Key Findings and Strength of Evidence

Pharmacotherapy (particularly second-generation antidepressants [SGAs]) is the primary intervention for treating patients with major depressive disorder (MDD) in primary care. Nonetheless, primary care patients and clinicians may prefer other options (or at least want to be able to consider them). These include psychotherapeutic interventions, complementary and alternative medicine (CAM) options, exercise, or a combination of these treatments. Our report provides a comprehensive summary of the available evidence on comparative effectiveness and risk of harms of commonly used pharmacological and nonpharmacological treatments for MDD.

In this review we focus on two key issues that primary care physicians commonly face:

1. How do different treatment options compare as an initial treatment choice, and how effective are SGAs compared with nonpharmacological interventions?
2. For patients whose depression did not achieve remission following initial treatment with an SGA, what is the comparative effectiveness of alternative pharmacological and nonpharmacological options?

Overall, the available evidence was characterized by substantial methodological shortcomings and a lack of adequate assessment of harms. In addition, outcomes reporting bias often appeared to be an issue. For example, publications frequently did not report remission and adverse events, yet trials on treatment of patients with MDD are unlikely to fail to assess these outcomes.

The available evidence base has some clear limitations. Some nonpharmacological interventions have never been compared with any SGAs. Very limited evidence is available to address the comparative effectiveness of second-step therapies (i.e., treatment options for patients who did not achieve remission after an initial treatment trial). Further, the role of depression severity as a moderator of comparative treatment effectiveness, whether for first- or second-step therapies, has received very little direct testing in head-to-head trials.

Nevertheless, we were able to draw some conclusions. Because reliable evidence supports similar effectiveness within the class of SGAs, our conclusions are likely valid for the entire class of SGAs.

Comparative Benefits and Harms of Treatment Options for Initial Treatment of Patients With Major Depressive Disorder

Across all interventions, we graded the strength of evidence (SOE) as moderate for only one comparison: namely, SGAs compared with cognitive behavioral therapy (CBT). Results from trials of this comparison indicate that SGAs and CBT have similar effectiveness regarding symptomatic relief in patients with mild to severe MDD. For risk of harms, we graded the SOE as moderate for some outcomes of three comparisons, namely SGAs compared with CBT, acupuncture, and St. John’s wort. For all three comparisons, patients treated with SGAs had a higher risk of experiencing adverse events or discontinuing treatment because of adverse events than patients treated with CBT, acupuncture, and St. John’s wort. The evidence is insufficient to draw conclusions about differences in serious adverse events such as suicidal ideas and behavior.

Our confidence in findings from the remaining comparisons of SGAs with other treatment options was low or insufficient, indicating that these bodies of evidence had major or unacceptable deficiencies. Nevertheless, for most comparisons the overall findings did not show a statistically significant difference in effectiveness but did indicate a lower risk of adverse events for nonpharmacological treatment options. Notable exceptions are omega-3 fatty acids, which appear to have lower effectiveness than SGAs, and the combination of SGAs with acupuncture, which appears to have greater effectiveness than SGA monotherapy. Our confidence in these findings, however, is low and results have to be interpreted cautiously. In addition, for many comparisons that are limited to single trials, determining whether similar treatment effects between SGAs and other interventions are based on similar effectiveness or high placebo response rates is impossible.

The available data offers no conclusions on how selection of treatment strategies might differ based on a patient’s severity of depression. Overall, data do not indicate differences in the comparative effectiveness between SGAs and nonpharmacological interventions for patients with severe MDD. This important question concerning MDD severity, although, raised by a few systematic reviews,^22–24 remains without a clear answer.

Beyond the two articles identified comparing switching and augmentation strategies employing a limited number of medication options or CT, the absence of relevant comparative data about which treatment options are most effective for those needing second-step treatment (about 70% of patients with MDD)^26,27 was striking. Table 30 summarizes our main findings and the respective certainty that we have about these findings, presented as SOE grades (high, moderate, low, or insufficient).⁸³ In this table, we do not present comparisons for which we found no studies whatsoever or for which we were unable to estimate the comparative effectiveness with network meta-analyses. We discuss the summary of findings in more detail below.

Table 30

Summary of findings with strength of evidence: Comparative benefits and harms of second-generation antidepressants and other treatment options as an initial choice for the treatment of patients with major depressive disorders.

For psychotherapies, the available evidence based on 20 randomized controlled trials (RCTs) with 3,000 patients suggests similar beneficial treatment effects of SGAs and psychotherapies, either alone or in combination. Except for SGAs compared with CBT, however, the SOE was low or insufficient, indicating a strong uncertainty associated with these findings.

With respect to risk of harms, SGAs often had higher rates of adverse events or discontinuation rates because of adverse events than psychotherapies. For most of these comparisons, the SOE was also low or insufficient. For example, the evidence was insufficient to draw any conclusions about the comparative risk for serious adverse events. Reasons for low or insufficient SOE grades reflected mainly levels of risk of bias for individual trials and lack of precision of results that encompassed substantial benefits for both interventions.

Many trials had methodological shortcomings such as high dropout rates or lack of blinding of outcome assessors that reduced our confidence in the results. In addition, few trials adequately determined or reported differences in harms. Some comparisons were based on single trials with small sample sizes, which led to indeterminate results because of wide confidence intervals that encompassed appreciable benefits for both comparators. The best available evidence for psychological interventions with moderate SOE was SGAs compared with CBT monotherapy. We found no statistically significant difference in treatment effects on response or remission in our analysis of trials that we rated as low or medium risk of bias trials, although a sensitivity analysis of remission that included three trials that we rated high risk of bias yielded a result that favored SGAs.

For the comparison of SGAs with CAM interventions, we identified 20 RCTs including 2,600 patients comparing an SGA with one of six CAM therapies for treating patients with MDD. Individual trials faced the same methodological issues as trials for psychological interventions. In addition, all trials of CAM interventions used either moderate or low SGA doses as comparators. We rated nearly half of them as high risk of bias (nine trials). Few trials adequately assessed and reported the risk of harms. Because of the lack of evidence and the methodological limitations of many head-to-head trials, we relied on both direct evidence and network meta-analyses to draw conclusions. With the exception of omega-3 fatty acids, beneficial effects appeared to be similar for SGAs and CAM interventions; however, results for comparisons of SGAs with acupuncture, S-adenosyl-L-methionine (SAMe), and St. John’s wort are limited to low SOE, indicating substantial uncertainty of findings. Network meta-analyses resulted in higher response rates for SGAs than omega-3 fatty acids.

Based on two RCTs with low SOE, we found that the beneficial treatment effects of SGAs and exercise, either alone or in combination, were not significantly different. In one trial, patients in the exercise groups reported a slightly lower risk of side effects (diarrhea) than those treated with SGAs.

We did not find any trials on behavior therapy and behavior modification, meditation, or yoga that met our eligibility criteria.

Comparative Benefits and Harms as a Function of Baseline Depressive Severity

The evidence was insufficient to draw any firm conclusions about comparative differences in benefits and harms among interventions of interest as a function of depressive severity. Table 31 summarizes our findings and the respective certainty that we have about these findings, presented as SOE grades (high, moderate, low, or insufficient).⁸³

Table 31

Summary of findings with strength of evidence: Variation in effectiveness by severity for second-generation antidepressants compared with other treatments for patients with major depressive disorder.

Comparative Benefits and Harms of Alternative Pharmacological and Nonpharmacological Options for Patients Whose Depression Did Not Achieve Remission Following Initial Treatment With a Second-Generation Antidepressant

Table 32 summarizes our findings and the respective certainty that we have about these findings, presented as SOE grades (high, moderate, low, or insufficient).⁸³ Comparisons only involved medications and CT; no eligible trials involving CAM or exercise interventions were identified.

Table 32

Summary of findings with strength of evidence: Comparative benefits of second-generation antidepressants and other treatment options as a second-step choice for the treatment of major depressive disorder (KQ 2a).

Two trials involved 1,992 patients and provided data for four comparisons. All findings suggested little difference in benefit for depression regardless of whether a switch or augmentation strategy was used or whether medications or cognitive therapy (CT) were involved. Both trials suffered from attrition rates of more than 20 percent, and all comparisons other than SGA switch compared with SGA switch were based on data from one study. For all the comparisons except one, the SOE was low, indicating limited confidence that the estimate of effect lies close to the true effect for these outcomes.

Comparative Benefits and Harms of Second-Step Therapies as a Function of Baseline Depressive Severity

The evidence was insufficient to draw any conclusions about differences in benefits and harms among second-step interventions of interest as a function of depressive severity. Table 33 summarizes our findings.

Table 33

Summary of findings with strength of evidence: Second-generation antidepressants compared with other treatments for major depressive disorder—Does effectiveness vary by severity? (KQ 2b).

Findings in Relationship to What Is Already Known

Our findings are consistent with several prior systematic reviews and meta-analyses that compared SGAs with nonpharmacological interventions. Most of these reviews, however, included populations that were not eligible for our review, such as patients with minor depression, bipolar disorder, or dysthymia.

For psychological treatments, one meta-analysis found that serotonin-specific reuptake inhibitors (SSRIs) were more effective than psychotherapy in treating patients with depressive disorders; however, this effect was small and potentially clinically insignificant.¹⁴⁵ Another meta-analysis found that SGAs and psychotherapy have equivalent efficacy in the short term after 6 to 26 weeks of treatment.⁴² Our finding that SGA monotherapy, CBT, interpersonal therapy, and PSYD may all have equivalent effects in the short-term treatment of depressed patients is consistent with those results.

Our results are also consistent with the recommendations of both the American Psychiatric Association²⁰ and the U.S. Department of Veterans Affairs/Department of Defense.¹⁴⁶ These two groups consider both pharmacotherapy and psychotherapy to be appropriate individual first-step treatments for mild to moderate MDD. Furthermore, they state that pharmacotherapy plus psychotherapy may be a useful initial treatment for patients with moderate to severe MDD and for those with MDD and comorbid conditions. Although our results are consistent with the recommendations from these two entities, a case could be made for preferring psychological interventions as the first-step treatment for patients with mild to moderate MDD because psychological interventions have fewer to no side effects, and cognitive-behavioral interventions may have enduring effects that reduce subsequent risk.¹⁴⁷ Our results diverge from the APA and Department of Veterans Affairs/Department of Defense guidelines in that CBT had similar levels of effectiveness of symptomatic relief as SGAs, suggesting that there is no evidence-based reason to prefer SGAs over empirically supported psychological interventions. Again, this may be especially relevant given that the overall risk for adverse events or discontinuation of treatment because of adverse events, is lower for psychological therapies than with SGAs.

Several reviews have been done of CAM therapies for treating MDD patients; these include an APA Task Force Report, Clinical Guidelines from the Canadian Psychiatric Association, and a systematic review from the U.S. Department of Veterans Affairs.^50,51,148 Additionally, many reviews of individual CAM therapies have been published for the treatment of MDD,^{55,144,149,150} including reviews by the Cochrane Collaboration.^48,151,152

Although one systematic review of acupuncture concluded that it had efficacy comparable with that for antidepressant medications,¹⁴⁴ a Cochrane review⁴⁸ and reviews from the American Psychiatric Association, Canadian Network for Mood and Anxiety Treatments, and Department of Veteran Affairs agree that the evidence is insufficient to recommend acupuncture as monotherapy or combination therapy for treating MDD patients. Some reports recognize that risk of harms for acupuncture may be low. Nevertheless, most reports note that current trials often have high risk of bias. Similarly, we found few high-quality trials to support the use of acupuncture for MDD. Nevertheless, we found that a few RCTs, in addition to network meta-analysis, may indicate (a) similar effectiveness for acupuncture monotherapy compared with SGA and (b) better treatment response for a combination of acupuncture with SGA compared with only SGA. However, we concluded that the SOE for these associations was low due to the relative paucity of trials and high risk of bias among those trials we identified. It is also important to note that all trials of acupuncture we identified were conducted in China, where publication bias for trials of acupuncture continues to be problematic.^153,154

Both the U.S. and Canadian reviews recommend omega-3 fatty acids as augmentation for treating patients with mild to moderate MDD, noting modest evidence of efficacy and low risks of harm. However, a well-done systematic review and meta-analysis comparing omega-3 fatty acids with placebo found only a small, non-significant benefit that was largely attributable to publication bias.⁵⁵ Currently, the Cochrane Collaboration is conducting a systematic review on its use for treatment of MDD.¹⁵² Our network meta-analyses clearly favored treatment with SGAs over omega-3 fatty acids monotherapy.

Although the Canadian guidelines recommend the use of SAMe as monotherapy for mild to moderate MDD, the U.S. report calls for more studies to determine its efficacy.^50,51 Most studies of SAMe are limited to parenteral administration of the supplement, which appears to have better efficacy than a placebo.¹⁵⁵ However, few studies evaluate oral preparations, and little is known about optimal SAMe dosing.¹⁴⁹ We found only one trial to evaluate comparative effectiveness and concluded evidence was insufficient to make a recommendation for (or against) use of SAMe.

St. John’s wort is perhaps the most commonly evaluated CAM therapy for MDD patients. Both the U.S. and Canadian guidelines recommend St. John’s wort for first-step treatment of mild to moderate MDD, whereas there is less consensus on its use for severe MDD. A Cochrane review evaluating 18 RCTs comparing St. John’s wort with placebo concluded there was superior efficacy for St. John’s wort but noted high heterogeneity among trials. However, their analysis of 17 head-to-head RCTs comparing St. John’s wort with both tricyclic antidepressants and SSRIs demonstrated similar treatment effectiveness for patients with mild to moderate MDD.¹⁵¹ The Cochrane study authors concluded that “…an attempt at treating mild to moderate major depression with hypericum…is clearly justified.” In contrast, although we found no difference in treatment outcomes between SGAs and St. John’s wort, we concluded the SOE supporting these findings was low. Although our analyses used the same studies found in the Cochrane review, there are two important differences in our methods. First, we rated individual study quality using the Cochrane Risk of Bias tool,⁷⁵ which was not available to the Cochrane reviewers who instead used a combination of the Jadad and Internal Validity scales. While the Cochrane reviewers noted that “…the majority of trials were of high quality...,” we rated several trials as low quality, which were subsequently excluded from our primary analyses. Second, we noted that almost all trials of St. John’s wort were conducted using low antidepressant doses,^68,69,156 which led us to decrease the SOE for all treatment outcomes. Although in sensitivity analyses we did not show a difference in outcomes between trials using low versus moderate antidepressant doses, no trial compared St. John’s wort to fully adequate antidepressant doses, which might falsely bias our analyses towards concluding there were no treatment differences between the two agents.

Two additional differences between our conclusions and those of the Cochrane review are worth mentioning. First, we noted that participants in included trials had moderate to severe MDD according to reported baseline HAM-D scores. The definitions of depressive severity have varied, with limited empirical research available to define distinctions between mild, moderate, and severe.⁶⁹ Our analyses apply definitions of mild, moderate, and severe that compare HAM-D ranges to semistructured interviews and disease severity assessments;⁶⁹ these thresholds define our study population as moderate to severe rather than mild to moderate as previously reported. Second, in their review, the Cochrane authors noted that trials conducted in Germany tended to demonstrate more favorable results for St. John’s wort compared with trials conducted outside of Germany. However, their analysis held for the placebo trials but not for the direct head-to-head comparisons. Likewise, we did not find any difference in outcomes between head-to-head studies conducted in Germany versus elsewhere. Therefore, we concluded there was no clear bias based on study country. In summary, we concluded that in patients with moderate to severe MDD, we did not show differences in treatment outcomes between SGAs and St. John’s wort, but the SOE for these findings is low largely due to moderate to high risk of bias in many studies and comparisons using inadequately dosed antidepressants.

Numerous systematic reviews and meta-analyses have been done on exercise for depression.^{49,63,64,157–161} These reviews have examined a variety of types of exercise, including walking, aerobic and nonaerobic forms of movement, and strength training, using randomized and nonrandomized designs and various comparison groups, including no treatment, wait-list controls, and active treatments. Overall, exercise has been found to have a small to moderate clinical benefit when compared with no treatment, wait-list, or placebo and comparable benefit when compared with other active treatments, including SGAs. Our findings are consistent with the recent Cochrane Review by Cooney et al.⁴⁹ that included a separate analysis of SGAs versus exercise and found that the SGA (sertraline) was no more effective than exercise for reducing depression. The Cooney et al. report included four studies—we included two in our review and excluded the other two; for the latter, one was excluded because the population was older adults with minor depression rather than MDD¹⁶² and the other because the population was patients with coronary artery disease.⁶⁵

Current literature suggests that depression severity is an important factor to consider when deciding to treat with an antidepressant. In particular, patients with higher severity MDD respond better to medication than those with lower severity depression,²² possibly because those with low depressive severity respond well to a placebo arm (making it more difficult to detect a statistically significant difference in treatment response between drug and placebo). In any case, based on trials that met the eligibility criteria for our report, we could not draw any firm conclusions about whether depression severity influences the comparative benefits and harms of SGAs and psychological interventions or CAM treatments.

Applicability

The scope of this review was limited to trials that enrolled adult patients with MDD. We did not attempt to review literature on interventions for children with MDD or for patients with subthreshold depression, dysthymia, psychotic depression, or perinatal depression. Because of the serious methodological limitations of some trials, the degree of applicability of some of our findings to real-world settings might be compromised, grades of low or insufficient for SOE also reflect that problem.

The included trials covered populations with mild, moderate, and severe MDD. Most trial populations, however, excluded patients with medical comorbidities or suicidal ideas and behaviors; few trials included elderly patients. Furthermore, most trials were conducted in clinical settings. Results from samples of patients attending a clinic might not apply to members of the general community who suffer from MDD of the same type. Similarly, we did not find evidence to confirm or refute whether treatments are more or less efficacious for various subgroups: patients characterized by sex, race, or ethnicity or individuals with coexisting psychiatric conditions. The samples in many trials had some subjects with the aforementioned subgroup characteristics, even if the main focus was on a different population. For instance, the trials may have included individuals with a history of psychiatric comorbidities but did not report whether interventions were similarly efficacious (or not) for such individuals. Finally, many trials, particularly for CAM interventions, were conducted outside the United States. Whether and how differences in ethnic or cultural backgrounds and health systems affect the applicability of results to U.S. populations remains uninvestigated and unanswered. For example, most of the acupuncture trials were conducted in China, where acupuncture is commonplace, and the effects of acupuncture treatment expectancy may differ substantially between such populations and Western populations.¹⁶³

With few exceptions, interventions in included trials were in line with clinical practice. Except for many CAM trials in which patients received SGA dosages at the lower end of the recommended range, prescribing patterns and doses in the SGA arms of our evidence base were consistent with clinical practice. Some newer SGAs such as desvenlafaxine, levomilnacipran, vilazodone, or vortioxetine, however, have never been compared with psychological or CAM treatments or exercise. Nevertheless, reliable evidence indicates that the comparative effectiveness of SGAs is similar.³³ Consequently, we believe that our findings are applicable across the class of SGAs.

As noted previously, detecting no statistically significant difference does not necessarily mean that the treatment options are equivalent. The studies involved were designed to test whether an outcome for one intervention was different from the outcome for another rather than to test equivalence, which would generally require a larger sample size. This point is especially relevant for those findings with a low SOE. Further, while comparative effectiveness at a group level did not show a difference between SGA and CBT, how best to tailor this information to an individual patient is still not clear. Indeed, other potentially relevant indicators (e.g., depressive severity, comorbid psychiatric illness) may favor one over another, but the current evidence base (as indicated in the KQ 1b and 2b findings) is quite limited.

The number and length of sessions of the various psychological interventions were generally consistent with clinical practice and likely represent an adequate course of treatment. As is generally the case when comparing the effectiveness of psychological treatments with other psychological interventions or other types of treatment, heterogeneity of the content and delivery of the identified intervention is problematic. Many of the psychological interventions in our evidence base provided broad descriptions of the type of intervention; others used a manualized protocol. Both of the included studies that used CBT or CT followed a manualized protocol.

Further, variability among the trials was high with respect to the degree to which treatment fidelity was assessed and adhered to. Type, training, and experience of the providers of the various interventions were also quite heterogeneous. Although clinician characteristics may be less problematic than the content of the intervention for understanding comparative benefits or harms, unlike the case with SGAs that are broadly equivalent and have standardized dosing, the cumulative effect of the various sources of heterogeneity within and across psychological interventions may limit the applicability of our findings. Clinician type, training, experience and degree of treatment fidelity are likely to be even less in routine clinical practice than in the studies included in this review. Along with psychotherapist availability, these are important factors for clinicians to consider when recommending psychological treatment and interventions.

For acupuncture, treatment protocols were so varied as to preclude definitive conclusions about any single acupuncture intervention. For these reasons, we find it difficult to recommend any single type of acupuncture, or acupuncture more generally, as a substitute for treatment with antidepressant medications.

For St. John’s wort, use of standardized extracts may be broadly applicable with certain caveats. Although several different St. John’s wort preparations were represented among the trials we found, many of the trials used St. John’s wort doses that were consistent with current recommendations (i.e., 900 mg daily, standardized to 0.1 percent to 0.3 percent hypericin).^43,44,164 Furthermore, high quality, standardized St. John’s wort extracts are now commonly available.

An important concern about the use of St. John’s wort is its potential to interact with other medications. St. John’s wort is well known to cause substantial changes in plasma concentrations of drugs metabolized by cytochrome P450 3A4, which includes SSRIs, tricyclic antidepressants, and many drugs used to treat common conditions such as heart disease, hypertension, hypercholesterolemia, HIV, and many cancers.^61,165 Therefore, St. John’s wort should not be recommended to patients taking any pharmaceutical medications without the advice of a medical provider or pharmacist with expertise in evaluating herb-drug interactions.

Doses in the exercise arms were within the dose range suggested for exercise programs for middle-age to older adults. For example, the guideline for depression from the National Institute of Health and Care Excellence recommends structured, supervised exercise three times per week.³⁶ However, the small numbers of trials that have examined dose-response of exercise for depression indicate that higher intensity and frequency of exercise may be more helpful in alleviating depression.⁴⁹ Although the two Blumenthal et al. trials, reported reasonable compliance rates for both the SGA and exercise groups, in clinical practice, particularly when exercise is prescribed in less structured formats, depressed patients may well have more difficulty in initiating exercise regimens or staying motivated to exercise, because depression is known to be associated with lower levels of physical activity.¹⁶⁶ Although our report had insufficient data to determine whether depression enhances quality of life, we did find that aerobic capacity increased significantly more among the exercise group. Because both of these trials targeted middle-age and older adults, the results cannot be generalized to younger age groups. Additionally, their generalizability to a typical primary care or psychiatric population of depressed adults is unclear because these trials only included patients free of medical comorbidities that would restrict their ability to follow the prescribed exercise regimens.

Most trials did not assess quality of life or functional capacity as outcomes. Conceivably, response to treatment and remission does also improve quality of life and functional capacity.

The lack of assessment of harms in many trials poses a serious threat to the applicability of findings to typical clinical settings or patient populations. The comparative balance of benefits and harms among treatment options is impossible to determine when harms are not assessed and reported reliably. In clinical trials of SGAs with close adverse events surveillance, up to 60 percent of patients experienced adverse events.³³ For some patients, these adverse effects were tolerable; for others, they led to discontinuation of treatment. In the body of evidence for this report, neither harms for SGAs nor harms for nonpharmacological treatments were assessed adequately. In particular, when studies comparing SGAs with psychotherapy alone or in combination with SGAs reported the occurrence of harms, they only infrequently considered potential harms that can specifically stem from psychotherapy (e.g., increased conflict with partners or negative consequences resulting from behavioral therapy). This limitation of our evidence base reflects a general lack of information about how to classify and measure potential harms that can result from psychotherapy.¹⁶⁷ For these reasons, we could not draw any conclusions about applicability.

Implications for Clinical and Policy Decisionmaking

Our systematic review of head-to-head trials detected no statistically significant differences in effectiveness between an SGA medication or CBT in treating MDD. These findings suggest that either approach can serve as a reasonable starting place for treatment of MDD. We caution, however, that it remains unknown whether the severity of depression should influence decisions about the initial treatment strategy.

Health care reform around the world reflects a trend toward integrative care as a remedy for the current, fragmented delivery of health and social services common in many health care systems. Given that both SGAs and psychotherapies can have equal merit in treating MDD, locating clinicians who render mental health care in primary care settings needs to be part of this trend. Doing so would likely increase patient access to psychiatric consultation and therapy, and it would enhance coordination of care between primary care clinicians and mental health professionals. Further, we know that approximately 20 percent of patients do not fill their prescriptions for antidepressant medication; even if they start a course of treatment, they may discontinue early before receiving an adequate course.¹⁶⁸ Having access to nonpharmacological interventions in the primary care setting might enhance treatment adherence and improve treatment outcomes for patients with MDD. It may also have additional downstream effects in reducing the stigma associated with mental illness in general, empowering patients to address the symptoms and issues associated with not only depression but also other mental health–related concerns, and encouraging them to seek and maintain treatment more quickly at an earlier stage of their illness.

Related to this, access to psychotherapy should not be financially prohibitive. Some insurance plans in the United States charge different rates for psychotherapy and other mental health services than they do for generalized medical care. Decision- and policymakers need to make sure that fees associated with accessing these interventions do not make them unaffordable for patients that need and would benefit from these services the most.

Similarly, one great difficulty for CAM therapies, for both patients and providers, is how to pay for them. For most patients, their insurers do not cover CAM services. This lack of coverage is particularly vexing for patients and providers, especially when the weight of evidence addresses the efficacy of CAM treatment compared with placebo. In many of these instances, patients need to pay for these treatments out of pocket, which creates disparities in care by limiting access to proven treatments for patients who cannot manage those out-of-pocket costs.¹⁶⁹

Although SOE is low, findings regarding the lack of statistically significant differences in effectiveness of SGAs and exercise, combined with the low adverse effects generally found in exercise trials, can provide clinicians with some indication as to how to guide their patients in clinical practice. In terms of clinical decisionmaking, the information in this review can be helpful to physicians because they can provide a summary of the available evidence base indicating the advantages and disadvantages of these options, and patients can identify which intervention they would prefer. Some options, such as medication and St. John’s wort, would require close physician supervision and monitoring given potential side effects and drug interactions. Moreover, those patients who would like to maintain or start an exercise regimen in addition to undergoing SGA therapy can be encouraged to do so. The enhanced potential for increasing physical well-being as well as expanding social interactions may be an added incentive to encourage an exercise regimen.

Limitations of the Comparative Effectiveness Review Process

To find relevant studies, we employed an intensive search process in multiple electronic databases; we also conducted searches for grey literature. Because of time and monetary limitations, however, we limited eligible studies to those published in English, German, and Italian. Methods research indicates that such an approach can introduce language bias; in general, however, it may also lead to overestimates of the effectiveness of interventions.

For KQ 2, we extended eligibility criteria after we realized that we would not find sufficient evidence to answer this KQ. Despite re-reviewing more than 6,000 abstracts, we could still not find reliable evidence to address the question about the best treatment option for patients who did not achieve remission during an initial treatment trial.

For harms, studies conducted in other patient populations (e.g., those with subthreshold depression or dysthymia) might have yielded useful information. Many studies using psychological or CAM therapies included populations suffering from any form of depression, not just MDD. In addition, studies with placebo or waiting list control groups could have provided important information about adverse effects of interventions. We lacked the resources to explore such a broad evidence base just to assess harms.

Because we dealt with study-level data, we could not reliably assess the impact of severity of MDD on the comparative benefits and harms of interventions. Such a question would best be addressed with individual patient data from trials and individual patient data meta-analyses.

If information in full-text articles was unclear or missing, we attempted to contact authors for clarification. The yield of this effort, however, was small. Despite multiple attempts to contact authors, few replied or were able to provide missing information.

Finally, publication bias and selective outcome reporting are potential limitations. Although we searched for grey and unpublished literature, the extent and impact of publication and reporting bias in this body of evidence is impossible to determine.

Limitations of the Evidence Base

Overall, several major limitations characterize this body of evidence. First, no reliable evidence was available assessing the effectiveness or risk of harms of many of our eligible interventions. Particularly for KQ 2 on populations who did not achieve response to an initial treatment attempt, we found no eligible switch trials directly comparing SGAs with CAM or exercise; neither did we find any eligible augmentation trials comparing SGAs with CAM or exercise. We also found no direct comparisons of switching strategies versus augmentation strategies. Likewise, the role of depressive severity as a moderator of the comparative effectiveness of both first- and second-step therapies has received very little planned, prospective study.

Second, even when evidence was available, the small number of trials and the small sample sizes posed considerable limitations. Much of the evidence base directly comparing treatments was powered to test whether one treatment was superior to the other. Failure to find such a difference is not equal to concluding that the interventions are equivalent. In addition, for some trials we had concerns about adequate dosing of SGAs. For example, three of eight trials compared St. John’s wort to either fluoxetine 20 mg or sertraline 50 mg, the lowest recommended doses of these drugs. Considering that mean baseline depressive severity for most trials fell in the severe range (HAM-D scores 19 to 23), patients in the SGA arms were undertreated. The extent to which this affects the comparative benefits between SGAs and St. John’s wort remains unclear.

Third, available evidence was frequently fraught with methodological shortcomings. Of the 44 trials meeting our eligibility criteria, we rated 16 as high risk of bias and only 4 as low risk of bias. Trials assessed as high risk of bias have significant flaws of various types (e.g., stemming from serious errors in design, conduct, or analysis) that may invalidate their results. Consequently, the evidence base for most critical outcomes was insufficient to draw conclusions. The SOE could be rated as low or moderate for only a few outcomes; the latter indicates reasonable confidence in the effect estimates from those trials.

Fourth, even when trials assessing the comparative effectiveness of interventions were available, they often did not assess harms or did not assess harms adequately. Of the 44 included trials, only one trial used an objective scale to assess harms. Most trials combined spontaneous 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 trial durations and small sample sizes also limited the validity of adverse event assessment in many trials. No trials were designed to assess specific adverse events as primary outcomes.

Fifth, of the limited body of evidence, most trials were explanatory (i.e., designed to show if a treatment could work in ideal circumstances rather than in everyday practice) rather than pragmatic trials and provided information for the acute phase of treatment. These factors may well compromise the applicability of findings and do not inform management in the continuation or maintenance phases of treatment.

Sixth, few studies explored the role of treatment expectancy on outcomes. In a notable exception, an independent group of researchers reanalyzed the U.S. Hypericum Depression Trial.⁹² In this three-arm study comparing sertraline, St. John’s wort, and placebo, they concluded participant beliefs regarding treatment assignment were more strongly associated with clinical outcome than the actual treatment received,¹⁷⁰ a finding echoed in other studies of MDD.^171,172 Expectancy may play a larger role for CAM intervention studies conducted in countries where the treatment is commonly accepted, such as acupuncture in China or St. John’s wort in Germany. Finally, it was not always clear how the diagnosis of MDD was ascertained in individual studies. Some studies used structured interviews based on DSM criteria, but others did not report the method of ascertainment. Finally, it was not always clear how the diagnosis of MDD was ascertained in individual studies. Some studies used structured interviews based on DSM criteria, but others did not report the method of ascertainment.

Research Gaps

Across all comparisons of interventions, major research gaps pertain to information about patient-centered outcomes, such as functional capacity and quality of life, and the comparative risk of harms. For patients and clinicians, balancing benefits and harms based on objective information is crucial. Lack of information about harms can lead to a biased knowledge base and the potential for decisions that cause more harm than good. Findings from the STAR*D study suggest that factors other than depression severity (e.g., comorbid medical disorders, employment status)¹⁷³ contribute significantly to the health-related quality of life of outpatients with MDD. A comprehensive assessment of quality of life outcomes is, therefore, paramount for informed decisions about treatment options.

We found no eligible studies that compared SGAs with behavior therapy or behavior modification, humanistic therapies, yoga, or mindfulness interventions. Given the wide use of these types of psychotherapies in clinical practice, further research into their comparative effectiveness with SGAs in treating MDD patients is desirable. For many psychotherapies and CAM therapies that have been evaluated against an SGA, the data were insufficient because trials did not report important outcomes, most notably quality of life and functional capacity. Future studies should assess remission, response to treatment, quality of life, and functional capacity using standardized measures to allow for more direct comparisons across studies using the same or similar SGAs and psychological interventions.

These same deficiencies in the literature extend to the comparative effectiveness of SGAs and both psychological and CAM interventions for treating MDD as a function of depression severity. Only a single trial evaluating SAMe and no trials assessing psychological interventions or other CAM therapies were designed to address the question of whether depression severity affects the comparative effectiveness of SGAs as compared with these interventions.

Research comparing an SGA with exercise, either alone or in combination with an SGA, is also limited. We found only two trials comparing SGAs with exercise that met our criteria, and these both used aerobic exercise, in which individuals were assigned continuous walking or jogging that would maintain heart rate from 70 to 85 percent of their heart rate reserve. Missing from the literature were any studies meeting our criteria using other forms of exercise (e.g., strength training or mindful exercises such as yoga, tai chi, or qigong). Moreover, we found no studies in which an SGA was systematically compared with differing intensities and frequencies of exercise (this research could be helpful, because there is indication from non-SGA studies of better treatment outcomes with high-dose versus low-dose exercise regimens).¹⁷⁴ Changes in aerobic capacity were reported in both our included trials; more trials, however, should include standardized measures of quality of life and functional capacity. Having such data might then enable reviewers to compare results across trials. Trials that include a wider age range of participants would also be helpful in determining whether different types and intensities of exercise are more effective for patients of different ages; preferences and usefulness of various types and intensities of exercises may differ by sex or ethnic or cultural variables. Research should also investigate how baseline depression severity affects patient preferences, adherence, and outcomes of prescribing an SGA versus exercise or exercise-SGA combination.

One primary challenge for studies of CAM therapies is defining the proper dose of the therapy being tested. Although experts tend to agree about dosing of St. John’s wort,^43,44,164 only scant evidence informs dosing regimens of SAMe, and dosing practices for omega-3 fatty acids differ widely. Future studies of natural products should be based on dosing regimens that are supported by investigations of their pharmacokinetic and dose-response properties. Similar problems exist for acupuncture dosing, but this particular issue is even more complex because of the heterogeneity of point selection, needle stimulation, session duration, and number of treatments for acupuncture interventions.

The limited amount of comparative intervention data addressing whether depressive severity moderates outcomes provides little guidance on how selection of treatment strategies might differ based on whether a depression is on the milder end of the spectrum compared with the more severe end. This question, raised by a number of systematic reviews,^22–24 remains without a clear answer.

Finally, beyond the two articles identified comparing switch and augmentation strategies employing a limited number of medication options or CT, the absence of relevant comparative data about which treatment options are most effective for those needing second-step treatment (about 70 percent of patients with MDD)^26,27 was striking. Further, no second-step therapy data at all exist that compare SGAs with CAM or exercise treatments. This void in the evidence base is a major one that will perplex and confound clinicians, patients, policymakers, and guideline-developers alike.

Conclusions

Available evidence indicates that SGAs and CBT do not differ significantly in effectiveness as first-step treatments for adult outpatients with mild to severe MDD. The SOE for this finding is moderate, which means that the body of evidence has some deficiencies, but we believe that the findings are likely to be stable as new studies emerge. Most comparisons of SGAs with other treatment options also did not show statistically significant differences. Exceptions, however, are omega-3 fatty acids that appear to have lower effectiveness than SGAs, the combination of SGAs with acupuncture, and the combination of SGAs with interpersonal psychotherapy, which appear to have greater effectiveness than SGA monotherapy. These findings, however, have to be interpreted cautiously because of methodological limitations. Our confidence in these results was low or evidence was simply insufficient. We believe that future studies will have a substantial impact on results. In addition, populations with MDD are known to have high response rates to placebos. For many comparisons that are limited to single trials, determining whether similar treatment effects between SGAs and other interventions are based on similar effectiveness or high placebo response rates is impossible.¹⁷⁵

Interventions other than SGAs usually have a lower risk for harms. One exception is St. John’s wort, which is known to interact with many important medications and should not be taken without the supervision of a provider experienced in its use. It should be noted, however, that treatment side effects appear to be fewer with St. John’s wort compared with SGAs. Some nonpharmacological interventions, however, require more personal engagement or costs than others, which could affect patient adherence.

The choice of the initial treatment of MDD should, therefore, consider patient preferences following a discussion of the advantages and disadvantages and the feasibility (e.g. costs, likely adherence) of each treatment option. Differences with respect to adverse events, personal engagement, and costs may be taken into consideration for the choice of a first-step treatment. Such shared and informed decisionmaking might enhance treatment adherence and improve treatment outcomes for patients with MDD, especially because treatment continuity is one of the main challenges in treating such patients.¹⁷⁶

For second-step therapies, although evidence is limited, no clear benefit emerges to suggest either switching to a particular SGA or to CT or augmenting with a particular medication or CT. Available data suggest that switching to another SGA, switching to CT, or augmenting with a particular medication or cognitive therapy are all reasonable options. The more important decision appears to be simply to try a different evidence-based approach.