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

Sun X, Patnode CD, Williams C, et al. Interventions to Improve Patient Adherence to Hepatitis C Treatment: Comparative Effectiveness [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2012 Dec. (Comparative Effectiveness Reviews, No. 91.)

Cover of Interventions to Improve Patient Adherence to Hepatitis C Treatment: Comparative Effectiveness

Interventions to Improve Patient Adherence to Hepatitis C Treatment: Comparative Effectiveness [Internet].

Show details


Literature Search

Our search of English-language publications yielded 1,629 citations. From this body of literature, we provisionally included 85 articles for full-text review based on abstracts and titles (Figure 2). After screening full text articles against our inclusion/exclusion criteria (Table 3), we excluded 73 for various reasons, such as having no relevant outcomes (k=26), including a population not undergoing combination therapy of pegIFN-α plus ribavirin (k=19), or not evaluating hepatitis C treatment adherence (k=12). While we also searched for non-English publications and identified 99 potentially relevant studies, evaluating the non-English studies was not within the scope of this review. The full list of English-language excluded studies (including reasons for exclusion) is provided in Appendix C.

This figure gives an overview of our literature searching methods. The details regarding the number of included articles and the reasons publications were excluded is provided.

Figure 2

Literature flow diagram.

Characteristics of Included Studies

Twelve studies61,64,65,9098 met the inclusion criteria for at least one of our Key Questions (Table 5). About half of these studies were RCTs of fair97 or poor quality.61,64,94,96,98 The remaining studies were cohort studies rated as good,91,93 fair,65,90 or poor quality92,95 (see Appendix D and E for individual study quality ratings for the RCT and cohort studies, respectively). Most of these studies were conducted in United States clinic-based settings, although two were conducted in hospital-based settings in Italy and two were multi-site studies conducted in France. Six primarily poor-quality studies had sample sizes less than 50,64,92,9598 while three primarily poor-to-fair-quality studies enrolled 100 to 250 patients.61,65,90,91,93 Only two studies measured patient-important health outcomes,61,64 while the remaining studies measured intermediate disease management outcomes (e.g., EVR, SVR) and/or treatment adherence.

Table 5. Included adherence interventions and comparisons.

Table 5

Included adherence interventions and comparisons.

We included studies that evaluated a variety of adherence approaches (Table 5), including one fair- and two poor-quality studies examining interventions targeting system-level factors,64,65,98 one fair-quality study targeting regimen- or therapy-related factors,90 two good- and two poor-quality studies addressing patient-level factors,9194 and three fair- and one poor-quality studies accessing the direct management of adverse events.61,9597 We did not include any studies that included interventions targeting policy- or provider-level factors. All of the trials, except one,96 compared an adherence intervention with usual care. This single trial was conducted by Morasco and colleagues96 comparing the use of citalopram to placebo in decreasing therapy-induced depression. None of the studies defined what “usual care” consisted of in the study’s respective setting. All of the included cohort studies compared the presence (or absence) of exposure to the specific intervention being investigated among study participants that were intended to otherwise be comparable. In all of these instances, the usual care condition represented a minimal standard of adequate medical care, and thus all studies are comparative effectiveness. Even though there were three-to-four studies comparing intervention approaches within one intervention category (e.g., system-level or adverse event management interventions), none of these within-category studies tested the same adherence interventions. Thus, the body of evidence is generally limited to single studies of different intervention types and is further limited by the noncomparability of enrolled study populations, as described next.

Study participants varied widely across studies in important ways that may impact the probability of treatment response (i.e., SVR) and/or affect treatment adherence, which were the main outcomes available from these studies (Table 6). Response to dual therapy (the only therapy examined in these adherence studies) is primarily affected by the genotype of the HCV infection and by previous treatment history (Table 1). Most studies included several HCV genotypes (with varying probabilities of response to dual therapy)64,65,92,9498 or did not report HCV genotypes.93 Three studies limited their study participants to a single genotype (e.g., genotype 1)61,90 or to genotypes 2 or 3, which are similarly responsive to treatment.91 Two of the larger studies targeted those naive to treatment, who are most likely to respond to treatment65,91 and many did not report this important participant characteristic.61,92,93,96,98 Other characteristics that may affect likelihood of treatment adherence were similarly variable across studies (Table 7).

Table 6. Study characteristics.

Table 6

Study characteristics.

Table 7. Additional study characteristics.

Table 7

Additional study characteristics.

Results of Included Studies

We discuss the results of the four different types of comparisons separately: system-level interventions compared withusual care (Table 8), regimen-related interventions compared with usual care (Table 9), patient-level interventions compared with usual care (Table 10), and adverse event management interventions compared with usual care or placebo (Table 11). Studies reported highly variable outcomes (Table 6). In addition, the definition each study used for adherence and the specific methods for measuring adherence varied. We did not include reports that clearly reflected discontinuation or dose reductions initiated by a physician. In terms of health outcomes, no studies reported morbidity, mortality, or HCV transmission. Only two studies61,64 reported quality-of-life outcomes. Additionally, only two studies reported harms related to the adherence intervention.61,96 We present the results of Key Questions 1 (intermediate and health outcomes) and 2 (adherence) together due to the paucity of data for all outcomes.

Table 8. Outcomes of system-level interventions.

Table 8

Outcomes of system-level interventions.

Table 9. Outcomes of regimen-related interventions.

Table 9

Outcomes of regimen-related interventions.

Table 10. Outcomes of patient-level interventions.

Table 10

Outcomes of patient-level interventions.

Table 11. Outcomes of adverse event management interventions.

Table 11

Outcomes of adverse event management interventions.

Key Question 1 (Intermediate and Health Outcomes) and Key Question 2 (Treatment Adherence)

Key Question 1. In adult patients with chronic HCV infection undergoing antiviral therapy, what is the comparative effectiveness of treatment adherence interventions in improving intermediate (e.g. sustained viral response, histological changes, drug resistance, relapse rates, and treatment side effects) and health outcomes (e.g., disease-specific morbidity, mortality, QOL, transmission of HCV)?

Key Question 2. What is the comparative effectiveness of treatment adherence interventions in improving treatment adherence (e.g., medication adherence, medical plan adherence)?

System-Level Interventions Versus Usual Care

Key Points

  • Three, small fair- or poor-quality studies compared the effectiveness of system-level HCV treatment adherence interventions versus usual care, and none of these reported on important health outcomes (e.g., morbidity, mortality, or the transmission of HCV). (Strength of evidence = insufficient)
  • One poor-quality trial evaluated how a system-level treatment adherence intervention affected health-related QOL. Hepatitis-specific limitations and distress improved over time in the intervention group, but not in the control group. Data wereinsufficient to draw conclusions, however, due to high risk of bias and no statistical test of group differences. (Strength of evidence = insufficient)
  • Three studies examined the effectiveness of system-level treatment adherence interventions compared with usual care on SVR, adherence, or both. System-level interventions had an imprecise impact on SVR. In two studies, more methadone-maintenance patients receiving DOT achieved SVR compared with controls, while fewer patients receiving care at a specialty pharmacy, achieved SVR than those receiving usual pharmacy care. However, no results were statistically significant. Findings were further limited by moderate-to-high study-level risk-of-bias and the fact that we could not compare interventions across studies. (Strength of evidence = insufficient)
  • One fair-quality cohort study reported no benefit of specialty pharmacy care compared to usual pharmacy care for patient self-discontinuation of treatment. (Strength of evidence = insufficient)

Two poor-quality RCT64,98 and one fair-quality retrospective cohort study65 evaluated the effectiveness of a system-level intervention on QOL, SVR, EVR, and/or adherence, compared to usual care (Table 8).

A fair-quality retrospective cohort study by Cohen and coauthors65 compared the effects of patients’ use of specialty care pharmacies with patients’ use of standard retail pharmacies on SVR and adherence. Data were collected from the medical charts at a single academic institution. Pharmacies self-designated as either a specialty or standard retail pharmacy. Patients were placed into the study arm according to where they filled their prescriptions. This study included 197 patients: 95 in the specialty pharmacy group and 102 in the standard pharmacy comparison group. Sixty-three percent of all patients were naive to prior HCV therapy and the majority of patients (63%) were genotype 1. While no significant differences existed between groups in terms of HCV genotype, there were significant differences in terms of ethnicity—more Caucasian than African American patients used standard retail pharmacies and more African American patients than Caucasians used specialty pharmacies. All but three patients (all in the standard pharmacy comparison group and receiving pegIFN monotherapy) received pegIFN plus ribavirin. This study’s major threats to validity include the reliability and validity of the designation of pharmacy type (pharmacy self-reported) and the fact that the analysis did not adjust for potential confounders in the analysis of adherence outcomes.

A poor-quality RCT by Bonkovsky and colleagues64 included 48 patients enrolled in methadone maintenance programs for at least 3 months prior to study inclusion. All patients agreed to abstain from illicit drugs and alcohol throughout the study period. Patients were excluded if they had any significant medical comorbidity or a history of severe psychiatric disease. Most participants (83% of the intervention group [IG] and 67% of the control group [CG]) were male and 79 percent of the combined sample was Caucasian. Sixty percent of enrolled patients were genotype 1 (54% in the IG, compared to 67% in the CG, statistical differences not presented) and all were naïve to HCV treatment. Genotype 1 patients were treated with pegIFN-α2a (alpha 2a) (180 μg/week) and ribavirin (1,000/1,200 mg/day) for 48 weeks. Patients with genotypes 2 and 3 were treated with pegIFN-α2a (180 μg/week) and ribavirin (800 mg/day) for 24 weeks. Patients were randomized to receive supervised (i.e., DOT) pegIFN-α2a at methadone clinics once weekly (n=24) compared to self-administration of pegIFN-α2a (n=24). Self-administration patients received their first injection at the study site and subsequent injections were self-administered. Ribavirin was self-administered in both groups. The majority of patients in both groups also received methadone on nearly all of the study treatment days. This study followed patients for 24 weeks after treatment. Seventy-seven percent of patients completed their full length of therapy and group differences were not statistically significant. In addition to completion rates, the study assessed SVR and health-related QOL using a validated, self-administered survey that measured physical, psychological, and general health, and hepatitis-specific QOL domains (i.e., the Hepatitis QOL Questionnaire). The hepatitis-specific domains measured limitations and health distress due to hepatitis C. This study included an unadjusted analysis on the intent-to-treat population to assess the effect of the DOT intervention compared to self-administration on SVR. The authors also report change in QOL measures over time. This study had several quality concerns, including the lack of reporting on the method of randomization and whether or not group assignment was concealed. In addition, patients were not blinded to their condition, which could have influenced self-reported responses, including QOL measures.

Another small, poor-quality RCT 98 similarly randomized patients receiving methadone maintenance to either modified DOT (n = 12) or self-administered therapy (n = 9). All patients initiating HCV therapy received pegIFN-α2a (180 μg/week) and weight-based ribavirin. Subjects randomized to the intervention group received their methadone as part of their HCV therapy whereas control group participants received their methadone elsewhere. The major risk of bias of this trial is the high attrition in the control group (data were unavailable in 5 out of 9 patients). Whereas all participants in the DOT group patients started HCV treatment, only 44.4 percent (4/9) of patients in the self-administered group started HCV therapy. Data on EVR and SVR from this study are reported as being preliminary; it appears that more patients will be enrolled in this study.

Quality of Life

Only one poor-quality RCT64 reported quality-of-life outcomes. There was an improvement in hepatitis-specific limitations mean score from baseline in the supervised DOT treatment group (84.2 at the end of followup vs. 74.5 at baseline), whereas these self-reported limitations became worse in the self-administered control group (mean score of 68.9 at followup vs. 76.8 at baseline). Similarly, the mean score on self-reported health distress was improved at followup in the intervention group from baseline (81.6 vs. 63.8). There was a very small change in the self-administered treatment group (67.3 vs. 69.8). The study did not report statistical tests of changes over time or of differences between groups.

Sustained Viral Response

All three studies reported the adherence intervention’s effect on SVR with imprecise, nondefinitive results. In the cohort study,65 48 percent (46/95) of patients using specialty pharmacies achieved a SVR, compared with 56 percent (56/102) of those using a standard retail pharmacy. This difference was not statistically significant in unadjusted or adjusted analysis that accounted for age, sex, ethnicity, genotype, and prior treatment (ORadj 0.69, 95% confidence interval [CI], 0.37 to 1.30). One poor-quality RCT64 reported a higher achievement of SVR in 54 percent (13/24) of patients enrolled in the supervised DOT treatment, compared with 33 percent (8/24) using self-administered treatment (unadjusted OR 2.36, 95% CI, 0.73 to 7.60). Among patients with genotype 1, the SVR rate did not differ between groups. However, among patients with genotypes 2 or 3, SVR was achieved in 91 percent (10/11) of patients in the DOT group as opposed to 25 percent (2/8) of patients in the self-administration group. The other RCT found that 6 out of 12 patients (50%) receiving modified DOT of pegIFN-α2a and ribavirin versus 1 of 9 patients (11%) randomized to the self-administration group achieved a SVR. Five patients in the control group did not initiate HCV treatment.98

Early Viral Response

Only one poor-quality RCT98 reported data on EVR. In this study, 10 out of 12 patients (83%) in the modified DOT group versus 3 out of 9 patients (33%) in the control group achieved early viral response.


Both RCTs reported no adherence data.64,98 In the cohort study65 ten patients included in the specialty pharmacy group self-discontinued treatment, compared with 4 in the control group (calculated OR 0.35, 95% CI, 0.11 to 1.15). Physician-directed reasons for discontinuation of therapy included nonresponse or breakthrough.

Regimen-Related Interventions Versus Usual Care

Key Points

  • No studies evaluated the effect of regimen-related interventions on health outcomes or the intermediate outcomes of SVR or EVR. (Strength of evidence = insufficient)
  • A single fair-quality cohort study that compared packaging to reduce pill burden for ribavirin (RibaPak) with regular ribavirin reported the intervention effects on adherence, which the study measured three ways (duration of treatment, proportion of prescribed doses taken, and proportion taking at least 80% of prescribed doses). This study reported improved adherence in the reduced-pill-burden intervention on all three measures at 24 weeks and on two of three measures at 12 weeks. (Strength of evidence = low)

One fair-quality prospective cohort study,90 addressed the effect of regimen-related interventions on adherence (Table 9). No other outcomes were reported for this study. The “Accurate Dosing in Hepatitis C: Examining the RibaPak Experience” or ADHERE study was a 24-week, United States study at 33 sites in adults with HCV. The primary aim of the study was to evaluate the treatment adherence of patients who were prescribed RibaPak, available in 400 mg and 600 mg ribavirin tablets (i.e., reduced pill burden), compared with patients prescribed 200 mg ribavirin tablets. All patients were concurrently receiving weekly pegIFN injections. Patients were identified by their treating physician. Five-hundred and three patients were enrolled at a ratio of 3:1 (RibaPak vs. regular ribavirin); all patients were genotype 1 (personal communication, I. Alam, May 30, 2012). Participants in both groups were similar in terms of age, gender, race, BMI, and baseline viral load. The analysis did not adjust for other potential confounders, however, such as previous HCV treatment, mental health status, or substance abuse history. Data were collected at 4 weeks, 12 weeks, and 24 weeks from the start of treatment with each followup time point specific to the 4 weeks prior to the assessment.


Adherence was assessed in three ways: (1) the proportion of patients remaining on treatment at each followup, (2) the proportion of prescribed doses taken among those remaining on treatment, and (3) the proportion of patients who took at least 80 percent of their prescribed dose. The proportion of prescribed doses taken was measured objectively based on pill counts at each visit. Left over pills were counted by site personnel and were compared with the number of pills that should have been left over based on the prescribed daily dose and the number of days in the treatment period.

A greater proportion of RibaPak patients than patients taking traditional ribavirin remained on treatment at both 12 weeks (86.4% compared with 77.7%, p = 0.01) and 24 weeks (71.4% compared with 62.4%, p = 0.045). There was no significant difference between the groups in the mean number of doses missed at 12 weeks. At 24-weeks, there was a statistically significantly greater mean number of missed doses among the traditional ribarivin patients (1.12 missed doses) than the RibaPak patients (0.36 missed doses) (p = 0.01). At both 12 and 24 weeks, patients using RibaPak were statistically significantly more likely to have taken at least 80 percent of their prescribed medication than those using traditional ribavirin (12 weeks: 94% vs. 84%, OR 2.28, 95% CI, 1.54 to 3.38; 24 weeks: 98% vs. 89%, OR 1.90, 95% CI, 1.30 to 2.78) (data reported in Table 9).

Patient-Level Interventions Versus Usual Care

Key Points

  • No patient-level adherence intervention studies reported health outcomes. (Strength of evidence = insufficient)
  • Three studies (one good-quality cohort, one poor-quality cohort, and one poor-quality RCT) comparing patient-level adherence interventions with usual care all tended toward increased proportions achieving SVR among patients receiving enhanced patient education and support, although no differences were statistically significant. (Strength of evidence = low)
  • Four studies (two good-quality cohort studies, one poor-quality RCT, and one poor-quality cohort study) comparing patient-level adherence interventions with usual care all tended toward better adherence at the end of treatment among patients receiving the adherence interventions. (Strength of evidence = moderate)

Three cohort studies (two good-quality,91,93 one poor-quality92) and one poor-quality RCT94 compared the effect of a patient-level intervention with usual care among adults with HCV on SVR and adherence (Table 10).

One good-quality prospective cohort study91 in France included 674 HCV patients infected with genotype 2 or 3. Patients undergoing HCV dual therapy with pegIFN-α2b and ribavirin were compared according to whether they received therapeutic education from a third party (health care professionals other than the prescribing physician) (n=370) or no therapeutic education (usual care) (n=304). Therapeutic education was provided at the discretion of the treating physician and included the distribution of education materials during individual sessions. Patients were considered to have adhered to pegIFN if they received three of four injections during the past 4 weeks, and to have adhered to ribavirin if they had taken at least 22 (200 mg) capsules over the past week. Patients were considered to have adhered to the full therapy if they had adhered to both drugs for at least 20 of the 24 weeks of treatment. Patients receiving therapeutic education had statistically significantly higher rates of depression, psychiatric disorder, drug use, and significant liver fibrosis. In order to account for these baseline differences, adjusted analyses were conducted to evaluate the association between exposure to therapeutic education and adherence and SVR. Twelve variables were used for the adjusted analyses, including sex, weight, BMI, educational level, history of depression, psychiatric disorders, alcohol consumption, drug abuse, duration of HCV infection, previous antiHCV treatment, HCV genotype, and pegIFN dose prescribed at treatment initiation.

One good-quality retrospective cohort study93 used propensity scoring methods to compare the “Be in Charge” (BIC) program, a patient-support program provided by the manufacturer of pegIFN-α2b (alpha 2b), with usual care. The BIC program was designed to improve patient adherence. Patients prescribed pegIFN-α2b plus ribavirin could join the program at any point during the course of their HCV therapy. Those enrolled in the program received personalized nursing support by telephone and/or mailed educational materials and motivational letters throughout therapy. The patients chose what level of intervention intensity they wished to receive, which ranged from 24 hour/7 days a week (24/7) access to a registered nurse to 24/7 access plus regular outbound telephone calls, motivational letters and other requested mailings. The study applied propensity scores based on observed covariates believed to be associated with the likelihood of enrolling in the BIC program such as age, sex, use of other HCV medications used in the 6 months prior to pegIFN initiation, and history of several comorbid conditions to match patients in the intervention group with those not enrolled in the intervention at a ratio of 1:1. A total of 1,560 patients (780 in each group) were included in the analyses. This study did not report data on HCV genotype or history of prior HCV treatment. Adherence data, which was defined as proportion of patients who filled all doses of the prescribed pegIFN-α based on pharmacy claims data, were collected at 12 weeks, 24 weeks, and 48 weeks. Of the 1,560 included patients, data at 48 weeks were available in only 666 patients (the main quality concern with this study). This study reported no other outcomes.

The poor-quality RCT94 took place in France. Two-hundred fifty patients were randomized to either therapeutic education by a nurse (n=123) or conventional clinical followup with the investigating physician (i.e., usual care) (n=121). The method of randomization was not reported including whether allocation was concealed. The intervention included regular consultation with a nurse who evaluated the patients’ understanding of the disease and side effects oftreatment and aimed to increase adherence. Nurse consultation took place in addition to medical consultation with the physician at the beginning of treatment and weeks 4, 8, 12, 24, and 36 (among those completing 48 weeks of treatment). Just over half (54.9%) were genotype 1-infected patients and the majority (59.8%) were treatment naïve. Groups were similar at baseline on a number of characteristics including age, sex, BMI, genotype, and treatment history. All patients received pegIFN-α2a (180 μg/week) and twice-daily ribavirin weight-based dosing (<75 kilograms [kg], 1000 mg/day; >75 kg, 1200 mg/day), for 24 or 48 weeks depending on genotype, viral load, and previous treatment. Reported outcomes included adherence to treatment and SVR. It was unclear whether the measurements, particularly around treatment completion and patient adherence were equal between groups or valid measures. In addition, patients were not blinded to their condition, a factor that could have influenced reporting. Analysis was conducted on an intent-to-treat analysis.

Finally, one poor-quality prospective cohort study,92 conducted in Italy, evaluated the “Together To Take Care” (TTTC) program, a multidisciplinary educational intervention in which patients who had a history of substance abuse received counseling on the risks of HCV infection and psychological support to help them modify their behavior. A case manager was assigned to each patient to coordinate treatment and counseling regarding the disease itself, addiction, and mental health. This study included a total of 48 patients: 16 patients in addiction therapy who received the TTTC intervention and 32 control group patients also in addiction therapy, who were consecutively pair matched 2:1 for age, sex, and time of HCV infection at enrollment. Though control patients were pair matched, patients in the TTTC intervention group were generally older at the time of infection than control participants (32.5 years compared to 28 years, respectively), although the authors report this as “presumptive”). Baseline data for both groups are presented at the individual patient level, making it difficult to make direct comparisons. It appears that the majority of patients in the intervention group were genotype 1 or 3, while several of the participants in the control group were reported to have “nondetermined” genotypes. It is not clear what proportion of patients in each group had received prior treatment for HCV. Control group patients were being treated with dual therapy for HCV at other health centers, but receiving treatment for drug addiction at the same center as the intervention group patients. While control group patients received care at the same drug addiction center, they did not receive the same “progressive and constant monitoring” that treatment group patients did via their case manager; however, there is some risk that control group patients may have also received enhanced education or psychological support from operators at the drug addiction center.

Sustained Viral Response

Three studies91,92,94 reported data on SVR. All three of these studies consistently showed that patients enrolled in interventions targeted patient-level factors (e.g., therapeutic education) achieved a higher level of SVR than usual care. The difference was statistically significant in the poor-quality RCT evaluating a nurse-led therapeutic education intervention compared to usual care (38.2% vs. 24.8%; unadjusted OR 1.88, 95% CI, 1.08 to 3.25),94 but not in the prospective observational study of therapeutic education (77% vs. 70%; ORadj 1.54, 95% CI, 0.99 to 2.40),91 or the multidisciplinary patient-support program (68.7% vs. 45.8%; OR 2.6, 95% CI, 0.69 to 9.81).92

Early Viral Response

Of the four studies included in this group, only the RCT reported data on EVR. This study reported that patients enrolled in the nurse education intervention were more likely to achieve EVR (72.8% vs. 57.6%; p < 0.01).94


All four studies reported data on adherence. Two studies reported data at 12 weeks, 24 weeks, and 48 weeks.93,94 All studies consistently showed that patient-level interventions improved adherence, despite variability in study designs, study quality, adherence definitions, and analytical techniques (Table 10). Patients in the intervention groups generally had approximately 50-percent higher odds of adhering to therapy or continuing with treatment at 24–48 weeks compared with control groups. One poor-quality study92 showed a statistically significant OR of 4.38 when comparing the intervention group with the usual care. Although the level of adherence decreased over time in all studies, data from studies reporting multiple time-points of followup suggested that the effect size (or difference between patient-level adherence interventions compared to usual care) tended to increase over time (e.g., 48 weeks vs. 24 weeks). In the good-quality prospective cohort study by Cacoub and colleagues, for example,91 66 percent of patients receiving therapeutic education were adherent to both drugs at 12 weeks, compared to 63 percent of patients in the control group. This difference was nonsignificant. At 24 weeks, however, the difference was statistically significant: 61 percent of the exposed were considered adherent to both drugs, compared to 47 percent of the nonexposed group (ORadj 1.58, 95% CI, 1.02 to 2.46). In the Hussein study,93 the proportion of patients who refilled the maximum number of pegIFN-α2b decreased from 72 percent in the intervention group at 12 weeks to 22 percent at 48 weeks. This proportion fell from 64 percent in the control group at 12 weeks to13 percent at 48 weeks. The odds of having refilled their injections among BIC enrollees was 1.77 (95% CI, 1.20 to 2.62) at 48 weeks compared to controls.

Adverse Event Management Interventions Versus Usual Care/Placebo

Key Points

  • There were no studies of the effects of adverse event management interventions on health outcomes besides QOL. (Strength of evidence = insufficient)
  • One small, fair-quality RCT found greater improvements in QOL (as measured by increased energy and activity) in dual-therapy–treated, genotype 1 HCV patients with anemia who received epoetin, an agent to reduce anemia, compared with those whose anemia was managed by a reduction in ribavirin. Patients receiving epoetin showed a significant increase in hemoglobin serum levels over the course of treatment whereas those just receiving a reduction in ribavirin did not. Improvement in SVR was also reported in the epoetin-treated group, compared with the ribavirin reduction group. (Strength of evidence = insufficient)
  • Two studies of depression prevention (citalopram, an antidepressant) or management (antidepressants for documented symptoms) to improve adherence in dual-therapy-treated HCV patients did not provide clear evidence about the effect on SVR due to reporting or risk-of-bias limitations. The study of prophylactic citalopram found greater EVR at 12 weeks, particularly in genotype-1 patients. (Strength of evidence = insufficient)
  • One study comparing prophylactic citalopram with placebo and one study comparing cognitive behavioral therapy (CBT) with usual care showed no statistical difference between groups in terms of treatment completion or adherence. The CBT intervention participants were less likely to be adherent to their pegIFN-α therapy than control participants, although the difference was not significant. (Strength of evidence = insufficient)

Three small, fair- and poor-quality RCTs61,96,97 and one poor-quality retrospective cohort study95 assessed the effect of interventions to prevent or manage adverse events (e.g., anemia, depression) related to HCV treatment on health outcomes (i.e., QOL) or intermediate outcomes (i.e., SVR, EVR, and/or adherence) (Table 11).

The first, a fair-quality RCT,97 randomized 29 HCV-treatment-naive patients with multiple genotypes to receive either eight 50-minute individual sessions of CBT in addition to standard HCV dual therapy or usual care. All patients were enrolled in methadone maintenance treatment program for at least six months. The authors report that the sample was recruited from an urban hospital-based primary care clinic among those seeking antiviral treatment. Of the original 117 patients deemed to be provisionally eligible, 88 were excluded (primarily for antidepressant use). No statistically significant differences existed between groups at baseline for a number of demographic variables, illicit drug use, and depression scores. The distribution of patients according to genotype did not appear to differ significantly (e.g., seven patients were genotype 1 in the CBT group, compared to 5 in the control group). The CBT included training in skills for depression management, such as mood monitoring, pleasant activities, constructive thinking, social skills, and assertiveness. This trial also included specific counseling regarding the unique needs of patients on antiviral medication for HCV, such as regular mood ratings to track depressive symptoms and addressing strategies for coping with drug cravings. This trial excluded patients taking antidepressant medication. Adherence was defined as receiving at least 24 pegIFN injections over 24 or 48 weeks of treatment. Data were abstracted from medical charts. Five (18%) patients were lost to followup in this study. The analysis of adherence was conducted on an intent-to-treat basis.

In the second, a poor-quality RCT,61 134 HCV-infected, genotype-1 patients treated with dual therapy who were experiencing a therapy-induced reduction in hemoglobin (Hb) levels (i.e., anemia) were randomized to receive epoetin alpha (epoetin) (group 1, n = 67) or to receive a reduction of ribavirin (800–1,000 mg/day) (group 2, n = 67) for 48 weeks. In this study, 214 patients were enrolled and started HCV dual therapy with the standard doses of subcutaneous pegIFN-α2a plus weight-based doses of ribavirin. This study only randomized patients who experienced a Hb reduction of greater than 2 g/dL at week 12 to the two groups. At week 12, no significant statistical difference was found between the groups concerning total bilirubin, platelets count, Hb, ferritin, and albumin serum levels. This study presented no other baseline comparisons by group. The study analyzed data on SVR 6 months after the end of treatment (week 72) based on an intent-to-treat basis. No patient-related adherence data were reported. QOL was assessed using the Linear Analogue Self-Assessment (LASA) scale at baseline and 36 weeks. This was an open-label trial with no blinding of patients or providers. While the authors state that randomization was performed using a computer program, it is unclear if the method was valid and whether or not the allocation was concealed. As previously stated, this randomization occurred after the assessment of EVR.

The third, a poor-quality RCT,96 evaluated the efficacy of taking citalopram in preventing the development of pegIFN-α-induced depression and improving treatment completion among HCV patients. Thirty-nine patients with HCV genotypes 1, 2, or 3 were randomized to receive prophylactic citalopram (20-mg tablets) (n = 19) or placebo pills (n = 20), which were dispensed to participants blindly. Participants who experienced increasing depression scores (according to the Beck Depression Inventory-II) were given a dose increase of 20 mg/day of citalopram or up to three additional placebo tablets. Participants with moderate-to-severe depression or suicidal thoughts were placed into a rescue arm of the study. While this study reported that participants were excluded if they had ongoing depression or active psychotic symptoms during the prior 3 months or current antidepressant use, a mean score for baseline current depression severity (indicating current depression) was presented for the full sample. No significant baseline differences existed between groups on any demographic or medical-related variables. After 24 weeks, blinding for treatment assignment was broken and all patients who continued therapy for 48 weeks (genotype 1) were offered citalopram for the duration of their treatment. While this study was originally powered to detect significant differences among groups for the development of pegIFN-induced depression, small sample sizes and the low rate of depression among both groups limited the ability to detect differences between groups. This study defined adherence as the completion of the recommended course of treatment. Unadjusted analysis was conducted to assess the association of citalopram use with adherence and SVR.

The poor-quality retrospective cohort study95 examined the effect of the use of antidepressants among those experiencing or not experiencing depressive symptoms during HCV therapy. Patients were categorized as having depression if there was at least one mention of depressive symptoms in their medical chart during the course of their HCV therapy, regardless of any followup treatment. This study compared four treatment groups: (1) no depressive symptoms experienced; (2) depressive symptoms experienced, but no antidepressant treatment received; (3) pre-existing and/or prophylactic antidepressant use before therapy; and (4) on-demand therapy for depressive symptoms. For the purpose of our review, we only compared two relevant strategies—on-demand psychiatric therapy (group 4, n = 25) compared with no antidepressant treatment in the presence of depressive symptoms (group 2, n = 17). This study made no comparisons by group according to important patient characteristics and none of the analyses were adjusted for potential confounders, which presents a major risk of bias in this study. In addition, although the percent of patients who completed treatment was presented by group, this outcome reflected physician-directed discontinuations in treatment not patient-directed lack of adherence.

Quality of Life

One study61 applied the LASA scale to assess the change in QOL from baseline in patients using epoetin compared with those receiving a reduction in ribavirin. The LASA scale includes scores for energy-and activity-related QOL. At 36 weeks, improvements were apparent in both scores from baseline in group 1 patients using epoetin (energy score change, 18 ± 17.3; activity score change, 20 ± 18.5) and in group 2 patients (with weight-based reduction in ribavirin (energy score change, 12.2 ± 21.6; activity score change, 7 ± 18.7). These changes were statistically significantly larger in the epoetin group (p < 0.05 for energy score, and p < 0.01 for activity score) than the ribavirin-reduction comparison group (Appendix F).

Sustained Viral Response

Three studies61,95,96 reported SVR. Of these, one RCT96 did not report sufficient data to allow calculation of effect estimates. In the comparative effectiveness trial that compared epoetin with a reduction of ribavirin dosing, patients on epoetin were statistically significantly more likely to achieve SVR (59.7% vs. 34.4%; OR 2.83, 95% CI, 1.40 to 5.72).61 While the use of antidepressants appeared to reduce SVR compared with usual care (36% vs. 53%; OR 0.5, 95% CI, 0.14 to 1.75),95 this result was based on a poor retrospective study.

Early Viral Response

One study96 reported EVR for genotype 1 and genotypes 2/3. In both patient genotype cohorts, a higher proportion of patients on citalopram than patients receiving a placebo achieved EVR (75% vs. 44.4% in genotype 1; 85.7% vs. 81.8% in genotypes 2/3). These differences, however, were not statistically significant.


Two studies96,97 reported adherence outcomes. In study by Morasco and colleagues,96 84.2 percent of patients receiving citalopram completed their recommended course of treatment, compared with 75.0 percent of patients receiving placebo, although this difference was not statistically significant (OR 2.13, 95% CI, 0.34 to 13.24). The reasons patients did not finish recommended treatment did not differ between the two groups and included medical factors (n = 3) and noncompliance (n = 1). In the RCT by Ramsey and colleagues,97 50 percent of the CBT-intervention group were considered to be adherent (i.e., received at least 24 pegIFN-α injections over the course of their therapy), compared with 80 percent of the control group. Again, this was not a statistically significant difference (ORadj, 0.19 95% CI, 0.03 to 1.15).

Does the Comparative Effectiveness of Treatment Adherence Interventions Differ by Patient Subgroups?

None of the included studies assessed whether the comparative effectiveness of adherence interventions on adherence differed by patient subgroups.

Key Question 3. Harms

Key Question 3. What are the harms associated with hepatitis C antiviral treatment adherence interventions?

Only two poor-quality RCTs61,96 reported information on harms related to an adherence intervention. Both studies evaluated the use of medications (i.e., epoetin and citalopram) to prevent or manage the side effects related to antiviral treatment. Although neither study found adverse events associated with the use of epoetin or citalopram, both studies were quite small and short-term. In addition, the relatively small trial (n = 29) comparing the effect of CBT with usual care found that more participants in the usual care control group than in the intervention group received at least 24 pegIFN-α injections at 24 weeks (i.e., were considered adherent). This effect was also not statistically significant.

PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.0M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...