Adherence to Hepatitis C Treatment Interventions: A Comparative Effectiveness Review

Adherence to Hepatitis C Treatment Interventions: A Comparative Effectiveness Review
.
July 11 2012
.
This report is based on research conducted by the Evidence-based Practice Center (EPC) under contract to the Agency for Healthcare Research and Quality (AHRQ), Rockville, MD. The findings and conclusions in this document are those of the author(s), who are responsible for its content, and do not necessarily represent the views of AHRQ. No statement in this report should be construed as an official position of AHRQ or of the U.S. Department of Health and Human Services.

The information in this report is intended to help clinicians, employers, policymakers, and others make informed decisions about the provision of health care services. This report is intended as a reference and not as a substitute for clinical judgment.

This report may be used, in whole or in part, as the basis for the development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products or actions may not be stated or implied.

This document is in the public domain and may be used and reprinted without special permission. Citation of the source is appreciated.
.
Summary and Discussion ONLY
Overview of Main Findings
.
Download Full Text-PDF
.
We identified 11 studies—including five RCTs and six cohort studies—that addressed the comparative effectiveness of adherence interventions on health outcomes, surrogate markers, and patient adherence in hepatitis-C patients treated with the standard dual combination viral therapy. This existing body of literature, however, had substantial methodological and clinical heterogeneity.

The five included RCTs were rated as primarily poor quality and all included small sample sizes (29-250). While two good-quality cohort studies88,90 included a relatively large number of patients (674 and 1560, respectively) and reported effect estimates that adjusted for the influence of potential risk factors, the remaining cohort studies had serious methodological limitations and generally had small sample sizes. We also found important variations in patient populations in all of the included studies, such as including patients with differing genotypes, history of substance abuse, and history of antiviral treatment. These factors may represent potentially important risk factors for treatment response and/or adherence (see Table 1). Patient populations also differed in racial and ethnicity distribution, as well as patient comorbidities.

How these studies evaluated adherence interventions was another source of heterogeneity. While grouped into four general categories, studies within a single category often investigated interventions that differed in their components and intensity. Interventions for managing adverse effects, for example, included medications addressing different conditions (e.g. epoetin for preventing anemia vs. antidepressants for depression), the use of antidepressants to prevent or to manage depression once it occurred, and cognitive behavioral therapy to prevent depression. Similarly, the two system-level interventions had very different approaches. One intervention evaluated the effect of specialty compared with standard pharmacy services and the other evaluated direct observation treatments on QOL or intermediate outcomes. The most consistent grouping was among the four patient-level interventions that enhanced patient education and/or support in order to improve adherence. Despite this, we were not able to identify the most successful intervention components given the lack of detailed descriptions, differences in intervention providers (e.g., nurses vs. physicians vs. psychologists), and approaches in the various interventions.

The included studies rarely reported health outcomes, which hampered our ability to directly interpret the evidence. Even among intermediate outcomes, we were unable to pool these outcomes due to differing definitions and measurement methods for adherence. Although the completion of HCV treatment is a commonly used definition, studies used different thresholds for defining treatment completion. We encountered additional issues to cross-trial comparisons for these studies, including studies that may target the completion of different antiviral agents (i.e., ribavirin vs. pegIFN-α, vs. both) or fail to clarify which antiviral agents they measured. The methods of measuring adherence included self-reported questionnaire, one-on-one interviews, pill counts, treatment administration records, or chart reviews. Several studies did not report this information. While SVR was commonly reported, this outcome was generally not comparable across studies due to diverse patient populations (with different likelihood of responding to treatment) across the body of evidence. 44

Outcomes of adherence interventions
There is a paucity of evidence assessing the effect of adherence interventions on health outcomes, particularly hepatitis C complications and mortality. Only two small, poor-quality studies56,60 reported data on QOL. Both studies suggested a tendency towards improved QOL in the adherence intervention groups, compared with usual care, despite the interventions reflecting completely different approaches in very different patient populations: the use of epoetin to manage treatment-associated anemia in 67 patients56 and the use of DOT in methadone maintenance clinic attendees.60We cannot eliminate the possibility that these positive findings are affected by publication, reporting, or other biases. Nonetheless, the fact that the few studies that reported any health outcomes tended towards benefit and also did not indicate a decrement in intermediate measures of adherence and treatment response (i.e., SVR) should be encouraging to patients, clinicians, and researchers as this would be consistent with overall potential health benefit.

The association of adherence interventions with virological response, particularly SVR, was the most commonly investigated outcome in the available literature. In general, adherence interventions tended to result in greater proportions of patients achieving a SVR (and EVR where reported), but few studies showed statistically significant differences between groups. When considered by intervention type, the evidence for increased SVR was most consistent for patient-level adherence interventions. Whether viewed by intervention type or considered as a whole, however, the available evidence is very weak in suggesting a clear improvement in SVR through adherence interventions.

Almost all included studies that measured adherence showed that interventions tended to improve adherence, despite the varying quality, interventions, definitions, and measurements. Additionally, the magnitude of the association remained consistent (or increased) over time (12 vs. 24 vs. 48 weeks) in those studies reporting adherence data in multiple follow-up time points.87,88,90 The two fair-quality studies – one evaluating the effect of specialized pharmacy care61 and the other evaluating the effect of cognitive behavioral therapy94 – that showed no impact on adherence (and suggested a possible increase in nonadherence) after the interventions were imprecise in their estimates and relatively small. The existing body of literature offers little data about the harms associated with adherence interventions.

Strength of Evidence
We present the strength of the evidence for health outcomes for all studies and by intervention group in Table 12. The strength of the evidence for intermediate outcomes for all studies and by intervention group are presented in Table 13. We summarize this information by outcome and intervention group in narrative below.

Health Outcomes
Overall, we found insufficient evidence to determine the effect of adherence interventions on health outcomes. No studies reported morbidity, all-cause mortality, or HCV-specific mortality. In addition, no studies reported on HCV transmission. One poor-quality RCT and one poor-quality cohort study provided insufficient evidence for quality-of-life improvements that resulted 45

from patient adherence interventions due to risk of bias, imprecision, and lack of sufficient number of studies.

Two poor-quality RCTs with a high risk of bias provided insufficient evidence for harms related to adherence interventions. Both of these studies tested the effect of medications (e.g., epoetin and citalopram) to help manage side effects related to HCV treatment. Both studies reported that no patients showed adverse effects related to the use of these medications, but provided no additional details.
;
Intermediate Outcomes
The strength of evidence is insufficient-to-low for SVR achievement through adherence interventions that manage adverse effects, provide patient education and support, or directly oversee HCV therapy in patients at high risk for nonadherence (methadone maintenance clinic patients). This rating is due to medium-to-high risk of bias, imprecision, and lack of sufficient numbers of comparable studies.

We also found insufficient evidence on how interventions affected EVR based on two RCTs with high risk of bias. One study presented inadequate data, which precluded us from determining estimates of overall consistency and precision.

We deemed the strength of evidence to be insufficient (based on one fair- and two poor-quality RCTs) or low (based on five primarily fair-to-good quality cohort studies) for improved adherence as a result of various types of interventions. In general, the cohort studies found that adherence interventions had a consistent benefit on patient adherence.
.
System-Level Interventions versus Usual Care
We found insufficient evidence regarding the impact of system-level interventions on QOL, SVR, and adherence. No evidence exists regarding mortality and morbidity.
.
Regimen-Related Interventions versus Usual Care
We found insufficient evidence on the association between regimen-related interventions and patient adherence. We found no evidence about other outcomes.

Patient-Level Interventions versus Usual Care
We judged the strength of evidence for the association between patient-level interventions and the achievement of SVR to be low. We made this valuation based on a medium risk of bias across three studies with consistent effects, despite imprecise estimates and the fact that these outcomes were indirect.

The studies provided generally consistent and precise effect estimates related to patient adherence. We judged the strength of evidence to be moderate given the relatively few studies (four) with overall medium risk of bias and the indirectness of the outcome. More research in this area may affect this estimate and our confidence in the effect estimate. Only one study examined the effect of a patient-level intervention on EVR. As a result, we found the strength of evidence to be insufficient. There was no evidence regarding health outcomes, including harms related to patient-level adherence interventions. 46

Adverse Effect Management Interventions versus Usual Care/Placebo

The strength of evidence on QOL was found to be insufficient, based on a relatively small poor-quality RCT. The evidence on harms was also insufficient given the high risk of bias and the lack of detail provided. Similarly, we judged the evidence on SVR, EVR, and adherence to be insufficient due to high risk of bias, the inconsistency and imprecision of the effects, and the indirectness of the outcomes. Again, no evidence addressed the effects of the intervention on mortality or morbidity.

Click Table To Enlarge

Table 12. Strength of evidence for final health outcomes

Click Both Tables To Enlarge

Table 13. Strength of evidence for intermediate health outcomes

Applicability of the Evidence to United States Health Care System
The findings from included studies have generally good applicability to HCV patients in the United States receiving standard (dual) combination therapy of pegIFN-α and ribavrin. However, given the recent recommendation for adding protease inhibitors to the existing combination therapy for patients with genotype 1 HCV, which represents the preponderance of HCV infections in the United States,16 the available evidence is unlikely to be directly applicable to the present patients with genotype 1 HCV.30 In general, patient adherence to medication regimens often decreases as the complexity of the treatment regimen increases. It is plausible that the addition of a third agent administered multiple times per day is likely to further impact patients’ ability and likelihood of complying to treatment. In June 2012, the CDC called for universal HCV screening of the "baby boomer" population (i.e., individuals born between 1945 and 1965).96 Such screening could result in a rapid increase in the number of individuals being treated for HCV and subsequently struggling with adherence.

Seven of the 11 included studies were conducted in the United States. The remaining trials were conducted in France (k=2) or Italy (k=2). Two studies enrolled patients from a primary care setting,93,94 two from specialized hepatology units,56,61 two from addiction management centers,60,89 and four from multiple clinics.87,88,91,92 The other trial did not specify study setting . These studies included both academic and nonacademic centers.

Most studies had wide inclusion criteria, although a number of studies excluded those presumed to be less responsive to therapy (i.e., with coexisting infections or previous history of HCV treatment) or those at risk for poor adherence (i.e., with psychological illnesses or current or previous abuse of substance). Patients coinfected with HBV, HIV, and/or hepatitis D virus (HDV) were excluded in five studies,56,60,61,87,91 those with ongoing depression were excluded in two,91,93 and patients having a history of and/or active substance use were excluded in two studies.92,93 Across all studies, there were a larger proportion of males than females and the majority of patients were Caucasian. Patients with HCV genotype 1 to 4 were the primarily studied population, and the majority of patients had genotype 1 HCV in seven of the 11 included studies. Three studies60,89,94 exclusively enrolled patients currently abstinent from drugs and other substances, but seeking treatment for drug abuse in methadone maintenance or other addiction centers. These data, although limited, suggest that patients at risk for poorer adherence may be appropriate candidates for HCV therapy coupled with effective adherence interventions. Generally, patients included in those studies were representative of the prevalent HCV population in the United States.

Patients in the included studies exclusively used standard doses of combination antiviral therapy of pegIFN-α and ribavirin. The intended duration of treatment in all studies was 48 weeks for patients with genotype 1 or 4, and 24 weeks for those with genotype 2 or 3. Again, although the antiviral therapy was consistent with the current recommendations for patients with genotypes 2, 3, or 4, the currently recommended treatment for patients genotype 1 has shifted from the standard combination therapy to the triple therapy, in which a protease inhibitor is added to the combination of pegIFN-α plus ribavirin.11

A wide variety of adherence interventions were investigated in the included studies. These interventions included simplifying dosing, the use of medications or counseling for managing adverse effects, patient education and support by various parties to motivate antiviral medication use or help manage adverse effects, and provision of care within specialized care delivery systems (e.g., specialized pharmacies, methadone clinics). We found no studies that directly 50 compared the effectiveness of one type of intervention to another type of intervention. In addition, very little detail was given in the majority of the studies regarding the specific intervention components, messages, frequency, and duration. Thus, it is unclear how feasible or effective these interventions would be in real-world settings.

Limitations
Potential Limitations of our Approach
Our approach has a number of potential limitations. Our systematic review methodology may not be the ideal method to synthesize findings across studies that are predominately poor quality, with a high level of heterogeneity. Additionally, there are likely major limitations in determining the effect of treatment adherence interventions on both intermediate and final health outcomes because of multiple confounding factors that also affect response to treatment (e.g., age, genotype, body mass index, viral load). Because we are limited to the data that are presented in the primary studies, we were unable to adjust for many of these potential confounders. We discuss other limitations of the literature below.

We also excluded studies with length of followup shorter than 12 weeks. Although these short-term results may be of interest, such studies can only provide evidence on rapid virological response and possibly early virological response, both of which were judged as much less important intermediate outcomes than SVR.

We did not include non-English language studies, and thus may be missed some relevant data. Our search found only 99 citations for potentially relevant studies that were published in languages other than English. The majority of these studies were written in Spanish, French, and German. More importantly, the vast majority of non-English studies may be less applicable to the United States health care system. Therefore, their findings may be of very limited value to the context of our review.

Limitations of the Literature
There are several major limitations of the available literature. First, the studies are limited to relatively small sample sizes and are of suboptimal quality. Three of the five RCTs had sample sizes smaller than 50, and the other two included 134 and 250 patients, respectively. One RCT was of fair quality, and the other four were considered poor. The quality of cohort studies varied. In the only two good-quality studies,88,90 a relatively large number of patients (674 and 1560) were included. Other cohort studies were generally small and had important methodological limitations, including the fact that almost all failed to adjust for the influence of potentially important confounding factors. Additionally, the subpopulations varied substantially in terms of their risk for nonadherence and nonresponse to treatment across studies, which hamper our ability to pool data or results across studies.

Second, inadequate reporting of details about study design and conduct was prevalent across all studies. This resulted in substantial difficulties collecting data and determining the quality and applicability of study findings. For example, limited information was available about the intensity and length of interventions and the parties that carried out interventions. Collectively, these issues represent particularly important potential limitations because most interventions were behavior-based, and lack of implementation details makes it challenging to judge the fidelity, comparability, and applicability of study findings. In another example, many cohort studies, particularly retrospective studies, failed to detail the sources of data, the approaches to 51 acquiring and measuring data, and strategies for controlling the influence of bias. Data on loss to followup were also inadequately reported. There was a significant and disproportionate loss to followup between intervention and control groups in three studies,60,90,94 which impedes our ability to interpret the true effect of interventions.

Third, there were several serious variations and ambiguities in the definition of adherence used across studies. For example, two studies61,89 defined adherence as "completion of treatment." However, it was unclear which agent or agents (pegIFN-α vs. ribavirin vs. both) this referred to, whether it allowed for any missed doses over the course of treatment, and to what extent it reflected patient- versus physician-initiated changes in treatment. Of the eight studies reporting adherence data, at least five different definitions were used (Tables 8-11). The widely varying definitions of adherence used by study investigators created a major obstacle in our ability to compare findings across studies; this also hampers the ability of clinicians, patients, and policy-makers in using the evidence for practice and decision making.

Many studies failed to distinguish between physician-initiated reductions in dosage or therapy duration and patient-directed nonadherence. Physician-initiated dose-modification or even discontinuation generally represents individualized patient care, which should not be considered as nonadherence. Patient-directed dose-reduction and discontinuation may be due to toxic effects, and many other reasons (e.g., patients not remembering dosing schedule, having difficulties in using pegIFN).97 Although debate continues about the inclusion of physician-directed treatment discontinuation or modification in defining "nonadherence,"42 for this review we decided that patient-directed nonadherence was the primary focus. Thus, we excluded many studies that did not present patient- and physician-directed treatment discontinuation separately in their analyses.

Populations varied substantially in terms of their risks for nonresponse to treatment (e.g., what genotypes, previous treatment history, or ages were represented) and their risks for potential nonadherence (i.e., current or past drug users). Within studies, these potentially important factors were not generally assessed for baseline comparability or controlled for in analyses. This was particularly true in prospective and retrospective cohort studies. Of the five cohort studies, only two adequately adjusted for the influence of confounding factors.88,90 Other studies either failed to adjust for or inadequately controlled for the influence of other important factors.

Another important limitation in this literature is the fact that all identified studies relied on surrogate outcomes. Likewise, none reported long-term health outcomes besides two that reported on quality of life. The goal of adherence interventions is to improve treatment response, typically SVR, and ultimately improve hepatitis C complications, such as cirrhosis and hepatocellular carcinoma. However, no evidence has examined whether interventions for adherence improve those final long-term health outcomes. Additionally, available evidence assessing the comparative effectiveness of interventions for intermediate outcomes such as SVR is very weak.

Finally, while treatment standards for HCV have been rapidly evolving, available studies have only included patients receiving dual therapy through a standard combination of pegIFN-α and ribavirin. Further research is needed to determine how patient adherence may change with the addition of a third antiviral agent into the standard treatment regimen, and how adherence interventions should be designed to incorporate the new class of drugs. Prior reviews examining medication adherence have found that patient adherence decrease as treatment regimens become 52

more complex.54 However, it is unclear how adherence may change in patients undergoing antiviral therapy for chronic HCV infection with the new therapy regimen.
 
Clinical Implications
Available evidence does not provide a clear direction for clinical practice to improve adherence in hepatitis C treatment. The included studies suggest that adherence interventions tended towards improved adherence and/or SVR, particularly those focused on reducing pill burden or providing additional patient education and support. This result was evident despite including various patient populations, use of diverse interventions, and suboptimal quality. Our review found four studies with an overall moderate strength of evidence supporting patient-level interventions to improve treatment adherence. However, it continues to be uncertain which specific interventions are effective and what degree of improvement could be expected in current practice, particularly considering the recent updated recommendation for triple therapy in genotype-1 patients.

In general, the available evidence on guiding efforts to improve adherence to recommended treatments of patients with chronic hepatitis C remains very limited. While the uncertainty continues, care providers may not be compelled to formally initiate interventions in order to achieve a higher level of patient adherence to hepatitis C antiviral treatment. Nonetheless, general principles such as patient education and support and reducing pill burden that have been shown to increase patient adherence to treatment may be considered, since existing epidemiological studies suggest a consistency in the association between a higher level of adherence and an improved SVR.41,69,72

Evidence Gaps
Substantial gaps exist for all types of adherence interventions. Across all trials, no trials investigated the impact of adherence interventions on long-term health outcomes, such as decompensated cirrhosis, hepatocellular carcinoma, and mortality. Nearly all studies included genotype-1 HCV patients that received the standard combination antiviral therapy. Therefore, the results may not be applicable to current clinical practice.

For system-level interventions, evidence was inconsistent regarding SVR and substantial uncertainty remains regarding adherence. While it appears that dose simplification is an effective regimen-related strategy to improve adherence, the evidence on SVR is lacking. While generally low, the evidence of patient-related interventions suggested a trend of improvement in SVR and adherence. The evidence for adverse effects management is conflicting, although studies with fair-quality RCTs suggest a trend of improvement in SVR.

We identified no studies that evaluated the effect of an intervention that targeted two or more levels of influence (e.g., system-level changes plus patient counseling). It is likely that the most effective interventions would include a combination of changes made to the systems and settings in which HCV care is received, the packaging and delivery of medications, the support and education provided to HCV patients including strategies to helping patients manage HCV treatment-related side effects through medications or counseling methods. There is a need in the HCV literature to design and test such comprehensive approaches. 53

Future Research
Future research should use more rigorous methods in the design and conduct of hepatitis C adherence intervention studies. Although various designs can assess the comparative effectiveness of adherence interventions, RCTs remain the optimal approach for hypothesis testing.
98 While cohort studies may be used, they are susceptible to selection bias and are less able to account for unknown prognostic factors than RCTs,99 despite the use of novel approaches such as propensity scoring.100 Future studies should have sufficient power for testing hypotheses, with longer follow-up periods to include long-term health outcomes. As noted earlier, the quality and design of the available literature was a serious limitation in our review.

Studies should also strive to use direct health-related outcomes such as HCV-morbidity, mortality, and quality of life, in addition to the surrogate outcomes that are most often reported in the current literature. While these outcomes will require longer follow-up and may be challenging when conducting studies, reporting these outcomes will improve the applicability of study findings to clinical practice. Longer-term outcome data, such as cirrhosis and hepatocellular carcinoma, are less readily available in RCTs. Nonetheless, it is possible to use cohort studies that rely on patient registries to address this issue.

The recommended treatment for genotype-1 patients has shifted from the standard combination therapy of pegIFN-α plus ribavirin to triple therapy including protease inhibitors.11 As such, the available evidence is of very limited value to the treatment of genotype 1 HCV. Although the available literature base may provide indirect evidence regarding interventions for this population, it is unclear how adding new antiviral agent will affect patient adherence. In particular, the administration of the protease inhibitor is complex, and adding this agent to the standard combination therapy will further complicate the treatment. Uncertainty will remain until well designed and conducted trials are available, and adequately powered RCTs testing adherence interventions incorporating this new treatment regimen are conducted.

There is also a strong need for standardizing the definitions of adherence in the context of chronic hepatitis C treatment. Multiple components—including treatment duration, dosing, timing, and intensity—are used in the varying definitions of adherence that we found, and treatment adherence can be associated with one or more antiviral agents in hepatitis C treatment. The multiplicity of domains and components may result in many variants in the definitions about adherence to hepatitis C treatment. The "80/80/80" criterion is often used in hepatitis C literature but has two major limitations. First, this definition will no longer be applicable to the triple antiviral therapy for genotype 1 HCV patients. Second, there seems a continuous relationship between the level of adherence and the treatment response 41 so defining adherence vs. nonadherence based on an arbitrary threshold may thus be suboptimal.
.
Future studies should clearly distinguish physician-initiated dose-reduction or discontinuation from patient nonadherence to treatment. Although physician initiated dose-reduction or discontinuation seems related to adherence, this treatment change is typically due to vital adverse effects associated with antiviral therapy, and is based on the treatment protocol. The nature of this change differs from patient nonadherence, in which patients fail to match agreed treatment plan probably because of difficulties in remembering taking medications or following the complex treatments, unwillingness to continue the treatment, and reduced quality of life.

In our exploration of risk factors associated with treatment response, we have found a number of potentially important factors associated with treatment response and patient nonadherence (Table 1). Future studies, particularly observational studies, should consider the 54 issue about patient comparability in exposure and nonexposure groups. Efforts are needed to adequately adjust for the influence of those factors.

Finally, as noted earlier, many of the studies we found were of poor quality, with inadequate reporting of study design and intervention details. Future studies should include clearer and more detailed reporting of study design and conduct. Studies need to provide sufficient information about how adherence interventions are undertaken, including the parties of undertaking intervention, such details of interventions such as intervention components, intensity, and duration. Studies should also describe methodological characteristics in more details. RCTs should report details on patient selection, allocation, and followup. In the results, the data on loss to followup should be clearly reported. Cohort studies should provide detail on collected variables, sources of data, accuracy of measurements, and approaches that are used to minimize bias. In addition, studies should be more explicit and clear in defining and measuring adherence. Ideally, study reports should include a section to describe the definition and measurement of adherence.

http://www.effectivehealthcare.ahrq.gov/ehc/products/326/1185/Hep-C-Tx-Adherence_DraftReport_20120711.pdf