Interferon-Based Therapies for Hepatitis C: Utilization, Costs, and Outcomes
This study illustrates the healthcare utilization patterns, direct costs, and sustained virologic response rates associated with interferon-based therapies for hepatitis C patients.
Since the hepatitis C virus (HCV) was fi rst identifi ed in 1989,1 it has become recognized as a common cause of chronic liver disease around the world. Between 130 and 170 million people are chronically infected with HCV worldwide and more than 350,000 die from HCV-related liver disease annually.2 In the United States alone 1.6% of the general population is seropositive for HCV antibodies, and 1.3% (3.2 million individuals) have chronic hepatitis C.3 Currently, 36% of patients wait-listed for liver transplant in the United States have chronic HCV infection.4 Over the next decade, it is estimated that US direct medical care costs for hepatitis C will range from $7 to $14 billion.5
Veterans have higher rates of HCV infection than the general population.6 Patients with chronic hepatitis C in the Veterans Health Administration (VA) system represent a significant number of patients; more than 147,000 active patients with documented HCV viremia were in the VA HCV Clinical Case Registry as of 2008.7 A 1999 national crosssectional study of more than 26,000 veterans receiving care through the VA found a 6.6% point prevalence of HCV seropositivity.8 A similar study conducted in 2001 reported that 5.4% of VA users were seropositive for HCV antibodies.9 Antiviral treatment for HCV is complex and has evolved over time. In 1991, initial treatment consisted of interferon alfa monotherapy (interferon alfa-2b, interferon alfa-2a, or interferon alfacon-1). Dual combination therapy with interferonalfa and ribavirin became available in 1998,10 and the combination of pegylated interferon and ribavirin became available beginning in 2001.11,12 Recently 2 new direct-acting antiviral protease inhibitors, telaprevir and boceprevir, have been approved for the treatment of patients with HCV genotype 1 infection in combination with pegylated interferon and ribavirin.13-17
To date, treatment success is largely dependent on the HCV genotype and duration of treatment. Six strains of HCV have been identified. Of these, genotype 1 is the most prevalent in the United States, followed by genotypes 2 and 3 (collectively referred to as non—genotype 1).18 Genotype 1 is the most diffi cult strain to treat and generally requires treatment duration of 48 weeks. The recommended treatment duration for patients with genotype 2 or 3 is 24 weeks.19,20 With currently available pegylated interferon and ribavirin therapies, the sustained virologic response (SVR) rates from registration trials associated with treatment of genotype 1 is 42% to 46%, and with treatment of genotypes 2 and 3 it is 76% to 82%.11,12 In contrast, the actual SVR rate achieved in US veteran patients with HCV genotype 1 is 20%, while the SVR rates for genotypes 2 and 3 are 52% and 43%, respectively.21 The newly approved therapies telaprevir and boceprevir will require 24 to 48 weeks of treatment and are expected to increase SVR rates.
In response to the rising number of HCV diagnoses, the VA initiated a directive for implementing broad clinical care programs for HCV, including routine screening of all VA patients.22,23 As VA medical centers are reimbursed for caring for complex patients treated for HCV, the cost associated with screening, treatment, and followup for HCV patients is important to VA administrators. Identifying strategies aimed at improving the quality and cost-effectiveness of care provided to HCV patients is an important administrative goal.
A number of studies in general populations of HCV patients have found that dual therapy with interferon and ribavirin can be considered cost-effective.24-28 Very few studies have reported the actual costs incurred while treating patients diagnosed with HCV infection,29 and no studies have focused on VA costs. The most accurate method to determine cost components is a “top down” approach directly following a cohort and collecting actual longitudinal frequency and unit cost data.30 The purpose of this study wasto examine the treatment patterns, direct costs, and health outcomes associated with HCV antiviral therapy in an actual VA HCV treatment population. These data can then be used as baseline data for future cost-effectiveness studies comparing current treatment strategies with newly approved therapies.
A consecutive cohort of patients with chronic hepatitis C receiving antiviral treatment between January 1, 2000, and December 31, 2001, were identifi ed. During this time period, 681 VA patients had a positive HCV enzyme- linked immunoassay test at a midwestern Department of VA Medical Center (Figure). Of these patients, 557 did not receive antiviral therapy and were excluded from the analysis. Patients who received experimental or alternative therapies such as gamma interferon (n = 5) or peginterferon antiviral monotherapy (n = 1) were also excluded. Finally, additional patients were excluded due to a lack of cost data (n = 17) or the inability to determine their genotype (n = 2), resulting in a fi nal sample size of 99 patients treated for HCV with interferon-based therapies.
Because some patients had been diagnosed prior to 2000, we identified the earliest diagnostic date in the patients’ administrative records going back to 1997, thus ensuring that the analyses pertained to patients who were treatment naïve. Treatment start date for antiviral therapy was based on the date of the first prescription and the treatment duration of the last prescription written by the physician.
Hepatitis C virus—related outpatient encounters, labs, and biopsies were identifi ed using VA clinic stop code 307 for gastroenterology and the Current Procedural Terminology (CPT) codes listed in Table 1. Each patient’s HCV antiviral genotype, therapy combination, and treatment length were examined. Patients were classifi ed as either genotype 1 or non—genotype 1 patients. Three therapy combinations were identifi ed: interferon and ribavirin (IR); interferon, pegylated interferon, and ribavirin (IPR); and pegylated interferon and ribavirin (PR). These included the following medications: interferon alfa-2b, ribavirin, combination of interferon alfa-2b and ribavirin, pegylated interferon alfa-2a, and pegylated interferon alfa-2b.
Using the VA’s clinic stop code 307 for gastroenterology, the timeline used to identify associated HCV clinic visits was defi ned as occurring 7 days prior to initiation of antiviral therapy and 1.5 years after the prescription treatment end date. The 1.5-year follow-up period was based on the following clinical treatment regimen: HCV patients returned for a follow-up visit 6 months after completion of antiviral therapy. Negative results indicated SVR achievement, and patients came in for a second follow-up visit 1 year after the initial 6-month follow-up visit. If the 6-month HCV test was positive, patients were reevaluated. Decisions regarding a subsequent round of antiviral treatment depended on the patients’ prior response and their ability to tolerate side effects. Physician recommendations regarding other behavioral and medical changes that would reduce fibrosis progression were provided.
In order to accommodate the range of therapeutic duration, treatment lengths of 1.5 times the expected number of weeks of medication therapy were used to distinguish between patients undergoing 1 versus more than 1 course of treatment. Thus, genotype 1 patients were classified as having either “greater than or equal to” or “less than” 48 times 1.5 = 72 weeks of treatment. Non—genotype 1 patients were classified as having either greater than or equal to or less than 24 times 1.5 = 36 weeks of treatment.
Finally, regression analyses were used to address factors predictive of differences in costs across the 3 treatment groups. Because cost data are typically skewed and not normally distributed, we estimated log expenditure models that included indicator variables for the IPR and PR groups. The largest group of IR patients served as our reference category. Demographic data included age, sex, marital status, service-connected status, genotype, and treatment group. Service connection is determined by the VA as an injury or disease incurred while serving on active duty. Service-connected ratings vary from 0% to 100%, with higher ratings generally indicative of increasing disease severity.
For calendar years 1997 to 2004, clinical encounters and laboratory tests were extracted from the VA’s Patient Treatment and Outpatient Care files using VA clinic stop codes and CPT codes. Patients’ demographic characteristics, SVR status, genotype, and whether veterans were service connected for injuries or disease incurred or aggravated in military service were obtained by reviewing patient medical records.
Outpatient costs were extracted from the Health Economics Resource Center’s (HERC) Average Outpatient Cost database. Because the HERC’s data sets are available only from VA fiscal year 1999 onward (ie, October 1998 onward), costs for outpatient encounters occurring prior to October 1998 were estimated using similar outpatient costs for encounters occurring after October 1998. Because the HERC’s outpatient cost data are based on Medicare reimbursement rates, the relative costs of clinic visits experienced by veterans in each treatment regimen are comparable to those in Medicare. Because VA costs are adjusted to refl ect the VA’s overall annual budget, the actual dollar amounts for specific healthcare events differ from those of Medicare.
Prescription medications and costs were derived from the Veterans Health Information Systems and Technology Architecture database. The costs for individual prescriptions were calculated by multiplying the medication’s unit price by the quantity supplied for that particular prescription. All costs (medication and clinic visit) were converted to 2010 prices using the Consumer Price Index for medical care.31
This project was approved by an institutional review board. The analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).
The percentage of the study population who were white (81%) and male (96%) closely resembles the general VA patient population (Table 2); the study sample of HCV patients was signifi cantly younger (49 years) than the national 2001 VA average of 58 years, and more likely to be divorced.32 As is characteristic of the general US population of HCV patients,33 most veteran patients (71/99, or 72%) were infected with the genotype 1 strain of HCV.
Antiviral Medications and Costs
Genotype 1 patients receiving IR therapy (n = 28) had the lowest total medication cost ($12,188 per person), with an overall duration of antiviral therapy of 63 weeks (Table 3). Although patients receiving PR therapy (n = 24) had higher medication costs ($14,716) but shorter antiviral treatment duration (32 weeks) compared with the IR group, this difference was not signifi cant. On average, genotype 1 patients requiring multiple treatments in the IPR group (n = 19) had the longest antiviral treatment duration (114 weeks) as well as the highest medication cost of all 3 treatment groups ($23,517 per person) (P <.01).
Concurrent with treatment guidelines, the treatment regimen for non—genotype 1 patients was generally half as long as that for genotype 1 patients. Subsequently, medication costs for non–genotype 1 patients were 15% to 24% lower than those of genotype 1 patients.
Across both genotypes, while 1 course of IR therapy was typically longer than 1 course of PR therapy, IR medication costs were 24% to 47% less expensive than PR therapy (for genotype 1 patients, IR medications cost $11,249 and PR medications cost $14,716; for non—genotype 1 patients, IR medications cost $6681 and PR medications cost $12,509). The extended duration of IR therapy reflects the attempt to increase SVR rates, as well as the need for some patients with interrupted treatment regimens to repeat the course of treatment. Independent of genotype, patients experiencing more than 1 course of IR therapy had higher costs than patients experiencing PR therapy. Considering the additional medication and clinical costs, multiple courses of IR therapy were the second-most expensive treatment regimen overall ($22,439 for genotype 1; $21,642 for non–genotype 1).
The IPR subgroup primarily represents patients with IR treatment relapses or failures who were transitioned to PR therapy. Approximately 75% of patients (74% [14/19] of genotype 1 patients; 80% [4/5] of non—genotype 1 patients) experienced a signifi cant lapse in their treatment regimen before transitioning to PR, while the remaining 25% did not experience treatment delays while transitioning. The extended treatment duration along with higher medication costs and more intensive clinical encounters resulted in IPR patients comprising the most expensive treatment group (P <.01).
Average Total Costs of Prescriptions and Outpatient Clinical Encounters
Over all treatment regimens, the average total cost of antiviral therapy and HCV-related clinical encounters was higher for genotype 1 patients ($19,660 per person) than for non—genotype 1 patients ($14,373 per person) (not shown). For each treatment regimen, the average total cost of genotype 1 patients treated with IR, IPR, or PR therapy was $15,647, $28,187, and $17,591 per person, respectively, compared with $12,713, $20,856, and $14,626 per person, respectively, for non–genotype 1 patients. Differences in cost were significant between the IPR and IR groups as well as between the IPR and PR groups (P <.01 for both comparisons).
Cost per Sustained Virologic Response
The overall HCV study population achieved an SVR rate of 38% (38/99) (Table 4), resulting in an overall cost per SVR of $47,324. Concurrent with the literature, non—genotype 1 patients achieved a higher SVR rate than the genotype 1 patients (57% [16/28] vs 31% [22/71], respectively). Combined with lower treatment costs, it was generally more cost-effective to treat non–genotype 1 patients, with an overall cost per SVR of $25,152. By contrast, the overall cost per SVR for genotype 1 patients was $63,448.
The SVR rates for genotype 1 patients were similar for IR and PR treatment groups. Interestingly, the SVR rate for the IPR group was higher, possibly due to the longer duration of therapy received; however, definite comparisons cannot be made due to the relatively small numbers of patients in each group. Across treatment groups, costs per SVR for genotype 1 patients were $70,364 (PR), $68,250 (IPR), and $54,768 (IR).
Costs per SVR for non—genotype 1 patients were $34,761 (IPR) and $19,559 (IR). Note that there were too few non–genotype 1 patients treated with PR to definitively evaluate SVR rates.
Adjusting for differences in patient characteristics, Table 5 lists the results for estimating the log expenditure models. In the adjusted analyses, the IPR (P = .07) and the PR (P = .03) groups incurred a significantly higher level of medication costs than the IR group. While the costs associated with outpatient visits were not found to significantly differ across the 3 treatment groups, the higher level of medication costs contributed to higher total (outpatient and medication) costs for the IPR (P =.06) and the PR (P = .07) groups relative to the IR group. However, these differences were not signifi cant at the P = .05 level. Generally speaking, across all treatment groups, longer treatment duration was associated with higher medication, outpatient, and total costs (P <.01).
Antiviral treatment of patients with HCV is intense and complex, requiring many visits over the initial treatment period of 24 to 48 weeks, as well as subsequent follow-up visits and/or re-evaluation. Our data indicate that the total overall costs related to treatment of patients with HCV are highly dependent on genotype, medication costs, and duration of therapy. Patients treated during periods that include conversion of actively treated or previous treatment failures to newer therapies, as represented by the IPR group, have the longest duration and highest costs. Despite these higher costs, the HCV genotype 1 patients in this group demonstrated relatively high SVR rates, indicating that there is a benefi t from more intensive care during a period of changing protocols.
The greater number of visits (n = 33 visits) observed with IR patients undergoing 1 course of therapy (<72 weeks) may be due to more favorable tolerability of this regimen compared with PR (n = 26 visits) and an attempt at a longer duration of treatment. Overall SVR rates of 29% and 25% for IR and PR genotype 1 patients, respectively, were comparable and better than national estimates of 20% reported elsewhere.21
Largely due to the longer duration of the antiviral therapy, it is more expensive to treat genotype 1 patients than non—genotype 1 patients. Additionally, the higher SVR rate in the non–genotype 1 group resulted in a lower cost per cure ($25,152 per SVR) than that in the genotype 1 group ($63,448 per SVR). These results are comparable to those reported by Malone et al.34 Comparing the outcomes of patients treated with pegylated interferon alfa-2b plus ribavirin with the outcomes of patients treated with pegylated interferon alfa-2a plus ribavirin, Malone et al estimated the cost per SVR for treatment-naïve patients, 75% of whom were genotype 1 patients, as $37,638 and $46,717, respectively (ie, $47,146 and $58,518, respectively, in 2010 dollars).34
Finally, given that the annual cost of HIV treatment in 1999 was $20,280 per patient (ie, $31,434 in 2010 dollars)35 and the 1-year event and angina-free cost of coronary artery bypass grafting in 2001 was $14,095 (ie, $20,070 in 2010 dollars),36 these results support the primary hypothesis that the direct cost of medical care associated with the treatment and follow-up of HCV patients is comparable to that of other chronic disorders.
Because our analysis of utilization and costs relied on the use of secondary administrative data sets, assumptions were made regarding the acceptability of time lapses between multiple antiviral treatments and the time frame for applicable clinic visits. In addition, we made the assumption that all visits to the gastroenterology clinic were related to HCV care, whereas some of these encounters may have only been marginally related. Thus, the cost per cure for the treatment regimens considered in our analyses may be overstated. The costs per cure could have been lower if some of the visits had been determined not to be HCV related and were excluded from our analyses. We also made the assumption that our study population received care only at the VA. Because the medications prescribed for patients diagnosed with HCV are expensive, the treatment regimen is intense and complex, and the patients tracked in this study received care at the VA Medical Center throughout the duration of the study, we believe that the likelihood of patients receiving care related to HCV outside the VA was minimal. Thus, any bias associated with the use of non-VA care was negligible. Because clinical encounters accounted for only approximately 20% of the total cost of the care provided, we believe that whatever bias was introduced by the above limitations was minimal and did not affect our conclusions. Moreover, our results appear to be within the range of that estimated by other studies.
Our data are limited to inclusion of patients with HCV who are patients at 1 VA Medical Center and may not be generalizable to the nonveteran population. The VA serves a fairly unique sector of the general population in that veterans are largely white and male, and veterans eligible to use VA services have open access to care with no restrictions placed on referrals, specialists, or number of visits.37 In addition, the VA receives larger pharmaceutical discounts than the private market.38 Thus, medication costs will likely be higher in the private sector. To illustrate this point, Yeh et al used data from a VA national pharmacy database and from indirect expert opinion regarding the number of visits needed, and estimated that the weekly costs of pegylated interferon plus ribavirin in the VA system in 2005 ranged from $227 to $247 (ie, $273-$297 in 2010 dollars), which was less than costs of this regimen used in non-VA cost utility analyses ($433- $817) (ie, $520-$982 in 2010 dollars).39 Because medication costs comprise approximately 80% of the cost of HCV therapy, the relative cost-effectiveness of these therapies may differ in the VA compared with the private sector. Given these limitations, the results of this study may be more applicable to patients in other large managed care organizations receiving similar medication cost discounts.
Finally, inherently problematic in all observational studies is the issue of self-selection bias. Differences in costs and SVR could be due to latent differences among the respective patient groups themselves rather than to the treatment options. Although our study’s sample size limitations do not lend themselves well to the traditional modes of correcting for selection bias, at a minimum we conducted regression analyses to account for measurable (observed) differences using the data that we have on patient demographics, genotype, duration of treatment, and treatment type. Although these analyses did not attempt to adjust for unobserved differences in patient treatment groups, we believe that the analyses reflect what the consulting provider considered an appropriate treatment regimen given the patient’s history and diagnosis. In the absence of a randomized controlled trial, we believe that this analysis is useful in that it assesses the observed cost of caring for HCV patients in one of the largest managed care organizations in the United States.
In 2011, telaprevir and boceprevir became available for treatment in combination with the standard of care, PR, for treatment of patients with HCV genotype 1. Current phase III clinical trials indicate an up to 2-fold improvement in SVR rates to 68% to 75% for genotype 1 HCV patients, and a 2- to 5-fold increase in SVR rates to 53% for genotype 1 nonresponders to standard PR treatment. 13-17 The timing of visits and laboratory monitoring for these regimens are equivalent to visits required for current interferon-based treatments, and the majority of patients treated will require 24 weeks of treatment rather than 48 weeks.13-17 New cost-effectiveness analyses will be needed to take into account the improved effi cacy and the increased costs associated with these novel drugs. The current data regarding the actual treatment-related costs of resource use for patients receiving 24 to 48 weeks of interferon-based treatments are relevant to future costeffectiveness analyses using these new combinations.