I score of 1 or 2 during the last 3 weeks of the acute phase) were randomized to placebo (n=213) or continuation treatment with the same dose of bupropion SR (n=210). After 44 weeks, relapse rates were statistically significantly lower for patients on bupropion than for those on placebo (37 percent vs. 52 percent, respectively; P=0.004). The median time to relapse, as defined by the need for treatment intervention after randomization into the double-blind phase, was 24 weeks for placebo and at least 44 weeks for bupropion.

Other Second-Generation Antidepressants: Nefazodone Versus Placebo

One relapse-prevention trial158 and one recurrence-prevention trial169 evaluated nefazodone. In the relapse-prevention study, investigators randomized patients in remission (HAM-D≤10) to 36 weeks of double-blind treatment with nefazodone 400–600 mg/day (n=65) or placebo (n=66).158 Statistically significantly fewer nefazodone-treated than placebo-treated patients relapsed (2 percent vs. 18 percent, respectively; P=0.009). The recurrence-prevention study openly treated 681 patients with chronic or recurrent major depression for 12 weeks with nefazodone 300-600 mg/day.169 Patients who responded (≥50 percent improvement in HAM-D score from baseline) continued open-label nefazodone for an additional 16 weeks, and patients who maintained a response after this 16 weeks of continuation treatment were randomly assigned to 1 year of double-blind treatment with nefazodone (n=76) or placebo (n=84). The rate of recurrence was statistically significantly lower for patients on nefazodone than for those on placebo (30 percent vs. 48 percent, respectively; P=0.043).

Achieving Response in Unresponsive or Recurrent Disease: Overview

Trials relating to treating depressive disorders (MDD, dysthymia, or subsyndromal depression) in patients who had not responded to any acute-phase therapy—often referred to as treatment-resistant or refractory depression—or who suffered a relapse or recurrence focus on using drugs other than any medication first tried (KQ 2b). We review head-to-head evidence for treatment-resistant patients.

Six studies assessed differences among several alternative antidepressants in patients who had either not responded or could not tolerate an acute-phase treatment;191-197 all included venlafaxine as a comparison. This group of trials varied in design; they included two effectiveness studies191, 198 and four efficacy trials.192, 193, 196, 197

Achieving Response in Unresponsive or Recurrent Disease: Key Points

Of six comparative studies, the majority of studies did not report statistically significant differences among compared treatments.193-197 The best evidence comes from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial, which was a good-quality study that indicated no differences in effectiveness among venlafaxine XR, bupropion SR, and sertraline as second-line agents.194 Similar conclusions of no differences can be drawn based on three efficacy trials; one comparing citalopram with venlafaxine XR;196 one comparing fluoxetine with venlafaxine XR;197 and one comparing venlafaxine, mirtazapine, and paroxetine.193 One efficacy trial comparing venlafaxine with paroxetine192 and one open-label Spanish effectiveness study comparing venlafaxine XR, citalopram, fluoxetine, mirtazapine, paroxetine, and sertraline both contradict these findings.191 However, the efficacy trial was only 4 weeks long, which could limit the ability to observe full medication effects. Further, statistically significant differences were noted on the remission but not the response outcome measure.192 Although the effectiveness study191 was larger and potentially more generalizable, the magnitude of differences was relatively small and may not be clinically significant.

Overall, the body of evidence suggests that meaningful differences likely do not exist among compared agents for treatment-resistant depression. However, the degree of conflicting evidence as well as the lack of consistency in statistical significance led us to rate the overall strength of the evidence as low. The body of evidence was limited to relatively few comparisons, and additional studies could influence our overall conclusions of no differences.

Achieving Response in Unresponsive or Recurrent Disease: Detailed Analysis

Six studies assessed differences among alternative antidepressants in patients who either had not responded to or could not tolerate an acute-phase treatment (Table 29).191-197 They covered several antidepressants, but all included venlafaxine (an SNRI) as a comparison. Three efficacy trials compared an SSRI with venlafaxine (an SNRI) in patients with treatment-resistant depression; comparisons included citalopram,196 fluoxetine,197 and paroxetine.192 An additional trial compared venlafaxine with paroxetine but also included a mirtazapine arm.193 Of two effectiveness trials, one compared venlafaxine XR with citalopram, fluoxetine, mirtazapine, paroxetine, and sertraline in patients failing venlafaxine XR 75–225 mg/day or with some other conventional antidepressant therapy.191 A second effectiveness trial compared bupropion SR, sertraline, and venlafaxine XR in patients failing aggressive management with citalopram. This trial also included augmentation strategies that added bupropion SR or buspirone to the citalopram.194, 195, 198 Details on these comparisons are provided here.

Table 29. Head-to-head trials of treatment-resistant and recurrent depression.

Table 29

Head-to-head trials of treatment-resistant and recurrent depression.

Citalopram Versus Venlafaxine XR

One efficacy trial assessed differences between citalopram 20–60 mg/day and venlafaxine XR 75-300 mg/day among 406 patients from Europe and Australia who had not experienced a response to 8 weeks of monotherapy with an adequate regimen of an SSRI other than citalopram.196 After 12 weeks, similar numbers of patients met criteria for remission (HAM-D21≤7; approximately 27 percent for citalopram and 36 percent for venlafaxine; P=0.95).

Fluoxetine Versus Venlafaxine XR

Another efficacy trial compared fluoxetine with venlafaxine in 119 patients who had failed to achieve satisfactory response to at least 6 weeks of SSRI treatment at a therapeutic dose.197 This trial also included treatment arms for olanzapine (an atypical antipsychotic) and olanzapine plus fluoxetine combination, although we did not consider these comparisons. After 12 weeks, a larger percentage of patients treated with venlafaxine than with fluoxetine had a response (≥50 percent improvement in MADRS total score from baseline; 50 percent vs. 34 percent) or went into remission (MADRS≤8; 22 percent vs. 18 percent); statistical significance was not reported for these comparisons.

Paroxetine Versus Venlafaxine XR

A third efficacy trial compared paroxetine with venlafaxine in patients with major depression who either had not responded to or could not tolerate at least two previous treatments for their current depressive episode.192 Patients were to be no more than minimally improved (CGI-I≥3) with their second treatment. The investigators enrolled 123 patients in the study—61 on venlafaxine 200-300 mg/day and 62 on paroxetine 30-40 mg/day—and followed them for 4 weeks. At endpoint, statistically significantly more venlafaxine-treated patients than paroxetine-treated patients were classified as having responded to treatment (≥50 percent improvement in HAM-D from baseline; 45 percent vs. 36 percent, respectively; P=0.07) and being in remission (HAM-D<10; 37 percent vs. 18 percent, respectively; P=0.02).

Paroxetine Versus Venlafaxine XR Versus Mirtazapine

A Chinese trial randomized patients with MDD who had failed two consecutive antidepressant trials to fixed-dose treatment with venlafaxine 225 mg/day (n=50), mirtazapine 45 mg/day (n=55), or paroxetine 20 mg/day (n=45).193 After 8 weeks, response (HAM-D17 reduction from baseline≥50 percent) and remission (HAM-D17≤7) rates were similar across all treatment groups. For response, the figures were 64 percent, 58 percent, and 67 percent, respectively (P=0.664); for remission, the figures were 42 percent, 36 percent, and 47 percent, respectively (P=0.578).

Sertraline Versus Venlafaxine XR Versus Bupropion SR

One study, the STAR*D trial, had several different treatment comparisons. We rated the quality of this trial as good and classify it as an effectiveness trial. Aspects of this trial have been reported in multiple manuscripts; we focused on the randomized medication-switch comparisons in level 2 (i.e., following failure of open-label citalopram) because these were the only direct comparisons of antidepressants included in this review.194, 195 However, we also briefly mention the augmentation comparisons that included second-generation antidepressants.198

The STAR*D trial assessed differences in effectiveness in patients with MDD who had not gone into remission (Quick Inventory of Depressive Symptomatology—Clinician version [QIDS-C-16]≤5) or could not tolerate citalopram during acute-phase treatment.194, 195, 198 Participants eligible for second-step treatment had the option of switching to an alternative medication, cognitive behavioral therapy, or augmentation therapy. To mimic clinical practice, patients could opt to exclude certain second-step treatment options, and they were then randomized to an acceptable treatment option. The investigators compared only the treatments for which patients had accepted randomization. The primary outcome measure was the Quick Inventory of Depressive Symptomatology—Self Report (QIDS-SR).

Of the 727 patients randomized to second-step medication switch, 239 received bupropion SR 150–400 mg/day, 238 received sertraline 50–200 mg/day, and 250 received venlafaxine XR 37.5-375 mg/day. The investigators adjusted doses based on clinical judgment and side effect rating scales. Second-step treatment was continued for up to 14 weeks. At endpoint, response and remission rates were not statistically significantly different among bupropion SR, sertraline, and venlafaxine XR. For response, the figures were 26 percent, 27 percent, and 28 percent, respectively (P>0.05); for remission, the figures were 21 percent, 18 percent, and 25 percent, respectively (P=0.16). Treatments also differed only minimally with respect to tolerability and adverse events.

Level 2 of the STAR*D trial also included a randomized comparison of patients receiving citalopram plus augmentation with either bupropion SR 200–400 mg/day or buspirone 15–60 mg/day. (Buspirone is a psychoactive medication used principally as an anxiolytics; it does not belong to the SSRI/SNRI drug classes.) After 12 to 14 weeks, the percentage of patients with a QIDS-SR response or remission was not statistically significantly different between the patients receiving bupropion SR and buspirone augmentation (P=0.21 and P=0.13, respectively).

Citalopram Versus Fluoxetine Versus Mirtazapine Versus Paroxetine Versus Sertraline Versus Venlafaxine XR

The effectiveness trial randomized 3,502 patients with major depression, dysthymia, or minor depression who had shown inadequate response or intolerance to at least 4 weeks of previous antidepressant treatment with venlafaxine XR 75–225 mg/day or with some other conventional antidepressant therapy.191 Conventional therapy selection was at the discretion of the treating psychiatrist; it included citalopram 20–40 mg/day (n=333), fluoxetine 20–40 mg/day (n=292), mirtazapine 30–45 mg/day (n=133), paroxetine 20–40 mg/day (n=361), sertraline 50–150 mg/day (n=299), and other miscellaneous drug treatments (n=254).

After 24 weeks of treatment, venlafaxine-treated patients had a statistically significantly better rate of response and remission than patients treated with conventional therapy. (For response, the figures were 78 percent vs. 71 percent, respectively; P<0.001; for remission, the figures were 59 percent vs. 52 percent, respectively; P<0.001.) Response and remission rates for venlafaxine XR were statistically significantly better than the rates for each of the individual drugs characterized as conventional therapy except for paroxetine. The response and remission rates in this study were much higher than those reported from the good-quality (STAR*D) effectiveness trial comparing bupropion SR, sertraline, and venlafaxine XR.194, 195 Although differences in measurement scales may partially explain response rates, the reason that remission rates differed remains unclear because both trials used a HAM-D cutoff point of 7 or less to classify persons in remission.

Finally, one systematic review and meta-analysis of five trials reported a greater odds of response (OR, 1.35; 95% CI, 1.19 to 1.52) and remission (OR, 1.35; 95% CI, 1.2 to 1.52) for venlafaxine than for bupropion, citalopram, fluoxetine, and sertraline.145 This analysis appeared to rely on the same data presented above, although we could not confirm which trials contributed to the meta-analysis.

Key Question 3. Efficacy or Effectiveness for Treating Symptoms Accompanying Depression

All Symptoms: Overview

For this issue, we focus on the comparative benefit of medications for patients with depression and an accompanying symptom cluster. We identified studies addressing seven symptom clusters: anxiety, insomnia, low energy, pain, psychomotor change (retardation or agitation), melancholia (a depressive subtype that is a severe form of MDD with characteristic somatic symptoms), and somatization (physical complaints that are manifestations of depression rather than of an underlying physical illness). This set does not represent a complete list of symptoms commonly accompanying depression. For example, we did not identify any studies addressing appetite change—a common accompanying symptom reported by depressed patients.199, 200

For each symptom cluster, we arrange our summary by how the data best addresses the Key Question. We identified 29 relevant studies (Tables 3036). Of these, 20 studies were head-to-head trials and one was a systematic review. Seven trials were placebo-controlled.

Table 30. Studies of adults with major depressive disorders and accompanying anxiety.

Table 30

Studies of adults with major depressive disorders and accompanying anxiety.

Table 31. Trials of adults with major depressive disorders and accompanying insomnia.

Table 31

Trials of adults with major depressive disorders and accompanying insomnia.

Table 32. Trials of adults with major depressive disorder and accompanying low energy.

Table 32

Trials of adults with major depressive disorder and accompanying low energy.

Table 33. Trials of adults with major depressive disorders and accompanying melancholia.

Table 33

Trials of adults with major depressive disorders and accompanying melancholia.

Table 34. Trials or other studies of adults with major depressive disorders and accompanying pain.

Table 34

Trials or other studies of adults with major depressive disorders and accompanying pain.

Table 35. Studies of adults with major depressive disorders and accompanying psychomotor change.

Table 35

Studies of adults with major depressive disorders and accompanying psychomotor change.

Table 36. Studies of adults with major depressive disorders and accompanying somatization.

Table 36

Studies of adults with major depressive disorders and accompanying somatization.

We identified 12 head-to-head trials on anxiety,43, 49, 52, 67, 80, 84, 99, 107, 113, 201-203 six on insomnia,43, 55, 76, 102, 103, 123 two on melancholia,85, 202 one on pain,87 and one each on psychomotor changes202 and somatization.43 Two head-to-head trials assessed more than one symptom subgroup.43, 202 We did not locate any head-to-head trials on low energy.

The open-label effectiveness trial addressing somatization did not meet our eligibility criteria because of the lack of double blinding.128 However, we report on its results because it was a well-conducted randomized controlled effectiveness trial and constitutes the only available evidence on effectiveness for somatization in depressed patients.

The remaining seven studies were placebo-controlled trials. Five addressed pain,204-208 one addressed only anxiety,209 and one addressed anxiety, low energy, and insomnia.210 Two studies reported on adjuvant eszoplicone for insomnia.211, 212

All but two studies52, 80 either were funded by or involved authors funded by pharmaceutical companies.

We rated all studies as fair quality. The fair rating was nearly universally a result of inadequate description of randomization and allocation concealment. A second common weakness was failure to report attrition rates, which occurred in several trials.201, 202, 204, 205, 209 Quality was rated not applicable for the effectiveness trial because it did not meet our initial selection criteria.128 No trial was rated good quality. We excluded five studies because of poor quality: one each on melancholia,213 anxiety,214 and insomnia,215 and the other two on pain.216, 217 Generally, the poor studies suffered high attrition either between treatment groups213 or high overall attrition.214, 216 We rated the insomnia study as poor because the authors failed to provide essential baseline information regarding patient characteristics and did not make clear whether they used an ITT analysis.215 Finally, we excluded a meta-analysis of studies of patients with MDD and pain because of an inadequate literature search, poor assessment of the internal validity of included studies, and poor description of included studies.217

We report on poor studies only if the available evidence was very limited. For any poor studies retained for use in this report, we required, at a minimum, that investigators had employed a randomization scheme and applied ITT analysis.

Detailed information on these poor quality studies can be found in the evidence tables in Appendix D. We included one systematic review and meta-analysis on depressed patients with pain.218 Our evidence tables are presented in Appendix C and provide information on systematic reviews and meta-analyses related to treating depression and accompanying symptoms.

Anxiety: Key Points

Seven head-to-head trials investigated treatment of depression in patients with accompanying anxiety symptoms.80, 84, 99, 113, 201-203 Eleven head-to-head trials43, 49, 52, 67, 80, 84, 99, 107, 113, 201, 203 and two placebo-controlled trials examined treatment of accompanying anxiety symptoms in patients with MDD.209, 210 Six of these trials addressed both treatment of depression in patients with accompanying anxiety symptoms as well as treatment of accompanying anxiety symptoms.80, 84, 99, 113, 201, 203

Of the 14 trials, six compared various SSRIs with each other, six compared an SSRI with an SNRI or another second-generation drug, and two compared an SSRI or another second-generation drug with placebo (Table 30). We rated the strength of evidence that antidepressants are equally efficacious in treating depression in anxious patients and in treating the accompanying anxiety as moderate.

Depression in Patients With Anxiety

Overall, seven head-to-head trials generally indicated that antidepressant medications do not differ in treatment efficacy for depressed patients with accompanying anxiety symptoms. Five trials analyzed a subgroup with identified high anxiety; only two used the same definition criteria (a HAM-D anxiety-somatization factor of 7 or more).99, 201

The head-to-head trials compared SSRIs with each other,201-203 venlafaxine,80, 84, 99 or bupropion SR.113 Studies appeared to compare similar doses of antidepressant medications. Two studies comparing SSRIs (fluoxetine, paroxetine, and sertraline) found no statistically significant differences in depressive improvement, response rates, or remission rates.201, 202 One study comparing escitalopram and paroxetine showed escitalopram to be superior to paroxetine in improving depressive symptoms in a subgroup of patients with high anxiety.203 Three studies comparing an SSRI and venlafaxine showed mixed results. One found a greater decrease in depressive severity and higher response rates with venlafaxine than with fluoxetine,80 and one found no statistically significant difference in depressive severity change, response rates, or remission rates between venlafaxine XR and sertraline,99 and venlafaxine XR and fluoxetine.84 One study comparing sertraline and bupropion SR found no significant differences in response or remission rates.113

Anxiety in Depressed Patients

Overall, results from 11 head-to-head trials and two placebo-controlled trials suggested that antidepressant medications do not differ in treatment efficacy for treating anxiety associated with MDD. Six trials analyzed a subgroup with high anxiety;80, 99, 113, 201, 203, 209 only two used identical definitions to identify the high anxiety group.99, 201 In addition, outcome definitions for anxiety varied. The studies compared similar doses of antidepressants.

The head-to-head trials compared SSRIs with each other, with SNRIs, and with other second-generation drugs (bupropion, nefazodone). Four studies comparing SSRIs (including escitalopram, fluoxetine, sertraline, and paroxetine) found no statistically significant differences for treatment of patients' anxiety symptoms.43, 49, 52, 201 One trial of escitalopram versus paroxetine demonstrated a superior improvement in anxiety scores for escitalopram compared with paroxetine in a subgroup of highly anxious patients.203 Three studies comparing an SSRI (fluoxetine, sertraline) with venlafaxine found mixed results. Two trials reported that venlafaxine produced a greater decrease in anxiety severity than fluoxetine,80, 84 whereas the other study reported similar anxiety reduction for venlafaxine XR and sertraline.99 One study comparing sertraline and bupropion SR found no difference in anxiety reduction.113 Two other studies found no difference in anxiety reduction between paroxetine and nefazodone,107 and between citalopram and mirtazapine.67

The two placebo-controlled trials examined two different antidepressant agents for the treatment of anxiety; they produced conflicting information about the efficacy of the active agent compared with placebo. One trial reported that venlafaxine treatment produced a statistically greater reduction in anxiety scores than placebo.209 In contrast, a trial of bupropion XL failed to demonstrate superiority over placebo for patients with depression and reduced energy, pleasure and interest.210

Anxiety: Detailed Analysis

Head-to-Head Evidence

We identified 12 head-to head trials comparing the efficacy of specific medications treating depressed patients with coexisting anxiety symptoms. Of these, one trial addressed only improvement in depression among persons with anxiety202 and seven studies addressed only improvement in anxiety as an outcome.43, 49, 52, 67, 107, 209, 210

Escitalopram Versus Fluoxetine

One trial compared low-dose escitalopram (10mg/day) with low-dose fluoxetine (20mg/day) over 8 weeks in 240 Chinese patients with MDD.43 Patients were not required to have anxiety for inclusion and no subgroup analysis of patients with anxiety was provided. Response rates for the two HAM-D items for psychological and somatic anxiety (items 10 and 11) showed no significant difference between escitalopram and fluoxetine (Anxiety: psychological 77 percent vs. 76 percent and Anxiety: somatic 75 percent vs. 79 percent, respectively).

Escitalopram Versus Paroxetine

One trial compared high-dose escitalopam (20mg /d) with high-dose paroxetine (40mg /d) over 24 weeks. The investigators retrospectively divided the patients in a larger trial into high and low anxiety subgroups (HAM-A≤20 or HAM-A>20) and the results for depression and anxiety scores were re-analyzed for each subgroup. Here we report the results for the high-anxiety subgroup (n=286). Patients randomized to escitalopram showed a statistically significant greater improvement in both anxiety (HAM-A) and depression (MADRS) scores than those randomized to paroxetine (HAM-A: -17.6 vs. -15.2, P<0.05; MARDS, -24.2 vs. -21.5, P<0.05).

Fluoxetine Versus Paroxetine

Two trials compared the efficacy of low-to-high doses of fluoxetine with similar doses of paroxetine for treatment of anxiety.49, 52 Neither study required high anxiety for inclusion in the analysis.

One trial compared fluoxetine (20–80 mg/day) and paroxetine (20–50 mg/day) in a 12-week trial involving 203 patients with severe MDD.49 Improvements on multiple measures of anxiety did not substantially differ between the two treatment groups.

The other trial compared fluoxetine (20–60 mg/day) and paroxetine (20–40 mg/day) over 6 weeks in 90 patients with severe MDD.52 Mean baseline anxiety severity was similar; each group had a moderate to severe degree of anxiety. Improvements in HAM-A scores were similar for the two groups.

Fluoxetine Versus Paroxetine Versus Sertraline

One RCT compared low-to-high dose fluoxetine (20–60 mg/day), low-to-high dose paroxetine (20–60 mg/day), and low-to-high dose sertraline (50–200 mg/day) over 10 to 16 weeks in patients with MDD of at least moderate severity and high anxiety (as defined by a score on the six-item HAM-D anxiety-somatization factor≥7 [range 0–18]).201 Analyses were performed in the subgroup with high anxiety (n=108 patients from a trial with 284 participants overall); the outcomes included both depressive measures and anxiety measures. Depression outcomes were similar for the three medications, as measured by three outcomes: (1) improvement in HAM-D total scores, (2) improvement in response rates (≥50 percent reduction in HAM-D score; fluoxetine, 73 percent, paroxetine, 77 percent; and sertraline, 86 percent, P=0.405); and (3) improvement in remission rates (HAM-D endpoint≤7; fluoxetine, 53 percent; paroxetine, 50 percent; and sertraline, 62 percent; P=0.588). Authors reported no difference among the three groups with respect to anxiety outcomes (measured by overall change on HAM-D anxiety-somatization factor score).

Fluoxetine Versus Sertraline

One trial compared low-to-medium doses of fluoxetine (20–40 mg/day) and sertraline (50–100 mg/day) over 6 weeks in patients with MDD of at least moderate severity who also had high anxiety as defined by a Covi Anxiety Score≥7.202 The outcome was depression response. Authors reported that response rates (defined by≥50 percent reduction in HAM-D total score) did not differ between the fluoxetine-treated group (48 percent) and the sertraline-treated group (47 percent).

Citalopram Versus Mirtazapine

One trial compared the efficacy of low-to-high dose citalopram (20–60 mg/day) and low-to-high dose mirtazapine (15–60 mg/day) over 8 weeks in 270 patients with MDD of at least moderate severity.67 The outcome was treatment effect on anxiety as measured by HAM-A scores. However, patients were not categorized by anxiety level, and the analysis included all patients with MDD, not merely those with anxiety. The improvement in anxiety symptoms did not differ between citalopram and mirtazapine (mean HAM-A change in both groups was approximately -13 points).

Fluoxetine Versus Venlafaxine

Two trials compared fluoxetine and venlafaxine.80, 84 One trial compared low-to-medium doses of fluoxetine (20–40 mg/day) with low doses of venlafaxine (75-150 mg/day) over 12 weeks in 146 moderately depressed patients with MDD who had a Covi Anxiety Scale score of 8 or higher (consistent with clinically relevant anxiety).80 The other trial compared low-to-high doeses of fluoxetine (20–60mg/d) with low-to-high doses of venlafaxine XR (75–225 mg/d) over 12 weeks in 386 patients with MDD and anxiety (Covi score≥8). Both trials reported depression and anxiety outcomes. The results for depression were conflicting. In the smaller trial, the improvement in depressive severity on the HAM-D was significantly greater in the venlafaxine-treated group than the fluoxetine-treated group (-14.4 points vs. -10.4 points, P=0.0048). In the larger trial no significant difference in depression response or remission was reported. In contrast, venlafaxine was superior to fluoxetine for anxiety response in both trials. In the larger trial there were significantly more HAM-A responders at week 12 in the venlafaxine group compared with the fluoxetine group (P=0.037) and in the smaller trial the mean reduction on the Covi Anxiety Scale was greater for venlafaxine than for fluoxetine (-5.7 points vs. -3.9 points, P=0.001).

Sertraline Versus Bupropion SR

One efficacy trial compared low-to-high dose sertraline with low-dose buproprion SR over 16 weeks in 248 patients with MDD of moderate severity.113 High anxiety patients were defined as those with scores in the top quartile on HAM-A (≥19, consistent with at least moderate anxiety). Outcomes included both depression (HAM-D21) and anxiety (HAM-A) measures. For the subgroup with high anxiety, depression response rates (≥50 percent reduction in total score, approximately 70 percent in each group) and remission rates (endpoint≤8, approximately 70 percent in each group) were similar. Likewise, in the high-anxiety subgroup, authors reported no difference in anxiety reduction (measured by mean change in HAM-A) between patients treated with sertraline (-10.0) and bupropion (-9.7).

Sertraline Versus Venlafaxine XR

One efficacy trial compared low-to-high dose sertraline (50–150 mg/day) with low-to-high dose venlafaxine XR (75–225 mg/day) over 8 weeks in a subgroup of 120 patients with MDD of at least moderate severity and accompanying anxiety, defined as a HAM-D anxiety-somatization score of≥7.99 Outcomes included both depressive (HAM-D17) and anxiety (HAM-A) measures. Authors reported no difference between treatment groups in mean depressive severity reduction (-17.3 for sertraline vs. -14.8 for venlafaxine XR, P=0.7), depression response rates (≥50 percent reduction in total score, 80 percent for sertraline vs. 69 percent for venlafaxine XR, P=0.26), or depression remission rates (endpoint≤7, 63.0 percent for sertraline vs. 54.1 percent with venlafaxine XR, P=0.44).

Anxiety symptom outcomes did not differ between treatment groups for the overall study population (n=163) or for the high anxiety subgroup (n=120). In the overall study population, the mean reduction in HAM-A was -14.1 for the sertraline-treated group and -12.9 for the venlafaxine XR-treated group (P=0.32). In the high anxiety subgroup, response on the HAM-D anxiety-somatization subscale (criteria not described) was similar for both treatment arms (83.3 percent for sertraline vs. 70.5 percent for venlafaxine XR, P=0.12).

Paroxetine Versus Nefazodone

One RCT compared the low-to-medium dose paroxetine (20–40 mg/day) with low-to-high dose nefazodone (200–600 mg/day) for treatment of accompanying anxiety symptoms over 8 weeks in patients with moderate to severe MDD.107 Inclusion in the analysis did not require high anxiety, and patients were not categorized based on anxiety level; the outcome was the mean difference between treatment groups in HAM-A improvement. Authors reported similar improvement in HAM-A for the treatment groups (-8.0 for paroxetine versus -6.5 for nefazodone, P=NS, 95% CI for difference between groups, -0.7-3.8).

Placebo-Controlled Evidence

Two trials examined the efficacy of a second-generation antidepressant only against placebo.

Venlafaxine Versus Placebo

One 12-week study randomly assigned patients with severe MDD to one of three doses of immediate-release venlafaxine or to placebo.209 Inclusion did not require a high anxiety score. Treatment effects on anxiety were analyzed in a subgroup of 346 patients with accompanying anxiety (defined as a score of≥2 [at least moderate] on the HAM-D anxiety-psychological item, range 0–4). Each treatment arm had an equivalent number of patients with high anxiety. All four treatment arms experienced a reduction in anxiety. Patients in all three venlafaxine groups had statistically significant greater improvement in HAM-D anxiety-psychological and anxiety-somatization scores compared with the placebo group. The three venlafaxine groups did not differ from each other in anxiety outcomes.

Bupropion XL Versus Placebo

One placebo-controlled trial randomized 274 patients with depression and reduced energy, pleasure, and interest to 8 weeks of 150 mg/day to 450 mg/day of bupropion XL or placebo.210 Investigators measured anxiety using the anxiety subset of the 30-item Inventory of Depressive Symptomatology- (Interactive Voice Response) Self Report scale (IDS-IVR-30). After 8 weeks study investigators did not see any difference in improvement in anxiety between the bupropion XL and placebo groups: bupropion XL -2.4 compared with placebo -2.1, P=0.16.

Insomnia: Key Points

We identified six head-to-head studies that compared the effects of medications on treatment of depression and accompanying insomnia (Table 31)43, 55, 76, 102, 103, 123 and one placebo-controlled trial.210 Three of these trials required insomnia for inclusion in the analysis.55, 102, 211 Five other trials did not require insomnia for inclusion but rather assessed sleep for all subjects.43, 76, 103, 123, 210 The studies that identified an insomnia group provided data addressing both effects on depressive symptoms and effects on insomnia.55, 102, 211 The other studies provided information solely on insomnia outcomes. Generally, antidepressants were equally efficacious for accompanying insomnia; however, two trials demonstrated that treatment with trazodone produced greater improvement in sleep scores than fluoxetine and venlafaxine103, 123and one trial showed that fluoxetine led to a worsening in sleep parameters and nefzodone to a slight improvement.102, 212 In addition, two trials showed fluoxetine plus eszopiclone to be superior to fluoxetine alone.211, 212 We rated the strength of evidence for depression outcomes in patients with accompanying insomnia as insufficient and for insomnia outcomes in patients with depression as low.

Depressive Episode in Patients With Insomnia

Two head-to-head studies provided evidence regarding comparative efficacy of medications for treatment of depression in patients with accompanying insomnia.55, 102 The studies showed no statistically significant differences in depressive outcomes for fluoxetine compared with paroxetine and sertraline55 or fluoxetine compared with nefazodone.102 Two trials of fluoxetine supplemented with eszopiclone compared with fluoxetine alone showed mixed results for the difference between the groups for depression scores when the sleep items were excluded from the analysis.211, 212

Insomnia in Depressed Patients

Six head-to-head trials provided mixed evidence about the effects of antidepressants on insomnia in patients with depression. Two trials reported greater improvement in sleep scores for trazodone than for fluoxetine103 and venlafaxine;123 however, neither of these analyzed a subgroup of patients with insomnia. One trial found that sleep scores worsened with fluoxetine treatment but not with nefazodone.102 One trial each found no statistically significant differences for patients on the following medications: escitalopram or fluoxetine;43 fluoxetine, paroxetine, or sertaline;55 and fluoxetine or mirtazapine.76 Two trials of fluoxetine supplemented with eszopiclone compared with fluoxetine alone in depressed patients with insomnia showed an improvement in sleep for those receiving concomitant eszopiclone.211, 212 A placebo-controlled study of bupropion XL found a small, statistically significant improvement in insomnia in those taking bupropion.210

Insomnia: Detailed Analysis

Head-to-Head Evidence

Six head-to-head trials addressed this issue.

Escitalopram Versus Fluoxetine

One trial compared low-dose escitalopram (10 mg/day) and low-dose fluoxetine (20 mg/day) over 8 weeks in 240 Chinese patients with MDD.43 The investigators did not require insomnia for inclusion, nor did they present trial results for a subgroup of patients with insomnia. Response rates for the three HAM-D items for initial-, middle-, and delayed-insomnia (items 4, 5, and 6) showed no statistically significant difference between escitalopram and fluoxetine (initial, 77 percent vs. 73 percent; middle, 61 percent vs. 64 percent; delayed, 70 percent vs. 69 percent, respectively).

Fluoxetine Versus Nefazodone

One trial compared low-to-medium doses of fluoxetine (20–40 mg/day) with low-medium doses of nefazodone (200–500 mg/day) in an 8-week trial of 44 MDD patients with insomnia.102 The authors assessed sleep disturbance and improvement using polysomnographic recordings and the sleep items of the HAM-D. Overall nefazodone resulted in significantly less worsening of sleep parameters than fluoxetine (e.g., sleep efficiency and number of awakenings, P<0.05) and more improvement in the combined HAM-D sleep items “sleep disturbance factor” (mean ±SE: fluoxetine 1.5 ±0.4; nefazodone 2.5 ±0.3, P<0.05). Improvement in HAM-D score was similar for the two groups (mean improvement from baseline and 95% CI for fluoxetine 10.3 ±1.35 and for nefazodone 11.5 ±1.41).

Fluoxetine Versus Paroxetine Versus Sertraline

One trial compared low-to-high doses of fluoxetine (20–60 mg/day), paroxetine (20–60 mg/day), and sertraline (50–200 mg/day) in a trial of MDD patients with at least a moderate degree of depression that lasted between 10 and 16 weeks.55 A secondary analysis evaluated depression outcomes in patients with insomnia, defined as a score of at least 4 points on the HAM-D sleep disturbance subscale (a 0 to 6 scale consisting of a summed score of three HAM-D17 sleep items [assessing initial, middle, and terminal insomnia], where higher scores indicated worse insomnia). For the 125 patients in this subgroup, the three SSRIs did not differ significantly on the HAM-D score (overall P=0.853).

This trial also assessed the effect of medications on insomnia. Again, treatment groups did not differ. Insomnia (measured as above on the 6-point scale) improved to a similar degree for all three groups (fluoxetine, -3.1; paroxetine, -2.9; sertraline, -3.1; overall P=0.852).

Fluoxetine Versus Trazodone

One trial compared low-dose fluoxetine (95 percent of participants took 20 mg/day) with low-to-medium dose trazodone (50–400 mg/day, median 250 mg) over 6 weeks in patients with major depression.103 Investigators did not require insomnia symptoms for inclusion and did not analyze an insomnia subgroup. Overall HAM-D sleep disturbance scores improved more in the trazodone group than in the fluoxetine group (-2.7 vs. -1.6; P=0.001).

Fluoxetine Versus Mirtazapine

One trial compared low-to-medium doses of fluoxetine (20–40 mg/day) with low-to-high doses of mirtazapine in an 8-week trial of patients with severe MDD.76 The investigators did not categorize subgroups of patients by the presence or absence of insomnia. They compared outcomes on the Leeds Sleep Evaluation Questionnaire for all trial participants. Total scores were not reported; efficacy on individual items did not differ in any substantial or consistent way between treatment groups.

Venlafaxine Versus Trazodone Versus Placebo

One trial compared low-to-medium doses of venlafaxine (75–200 mg/day) and trazodone (150–400 mg/day) over 6 weeks in patients with major depression.123 Investigators did not require insomnia symptoms for inclusion and did not analyze an insomnia subgroup. HAM-D sleep disturbance scores were better (lower) at endpoint in patients receiving trazodone than in those receiving either venlafaxine or placebo (score 1.42 for trazodone, 2.22 for venlafaxine, 1.95 for placebo; P<0.05). HAM-D sleep disturbance factor scores at endpoint did not differ between venlafaxine and placebo (P=NR).

Fluoxetine Versus Fluoxetine Plus Eszopiclone

Two trials compared fluoxetine (20–40 mg/day) with fluoxetine (20–40 mg/day) and concomitant eszopiclone (3 mg/day) over 8 weeks in depressed patients with insomnia.211, 212 In one trial, the investigators measured an improvement in insomnia using prospective sleep diaries (completed by patients) and the Insomnia Severity Index (ISI) score.211 The other trial used an interactive voice recording system to monitor sleep functions and depression symptoms.212 The adjusted odds ratio for an improvement of 6 points on the ISI for patients receiving fluoxetine plus eszopcilone compared with fluoxetine alone was 7.21 (95% CI, 1.51 to 34.4).211 In the second trial, the patients reported statistically significant improvements in total sleep time and sleep latency.212 Results regarding depressive symptoms were conflicting: there was no statistically significant difference between the two groups in improvement on the HAM-D when sleep items were excluded from the analysis in one trial,211 and in the other trial, the improvement in depression based on the HAM-D remained statistically significant even when insomnia items were removed from the subanalysis.212

Placebo-Controlled Evidence

One placebo-controlled trial randomized 274 patients with depression and reduced energy, pleasure, and interest to 8 weeks of 150 mg/day to 450 mg/day of bupropion XL or placebo.210 Investigators measured insomnia using the insomnia subset of the 30-item IDS-IVR-30. After 8 weeks, participants in the bupropion XL group demonstrated a significantly greater improvement in insomnia score (bupropion XL -2.1; placebo -1.5, P=0.023).

Low Energy: Key Points

One placebo-controlled RCT focused on patients with reduced energy, pleasure and interest (the authors combine low energy and anhedonia items in their analysis) (Table 32).210

The strength of evidence that bupropion XL is superior to placebo for treating depression in patients with low energy or for treating the accompanying low energy is insufficient. The strength of evidence for the comparative efficacy of other antidepressants for treating low energy in depressed patients is insufficient.

Low Energy: Detailed Analysis

One 8-week, placebo-controlled RCT of bupropion XL involved 274 patients with reduced energy and pleasure as determined by their subset score on the self-rated IDS-IVR-30 scale.210 Patients who received 150–450mg of bupropion XL showed a statistically significant greater mean improvement in their total IDS-IVR-30 score after 8 weeks than those who received placebo (bupropion XL -21.3 vs. placebo -17.6, P=0.018). Similarly, the bupropion XL group demonstrated a significantly greater improvement in the energy, pleasure, and interest subset of the IDS-IVR-30 scale after 8 weeks than those receiving placebo (bupropion XL -6.7; placebo -5.3, P=0.007).

Melancholia: Key Points

Two head-to-head studies examined whether, for patients with melancholia, medications differed in their effect on depressive symptoms (Table 33).85, 202 We rated the strength of evidence for the comparative efficacy and effectiveness of second-generation antidepressants (fluoxetine, sertraline, and venlafaxine) for treating depression in patients with melancholia as insufficient.

We found no evidence addressing the comparative efficacy and effectiveness of second-generation antidepressants for the treatment of accompanying melancholic symptoms; thus, the strength of evidence is insufficient.

Depressive Episode in Patients With Melancholia

Two head-to-head trial compared fluoxetine with sertraline202 or venlafaxine.85 One found a greater response rate in patients receiving sertraline than fluoxetine.202 The other reported no difference between the fluoxetine and venlafaxine groups in response and remission rates.85

Melancholia in Depressed Patients

We identified no trial addressing treatment of melancholic symptoms.

Melancholia: Detailed Analysis

Head-to-Head Evidence

We identified two 6-week, fair-quality, head-to-head studies.85, 202

Fluoxetine Versus Sertraline

One trial enrolled patients who were at least moderately depressed (either MDD or the depressed phase of bipolar disorder); patients were randomized to low-to-medium dose fluoxetine (20–40 mg/day) or sertraline (50–100 mg/day) for 6 weeks.202 In the subgroup with melancholia by DSM-III-R criteria, depression response rates (≥50 percent decrease in HAM-D) were significantly better for sertraline than for fluoxetine (59 percent vs. 44 percent, P<0.05).

Fluoxetine Versus Venlafaxine

One trial involved severely depressed hospitalized patients or outpatients with MDD and melancholia per DSM-IV criteria; patients were randomized to 6 weeks of either 60 mg/day of fluoxetine or 225 mg/day of venlafaxine.85 Authors reported no statistically significant difference in response rates (≥50 percent decrease in HAM-D21 or MADRS and CGI improvement score of 1 or 2) between groups (58 percent for fluoxetine, 65 percent for venlafaxine). Similarly, remission rates (final HAM-D score<7) did not differ significantly (fluoxetine, 35.8 percent; venlafaxine, 40.7 percent).

Pain: Key Points

We included one systematic review,218 one head-to-head trial87 and five placebo-controlled trials204-208 that assessed the efficacy of antidepressants for treatment of depression and accompanying pain symptoms (Table 34). The systematic review included studies that reported any pain-specific outcome.218 Two placebo-controlled trials required baseline pain for inclusion;204, 207 these studies provided data addressing both parts of this Key Question (depression outcomes in patients with accompanying pain; pain outcomes in MDD patients). The other four trials did not require pain for inclusion but rather assessed pain symptoms for all subjects; these trials provided information only for pain outcomes.87, 205, 206, 208

We rated all studies fair quality. The strength of evidence for the comparative efficacy of paroxetine and duloxetine for accompanying pain is moderate. The strength of evidence is insufficient for the superiority of duloxetine over placebo for treating the depressive episode, it is moderate for treating accompanying pain.

Depressive Episode in Patients With Pain

Two trials reported conflicting results regarding differences in efficacy between duloxetine and placebo for treatment of depression in patients with mild to moderate pain.204, 207 One RCT of 282 patients suggested similar efficacy for duloxetine and placebo;204 one RCT of 327 patients showed duloxetine to be superior to placebo in treating the depressive episode.207

Pain in Depressed Patients

Pooled results of four head-to-head studies in the systematic review and meta-analysis showed that improvement in pain scores was similar for paroxetine and duloxetine.218 Six studies provided mixed evidence for efficacy of active drugs compared with placebo for treatment of accompanying pain. Six trials compared duloxetine with placebo;87, 204-208 three of these reported statistically greater pain improvement in at least one duloxetine treatment arm.205-207 One study compared paroxetine with placebo;87 and found a statistically greater improvement for paroxetine compared with placebo. Overall, mean differences in pain scores between groups were small and may not be clinically meaningful.

For outcome measures, studies used a visual analog scale (VAS) for overall pain (0 mm to 100 mm scale, where higher scores indicate worse pain) or the Brief Pain Inventory (BPI) severity scale (0 to 10 scale, where higher scores indicate worse pain). No study reported percentages of patients with clinically important improvement in pain. All studies were funded by the maker of duloxetine.

Pain: Detailed Analysis

Head-to-Head Evidence

Paroxetine Versus Duloxetine Versus Placebo

Two multicenter trials compared the efficacy of duloxetine, paroxetine, and placebo. Neither trial required pain symptoms for inclusion; baseline pain severity was mild in both trials.

One systematic review of studies that included at least one pain-related outcome pooled the results of four head-to-head studies of paroxetine and duloxetine.218 The results indicated that the efficacy of duloxetine and paroxetine does not differ meaningfully for treating accompanying pain; the reviewers calculated a pooled weighted mean difference on the VAS of -0.8 mm, slightly favoring paroxetine over duloxetine (95% CI, -3.8 mm to 2.3 mm).

In addition, one trial compared two high doses of duloxetine (80 mg/day and 120 mg/day) to low-dose paroxetine (20 mg/day) and placebo.87 Improvement in overall pain (decrease in 100 mm VAS) was similar for both duloxetine formulations and paroxetine (duloxetine 80 mg/day, -11.2 mm; duloxetine 120 mg/day, -12.2 mm; paroxetine, -16.0 mm; P=0.77 for duloxetine 80 mg vs. paroxetine; P=0.66 for duloxetine 120 mg vs. paroxetine). Mean pain improvement was statistically significantly superior to placebo for paroxetine (P=0.035) but not for either duloxetine formulation (P=0.063 for duloxetine 80 mg vs. placebo; P=0.086 for duloxetine 120 mg vs. placebo).

Placebo-Controlled Evidence

Duloxetine Versus Placebo

Overall, five trials provide evidence on duloxetine versus placebo.204-208 Two trials randomized only patients with pain to high-dose duloxetine (60mg/day) for 7,204 or 8 weeks.207 In the 7-week, multicenter trial, participants were 282 outpatients who met DSM-IV criteria for major depression and reported accompanying pain, with a BPI average pain score of 2 or more at baseline. Patients who had “a primary pain complaint with a diagnosis such as arthritis, fibromyalgia, migraine headache, or acute injury” were excluded. Mean baseline pain severity was moderate (BPI average: 4.85 for duloxetine, 4.62 for placebo). The authors found no statistically significant difference between duloxetine and placebo on either depression or pain outcomes. Mean HAM-D17 improvement was similar for the groups (duloxetine, -10.9; placebo, -10.3; P=0.544). Depression response and remission rates did not differ between duloxetine and placebo (response 42 percent vs. 40 percent, P=0.901; remission 23 percent vs. 24 percent, P=0.887). Mean reduction in BPI average pain was similar for duloxetine and placebo (-2.32 vs. -1.80; P=0.066). Mean changes in BPI worst pain, least pain, and current pain intensity did not differ between treatment groups (P>0.10 for all comparisons). Mean changes in VAS overall pain did not differ between treatment groups (values NR, P=NR). In contrast, depressed patients with at least moderate pain (based on a BPI-SF score of 3 or more) receiving duloxetine in the 8-week RCT demonstrated a significantly better response to treatment than those receiving placebo for depression (MADRS total score: duloxetine -16.69; placebo -11.31, P≤0.001) and pain (BPI-SF average pain: duloxetine -2.57 vs. placebo -1.64, P=0.0008).207

Three trials compared the efficacy of high-dose duloxetine (60 mg/day) to placebo over 8 to 9 weeks for treatment of pain in patients with depression who met DSM-IV criteria for MDD but were not required to have pain.205, 206, 208 Mean baseline pain severity was mild (VAS for overall pain: 29.0, 25.4, and 30.1 for duloxetine, 28.2, 26.2, and 33.35 for placebo). All three studies reported differences in VAS overall pain improvement favoring duloxetine over placebo; in two cases this result reached statistical significance: -8.5 mm vs. -1.3 mm (P=0.019)205 and -11.0 mm vs. -6.4 mm (P=0.037).206

Psychomotor Change: Key Points

One head-to-head trial examined depression response in subgroups with psychomotor change (including psychomotor retardation or psychomotor agitation) (Table 35).202 We graded the strength of evidence for the comparative efficacy of fluoxetine and sertraline for treating the depressive episode in patients with accompanying psychomotor change as insufficient. We found no evidence for the comparative efficacy and effectiveness of second-generation antidepressants for the treatment of accompanying psychomotor symptoms; strength of evidence is insufficient.

Depressive Episode in Patients With Psychomotor Changes

One trial provided evidence that fluoxetine and sertraline have similar efficacy for treatment of depression in patients with psychomotor retardation. It also reported that sertraline was more efficacious than fluoxetine for treating depression in patients with psychomotor agitation.202

Psychomotor Changes in Depressed Patients

We identified no efficacy trials addressing treatment of psychomotor change symptoms.

Psychomotor Change: Detailed Analysis

Head-to-Head Evidence

Fluoxetine Versus Sertraline

One 6-week trial compared low-to-medium doses of fluoxetine and sertraline for treating depression in subgroups of patients with MDD or the depressed phase of bipolar disorder and psychomotor retardation or psychomotor agitation.202 The subgroup with psychomotor retardation comprised 47 patients with a score of 2 or more on HAM-D item 8 (retardation) and 1 or less on item 9 (agitation). In this subgroup, mean HAM-D scores improved similarly for fluoxetine- and sertraline-treated patients (-10.7 vs. -9.1 points, P=NR). Response rates (≥50 percent improvement on HAM-D-17 total score) were also similar for fluoxetine and sertraline (46 percent vs. 48 percent, P=NR). The same study evaluated depression response in a subgroup of 78 patients with psychomotor agitation, defined as a score of 1 or less on HAM-D item 8 and 2 or more on item 9. Among patients with psychomotor agitation, improvement in HAM-D total score was greater in patients receiving sertraline than in those receiving fluoxetine (-12.4 vs. -8.7 points, P=0.02). Response rates were also significantly better for sertraline than for fluoxetine (62 percent vs. 39 percent, P=0.04).

Somatization: Key Points

We identified one randomized, head-to-head trial and one open-label, head-to-head effectiveness trial that compared effects of medications on accompanying somatization in depressed primary-care patients (Table 36).43, 128 The strength of evidence that antidepressants demonstrate similar efficacy and effectiveness for the treatment of accompanying somatization is insufficient. We identified no trials that dealt with treating depression among patients with somatization; thus, the strength of evidence for this issue is insufficient.

Somatization in Depressed Patients

One RCT of escitalopram and fluoxetine found no difference in response rates on the somatization items of the HAM-D (items 12 and 13).43 One open-label effectiveness study found no difference in effectiveness among paroxetine, fluoxetine, and sertraline on a somatization severity scale measure.128

Somatization: Detailed Analysis

Head-to-Head Evidence

Escitalopram Versus Fluoxetine

One trial compared low-dose escitalopram (10 mg/day) with low-dose fluoxetine (20 mg/day) over 8 weeks in 240 Chinese patients with MDD.43 The investigators provided response rates for the two HAM-D items for gastrointestinal and general somatization (items 12 and 13). Escitalopram and fluoxetine did not differ significantly in efficacy detected.

Fluoxetine Versus Paroxetine Versus Sertraline

One open-label, head-to head trial compared the effectiveness of low-dose fluoxetine, paroxetine, and sertraline for the treatment of depression in primary care over 9 months.128 Somatization severity was measured using the Patient Health Questionnaire Somatization Severity scale (0–28 scale, where higher scores indicate worse severity). The report did not present analyses stratified by levels of somatization severity. The authors reported no statistically significant differences in somatization severity scores among treatment groups (-3.1 for fluoxetine, -3.2 for paroxetine, and -4.1 for sertraline, P=NR).

Key Question 4. Safety, Adverse Events, Adherence

This section has two parts: the first relates to comparisons among second-generation antidepressants in general (e.g., as in KQ 1), and the second relates to comparisons between immediate- and extended-release compounds. The basic issues are whether the medications differ in safety, adverse events, or adherence and persistence. Of interest, as before, are the following: SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, and sertraline); SSNRIs and SNRIs (desvenlafaxine, duloxetine, mirtazapine, and venlafaxine); and all other second-generation agents (bupropion, nefazodone, and trazodone).

As described in more detail in the Methods section, we included data from head-to-head trials, placebo-controlled trials, and observational studies for the assessment of the comparative harms of second-generation antidepressants. We included observational studies when the sample size was larger than 1,000 and the study duration at least 3 months.

The two main parts dealing with these issues are generally presented in the same way as the earlier sections: an overview of the articles included a summary of the key points and a detailed analysis of studies. Because specific harms or categories of adverse events are of particular significance, we generally focus on those in subsections. Tables in the subsections on detailed analysis present information as in the tables for KQ 1, with information about comparisons (SSRIs, then SSNRIs and SNRIs, then other antidepressants) from head-to-head trials first, then placebo-controlled trials, then other types of studies. For this purpose, we regard systematic reviews and meta-analyses as observational studies.

Key Question 4a. Comparative Harms and Adherence for Second-Generation Antidepressants

We structured this section in four parts: a general overview, a synthesis of the evidence on adverse events and discontinuation rates, a section on serious adverse events, and a section on adherence. We have distinguished adverse events from serious adverse events based on a Food and Drug Administration (FDA) classification. FDA defines adverse events as any medical occurrence associated with the use of a drug, whether or not considered drug related.220 A serious adverse event is any medical occurrence that results in death, is life threatening, requires hospitalization, results in persistent or significant disability or incapacity, or is a congenital birth defect.220

Adverse Events and Discontinuation Rates: Overview

Most of the studies that examined the efficacy of one drug relative to another also determined differences in harms. Methods of adverse events assessment differed greatly. Few studies used objective scales such as the UKU-SES (Utvalg for Kliniske Undersogelser Side Effect Scale) or the adverse reaction terminology from the World Health Organization (WHO). Most studies combined patient-reported adverse events with a regular clinical examination by an investigator. Determining whether assessment methods were unbiased and adequate was often difficult. Rarely did authors report whether adverse events were prespecified and defined. Short study durations and small sample sizes also limited the validity of adverse events assessment in many trials.

Few randomized controlled trials (RCTs) were designed to assess adverse events as primary outcomes. Most published studies were post hoc analyses or retrospective reviews of databases.

Detailed information on included studies can be found in the evidence tables in Appendix C; information on systematic reviews and meta-analyses on this topic appears in the evidence tables. Most studies were rated fair quality; those rated otherwise are noted in text.

Adverse Events and Discontinuation Rates: Key Points

We analyzed adverse events data of 92 head-to-head efficacy studies of 22,586 patients and 51 additional studies of both experimental and observational design.

In efficacy trials, on average, 63 percent of patients experienced at least one adverse event during treatment. Diarrhea, dizziness, dry mouth, fatigue, headache, nausea, sexual dysfunction, sweating, tremor, and weight gain were commonly reported adverse events. Overall, second-generation antidepressants led to similar adverse events; the frequencies of specific adverse events, however, differed among some second-generation antidepressants. These findings are generally consistent with results from observational studies. Specifically:

  • Venlafaxine was associated with an approximately 49 percent (95% CI, 22 to 82) higher incidence of nausea and vomiting than with SSRIs as a class. The strength of evidence is high.
  • Mirtazapine led to higher weight gains than comparator drugs.75-77, 90, 92, 118 Mean weight gains relative to pretreatment weights ranged from 0.8 kg to 3.0 kg after 6 weeks to 8 weeks of treatment. The strength of evidence for higher risks of weight gain with mirtazapine than with other antidepressants is high.
  • Sertraline led to higher rates of diarrhea than comparator drugs (bupropion, citalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, venlafaxine) in most studies.41, 56, 58-60, 64, 66, 96, 97, 112-114, 132, 133, 201, 221 The incidence was 8 percent (95% CI, 3 to 11) higher than with comparator drugs. Whether this finding can be extrapolated to comparisons of sertraline with the remaining second-generation antidepressants remains unclear. The strength of evidence that sertraline has a higher risk of diarrhea than other antidepressants is moderate.
  • Trazodone was associated with an approximately 16 percent (3 percent less to 36 percent higher) higher incidence of somnolence than comparator drugs (bupropion, fluoxetine, mirtazapine, paroxetine, venlafaxine).103, 104, 109, 119, 123, 124 Whether this finding can be extrapolated to comparisons of trazodone with the remaining second-generation antidepressants remains unclear. The strength of evidence that trazodone leads to higher rates of somnolence than comparator drugs is moderate.
  • Overall discontinuation rates were similar between SSRIs as a class and other second-generation antidepressants. The strength of evidence is high.
  • Discontinuation rates because of adverse events were also similar between SSRIs as a class and bupropion, mirtazapine, nefazodone, and trazodone. The strength of evidence is high. Duloxetine had a 67 percent (95% CI, 17 to 139) and venlafaxine an approximately 40 percent (95% CI, 16 to 73) higher risk for discontinuation because of adverse events than SSRIs as a class. The strength of evidence is high.
  • Discontinuation rates because of lack of efficacy were similar between SSRIs as a class and bupropion, duloxetine, mirtazapine, nefazodone, and trazodone. Venlafaxine had a 34 percent (95% CI, 47 to 93) lower risk of discontinuation because of lack of efficacy than SSRIs as a class. The strength of evidence is high.

Adverse Events and Discontinuation Rates: Detailed Analysis

Tables 3739 present data on the design, interventions, results, and quality ratings of studies we included to examine issues relating to key adverse events and discontinuation. We focused on general tolerability and discontinuation (including nausea and vomiting and selected gastrointestinal problems) (Table 37), weight change (Table 38), and discontinuation syndrome (Table 39). We rated the strength of evidence on general adverse events as high or moderate (depending on the specific measure) and on discontinuation rates as high.

Table 37. Studies assessing general tolerability and discontinuation.

Table 37

Studies assessing general tolerability and discontinuation.

Table 38. Studies assessing changes in weight.

Table 38

Studies assessing changes in weight.

Table 39. Studies assessing discontinuation syndrome.

Table 39

Studies assessing discontinuation syndrome.

Table 40 summarizes, by specific drug, the mean incidence and 95 percent confidence interval for six specific adverse events commonly reported in head-to-head trials. We calculated descriptive statistics based on data from efficacy studies. Comparisons across different drugs, however, should be made with caution given differences in assessment and reporting of adverse events across trials.

Table 40. Incidence of specific adverse events across head-to-head trials (mean percentage) (95 percent confidence interval).

Table 40

Incidence of specific adverse events across head-to-head trials (mean percentage) (95 percent confidence interval).

General Tolerability and Discontinuation

In efficacy trials, on average, 63 percent of patients experienced at least one adverse event during the course of a given study. Diarrhea, dizziness, dry mouth, headache, insomnia, nausea, vomiting, and weight gain were commonly reported adverse events. Several observational studies examined the comparative rates of adverse events among second-generation antidepressants223, 225, 236 Overall, no substantial differences among examined drugs were apparent. However, these studies did not investigate all currently approved antidepressants (Table 37).

The most extensive attempt came from a British study pooling data from prescription-event monitoring of general practitioners 6 months to 1 year after they had issued prescriptions.223, 236 Included drugs were fluoxetine, fluvoxamine, paroxetine, sertraline, venlafaxine, and nefazodone. The final cohort exceeded 10,000 patients for each drug. Demographics and indications were similar among study groups. Overall, the mean incidence of any adverse events per 1,000 patient-months for SSRIs was highest for fluvoxamine (fluvoxamine, 17.6; fluoxetine, 7.0; paroxetine, 7.6; sertraline, 6.2). Physicians, not patients, reported adverse events; the nonresponse rate was 40 percent. Therefore, measurement bias, selection bias, and potential confounding may compromise these results.

Nausea and Vomiting

In efficacy trials, venlafaxine had a consistently higher rate of nausea and vomiting than comparator SSRIs. In six studies, the difference reached statistical significance.72, 73, 81, 83, 86, 93 The rate of patients reporting nausea or vomiting ranged from 6 percent to 48 percent.

These findings are consistent with a British prescription-event monitoring study described earlier.223, 236 Nausea and vomiting were the two most frequent clinical reasons for withdrawal in the first month of treatment for all drugs.

Using data from efficacy trials, we compared the pooled relative risk (RR) of nausea and vomiting for venlafaxine with that for comparator SSRIs as a class (Figure 16). The RR was 1.49 (95% CI, 1.22 to 1.82). The corresponding number needed to harm (NNH) was nine (95% CI, 6 to 23).

Figure 16 shows the study statistics and forest plot for the random effects relative risk meta-analysis of nausea and vomiting with venlafaxine compared with SSRIs; relative risks larger than 1 indicate a higher risk for venlafaxine, relative risks smaller than 1 indicate a higher risk for SSRIs. Alves et al., 1999 report a relative risk of 1.59 (95% CI, 0.93 to 2.77). Ballus et al., 2000 report a relative risk of 3.57 (95% CI, 1.53 to 8.71). Bielski et al., 2004 report a relative risk of 3.92 (95% CI, 1.74 to 9.04). Clerc et al., 1994 report a relative risk of 0.75 (95% CI, 0.20 to 2.80). Costa e Silva et al., 1998 report a relative risk of 1.55 (95% CI, 1.07 to 2.24). De Nayer et al., 2002 report a relative risk of 1.31 (95% CI, 0.75 to 2.30). Dierick et al., 1996 report a relative risk of 1.97 (95% CI, 1.26 to 3.10). McPartlin et al., 1998 report a relative risk of 0.97 (95% CI, 0.72 to 1.32). Mehtonen et al., 2000 report a relative risk of 1.23 (95% CI, 0.78 to 1.98). Nemeroff et al., 2007 report a relative risk of 1.93 (95% CI, 1.34 to 2.83). Rudolph et al., 1999 report a relative risk of 1.77 (95% CI, 1.12 to 2.81). Schatzberg et al., 2006 report a relative risk of 2.16 (95% CI, 1.49 to 3.19). Silverstone et al., 1999 report a relative risk of 1.21 (95% CI, 0.87 to 1.70). Shelton et al., 2006 report a relative risk of 1.05 (95% CI, 0.51 to 2.17). Sir et al., 2005 report a relative risk of 0.92 (95% CI, 0.67 to 1.25). Tylee et al., 1997 report a relative risk of 2.01 (95% CI, 1.48 to 2.77). Tzanakaki et al., 2000 report a relative risk of 0.27 (95% CI, 0.08 to 0.83). Overall, the meta-analysis showed a pooled relative risk of 1.49 (95%CI, 1.22 to 1.82; I2 = 67%)

Figure 16

Relative risk of nausea and vomiting with venlafaxine compared with SSRIs.

In head-to-head trials, fluvoxamine also consistently exhibited higher rates of nausea than other SSRIs.

A pooled analysis of published and unpublished trials of duloxetine did not find significant differences in nausea between duloxetine (40–120 mg/day) and paroxetine (20 mg/day) or between duloxetine (120 mg/day) and fluoxetine (20 mg/day).226

Gastrointestinal Adverse Events

Two RCTs were designed primarily to detect differences in harms between fluvoxamine and citalopram40 and fluvoxamine and fluoxetine.47 A Dutch multicenter trial assessed gastrointestinal side effects from citalopram (20–40 mg/day) and fluvoxamine (100–200 mg/day).40 A total of 217 patients were enrolled for 6 weeks. Overall, 57 percent of patients reported adverse events. Significantly more patients in the fluvoxamine group than in the citalopram group had diarrhea (+13 percent; P=0.026) or nausea (+16 percent; P=0.017). However, the authors did not provide a baseline comparison of gastrointestinal illnesses between groups, so differences at baseline could bias results.

Another trial assessed differences in adverse events between fluvoxamine (100–150 mg/day) and fluoxetine (20–80 mg/day) in 100 patients over 7 weeks.47 No significant difference could be detected, except that patients on fluoxetine suffered nausea significantly more often than those on fluvoxamine (42.5 percent vs. NR; P=0.03).

In a Dutch prospective observational study (n=1,251), diarrhea occurred more frequently in the sertraline group than in patients on fluoxetine, fluvoxamine, and paroxetine (P<0.05).225 This finding is consistent with results from head-to-head efficacy studies. In most studies, sertraline led to higher rates of diarrhea than did comparator drugs (bupropion, citalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, and venlafaxine).41, 56, 58-60, 64, 66, 96, 97, 112-114, 132, 133, 201 Based on our own calculations from data of efficacy studies, the mean incidence was 8 percentage points (95% CI, 3 to 11) higher than with comparator drugs. Results from a Cochrane review confirm these findings; the pooled risk of diarrhea was significantly greater for patients on sertraline than patients treated with bupropion (OR, 3.88; 95% CI, 1.50 to 10.07) or mirtazapine (OR, 2.74; 95% CI, 1.52 to 4.97).221 Whether these findings can be extrapolated to comparisons of sertraline with other second-generation antidepressants remains unclear.

Changes in Weight

Consistently, studies comparing mirtazapine with other second-generation antidepressants reported higher weight gains for mirtazapine than for the comparator groups.75-77, 90, 92, 118, 119 In two RCTs, these differences reached statistical significance.90, 92 Mean weight gains ranged from 0.8 kg to 3.0 kg after 6 to 8 weeks of treatment. Standard deviations of these changes, however, were large, suggesting that some patients had substantially higher weight increases (Table 38).

Two placebo-controlled RCTs specifically assessed weight changes with fluoxetine treatment.155, 190, 229 Findings were mixed. One study, conducted in 671 patients older than 60 years,229 recorded a statistically significant weight loss for fluoxetine compared with placebo.229 The other study reported a weight gain.155, 190

A 32-week acute- and continuation-phase trial assessed differences in weight changes among patients treated with fluoxetine, paroxetine, and sertraline.55, 127 Paroxetine patients showed a significantly greater mean weight change (+3.6 percent) than those taking fluoxetine (-0.2 percent; P=0.015) and sertraline (+1.0 percent; P<0.001). With respect to weight gain of more than 7 percent, significantly more patients in the paroxetine group (25.5 percent) than in the fluoxetine group (6.8 percent; P=0.016) and the sertraline group (4.2 percent; P=0.003) had weight gains of this magnitude.

A pooled analysis of two RCTs comparing escitalopram and paroxetine reported a similar gain in body weight for both patient groups.227 After 27 weeks of followup, patients on escitalopram gained 1.68 kg and patients on paroxetine gained 1.64 kg.

A double-blinded, placebo-controlled, 52-week acute- and continuation-phase trial assessed weight changes during bupropion treatment.230 Patients receiving bupropion showed a modest but nevertheless significant decrease in body weight from baseline (-1.15 kg; P<0.001). The magnitude of weight change was closely related to the patient's body mass index (BMI). Patients with a higher BMI experienced greater weight loss.

A pooled analysis of 10 trials assessed the effects of duloxetine on body weight in patients with MDD.228 Both acute (8 to 9 weeks) and long-term (26, 34, and 52 weeks) studies were analyzed. In acute placebo-controlled studies, duloxetine-treated patients (doses ranging from 20 to 60 mg/day) lost significantly more weight from baseline to endpoint than did patients in the placebo group (-0.5 kg vs. +0.2 kg; P<0.001). The incidences of potentially clinically significant weight loss (≥7 percent) from baseline to endpoint or any time were significantly greater for patients receiving duloxetine compared with those on placebo treatment (P=0.035 and P=0.010, respectively). In acute studies that compared duloxetine with fluoxetine or paroxetine, respectively, no significant differences in weight changes was observed. During long-term treatment, weight changes in patients treated with duloxetine 120 mg (+0.9 kg) and paroxetine 20 mg (+1.0 kg) were similar but significantly greater than in placebo-treated patients (0.1 kg; P≤0.05 for each). A long-term (52 weeks) uncontrolled analysis of a dataset reported a mean weight change from baseline to endpoint of +1.1 kg for duloxetine-treated (80-120 mg) patients (P<0.001).

Discontinuation Syndrome

Withdrawal syndromes (e.g., headache, dizziness, lightheadedness, nausea, anxiety) commonly occur following the abrupt discontinuation of second-generation antidepressants. A systematic review with good reporting conducted by an Expert Working Group of the U.K. Committee on Safety in Medicines (CSM) assessed the frequency of discontinuation syndromes in second-generation antidepressants.233 Based on observational studies, spontaneous reporting data, and clinical trials data, discontinuation syndromes occurred in 0 percent to 86 percent of patients. Because of study durations, dosages, and different assessment methods, incidence rates could not be compared directly. Nevertheless, discontinuation syndromes occurred most commonly with paroxetine and venlafaxine and least commonly with fluoxetine (Table 39).

Four studies not included in the U.K. systematic review provide consistent results with the CSM report.231, 232, 234, 235 One head-to-head trial compared fluoxetine with paroxetine.231 Treatment interruption led to significantly fewer symptoms in the fluoxetine group than the paroxetine group (P=0.001) using the Discontinuation-Emergent Signs and Symptoms checklist (DESS). A placebo-controlled trial of fluoxetine did not find any differences in discontinuation syndromes between fluoxetine and placebo.234 A pooled analysis of six trials investigated the effects of abrupt discontinuation of duloxetine and placebo.235 Significantly more patients receiving duloxetine than receiving placebo reported discontinuation syndromes (44.3 percent vs. 22.9 percent; P<0.05). Finally, a pooled analysis of two RCTs reported more discontinuation-emergent signs and symptoms for patients who were treated with venlafaxine XR than escitalopram (DESS checklist: 5.0 points vs. 2.4 points; P<0.001).232

Discontinuation Rates

In efficacy trials, discontinuation rates because of adverse events were not substantially different.

Table 41 summarizes average discontinuation rates.

Table 41. Average rates of overall discontinuation, discontinuation because of adverse events, and discontinuation because of lack of efficacy.

Table 41

Average rates of overall discontinuation, discontinuation because of adverse events, and discontinuation because of lack of efficacy.

Using data from efficacy studies, we conducted meta-analyses to assess differences in the overall loss to followup, discontinuation rates because of adverse events, and discontinuation rates because of lack of efficacy of SSRIs as a class compared with other second-generation antidepressants (bupropion, duloxetine, mirtazapine, nefazodone, trazodone, and venlafaxine) in adult patients with MDD. Figures 17 through 19 depict relative risks of discontinuation rates comparing these agents with SSRIs as a class. The available data on desvenlafaxine were insufficient for such comparisons. According to our pooled analyses of relative risk, overall discontinuation rates did not differ significantly between SSRIs and bupropion, duloxetine, mirtazapine, nefazodone, trazodone, or venlafaxine (Figure 17). Duloxetine (RR, 1.67; 95% CI, 1.17 to 2.39) and venlafaxine (RR, 1.42; 95% CI, 1.14 to 1.77) had statistically significantly higher discontinuation rates because of adverse events than SSRIs as a class. (Figure 18). For venlafaxine, this finding was balanced by lower discontinuation rates because of lack of efficacy (RR, 0.66; 95% CI, 0.47 to 0.93) (Figure 19). A meta-analyses comparing discontinuation rates of fluoxetine with other SSRIs reported similar results as our analyses.222

Figure 17 summarizes the pooled results of random effects meta-analysis of the relative risks of overall discontinuation rates comparing SSRIs with bupropion, duloxetine, mirtazapine, nefazodone, trazodone or venlafaxine; relative risks greater than 1 favor SSRIs, relative risks smaller than 1 favor the comparator. The pooled analysis of bupropion versus SSRIs reports a relative risk of 0.86 (0.73 to 1.01). The pooled analysis of duloxetine versus SSRIs reports a relative risk of 1.26 (1.04 to 1.53). The pooled analysis of mirtazapine versus SSRIs reports a relative risk of 1.01 (0.79 to 1.21). The pooled analysis of nefazodone versus SSRIs reports a relative risk of 1.01 (0.72 to 1.41). The pooled analysis of trazodone versus SSRIs reports a relative risk of 0.95 (0.56 to 1.91). The pooled analysis of venlafaxine versus SSRIs reports a relative risk of 1.05 (0.93 to 1.19).

Figure 17

Relative risks of overall discontinuation.

Figure 18 summarizes the pooled results of random effects meta-analysis of the relative risks of discontinuation rates because of adverse events comparing SSRIs with bupropion, duloxetine, mirtazapine, nefazodone, trazodone or venlafaxine; relative risks greater than 1 favor SSRIs, relative risks smaller than 1 favor the comparator. The pooled analysis of bupropion versus SSRIs reports a relative risk of 1.08 (0.53 to 2.18). The pooled analysis of duloxetine versus SSRIs reports a relative risk of 1.67 (1.17 to 2.39). The pooled analysis of mirtazapine versus SSRIs reports a relative risk of 1.14 (0.68 to 1.92). The pooled analysis of nefazodone versus SSRIs reports a relative risk of 1.39 (0.85 to 2.27). The pooled analysis of trazodone versus SSRIs reports a relative risk of 0.75 (0.33 to 1.73). The pooled analysis of venlafaxine versus SSRIs reports a relative risk of 1.42 (1.14 to 1.77).

Figure 18

Relative risk of discontinuation because of adverse events.

Figure 19 summarizes the pooled results of random effects meta-analysis of the relative risks of discontinuation rates because of lack of efficacy comparing SSRIs with bupropion, duloxetine, mirtazapine, nefazodone, trazodone or venlafaxine; relative risks greater than 1 favor SSRIs, relative risks smaller than 1 favor the comparator. The pooled analysis of bupropion versus SSRIs reports a relative risk of 0.77 (0.42 to 1.43). The pooled analysis of duloxetine versus SSRIs reports a relative risk of 0.96 (0.39 to 2.34). The pooled analysis of mirtazapine versus SSRIs reports a relative risk of 0.87 (0.40 to 1.90). The pooled analysis of nefazodone versus SSRIs reports a relative risk of 0.63 (0.10 to 4.06). The pooled analysis of trazodone versus SSRIs reports a relative risk of 1.39 (0.44 to 4.38). The pooled analysis of venlafaxine versus SSRIs reports a relative risk of 0.66 (0.47 to 0.93).

Figure 19

Relative risk of discontinuation because of lack of efficacy.

Serious Adverse Events: Key Points

In general, trials and observational studies were too small and study durations too short to assess the comparative risks of rare but serious adverse events such as suicidality, seizures, cardiovascular adverse events, serotonin syndrome, hyponatremia, or hepatotoxicity. The strength of the evidence on the comparative risks of second-generation antidepressants on most serious adverse events is insufficient to draw firm conclusions. Long-term observational evidence is often lacking or prone to bias. Tables 42 to 46 summarize studies included for the assessment of serious adverse events: suicidality (suicidal thoughts and behavior) (Table 42), sexual dysfunction (Table 43), seizures (Table 44), cardiovascular events (Table 45), and other adverse events (Table 46).

Table 42. Studies assessing suicidality.

Table 42

Studies assessing suicidality.

Table 43. Studies assessing sexual dysfunction.

Table 43

Studies assessing sexual dysfunction.

Table 44. Studies assessing seizures.

Table 44

Studies assessing seizures.

Table 45. Studies assessing cardiovascular events.

Table 45

Studies assessing cardiovascular events.

Table 46. Studies assessing other adverse events.

Table 46

Studies assessing other adverse events.

An exception, however, is sexual dysfunction. Eight trials and a pooled analysis of two identical RCTs provide evidence that bupropion causes lower rates of sexual dysfunction than escitalopram,237 sertraline,110-112 and fluoxetine100, 101, 105 (Table 43). The NNT to yield one additional person with a high overall satisfaction of sexual functioning is seven. This treatment effect was consistent across all studies. The strength of evidence that bupropion has lower rates of sexual dysfunction than comparator drugs is high.

Compared with other second-generation antidepressants (fluoxetine, fluvoxamine, nefazodone, and sertraline), paroxetine frequently led to higher rates of sexual dysfunction (16 percent vs. 6 percent).55, 62, 108 The strength of evidence is moderate.

The strength of evidence about the comparative risk of second-generation antidepressants with respect to suicidality is insufficient.

Serious Adverse Events: Detailed Analysis

Suicidality

We found 15 studies that assessed the risk of suicidality (suicidal thinking or behavior) in patients treated with second-generation antidepressants.233, 238-251 Data on the comparative risk of suicidality among second-generation antidepressants are sparse. Results from existing studies do not indicate that any particular drug of interest has an excess risk compared with that of other second-generation antidepressants.239-242, 246, 249, 251 All these studies, however, were underpowered to detect a statistically significant difference between two drugs. Because suicides are a relatively rare event (about 1 in 8,000 psychiatric patients treated with second-generation antidepressants), to detect 20 percent increase in suicide risk, with 80 percent power and a 5 percent level of significance, a trial would need to have a sample size of 1.9 million participants.240 However, 1 in 166 patients reported suicidal feelings while being treated with a second-generation antidepressants.252

In addition, several large attempts were undertaken to determine whether second-generation antidepressants lead to a general increase in the risk of suicidality.239, 240, 249

A recent meta-analysis of observational studies in a combined population of more than 200,000 patients resulted in different findings.249 Results indicated that with the use of SSRIs the risk of attempted or completed suicide was decreased among adults (OR, 0.57, 95% CI, 0.47 to 0.70) and among people ages 65 or older, exposure to SSRIs had a protective effect (OR, 0.46, 95% CI, 0.27 to 0.79).252 These findings were consistent with an FDA data analysis on more than 99,000 participants in 372 trials. FDA pointed out that the risk of suicidality is increased in children and patients 18 to 24 years of age but not in other adult patients.

In 2004 the CSM working group investigated ongoing safety concerns about suicidal behavior with some second-generation antidepressants (citalopram, escitalopram, fluoxetine, fluvoxamine, mirtazapine, paroxetine, sertraline, venlafaxine) in patients with MDD.233 They used data from 477 published and unpublished RCTs on more than 40,000 individuals as well as spontaneous reporting data. These data, however, were limited to studies funded by the pharmaceutical industry.

A meta-analysis limited the CSM data to placebo-controlled trials of SSRIs in about 40,000 adults. Results did not yield any evidence that SSRIs either increase or protect against the risk of suicide (OR, 0.85; 95% CI, 0.20 to 3.40).240 The risk of suicide-related events was similar between second-generation antidepressants and active comparators, although some evidence of an increased risk of suicide attempts was detected (OR, 1.57; 95% CI, 0.99 to 2.55).

Another meta-analysis of published data on more than 87,000 patients in SSRI trials for various conditions reported a significantly higher risk of suicide attempts for SSRI patients than for placebo-treated patients (OR, 2.28; 95% CI, 1.14 to 4.55).238 Furthermore, an increase in the odds ratio of suicide attempts was observed for SSRIs compared with interventions other than tricyclic antidepressants (TCAs) (OR, 1.94; 95% CI, 1.06 to 3.57). No significant difference existed in the pooled analysis of SSRIs compared with TCAs (OR, 0.88; 95% CI, 0.54 to 1.42). The overall rate of suicide attempts was 3.9 (95% CI, 3.3 to 4.6) per 1,000 patients treated with SSRIs, for an incidence of 18.2 suicide attempts per 1,000 patient years.

In addition, the CSM group commissioned an observational study (a nested case-control study) using the General Practice Research Database (GPRD) to investigate the association between antidepressants and suicide attempts. This study used data on more than 146,000 patients with a first prescription of an antidepressant for depression.239 It did not find any evidence that the risk of either suicide (OR, 0.57; 95% CI, 0.26 to 1.25) or suicide attempts (OR, 0.99; 95% CI, 0.86 to 1.14) was greater in patients on second-generation antidepressants than in patients on TCAs.

Findings of other large observational studies and meta-analyses are similar.241-248, 253, 271 Most detected a correlation of SSRI use in suicide attempts and suicides compared with placebo. In general, no significant differences in risks regarding suicidality could be detected between second-generation antidepressants and TCAs.

Sexual Dysfunction

Multiple studies assessed the comparative risk of sexual dysfunction among second-generation antidepressants (Table 43).100, 105, 110, 111, 237, 254, 260 The largest study was a Spanish open-label, prospective observational study using the Psychotropic-Related Sexual Dysfunction Questionnaire (PRSexDQ) in 1,022 outpatients treated with various antidepressants.254 All patients had normal sexual functioning at study onset. Overall, 59 percent of patients experienced some type of sexual dysfunction. Among second-generation antidepressants, citalopram, paroxetine, and venlafaxine had the highest incidence of sexual dysfunction (73 percent, 71 percent, and 67 percent, respectively); mirtazapine and nefazodone had the lowest (24 percent and 8 percent, respectively). A cross-sectional survey of patients on second-generation antidepressants presented similar results.262 Paroxetine had the highest rate of sexual dysfunction; nefazodone and bupropion had the lowest.

Sexual dysfunction was also a commonly reported adverse event for SSRIs and SNRIs in efficacy trials. Most of these studies did not report the use of targeted questions for sexual dysfunction. Therefore, patient-reported numbers might not reflect the true incidence. Patients receiving paroxetine and sertraline frequently reported significantly higher rates of sexual dysfunction51, 64, 66, 96, 112, 114 than did patients in the active control groups. In one trial, significantly more patients on sertraline than on bupropion SR withdrew because of sexual dysfunction (13.5 percent vs. 3.3 percent; P=0.004).112 A pooled analysis of four efficacy trials comparing paroxetine and duloxetine reported significantly higher rates of sexual dysfunction for patients on paroxetine.259

Ten RCTs assessed the comparative risk of sexual dysfunction between two or more second-generation antidepressants as primary outcome measures.100, 105, 110, 111, 237, 257, 258, 260, 272 Table 47 summarizes results of RCTs about sexual dysfunction of patients treated with bupropion or SSRIs.

Table 47. Characteristics of trials comparing bupropion with SSRIs on sexual dysfunction.

Table 47

Characteristics of trials comparing bupropion with SSRIs on sexual dysfunction.

Citalopram Versus Sertraline

A subgroup analysis of a Swedish RCT examined the incidence of sexual dysfunction from citalopram (20–60 mg/day) and from sertraline (50–150 mg/day) in 308 study completers with MDD.272 Outcome assessment was conducted at baseline and at week 24. Citalopram and sertraline did not differ significantly in the magnitude and frequency of sexual dysfunction. Only one patient was lost to followup attributable to sexual dysfunction in this study.

Bupropion Versus SSRIs

A good meta-analysis including data on 1,332 patients with MDD compared sexual adverse events of bupropion and three SSRIs (fluoxetine, paroxetine, sertraline) as a class.261 We do not describe studies included in this meta-analysis individually.101, 110, 111, 260 The rate of sexual satisfaction was significantly higher in patients receiving bupropion than in those receiving SSRIs (RR, 1.28; 95% CI, 1.16 to 1.41). Table 47 summarizes studies comparing bupropion with SSRIs on sexual dysfunction.

Three additional trials were published since the meta-analysis described above had been conducted.100, 105, 237 An 8-week RCT compared efficacy and sexual dysfunction of bupropion SR (150–400 mg/day), fluoxetine (20–60 mg/day), and placebo in 456 outpatients with MDD.100 Findings were consistent with those from the earlier meta-analysis. Throughout the study, patients on bupropion SR experienced significantly less sexual dysfunction than those on fluoxetine. Moreover, beginning at week 1 until endpoint, significantly fewer patients on bupropion than on fluoxetine were dissatisfied with their overall sexual function (P<0.05). The NNT to gain one more patient with high satisfaction with sexual functioning is 6 (95% CI, 4 to 9).

Two identically designed 8-week RCTs compared efficacy and sexual functioning of bupropion XL (150–400 mg/day), escitalopram (10–20 mg/day), and placebo in 830 outpatients (pooled data) with MDD.237 In both of the individual studies and the pooled dataset, the incidence of orgasm dysfunction as well as the incidence of worsened sexual dysfunction at the end of the treatment period was lower with bupropion XL than with escitalopram. In the pooled dataset the incidence rates of orgasm dysfunction at endpoint were 15 percent for bupropion XL and 30 percent for escitalopram (P<0.01); the incidence rates of worsened sexual dysfunction were 20 percent for bupropion XL and 36 percent for escitalopram (P<0.01). Furthermore, at endpoint, escitalopram was associated with statistically significantly worse sexual functioning than bupropion XL in both individual studies and the pooled dataset.

An 8-week RCT evaluated sexual functioning in men and women with MDD receiving either bupropion SR (150–300 mg/day) or paroxetine (20–40 mg/day).105 Sexual functioning decreased significantly in male paroxetine patients, whereas no change in sexual functioning was observed in men receiving bupropion SR. No significant drug differences of sexual functioning were observed for women.

Duloxetine Versus Escitalopram

An 8-month RCT (8 weeks fixed-dose acute-treatment phase followed by a 24-week flexible-dose extension-treatment phase) compared efficacy and sexual functioning of duloxetine (60 mg/day), escitalopram (10 mg/day), and placebo in 684 outpatients with MDD.258 The incidence of treatment-emergent global sexual dysfunction was significantly higher for patients with escitalopram treatment compared with those receiving duloxetine. At the 8-week point, more male patients treated with escitalopram reported worsening in global sexual functioning compared with duloxetine-treated male patients (59.2 percent vs. 36.7 percent; P=0.019), whereas no differences in categorical assessment of changes in global sexual functioning were observed for females.

Sertraline Versus Nefazodone

In one RCT, the emergence of sexual adverse events in patients who experienced sexual dysfunction with sertraline treatment was significantly greater for those receiving sertraline than for those receiving nefazodone.257

Seizures

Evidence from controlled trials and observational studies was insufficient to conclude for or against an increased risk of seizures in patients taking any of the reviewed drugs, including bupropion (Table 44). Two open-label trials examined the rate of seizures during bupropion treatment.264, 265 Both trials reported that the rate of seizures was within the range of other marketed antidepressants, but we rate the strength of this uncontrolled, open-label evidence as low.

A recent review of medical charts on 538 patients with deliberate self-poisoning with antidepressants reported that seizures were more common in patients with venlafaxine overdose than in patients with TCA or SSRI overdose.263

Cardiovascular Events

A nested case-control study examined the risk of sudden cardiac death or near death in patients treated with citalopram, fluoxetine, or venlafaxine (Table 45).266 The study was based on the United Kingdom General Practice Research Database, which included data on more than 207,000 patients who initiated treatment with citalopram, fluoxetine, or venlafaxine for MDD or anxiety. The followup time was an average of 3.3 years. Within the cohort 568 cases of sudden cardiac arrest or near death occurred. These cases were matched with more than 14,000 controls. Results showed that no significant differences in risks for sudden cardiac death or near death were obvious among the examined medications. The adjusted odds ratio associated with venlafaxine relative to fluoxetine was 0.66 (95% CI, 0.38 to 1.14), of venlafaxine relative to citalopram was 0.89 (95% CI, 0.50 to 1.60).

Two case-control studies, not included in this review, indicated an increased risk of ischaemic stroke for SSRIs as a class.273, 274 Neither of these studies provide data on the comparative risks among second-generation antidepressants.

Likewise, a case-control study found no excess risk of idiopathic venous thromboembolism in SSRIs as a class.275 We did not include the study in this report because it does not provide any evidence on the comparative risks among antidepressants.

Other Adverse Events

Diabetes Mellitus

In a cohort of 165,958 patients with depression included in the U.K. General Practice Research Database, a total of 2,243 cases of incident diabetes mellitus and 8,963 matched comparison subjects were identified.270 This nested case-control study showed that recent long-term use (>24 months) of antidepressants in moderate to high daily doses was associated with an increased risk of diabetes (incidence rate ratio (IRR), 1.84; 95% CI, 1.35 to 2.52). The study investigated tricyclic and tetracyclic antidepressants, SSRIs, monoamine oxidase inhibitors, and other antidepressants. For users of SSRIs as a group, increased risk was observed only for recent long-term use of moderate to high daily doses (IRR, 2.06; 95% CI, 1.20 to 3.52). When individual antidepressants were analyzed, increased risk estimates only in long-term users were observed for recent use of fluvoxamine, paroxetine, and venlafaxine. Antidepressant treatment for shorter periods or with lower daily doses was not associated with an increased risk.

Fractures

A large, well-conducted case-control study, including 498,617 subjects (124,655 cases and 373,962 controls) from a Danish national prescription database, reported a significant dose-response relationship for citalopram, fluoxetine, and sertraline with respect to an increase of the risk of fracture (Table 46).267 Among SSRIs, high-dose citalopram, fluoxetine, paroxetine, and sertraline were associated with the highest risk for hip fracture (OR, 1.98, 95% CI, 1.82 to 2.16) and other fractures except fractures of the forearm and spine (OR, 1.38, 95% CI, 1.33 to 1.44). Evidence regarding the impact of the duration of use on the risk of fractures was mixed for second-generation antidepressants.

A Dutch case-control study that did not meet eligibility criteria reported an increase for nonvertebral fractures for SSRIs as a class.276

Increased Risk of Bleeding

Evidence from three case-control studies indicated an increased risk of upper gastrointestinal tract bleeding during SSRI treatment (Table 46).277-279 These studies did not meet eligibility criteria because they provided no information on the comparative risks among individual SSRIs.

Hepatotoxicity

Evidence from controlled trials and observational studies is insufficient to conclude for or against an increased risk of liver toxicity during nefazodone treatment (Table 46). Nevertheless, numerous case reports or prescription event monitoring studies not included in this report contain low quality but potentially important evidence citing an increased risk of liver toxicity during nefazodone treatment.280, 281

Hyponatremia

A retrospective cohort study that did not meet our eligibility criteria reported that hyponatremia in elderly inpatients (mean age 74 years) was significantly more common in patients treated with SSRIs or venlafaxine than in controls not on these drugs (OR, 3.5; 95% CI, 1.4 to 8.9) (Table 46).282 Otherwise, evidence from controlled trials and observational studies is insufficient to conclude for or against an increased risk of hyponatremia in patients treated with SSRIs.

Our methods for this comparative effectiveness review did not permit inclusion of case reports and case series. The published literature includes numerous case reports of hyponatremia and inappropriate secretion of an antidiuretic hormone as rare side effects.283 Even if this evidence is considered weak, such findings might be important in the absence of studies with the methodological strength to account for rare adverse events.

Serotonin Syndrome

Serotonin hyperstimulation syndrome is characterized by symptoms that include mental status changes, agitation, myoclonus, hyperreflexia, sweating, shivering, tremor, diarrhea, lack of coordination, and fever; it can lead to death (Table 46).224 Evidence from controlled trials and observational studies is insufficient to draw conclusions about differences in risk among second-generation antidepressants. The published literature has numerous case reports of serotonin syndrome.284

A postmarketing survey identified cases of the serotonin syndrome in British general practice among patients who received nefazadone.224 In a cohort of 11,834 patients, 19 cases met criteria for the syndrome (incidence=1 case per 1,000 patient-months of treatment with nefazodone). Similar rates of the syndrome were reported for fluoxetine, sertraline, paroxetine, and venlafaxine.

Toxicity

A database analysis in the United Kingdom on fatal toxicity of second-generation antidepressants found venlafaxine to have the highest fatal toxicity rate (13.2 per 1,000,000 prescriptions) among second-generation antidepressants (Table 46).268 A retrospective review of the charts of 2,428 nursing home residents did not detect differences in the risk of falls among residents treated with fluoxetine, paroxetine, and sertraline.269

Adherence and Persistence: Key Points

Adherence rates in efficacy trials range between 90 and 100 percent. Results from efficacy RCTs did not indicate any differences in adherence among second-generation antidepressants. The evidence, however, is limited to few comparisons for which the strength of the evidence is moderate. For the majority of possible comparisons among second-generation antidepressants, the strength of the evidence is insufficient to draw conclusions about the comparative adherence. Findings from highly controlled efficacy studies may have limited applicability to “real-world” practice especially because of the overall short duration of these trials. The evidence is insufficient to conclude on adherence and persistence in effectiveness studies.

Adherence and Persistence: Detailed Analysis

The published literature frequently uses the terms “compliance” and “adherence” interchangeably. Compliance has traditionally been used to describe a patient's ability to take medications as prescribed. Some authors argue, however, that adherence better represents the more complex relationship among patients, providers, and medications; it is meant to reflect the fact that following a medication regimen is not necessarily a simple choice.285 Given the lack of a clear definition, we use the term adherence.

Few efficacy studies reported rates of adherence. Lack of adherence, however, was often used as a reason to exclude patients from the study. Table 48 summarizes included head-to-head trials on adherence. The majority of RCTs that reported on the comparative adherence stated rates between 90 percent and 100 percent. We found 8 head-to-head trials that reported comparative data on adherence.41, 100, 106, 110-112, 124, 260 Overall, adherence rates in RCTs were similar. Most studies, however, provided little or no information on the methods of assessment. For example, a fair study reported that both treatment arms exhibited 100 percent adherence, but the investigators did not describe their method of determining adherence.76

Table 48. Head-to-head trials reporting adherence to second-generation antidepressants.

Table 48

Head-to-head trials reporting adherence to second-generation antidepressants.

None of the three effectiveness studies reported on adherence. To what extent results from highly controlled efficacy trials can be extrapolated to effectiveness settings remains unclear.

Persistence refers to the act of continuing the treatment for the prescribed duration.285 We did not find any studies on persistence.

Only 10 of 18 RCTs reported adherence rates for different treatment arms;41, 50, 100, 103, 109, 110, 136, 147, 286 of these, 8 were head-to-head comparisons (Table 48).41, 100, 106, 110-112, 124, 260 None of these studies noted a significant difference in adherence.

None of the three effectiveness studies reported on adherence. To what extent results from highly controlled efficacy trials can be extrapolated to effectiveness settings remains unclear.

We did not find any studies on persistence.

Key Question 4b. Comparative Harms, Adherence, and Persistence for Immediate- and Extended-Release Second-Generation Antidepressants

This part presents information on studies that examined differences in, first, harms or adverse events and, then, adherence and persistence. The medications of interest are bupropion, fluoxetine, fluvoxamine, mirtazapine, paroxetine, and venlafaxine, which can be administered in daily or weekly dosing regimens or have a variety of formulations, including immediate-release (IR), extended-release (XR), and controlled release (CR). (Some medications may use slightly different terminology or acronyms for long-acting formulations, such as XL for extended release or SR for sustained release).

Harms of Immediate- Versus Extended-Release Formulations: Overview

Of the five head-to-head studies that investigated the comparative efficacy (KQ 1c, above) of daily versus weekly dosing or IR versus ER formulations of various types, four also reported on differences in harms (Table 49).140-143

Table 49. Interventions, numbers of patients, results, and quality ratings of studies comparing harms of daily versus weekly and immediate- versus extended-release formulations.

Table 49

Interventions, numbers of patients, results, and quality ratings of studies comparing harms of daily versus weekly and immediate- versus extended-release formulations.

One study compared fluoxetine daily with fluoxetine weekly.139, 140 Two trials assessed paroxetine IR versus paroxetine CR;141, 142 and one study compared venlafaxine IR with venlafaxine XR.143 No studies of either fluvoxamine or bupropion (the remaining agents with these formulations) reported on harms.

Harms of Immediate- Versus Extended-Release Formulations: Key Points

One trial compared the harms of daily versus weekly dosing of fluoxetine.140 Overall, adverse events rates were similar between fluoxetine daily and fluoxetine weekly dosing regimens. The strength of evidence is moderate that no differences in adverse events exist between daily and weekly formulations of fluoxetine.

Three studies investigated differences in harms for IR versus ER formulations of two other second-generation antidepressants.141-143 Adverse event rates were similar between paroxetine IR and paroxetine CR, except for higher rates of nausea in patients treated with paroxetine IR than paroxetine CR. In addition, venlafaxine IR and venlafaxine XR had similar adverse event rates. The strength of evidence is low that paroxetine IR leads to higher rates of nausea than paroxetine CR.

We could not find any studies on IR and ER formulations of either fluvoxamine or buproprion that reported on harms.

Harms of Immediate- Versus Extended-Release Formulations: Detailed Analysis

Fluoxetine Daily Versus Fluoxetine Weekly

As described in KQ 1, no extended-release formulation of fluoxetine exists. Because of the long elimination half-lives of fluoxetine and its active metabolite norfluoxetine, investigators have explored different dosing regimens for fluoxetine during continuation-phase treatment. A weekly formulation of fluoxetine is administered as an enteric-coated medication.

One RCT determined the comparative harms between daily and weekly fluoxetine regimens.140 In it, the acute treatment period was open label and lasted 7 weeks. Patients who achieved response were randomized to double-blinded continuation treatment with fluoxetine once daily 20 mg, or fluoxetine once weekly 90 mg. During 25 weeks of followup, rates for most adverse event were similar for patients on daily or weekly treatments.

Paroxetine IR Versus Paroxetine CR

One double-blinded RCT142 and a pooled analysis of two identical RCTs141 compared the harms of paroxetine IR with those of paroxetine CR. These studies contained data on 639 patients. Overall, adverse events rates were similar for the treatment groups. One exception, however, was nausea, which occurred significantly more often in patients treated with paroxetine IR than CR during the first weeks of treatment (23 percent vs. 14 percent; P<0.05).141

Venlafaxine IR Versus Venlafaxine XR

One flexible-dose, placebo-controlled RCT compared the efficacy and safety of twice-daily venlafaxine IR (115–125 mg/day) with once-daily venlafaxine XR (124–140 mg/day) in 293 patients with acute-phase MDD.143 During 12 weeks of treatment, the groups did not differ significantly in adverse event rates.

Comparative Adherence and Persistence of Immediate- versus Extended-Release Formulations: Overview

Three studies assessed the comparative adherence of different formulations (Table 50).142, 287, 288 One compared fluoxetine daily with fluoxetine weekly; the other two evaluated paroxetine IR with paroxetine CR and bupropion SR with bupropion XL. We could not find any studies on fluvoxamine and venlafaxine.

Table 50. Interventions, numbers of patients, results, and quality ratings of studies comparing adherence of immediate versus extended release formulations.

Table 50

Interventions, numbers of patients, results, and quality ratings of studies comparing adherence of immediate versus extended release formulations.

We did not find any studies that directly investigated persistence.

Comparative Adherence and Persistence of Immediate- versus Extended-Release Formulations: Key Points

Three studies assessed the comparative adherence of immediate- and extended-release formulations.142, 287, 288 Based on one open-label RCT, adherence to fluoxetine weekly was higher than to fluoxetine daily.287 The strength of evidence is low.

The only double-blinded RCT available reported no significant differences in adherence between patients treated with paroxetine IR and those receiving paroxetine CR (93 percent vs. 96 percent) over a 25-week followup period.142 The strength of evidence is moderate.

A retrospective cohort study, based on U.S. prescription data, showed higher refill persistence for prescriptions of bupropion XL than for those of bupropion SR.288 The strength of evidence is low.

Comparative Adherence and Persistence of Immediate- versus Extended-Release Formulations: Detailed Analysis

Fluoxetine Daily Versus Fluoxetine Weekly

An open-label RCT randomized 109 patients who had responded to fluoxetine 20 mg during acute-phase treatment to fluoxetine 20 mg daily or fluoxetine 90 mg weekly for continuation treatment.287 During a follow-up period of 3 months, adherence to fluoxetine 20 mg daily was significantly lower than to fluoxetine 90 mg weekly (79.4 percent vs. 85.9 percent; P<0.01).

Paroxetine IR Versus Paroxetine CR

A double-blinded RCT of 319 patients compared their adherence to paroxetine IR, paroxetine CR, and placebo.142 Details of the study are described above. During the 12-week study period, adherence rates were similar for the paroxetine IR and paroxetine CR treatment groups (93.2 percent vs. 96.3 percent; P=NR).

Bupropion SR Versus Bupropion XL

A retrospective cohort study, based on a U.S. prescription database, compared refill rates as a proxy for persistence for twice-daily (bupropion SR) versus once-daily (bupropion XL) bupropion treatment for various indications.288 The database collated prescription data on more than 12,000 pharmacy retail chain outlets covering about one-third of all U.S. prescriptions. Over 1 year, data were available on more than 12,000 patients on bupropion SR and more than 257,000 patients treated with bupropion XL. The percentage of patients with more than one refill over a 1-year period was 51.3 percent for bupropion SR and 60.1 percent for bupropion XL (P<0.001). The percentage of patients with more than 6 refills over 1 year was 9.5 percent for bupropion SR and 25.3 percent for bupropion XL (P=NR). Whether prescription refills can be extrapolated to adherence to the dosing schedules remains unclear.

Key Question 5. Efficacy, Effectiveness, and Harms for Selected Populations

Overview: All Subgroups

For this Key Question, we focus on the comparative benefits and harms of second-generation antidepressants for treating a depressive disorder (major depressive disorder [MDD], dysthymia, or subsyndromal depressive disorder) in subpopulations. We focused on the following subgroups: older adults (55 years of age or older); demographic groups defined by sex or race/ethnicity; patients with medical comorbidities (Alzheimer's disease, arthritis, cancer, diabetes, multiple sclerosis, stroke, or cardiovascular disease); patients with psychiatric or behavioral comorbidities (alcohol/substance abuse, generalized anxiety disorder); and patients taking other medications.

We found no studies directly comparing the efficacy, effectiveness, or harms of second-generation antidepressants for any subgroup and the general population. However, a large number of studies conducted subgroup analyses or used subgroups as the study population. Currently, this is the strongest available evidence with which to address this Key Question.

Overall, we included 40 trials (44 articles)42, 48, 53, 58, 59, 65, 66, 68, 92, 105, 106, 119, 134-136, 142, 163, 174, 181, 286, 289-312 and one systematic review313 addressing a subgroup of interest.

We found 11 head-to-head trials that addressed efficacy in older adult populations with MDD; the evidence on older adults with dysthymia or subsyndromal depression was limited to placebo-controlled trials. We did not find any studies that met our eligibility criteria and assessed the comparative efficacy, effectiveness, or harms of second-generation antidepressants in different racial or ethnic groups. Only one randomized controlled trial addressed the general efficacy of sertraline in Latinos and blacks with MDD (and diabetes). For comorbid illnesses, evidence was limited primarily to placebo-controlled trials with the exception of one head-to-head trial that conducted a subgroup analysis in patients with co-occurring generalized anxiety disorder and one systematic review in patients with depression or depressive symptoms after myocardial infarction (MI). Detailed information on these studies appears in Appendix C in the evidence tables.

Because of the lack of evidence from included trials, in some cases we briefly summarize results of studies that did not meet our eligibility criteria but address this Key Question.

Age: Key Points

No studies directly compared the efficacy of second-generation antidepressants in older adults (55 years of age or older) and the general population. Fifteen trials (19 articles) provide mixed evidence on differences in efficacy, effectiveness, and harms in older adult patients treated with second-generation antidepressants.42, 48, 53, 58, 59, 65, 68, 92, 106, 119, 134-136, 142, 163, 174, 289-291 Table 51 (head-to-head) and Table 52 (placebo-controlled) present selected information on these studies.

Table 51. Head-to-head studies on efficacy and harms in older adults.

Table 51

Head-to-head studies on efficacy and harms in older adults.

Table 52. Placebo-controlled studies on efficacy and harms in older adults.

Table 52

Placebo-controlled studies on efficacy and harms in older adults.

Age: MDD

Head-to-head trials provided mixed results on differences in benefits and harms in older adults with MDD. The majority of the trials found no differences in efficacy but suggested some differences in adverse events. Two trials comparing fluoxetine, paroxetine, and placebo reported conflicting results. One trial comparing escitalopram with fluoxetine found a significant difference favoring escitalopram over fluoxetine for efficacy; however, this trial also found neither to be significantly better than placebo. Strength of evidence is moderate for comparative efficacy; strength of evidence is low for harms.

Age: Dysthymia

Two placebo-controlled trials examined the general efficacy of second-generation antidepressants in older adults with dysthymia. One found no difference in response rates between fluoxetine and placebo; the other found significantly greater improvement with paroxetine. Strength of evidence for comparative efficacy and harms is insufficient.

Age: Subsyndromal Depression

We found no head-to-head evidence of differences in elderly populations with subsyndromal depression. One placebo-controlled trial of paroxetine assessed efficacy and harms in a mixed population (dysthymia or subsyndromal depression). Strength of evidence for comparative efficacy, effectiveness, and harms is insufficient.

Age: Detailed Analysis

MDD: Head-to-Head Evidence

Escitalopram Versus Fluoxetine

One 8-week study compared escitalopram (10 mg/day), fluoxetine (20 mg/day), and placebo in 518 participants older than 65 years of age (mean age in each treatment group, 75 years).42 Outcome measures included the MADRS and the Clinical Global Impressions Severity Scale (CGI-S). Patients on escitalopram experienced greater improvement in MADRS total score at week 8 compared with those on fluoxetine (P<0.01). MADRS response rates showed that more escitalopram- than fluoxetine-treated patients achieved response (46 percent vs. 37 percent, P=not reported). Similar results were seen for MADRS remission rates and mean change in CGI-S scores. These efficacy results must be interpreted with caution because neither active treatment was significantly superior to placebo. For some efficacy measures, improvement in the placebo group was significantly greater than in the fluoxetine group. Adverse events were similar for both active treatment groups.

Fluoxetine Versus Paroxetine

Two trials (three articles) compared fluoxetine with paroxetine in patients older than 60 years old.48, 53, 290 One 6-week trial compared fluoxetine (20–60 mg/day) and paroxetine (20–40 mg/day) in 106 depressed patients ages 61 to 85 years (mean age 74 years).53, 290 Paroxetine-treated patients achieved significantly higher HAM-D response rates than fluoxetine-treated patients (P=0.03). Groups did not differ significantly in overall adverse events.

A 1-year Italian study enrolled 242 patients to compare the effects of fluoxetine (20–60 mg/day) and paroxetine (20–40 mg/day) on depressive symptoms, mood, and cognitive function in nondemented patients 65 years of age or older.48 In this long-term study, treatment groups did not differ significantly at study endpoint on HAM-D or CGI-S scores or on most cognitive scales (Blessed Information and Memory Test [BIMT], Mini-Mental State Examination [MMSE], Clifton Assessment Schedule [CLAS]). Severe adverse events were significantly more common in the fluoxetine group than in the paroxetine group (22 events vs. 9 events; P<0.002).

Fluoxetine Versus Sertraline

A 12-week study compared fluoxetine (20–40 mg/day) with sertraline (50–100 mg/day) in 236 participants ages 60 years and older.58 Outcome measures included MADRS, HAM-D, quality of life (Quality of Life Enjoyment and Satisfaction Questionnaire), and cognitive assessments (Shopping List Task [SLT], MMSE, and Digital Symbol Substitution Test [DSST]). Fluoxetine- and sertraline-treated patients demonstrated no significant differences on primary outcome measures (MADRS, HAM-D); HAM-D response rates (71 percent vs. 73 percent) and remission rates (46 percent vs. 45 percent) were similar. Quality of life and other patient-rated measures showed no differences between groups at endpoint. Sertraline-treated patients showed greater cognitive improvement than patients on fluoxetine on the DSST at endpoint (P=0.037). Adverse event rates were similar in the two treatment groups.

A subgroup analysis of this trial focused on 75 patients who were 70 years of age and older. Results demonstrated a greater HAM-D response rate for sertraline than for fluoxetine (58.5 percent vs. 42.4 percent, P=0.027).59 Tolerability was similar between groups with two exceptions. Reports of the adverse event “shaking” differed significantly between the fluoxetine and sertraline groups (0 percent vs. 14.3 percent, P=0.03). Fluoxetine-treated patients showed greater weight loss from baseline to endpoint than sertraline-treated patients (2.8 pounds vs. 0.6 pounds, P<0.05).

Fluvoxamine Versus Sertraline

A 7-week trial compared fluvoxamine (200 mg/day) and sertraline (250 mg/day) in 93 patients 59 years of age and older (mean age for both treatment groups, 68 years).65 HAM-D response rates favored fluvoxamine over sertraline but did not reach statistical significance (71.8 percent vs. 55.6 percent, P=0.12).

Paroxetine IR Versus Paroxetine CR

One 12-week trial compared the efficacy and tolerability of two formulations of paroxetine (paroxetine IR and paroxetine CR) and placebo in an elderly population (60 years of age or older).142 This trial enrolled 323 elderly patients with acute MDD, randomizing them to paroxetine IR (up to 40 mg/day), paroxetine CR (up to 50 mg/day), or placebo. The primary outcome measure was the change of HAM-D scores after 12 weeks of treatment. Patients in both active treatment arms showed similar changes in HAM-D scores (paroxetine IR, -12.3, paroxetine CR, -12.1). Likewise, response rates (65 percent vs. 72 percent) and remission rates (44 percent vs. 43 percent) were similar for the IR and CR groups.

Citalopram Versus Venlafaxine XR

A European 22-week study compared citalopram (10–30 mg/day) with venlafaxine XR (75–150 mg/day) for the treatment of depression in 151 elderly outpatients (mean age, 73 years).68 The investigators found no statistically significant differences at study endpoint in any efficacy outcome measures (MADRS, CGI-S, CGI-I). MADRS remission rates were 23 percent for citalopram and 19 percent for venlafaxine (P=not reported). Both treatment groups reached a 93 percent response rate at week 22 (response defined as a reduction of at least 50 percent in MADRS score). More spontaneously reported adverse events were reported by venlafaxine XR-treated patients than citalopram-treated patients (62 percent vs. 43 percent, respectively); tremor was more common in the citalopram group than the venlafaxine XR group, and nausea or vomiting was more common in the venlafaxine XR group than the citalopram group.

Fluoxetine Versus Venlafaxine

One study compared venlafaxine IR (37.5-225 mg/day) with fluoxetine (20–60 mg/day) and placebo in 300 elderly patients (mean age 71 years old).291 Both treatment groups experienced a significant reduction in HAM-D total scores at 8 weeks; however, the active treatment groups did not differ significantly in HAM-D, MADRS, or CGI scores at endpoint. Remission rates at 8 weeks were 27 percent for venlafaxine and 20 percent for fluoxetine. Venlafaxine-treated patients experienced significantly higher rates of nausea (45 percent vs. 23 percent, P<0.01), dry mouth (23 percent vs. 6 percent, P<0.01), and constipation (22 percent vs.10 percent, P<0.05) but significantly less anxiety (2 percent vs. 10 percent, P<0.005) than patients on fluoxetine

Paroxetine Versus Mirtazapine

One study compared paroxetine (20–40 mg/day) with mirtazapine (15–45 mg/day) in 255 elderly patients 65 years old and older; the trial included an acute phase (8 weeks) and a continuation phase (16 weeks).92 Although the two groups showed similar reductions in HAM-D scores at endpoint, mirtazapine led to a faster response. A Kaplan-Meier analysis showed a significantly faster time to response for mirtazapine than for paroxetine (mean 26 days vs. 40 days; P=0.016). The number needed to treat to yield one additional patient responding with mirtazapine at weeks 1 or 2 was 7. At study endpoint, the number of CGI responders was similar in the mirtazapine and paroxetine treatment groups (64 percent and 56.7 percent, respectively; P=0.267). Significantly more mirtazapine-treated patients reported dry mouth and weight gain (P<0.05). Paroxetine-treated patients reported a significantly higher rate of nausea, tremor, and flatulence (P<0.05).

Sertraline Versus Venlafaxine IR

One poor-quality 10-week trial compared sertraline (up to 100 mg/day) with venlafaxine IR (up to 150 mg/day) among 52 nursing home residents (61 to 99 years of age).314 We graded the quality of this study as poor because of high loss to followup (44 percent), but we note it here because it is the only study comparing these two agents. Venlafaxine-treated patients had a significantly higher rate of withdrawal because of severe adverse events (P=0.022) and withdrawal because of severe adverse events or side effects (P=0.005) than did the sertraline-treated patients.

Paroxetine Versus Bupropion SR

One trial examined the efficacy of paroxetine (10–40 mg/day) and bupropion SR (100–300 mg/day) over 6 weeks in 100 outpatients of ages 60 years and older (range 60 to 88 years).106, 289 This study found no significant differences in efficacy according to all outcome measures between treatment groups. Response rates (≥50 percent reduction in HAM-D scores) were similar in the paroxetine and bupropion SR groups (77 percent vs. 71 percent). Quality-of-life scales (Quality of Life in Depression Scale [QLDS], Medical Outcomes Study Health Survey-Short Form 36 [SF-36]) showed statistically significant improvements in both groups (P<0.0001), but they did not differ significantly between the groups.289 In addition, overall adverse events were similar in the two treatment groups.

SSRIs Versus Venlafaxine

In one study, investigators pooled data from eight randomized trials of venlafaxine IR (75–375 mg/day) or venlafaxine XR (75–225 mg/day), one of several SSRIs (fluoxetine, 20–80 mg/day; fluvoxamine, 100-200 mg/day; paroxetine, 20-40 mg/day), or placebo in the treatment of depression.315, 316 This study failed to meet our eligibility criteria for study design for this Key Question; however, we describe it because of the limited available evidence. The trials included in the analysis varied in length (6 weeks [three studies], 8 weeks [four studies], or 12 weeks [one study]) and included either outpatients (seven studies) or inpatients (one study). Four of the outpatient trials had a placebo arm. For venlafaxine-treated patients, neither age (<50 or≥50 years of age) nor sex affected remission rates.315 Among patients treated with SSRIs, however, a significant interaction was observed between treatment and sex (P=0.004): older women had a poorer SSRI response rate (28 percent) than younger women (36 percent) and both older (35 percent) and younger men (36 percent). Remission rates for older women treated with venlafaxine were higher than remission rates for older women treated with SSRIs (48 percent vs. 28 percent, P=0.0004). Hormone replacement therapy appeared to eliminate these differences. Additional analyses of age subgroups (≤40, 41–54, 55–64, and≥65 years old) and sex subgroups revealed that no significant age-by-treatment, sex-by-treatment, or age-by-sex-by-treatment interactions occurred; men and women of different ages within each treatment group had similar rates of remission, response, and absence of depressed mood.316 Among patients over 40 years old, the rates of adverse events were similar between the treatment groups, although venlafaxine-treated patients 55 to 64 years old reported significantly more nausea than placebo (P≤0.003), and placebo patients 41 to 54 years old reported significantly more headache than venlafaxine (P≤0.01).

Mirtazapine Versus Trazodone

One 6-week study compared mirtazapine with trazodone in patients with MDD older than 55 years old.119 Efficacy outcome measures in this trial favored mirtazapine, but differences did not reach statistical significance. More mirtazapine-treated patients discontinued treatment than did those on either trazodone or placebo. Both treatments were associated with more somnolence and dry mouth than placebo (P≤0.05); trazodone treatment was associated with significantly more dizziness and blurred vision compared with placebo (P≤0.05).

MDD: Placebo-Controlled Evidence

We did not include any placebo-controlled trials assessing response or remission in older adults with MDD because we found ample head-to-head evidence. However, we included placebo-controlled trials reporting maintenance of remission or prevention of relapse.

Escitalopram Versus Placebo

One trial assessed prevention of relapse in MDD patients 65 years of age and older.163 After 12 weeks of open-label treatment with escitalopram, patients who achieved MADRS remission were eligible for randomization to escitalopram (10 or 20 mg/day) or placebo for 24 weeks of double-blind treatment. Of the 405 patients who entered the open-label period, 305 were randomized to double-blind treatment. Over 24 weeks, a significantly higher proportion of placebo- than escitalopram- treated patients relapsed (33 percent vs. 9 percent, P<0.001). The estimated hazard ratio for time to relapse (based on Cox proportional hazard model) was 4.44 (95% CI, 2.41 to 8.17); P<0.001.

Sertraline Versus Placebo

A 100-week maintenance trial assessed the efficacy of sertraline (50–100 mg/day) compared with placebo in preventing depression recurrence in 113 elderly (65 years old and older) community residents.174 The trial found no statistically significant difference in the proportion of depression recurrence (HAM-D≥13 and met DSM-III-R criteria for MDD) between sertraline and placebo (45 percent vs. 54 percent, P=0.21). However, patients on sertraline experienced a longer time to recurrence than did patients on placebo (92 weeks and 48 weeks, respectively).

Dysthymia: Head-to-Head Evidence

We found no head-to-head trials satisfying our eligibility criteria that addressed efficacy or harms in older adults with dysthymia.

Dysthymia: Placebo-Controlled Evidence

Fluoxetine Versus Placebo

One randomized controlled trial of good quality examined the efficacy and harms of fluoxetine (20–60 mg/day) in dysthymia patients 60 years old and older over 12 weeks.136 Intention-to-treat results indicated that fluoxetine had limited efficacy. Response rates on the HAM-D favored fluoxetine over placebo, but the two groups did not differ significantly (27.3 percent vs. 19.6 percent; P<0.4).

Paroxetine Versus Placebo

A large, primary-care-based effectiveness study (two articles) randomized 656 patients with dysthymia or minor depression to 11 weeks of paroxetine, placebo, or behavioral therapy.134, 135 Participants were stratified into patients 60 years and older (n=415) and patients younger than 60 years (n=241) for ITT analysis. In the 60 or older subgroup, paroxetine-treated patients showed a greater change in HSCL-D-20 scores than placebo-treated patients (P=0.004).134 Effects were similar for patients with dysthymia and minor depression. For older dysthymia patients with high or intermediate baseline functioning scores, paroxetine significantly improved mental health functioning compared with placebo. Overall, however, improvements of mental health functioning were not statistically significantly different between dysthymia patients receiving paroxetine and those receiving placebo.

Among the younger patients, treatment groups did not differ significantly on the HSCL-D-20 scale.135 For dysthymia only, the remission rate of patients with at least 4 weeks of treatment was significantly higher in the paroxetine group than in the placebo group (80 percent vs. 44 percent; P=0.008). Paroxetine was not more efficacious than placebo in patients with minor depression.

Subsyndromal Depression: Head-To-Head Evidence

We found no head-to-head trials satisfying our eligibility criteria.

Citalopram Versus Sertraline

One nonrandomized trial evaluated citalopram (20 mg/day) and sertraline (50 mg/day) in the treatment of 138 nondemented elderly patients with minor depressive disorder and subsyndromal depression.137 Although this trial does not meet eligibility criteria because of the study design (because of flawed randomization, it is essentially a nonrandomized trial), we describe it here because it is the only comparative evidence in this population. Both treatments improved depressive symptoms (as measured by the HAM-D); HAM-D remission rates did not differ significantly at endpoint (53 percent vs. 42 percent, P=0.25).

Subsyndromal Depression: Placebo-Controlled Evidence

We found one trial (described above) providing evidence on elderly patients with dysthymia or subsyndromal depression.135

Race or Ethnicity: Key Points

No studies directly compared the efficacy, effectiveness, or harms of second-generation antidepressants among different races or ethnicities. One study compared sertraline with placebo in low-income minorities with comorbid diabetes to assess quality of life (Table 53).292 Strength of evidence is insufficient for comparative efficacy, effectiveness, and harms.

Table 53. Studies of efficacy, effectiveness, and harms for race or ethnicity subgroups.

Table 53

Studies of efficacy, effectiveness, and harms for race or ethnicity subgroups.

Race or Ethnicity: Detailed Analysis

Head-to-Head Evidence

No head-to-head trials on the efficacy, effectiveness, or harms of second-generation antidepressants compared different racial or ethnic groups.

Placebo-Controlled Evidence

Fluoxetine

One poor trial evaluated the efficacy of fluoxetine compared with placebo in the treatment of patients with comorbid HIV/AIDS.317 Owing to the scarcity of evidence examining race or ethnicity, we describe it here. A total of 118 patients were randomized to 8 weeks of treatment with either fluoxetine or placebo. Of all participants, 67 percent were white, 19 percent black, and 14 percent Latino; only 1.7 percent (n=2) were female. Loss to followup was significantly greater among Latinos (53 percent) than blacks (14 percent) and whites (28 percent) (P<0.05). Ethnicity was not associated with the total number of treatment side effects or dosage. Response rates among subjects who completed the study were higher in the fluoxetine group (white, 84 percent; black, 50 percent; Latino, 67 percent) than the placebo group (white, 43 percent; black, 36 percent; Latino, 80 percent). The differences were not significant; however, this may be because of the small sample size, particularly in the Latino group.

Sertraline

One trial randomized 89 low-income Latinos and blacks with diabetes to sertraline (50–100 mg/day) or placebo for 6 months.292 HAM-D scores decreased significantly in both groups but there was no difference between sertraline- and placebo-treated patients. Similar results were seen for quality of life subscales and scores—no differences between treatment groups.

Duloxetine

Two pooled analyses of seven placebo-controlled duloxetine trials assessed the efficacy and tolerability of duloxetine in Latino318 and black patients319 compared with white patients. We excluded both studies because they did not meet our study design eligibility requirements, but we describe them here because of the very limited available evidence on race or ethnicity. The first analysis included 1,342 white and 120 Latino patients and found no difference in efficacy outcomes.318 These two groups did not differ significantly in discontinuation rates due to adverse events or in the types or occurrence of specific adverse events. The second analysis of 1,300 white and 123 black patients also found no evidence for a differential effect of duloxetine in these subgroups for either efficacy or safety outcomes.319

Sex: Key Points

Two head-to-head studies provided limited evidence on differences in men and women (Table 54). Strength of evidence for comparative efficacy and effectiveness is insufficient. Strength of evidence for harms is low.

Table 54. Studies of efficacy, effectiveness, and harms for sex subgroups.

Table 54

Studies of efficacy, effectiveness, and harms for sex subgroups.

Sex: Detailed Analysis

Head-to-Head Evidence

Paroxetine Versus Sertraline

A Swedish randomized controlled trial compared paroxetine (20-40 mg/day) with sertraline (50–150 mg/day) in a 24-week study involving 353 patients.66 Paroxetine-treated women had significantly greater rates of decreased libido than sertraline-treated women (8.8 percent vs. 1.8 percent; P<0.05). Conversely, paroxetine-treated men had lower rates of decreased libido than sertraline-treated men; however, the differences were not statistically significant (12.7 percent vs. 3.8 percent; P=ns).

Paroxetine Versus Bupropion

One study randomized patients to paroxetine (20–40 mg) or bupropion (150–300 mg).105 Subgroup analysis found a significant difference in antidepressant-related sexual dysfunction in men but not in women. Women treated with paroxetine or bupropion did not differ significantly in sexual function. However, paroxetine-treated men reported a worsening of sexual function whereas bupropion-treated men had no significant change in sexual function (Sex FX total, P<0.002).

One 14-week retrospective cohort study of paroxetine (mean dose 30.7 mg/day), sertraline (99.0 mg/day), venlafaxine (151.6 mg/day), and moclobemide (a monoamine oxidase inhibitors drug; 485 mg/day) evaluated disturbances in sexual drive/desire and arousal/orgasm in depressed patients who completed 8 weeks of the study.256 This study did not meet our inclusion criteria; however, we describe it here because of the paucity of evidence on this topic. In this study, men reported greater impairment in drive/desire than did women (P<0.05). Men and women did not differ significantly on the arousal/orgasm scale (P=0.21). Rates of dysfunction in all treatment groups were similar for men; among women, sertraline and paroxetine appeared to be associated with greater dysfunction. All drugs appeared to be equally effective in reducing depressive symptoms (main effect for time, P<0.001); a favorable drug response was associated with less sexual dysfunction.

Placebo-Controlled Evidence

Duloxetine Versus Placebo

We briefly describe a study that did not meet our eligibility criteria. A pooled data analysis of seven placebo-controlled duloxetine trials assessed safety and tolerability of duloxetine for the treatment of MDD in 560 men and 1,062 women.320 No clinically meaningful differences were observed between men and women in safety and tolerability with duloxetine treatment. This analysis showed no significant differential sex effects for pulse rate, blood pressure, or weight. Withdrawals attributed to adverse events were similar for men and women. The only significant difference was in the occurrence of nausea; the nausea rate among placebo-treated patients was significantly greater in females than in males (10.7 percent vs. 3.7 percent, P<0.008).

Comorbidities: Key Points

We found no studies directly comparing the efficacy, effectiveness, and harms of second-generation antidepressants between depressed patients with comorbidities and the general population. However, numerous studies conducted subgroup analyses or used subgroups as the study population (Table 55). Strength of evidence is insufficient for comparative efficacy, effectiveness, and harms.

Table 55. Studies of efficacy, effectiveness, and harms for subgroups by comorbidity.

Table 55

Studies of efficacy, effectiveness, and harms for subgroups by comorbidity.

We present our findings differently in this section because we found just a handful of studies for each of the various subgroups with different comorbid illnesses. We note in the text whether the study addresses patients with MDD, dysthymia, or subsyndromal depression. In addition, the evidence is overwhelmingly placebo-controlled; therefore, we do not present the evidence under subheadings of head-to-head evidence and placebo-controlled evidence for each comorbid illness.

Comorbidities: Detailed Analysis

Alcohol/Substance Abuse

Fluoxetine Versus Placebo

One randomized 12-week trial evaluated fluoxetine and placebo in the treatment of depression in methadone-maintained opioid addicts.286 Among the entire sample (n=44), BDI (mean decrease for fluoxetine vs. placebo -8.0 vs. -4.7, respectively) and HDRS scores (mean decrease for fluoxetine vs. placebo: -6.0 vs. -7.7, respectively) decreased in both groups, but the treatment groups did not differ significantly. Among those subjects with major depression (n=31), the rate of change of depressive symptoms did not differ significantly by treatment group (fluoxetine vs. placebo) over time (BDI, -7.8 vs. -3.4; respectively; HDRS, -5.1 vs. -6.9, respectively).

Sertraline Versus Placebo

Three trials comparing sertraline and placebo in the treatment of patients with depression and co-occurring alcoholism had consistent findings.295, 297, 304 A 24-week study compared sertraline (50–150 mg/day) with placebo in recently detoxified alcohol-dependent patients with current depressive symptoms.295 Response (>50 percent decrease in MADRS score) was slightly higher in sertraline- than placebo-treated patients (44 percent vs. 39 percent). Both groups experienced significant improvements in HAM-D and MADRS scores during the study, but the two groups did not differ significantly. Relapse rates were higher in sertraline- than placebo-treated patients (31.8 percent vs. 23.1 percent), but the difference did not reach statistical significance (P=0.37). Adverse event rates were similar for the treatment groups.

A 12-week trial showed similar results.297 In this study, 82 currently depressed, actively drinking alcohol-dependent subjects were randomized to sertraline (50–200 mg/day) or placebo. The groups did not differ significantly in depression symptoms. However, in women, treatment with sertraline was associated with less depression at the end of treatment than placebo based on HAM-D scores (P=0.04) and BDI scores (P=0.005). There was no treatment group difference for men.

The third study was structured differently but produced similar results.304 This study randomized 345 patients with co-occurring MDD and alcohol dependence to sertraline (50–200 mg/day) or placebo for 10 weeks. After the run-in period, two groups of patients were randomized separately based on HAM-D scores: Group A scores were≥17; Group B scores were≤16. Mean reduction in HAM-D scores did not differ significantly between all sertraline-treated (-10.8) and placebo-treated (-9.6) patients (P=0.14). HAM-D response rates did differ significantly: in Group A, sertraline led to a significantly higher response rate than placebo (64 percent vs. 47 percent, P=0.022) whereas in Group B, sertraline patients had a significantly lower response rate than placebo patients (58 percent vs. 77 percent, P=0.018). Overall, the incidence of adverse events was similar for the two groups; however, significantly more sertraline-treated patients discontinued because of adverse events than did placebo-treated patients (P<0.05).

Nefazodone Versus Placebo

One randomized trial compared nefazodone and placebo in the treatment of depressed patients with comorbid alcohol dependence over a period of 10 weeks.296 Nefazodone was similar to placebo, as measured by improvement in depression on the HAM-D from intake to study endpoint (mean change in HAM-D score for nefazodone vs. placebo: -12.25 vs. -12.55, P=0.51).

Alzheimer's Disease or Dementia

Sertraline Versus Placebo

Two 12-week trials comparing sertraline and placebo in depressed patients with comorbid Alzheimer's disease provided mixed results.298, 310 One trial randomized 44 patients to sertraline (25–150 mg/day) or placebo and showed statistically significant improvement in efficacy in sertraline-treated patients compared with placebo, as measured by both the Cornell Score for Depression in Dementia (CSDD) (P=0.002) and the HDRS (P=0.01).298 More patients treated with sertraline than with placebo responded (38 percent vs. 20 percent). The groups did not differ in frequency of adverse events.

The other trial randomized 133 patients to sertraline (50–100 mg/day) or placebo and found no significant difference between groups in CSDD scores (P=0.97) or remission rates (OR, 2.06; 95% CI, 0.84 to 5.04).310 Also in contrast to the other trial, sertraline treatment was associated with more adverse events, but the groups did not differ significantly in occurrence of serious adverse events (P=0.23).

Arthritis

Duloxetine Versus Placebo

One trial evaluated the efficacy of antidepressants in depressed patients with comorbid arthritis.305 This study is a subgroup analysis of a larger placebo-controlled trial in elderly patients randomized to duloxetine (60 mg/day) or placebo.219 The subgroup analysis included 233 subjects with MDD and co-occurring arthritis, diabetes, and/or vascular disease. No statistically significant treatment-by-comorbidity interactions occurred for any comorbidity (P=0.266) in HAM-D, GDS, or SF-36 scores or in response or remission rates.

Cancer

Fluoxetine Versus Placebo

One study compared fluoxetine and placebo in cancer patients with accompanying depressive symptoms (subsyndromal or minor depression).312 Eligibility criteria stated that to qualify for this study, patients had to have at least some depressive symptoms. The aim of the study was to assess adherence to cancer treatment regimen and changes in quality of life by treating patients with fluoxetine before a determination of clinical depression. The study randomized 163 patients to fluoxetine (20 mg/day) or placebo for 12 weeks. Fluoxetine-treated patients showed significant improvements compared with patients on placebo.

Coronary Artery Disease

Citalopram Versus Placebo

One 12-week Canadian study assessed the efficacy and tolerability of citalopram (20–40 mg/day) and placebo in reducing depressive symptoms in patients with co-occurring coronary artery disease (CAD).308 Improvements in depressive symptoms were greater for citalopram than placebo. Mean HAM-D scores at endpoint showed significantly greater improvement in citalopram- than in placebo-treated patients (14.9 vs. 11.6, P=0.005); the between-group difference was 3.33 (95% CI, 0.80 to 5.85). Citalopram-treated patients also demonstrated significantly greater decrease in mean BDI-II scores at endpoint (P<0.05); between-group difference was 3.61 (95% CI, 0.58 to 6.64). The citalopram group had a lower overall withdrawal rate (13 percent vs. 30 percent, P=NR); however, withdrawals attributed to adverse events were similar between treatment groups.

Diabetes

Sertraline Versus Placebo

One study (described above in race/ethnicity) randomized 89 low-income Latino and black patients with diabetes to sertraline (50–100 mg/day) or placebo for 6 months.292 HAM-D scores decreased significantly in both groups: sertraline- and placebo-treated patients did not differ at the end of the study. Similar results were seen for quality of life subscales and scores—no differences between treatment groups.

Only one study assessed prevention of recurrence of major depression in patients with diabetes.181 In the induction phase, 351 patients with moderately severe and recurrent major depression and co-occurring type 2 diabetes were treated with sertraline for 16 weeks. Those who recovered (per DSM-IV criteria) were randomized to double-blind treatment with sertraline or placebo for 52 weeks or until recurrence of depression. Maintenance of response was significantly greater with sertraline (hazard ratio 0.51, 95% CI, 0.31 to 0.85; P=0.02).

Generalized Anxiety Disorder

Fluoxetine Versus Venlafaxine

A subgroup analysis of a trial described in KQ 184 assessed the efficacy of fluoxetine (20–60 mg/day), venlafaxine XR (75–225 mg/day), or placebo in 92 MDD patients with comorbid generalized anxiety disorder.306 Treatment with venlafaxine XR resulted in greater HAM-D response than treatment with fluoxetine or placebo.

Heart Failure

Sertraline Versus Venlafaxine

The Sertraline Against Depression and Heart Disease in Chronic Heart Failure (SADHART-CHF) trial randomized 469 patients with MDD and comorbid heart failure (left ventricular ejection fraction≤45 percent) to sertraline (50–200 mg/day) or placebo for 12 weeks.311 Both groups showed reduction in HDRS score, but the between-group reduction was not significant (P=0.89). Significantly more sertraline-treated patients withdrew because of adverse events believed to be study-drug-related than did placebo-treated patients (11.5 percent vs. 6 percent, P=0.03). The groups did not differ significantly in serious adverse events.

HIV/AIDS

Fluoxetine Versus Venlafaxine

Two placebo-controlled studies evaluated the efficacy of fluoxetine versus placebo in the treatment of patients with depression and comorbid HIV/AIDS.293, 294 The first study, a 12-week randomized trial, compared fluoxetine and placebo;294 the second, a 12-week, randomized trial, compared fluoxetine, testosterone, and placebo.293 In both studies, fluoxetine and placebo response rates (57 percent vs. 41 percent294 and 54 percent versus 44 percent293) did not differ significantly. However, these studies may not have been powered to detect a statistically significant difference.

Multiple Sclerosis

Paroxetine Versus Venlafaxine

We identified only one study assessing the efficacy and tolerability of antidepressants for depression with comorbid multiple sclerosis (MS).307 Forty-two MS patients diagnosed with MDD and/or dysthymia were randomized to paroxetine (10–40 mg/day) or placebo for 12 weeks. Although more paroxetine-treated patients achieved at least a 50 percent reduction in HAM-D scores (57 percent) compared with placebo-treated patients (40 percent), the difference was not statistically significant (P=0.354). Paroxetine- and placebo-treated patients showed improvement in secondary measures (CES-D, MFIS [Modified Fatigue Impact Scale], SF-36), but the treatment groups did not differ significantly on any of them. Paroxetine patients reported higher rates of nausea, headache, dry mouth, and sexual dysfunction.

Myocardial Infarction

One systematic review313 and three placebo-controlled trials299, 300, 309 addressed depression and comorbid myocardial infarction. Two of the trials were included in the systematic review.

SSRIs

AHRQ sponsored a systematic review of postmyocardial infarction (post-MI) depression; the authors concluded that SSRIs improved depression in post-MI patients.313 A good-quality 25-week trial randomized 54 patients to fluoxetine (20–60 mg/day) or placebo for the treatment of depression after a first MI.300 Another trial randomized patients to sertraline (50–200 mg/day) or placebo for 24 weeks for treating depression in patients with acute MI or unstable angina.299 In both trials, active treatment was associated with a significantly greater response rate than placebo (sertraline, 67 percent; placebo, 53 percent; P=0.01;299 fluoxetine, 48 percent; placebo, 26 percent; P=0.05300).

Mirtazapine Versus Venlafaxine

A study randomized 91 patients to mirtazapine (30–45 mg/day) or placebo for 8 weeks of acute treatment (and a 16-week continuation phase).309 After 8 weeks of treatment, mirtazapine was superior to placebo based on BDI and CGI scales but not HAM-D. The difference between treatment groups in mean decrease in HAM-D score was not significant at 8 weeks (standardized effect size [SES] 1.30 vs. 0.96). Based on change in HAM-D score at 8 weeks, more mirtazapine-treated patients were responders (57 percent vs. 40 percent), but the difference was not significant (P=0.18). Mirtazapine-treated patients showed a significantly greater decrease in BDI score at 8 weeks (-4.6 vs. -1.72, P=0.02). Decrease in CGI score was greater in mirtazapine-treated patients but the difference was not statistically significant (P=0.06). The differences between groups in decrease in HAM-D scores and BDI scores over 24 weeks was not statistically significant (P=0.36 and P=0.07). The difference in CGI scores over 24 weeks favored mirtazapine; the difference was significant (P=0.05). Mirtazapine patients experienced significantly more fatigue (P=0.02) and changes in appetite (P=0.02) over 24 weeks.

Stroke

Three placebo-controlled studies evaluated the efficacy of citalopram, fluoxetine, or sertraline in the treatment of patients with poststroke depression.301-303

Citalopram Versus Venlafaxine

A 6-week randomized trial evaluated the efficacy of citalopram (10–40 mg/day) and placebo in poststroke depression.301 Citalopram was associated with significantly greater improvements in depression than placebo on the HAM-D; mean improvements for citalopram compared with placebo were 8.0 vs. 7.2, respectively.

Fluoxetine Versus Venlafaxine

One 8-week trial compared fluoxetine (20–40 mg/day), an herbal supplement, and placebo in moderately to severely depressed patients after a stroke.303 Fluoxetine-treated patients showed a significantly greater HAM-D response rate than placebo-treated patients (65.5 percent vs. 21.4 percent, P<0.01). No serious side effects were reported in either group, and no patients withdrew because of adverse events.

Sertraline Versus Venlafaxine

A 26-week trial evaluated the efficacy of sertraline and placebo in the treatment of minor depression and less severe depression in stroke patients.302 Sertraline and placebo did not differ significantly in either response rates (week 6: 56 percent vs. 46 percent, respectively; week 26: 76 percent vs. 78 percent, respectively) or remission rates (week 6: 59 percent vs. 51 percent, respectively; week 26: 81 percent vs. 87 percent, respectively). However, at week 26, sertraline was associated with greater improvements in quality of life than placebo (effect size not reported, P<0.05).

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (7.4M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...