Treatment Patterns and Adherence among Patients with Chronic Hepatitis C Virus in a US Managed Care Population
Debanjali Mitra, Health Economics, RTI Health Solutions, 3040 Cornwallis Road, Research Triangle Park, NC 27709, USA. E-mail: firstname.lastname@example.org
Objective: The purpose of this study was to document real-world treatment patterns, medication adherence, and the impact of adherence on disease-specific and all-cause health-care costs among chronic hepatitis C virus (HCV) patients in a US managed care population.
Methods: Commercial insurance claims data between January 1, 2002 and December 31, 2006 from the Ingenix Impact (formerly Integrated Health Care Information Services) database were retrospectively analyzed. Chronic HCV patients with one or more prescriptions for an HCV-specific treatment within 6 months before or at any time after their first observed diagnosis of chronic HCV were selected. Prescribing patterns, treatment cost, and duration of treatment were assessed over the entire therapy period. Medication adherence rates and the relationship between adherence and health-care costs were assessed over the 24-week period after treatment initiation. The results were stratified by key clinical characteristics such as genotype, sustained virologic attainment, and disease severity.
Results: Results showed that peginterferon and ribavirin combination regimens were the most common treatments for chronic HCV. The patients underwent treatment for approximately 30–32 weeks on average, and treatment costs were over $20,000 per patient. Adherence to medication was suboptimal, especially among patients with severe disease. Adherent patients had higher pharmacy costs but significantly lower total costs when pharmacy was excluded.
Conclusions: New and improved treatments that promote better adherence and impose a lower cost burden on patients and payers are needed.
Hepatitis C virus (HCV) has been recognized over the past two decades as an emerging public health problem. Over 180 million people are infected with HCV worldwide . Estimates by the Centers for Disease Control and Prevention (CDC) suggest that it is one of the most common blood-borne infections in the United States and affects approximately 3.2 million people , with 17,000 new acute cases detected each year . At least three quarters of cases of HCV develop into long-term or chronic infections . Over time, approximately 20% of chronically infected patients develop cirrhosis of the liver  and other complications such as hepatocellular carcinoma, leading to an increased need for liver transplants.
The CDC estimates that nearly 8000–10,000 deaths are due to chronic liver disease each year . As the duration of infection increases among surviving HCV patients, the burden of disease is also likely to rise. It is estimated that the proportion of patients with cirrhosis will increase from 16% to 20% by 2020 and that hepatocellular carcinoma will increase by 81% and liver-related deaths will increase by 180% . Because the age group most frequently diagnosed with HCV, people aged 30–49 years, is likely to be employed and enrolled in an employer-based insurance plan, the burden of HCV in the United States falls largely on managed care payers .
Early diagnosis and treatment of chronic HCV can do the following: reduce hepatic inflammation; prevent progression to fibrosis, cirrhosis, and hepatocellular carcinoma by eliminating the virus in chronically infected patients; and control the spread of disease . Current treatments for chronic HCV include combination peginterferon (alfa-2a or alfa-2b) regimens with ribavirin, peginterferon monotherapies, combination interferon (alfa-2a or alfa-2b) regimens with ribavirin, interferon monotherapies, interferon alfacon-1 (consensus interferon) regimens with ribavirin, and consensus interferon monotherapy. Although there has been significant progress in the treatment of chronic HCV over the past two decades, there continues to be a large unmet need because of low rates of sustained virologic response (SVR) and a significant proportion of relapses and nonresponses. Current therapies also are associated with a high incidence of adverse events, leading to low adherence, therapy discontinuation, and/or dose reduction.
Despite the potential long-term burden of chronic HCV on managed care systems and the persistent unmet need in this population in spite of existing treatments, there have been no studies that provide a comprehensive understanding of real-world prescribing patterns of and adherence with key HCV therapies. This article seeks to address this gap by analyzing retrospective commercial claims data to examine treatment patterns and medication adherence among US managed care enrollees with chronic HCV. We also examine the impact of medication nonadherence on HCV-specific and all-cause health care costs. An understanding of current treatment patterns in this population may help inform payers and practitioners about the therapeutic landscape as novel antiviral and interferon therapies for chronic HCV become available.
Data were taken from the Ingenix Impact (formerly known as the Integrated Health Care Information Services) database containing medical (inpatient, outpatient, physician, ancillary) and pharmacy claims from a national sample of 40 managed care health plans covering approximately 50 million lives. The database has a representative distribution of age and sex compared with national enrollment in managed care plans. Geographic representation, however, is biased toward the East Coast. Claims records in the Ingenix Impact database contain information on patient characteristics that include the following: demographics eligibility for benefits; diagnoses (including primary and up to two nonprimary diagnosis codes); detailed information about hospitalizations, diagnostic testing, and therapeutic procedures; inpatient and outpatient physician services; prescription drug use; and cost data in the form of managed care reimbursed amounts for each service. Results for selected laboratory tests are also available for a subset of patients. Unique patient identification numbers in the Impact database allow patients to be tracked longitudinally.
Patients with a diagnosis of chronic HCV (International Classification of Diseases, 9th Edition, Clinical Modification [ICD-9-CM] codes 070.44, 070.54, 070.70, and 070.71) between January 1, 2002 and December 31, 2006, the five most recent years of Ingenix Impact data available for this analysis, were selected for initial study inclusion. The patients also were required to have no evidence of hepatitis B virus, identified by ICD-9-CM codes 070.2x and 070.3x (including all fifth digit modifiers) at any point during their claims history. The patients with acute HCV (ICD-9-CM codes 070.41 and 070.51) were excluded from the study sample if they did not also have at least one diagnosis of chronic HCV.
The patients were required to receive at least one HCV treatment at any time 6 months before or any time after the first observed HCV diagnosis. The patients were allowed to receive treatment before their first observed HCV diagnosis because we cannot identify the true date of HCV onset in our study data. Given the chronic nature of HCV, it is likely that many patients were diagnosed several years before the start of the study database or before enrollment in a health plan captured by the database. Our study sample, therefore, represents patients identified on the basis of a new episode of treatment who have current diagnostic evidence of HCV, rather than truly incident (i.e., new) HCV cases.
HCV treatments included combination peginterferon alfa-2a plus ribavirin (PEG2A + R), peginterferon alfa-2b plus ribavirin (PEG2B + R), peginterferon alfa-2a monotherapy (PEG2A), peginterferon alfa-2b monotherapy (PEG2B), combination interferon alfa-2a plus ribavirin (IFN2A + R), interferon alfa-2b plus ribavirin (IFN2B + R), interferon alfa-2a monotherapy (IFN2A), interferon alfa-2b monotherapy (IFN2B), consensus interferon regimen with ribavirin (IFN + R), and consensus interferon monotherapy (CIFN).
An index date was defined as the date of the first observed prescription for an HCV treatment. To ensure that each patient's index date was a reasonable marker for new treatment initiation, the patients included in this study sample were required to have at least 6 months of continuous health plan enrollment before their index date, with no evidence of prescriptions for HCV treatment during that time frame. To ensure that any observed lack of health-care events after treatment initiation was due to a true lack of medical activity and not cessation of insurance, the patients included in the study sample were required to have at least 12 months of medical and pharmacy benefits eligibility after their index date.
Baseline Patient Characteristics
Baseline characteristics that were measured at the index date included age, gender, geographic region, insurance payer type (e.g., commercial, Medicare, Medicaid), and health plan type (e.g., health maintenance organization, preferred provider organization). To assess overall comorbidity burden before the index date, we also calculated a Charlson Comorbidity Index (CCI) score for each patient. The CCI includes 17 categories of comorbidities, as defined by ICD-9 diagnosis codes, with associated weights corresponding to the severity of the comorbid condition of interest . A higher CCI score represents a higher overall comorbidity burden. A single CCI score was calculated for each patient based on the presence of the corresponding diagnosis during the 6-month period before the index date. In addition to baseline demographics and comorbidity burden, we estimated the distribution of genotype and disease severity in the subset of each patient sample for which the appropriate laboratory results were available.
We documented patterns of drug prescribing and treatment of all the key HCV therapies among patients with chronic HCV. The distribution of patients initiating each of the treatment regimens, the average number of prescriptions, the average duration of therapy, and the average cost of therapy were estimated and reported. For combination therapies, the number, duration, and cost of prescriptions were computed separately for the two drugs. The average number of prescriptions and average costs were computed across all prescription claims for that drug starting with the index (first observed) prescription. Duration of therapy was computed as the total number of days between treatment initiation and treatment completion or discontinuation. Therapy discontinuation was defined as an observed refill gap of at least 90 days between the end of one prescription and the start of the next prescription.
Adherence to therapy was measured by the patients' cumulative exposure to chronic HCV treatments over a 24-week period after treatment initiation. Because of the unavailability of genotype information for all the patients, we were unable to measure adherence over the entire treatment period because the intended treatment duration for the patients was not known in the absence of reliable genotype data. All of the patients, regardless of genotype, were expected to be treated for a minimum of 24 weeks. For the patients who were treated beyond 24 weeks, we also measured adherence over a 48-week period under the assumption that these patients, if treated more than 24 weeks, were genotype 1 or 4 and were intended to complete a full 48-week regimen.
We used the medication possession ratio (MPR) to capture treatment exposure. A recent systematic literature review  found MPR to be the most widely adopted measure (57% of all studies) in published claims-based analyses of medication adherence. MPR generally is defined as the proportion of days within an observation period covered by the total days' supply for a particular study drug within the observation period: MPR = sum of days' supply in observation period/days in observation period.
As noted in Andrade et al. , the observation period used in the MPR calculation is typically either a fixed number of days within a follow-up period (i.e., 24 weeks after treatment initiation) or the number of days between the first dispense date and end of the days' supply of the last refill for the study therapy of interest. Use of the latter denominator in the MPR definition, however, does not capture the negative adherence effect of early therapy discontinuation. To account for discontinuation, we therefore defined the denominator of the MPR formula as a fixed value of 158 days (i.e., 24 weeks) so that those patients who discontinue therapy completely before the end of the follow-up period will correctly have a lower MPR. For the patients who were treated beyond 48 weeks, we defined the denominator of the MPR formula as 336 days (i.e., 48 weeks).
The patients with an MPR of less than 0.8 (i.e., less than 80% adherence) were classified as nonadherent. The 80% threshold for identifying adherence has been the methodological basis of numerous adherence studies previously conducted [12–14]. In addition, the 80-80-80 rule, wherein patients who take 80% of their peginterferon therapy and 80% of their ribavirin therapy at least 80% of the time are more likely to achieve SVR, has been well-documented in the HCV literature .
To evaluate the direct economic impact of adherence, HCV-related and all-cause resource use and costs were calculated over the 24-week period after treatment initiation and stratified by adherent versus nonadherent patients. All cost data were adjusted to 2007 US dollars and represent reimbursed amounts paid by health plans to providers. Resource use was grouped according to the setting in which the service was delivered, including hospital admissions, skilled nursing facility admissions, emergency room, physician office, home health visits, laboratory services, prescription drugs, and a residual category capturing all other outpatient and ancillary services. The subset of all-cause resource use and costs specifically related to HCV were identified as those claims with a primary or nonprimary (i.e., up to the third code) ICD-9-CM diagnosis code for chronic HCV. HCV-related pharmacy utilization was identified as the subset of prescription drug claims for the HCV therapies described previously.
All analyses were stratified by relevant clinical characteristics, such as viral genotype, SVR attainment, and proxy indicators for disease severity, for subsets of patients with valid laboratory results, and other information required for constructing these analysis strata. The patients with available genotype results were stratified into genotype 1 versus genotype 2/3. For the subset of patients with available HCV ribonucleic acid (RNA) test results, SVR attainment status was assessed. SVR attainment is clinically measured as an undetectable HCV viral RNA test result at least 6 months after completion of treatment . In this study, the patients were identified as having attained SVR if an HCV RNA test performed at any point at least 6 months after therapy completion or discontinuation showed the viral load to be undetectable. If a patient had multiple, valid HCV RNA tests, the first valid result observed at least 6 months after therapy completion or discontinuation was used.
Because of lack of liver biopsy data in the claims database, the aspartate aminotransferase to platelet ratio index (APRI) score was used as a proxy measure for disease severity. The APRI score has been used in previous studies as a noninvasive marker of liver fibrosis and cirrhosis . APRI scores could be computed for the subset of patients with at least one valid laboratory result for both an aspartate aminotransferase and a platelet count test (occurring on the same day) during the 6-month period preceding the index date. For the patients with multiple, valid test results, the most recent result (i.e., closest to the index date) was used. The patients were classified into mild (APRI ≤ 0.5), moderate (0.5 < APRI ≤ 1.5), and severe (APRI > 1.5).
A total of 5086 patients with chronic HCV met all study inclusion criteria. Table 1 presents descriptive statistics on various key baseline demographic characteristics. Approximately two-thirds of the patients were male and had an average age of 48 years. More than 55% of all the patients were in the 45- to 54-year-old age group. Nearly 64% of all the chronic HCV patients were located in the Northeast region. Over 97% of all the patients were enrolled in commercial health insurance plans. Health maintenance organizations and preferred provider organizations were the most common plan types among these patients. The mean baseline CCI score for the sample was 1.77.
Table 1. Key demographic characteristics of the study sample
| Female||1715 (33.72)|
| Male||3371 (66.28)|
| Mean; standard deviation||48.34; 7.90|
| Range; minimum, maximum||6, 77|
|Age category|| |
| <18||17 (0.33)|
| 18–24||51 (1.00)|
| 25–34||171 (3.36)|
| 35–44||1076 (21.16)|
| 45–54||2847 (55.98)|
| 55–64||855 (16.81)|
| 65+||69 (1.36)|
|Geographic region|| |
| Northeast||3076 (60.48)|
| South||1191 (23.42)|
| Midwest||442 (8.69)|
| West||374 (7.35)|
| Unknown||3 (0.06)|
|Health plan type|| |
| HMO||2021 (39.74)|
| POS||842 (16.56)|
| PPO||2081 (40.92)|
| IND||97 (1.91)|
| Other||45 (0.88)|
|Payer type|| |
| Commercial||4972 (97.76)|
| Medicaid||94 (1.85)|
| Medicare||20 (0.39)|
|Charlson comorbidity score|| |
| Mean; standard deviation||1.77; 2.03|
| Range: minimum, maximum||0, 14|
Table 2 presents the distribution of genotype, disease severity, and SVR attainment among the subgroup of patients who had a valid laboratory test required for the classification of that characteristic. Additional tabular data are available upon request showing the distribution of various patient characteristics between the relevant categories of each clinical characteristic. Only 105 chronic HCV patients had a valid genotype, of which more than two-thirds were genotype 1. The patients were similarly distributed by gender, age, geographic region, and insurance characteristics regardless of genotype (P > 0.05 for genotype 1 vs. genotype 2/3 in all measures).
Table 2. Key clinical characteristics of the study sample
| Genotype 1||75 (71.43)|
| Genotype 2/3||30 (28.57)|
| Not available||4981 (–)|
|Disease severity (based on APRI scores)|| |
| Mild (APRI ≤ 0.5)||813 (74.94)|
| Moderate (0.5 < APRI ≤ 1.5)||222 (20.46)|
| Severe (APRI > 1.5)||50 (4.61)|
| Not available||4001 (–)|
|SVR attainment (6 months post-treatment completion/discontinuation)|| |
| Yes||336 (58.43)|
| No||239 (41.57)|
| Not available||4511 (–)|
Disease severity could be assessed for 1085 patients with chronic HCV. Nearly 75% of these patients had mild disease, 20% had moderate disease, and 5% had severe disease. The males comprised a substantially higher proportion of patients with moderate and severe disease (71% and 74%, respectively) compared with mild disease (59%) (P = 0.0008). The patients with mild disease were younger, on average, than the patients with severe disease (48 vs. 52 years; P < 0.0001), but no statistical difference in age was found when comparing mild with moderate disease (48 vs. 50 years; P = 0.2348). The proportion of the patients 44 years of age or younger was substantially higher in mild disease (27%) compared with moderate and severe (12% and 12%, respectively; all P < 0.0001).
SVR attainment could be examined for 575 patients with chronic HCV, of whom 336 (58%) attained SVR when measured 6 months or after treatment completion or discontinuation. Gender and age distributions did not vary significantly (P > 0.05 for all measures) by SVR status, but patients residing in the Northeast represented a substantially higher proportion of the patients who did not attain SVR (82%) compared with those who did attain SVR (62%) (P < 0.0001). The reason for this difference was not examined, but variations in region-specific practice patterns, formulary structures, demographic factors such as ethnicity and marital status or socioeconomic factors such as income could contribute to the difference either directly or through variation in compliance rates. For example, a larger proportion of patients living in the Northeast might be from ethnic backgrounds that have historically lower response rates. This should be evaluated in future studies.
Overall Treatment Prevalence
The number and percentage of the patients that initiated various HCV treatments are reported in Table 3. In addition, the number of months between first observed diagnosis and treatment initiation also is reported. Over 90% of the chronic HCV patients initiated treatment with PEG2A + R or PEG2B + R. IFN2A or IFN2B use (monotherapy or in combination with ribavirin) was the least common treatment used. Among the chronic HCV patients who initiated treatment within a 6-month period before first diagnosis, the average time between first observed diagnosis and treatment initiation was 2.2 months. Among those patients who initiated treatment after first observed diagnosis, the average length of time between diagnosis and treatment initiation was nearly 7 months. Although the patients initiated treatment up to 47 months after diagnosis, more than 90% initiated treatment within the first 12 months after diagnosis.
Table 3. Treatment prevalence
|Treatment initiated*|| |
| Peginterferon alfa-2a + ribavirin||2422 (47.62)|
| Peginterferon alfa-2b + ribavirin||2230 (43.85)|
| Peginterferon alfa-2a monotherapy||158 (3.11)|
| Peginterferon alfa-2b monotherapy||152 (2.99)|
| Interferon alfa-2a + ribavirin||1 (0.02)|
| Interferon alfa-2b + ribavirin||15 (0.29)|
| Interferon alfa-2a monotherapy||1 (0.02)|
| Interferon alfa-2b monotherapy||8 (0.16)|
| Interferon alfacon-1 (consensus interferon) + ribavirin||85 (1.67)|
| Interferon alfacon-1 (consensus interferon) monotherapy||14 (0.28)|
|Months between HCV diagnosis and treatment initiation (for patients initiating treatment before first observed diagnosis)|| |
| Mean; standard deviation||2.20; 2.29|
| Range: minimum, maximum||0, 6|
|Months between HCV diagnosis and treatment initiation (for patients initiating treatment after first observed diagnosis)|| |
| Mean; standard deviation||6.65; 6.94|
| Range: minimum, maximum||1, 47|
| Up to 6 months before index diagnosis||738 (14.51)|
| 1–6 months after index diagnosis||2897 (56.96)|
| 7–12 months after index diagnosis||864 (16.99)|
| 13–18 months after index diagnosis||289 (5.68)|
| 19–24 months after index diagnosis||132 (2.60)|
| 25–36 months after index diagnosis||128 (2.52)|
| Greater than 36 months after index diagnosis||38 (0.75)|
Treatment Patterns and Adherence
Table 4 displays the average number of prescriptions, the average prescription cost, the average duration of therapy, and the rate of discontinuation for the patients who initiated HCV therapies. Treatment patterns for IFN2A and IFN2B monotherapies, IFN2A + R and IFN2B + R combination therapies, and IFN monotherapy are not reported because of small sample sizes (n ≤ 15).
Table 4. Treatment patterns
|Mean number of prescriptions obtained (standard deviation)|| || || || || |
| Peg/interferon/infergen prescriptions||7.93 (4.81)||7.73 (4.64)||7.06 (5.36)||7.78 (7.84)||6.82 (6.95)|
| Ribavirin prescriptions||7.54 (4.64)||7.55 (4.79)||—||—||5.61 (4.38)|
|Mean total cost in $ (standard deviation)|| || || || || |
| Peg/interferon/infergen costs||12,656 (7,408)||12,074 (6,894)||11,244 (8,521)||12,511 (12,670)||10,141 (8,743)|
| Ribavirin costs||9,551 (5,823)||9,607 (6,071)||—||—||8,110 (6,145)|
|Duration (days) of therapy†|| || || || || |
| Mean (standard deviation)||224.19 (128.02)||213.71 (121.46)||233.08 (174.74)||283.84 (269.90)||187.49 (148.91)|
| Minimum, maximum||1, 1,249||1, 718||1, 762||1, 1,567||6, 551|
|Distribution|| || || || || |
| Duration < 4 weeks||180 (7.43)||180 (8.07)||17 (10.76)||16 (10.53)||12 (14.12)|
| 4 < duration ≤ 12 weeks||208 (8.59)||225 (10.09)||23 (14.56)||9 (5.92)||18 (21.18)|
| 12 < duration ≤ 16 weeks||134 (5.53)||125 (5.61)||10 (6.33)||16 (10.53)||8 (9.41)|
| 16 < duration ≤ 24 weeks||363 (14.99)||350 (15.70)||20 (12.66)||18 (11.84)||10 (11.76)|
| 24 < duration ≤ 40 weeks||599 (24.73)||558 (25.02)||34 (21.52)||32 (21.05)||11 (12.94)|
| 40 < duration ≤ 48 weeks||338 (13.96)||270 (12.11)||12 (7.59)||16 (10.53)||5 (5.88)|
| Duration > 48 weeks||600 (24.77)||522 (23.41)||42 (26.58)||45 (29.61)||21 (24.71)|
|MPR‡ at first 24 weeks (standard deviation)||0.76 (0.27)||0.74 (0.28)||0.66 (0.32)||0.67 (0.32)||0.57 (0.32)|
|% adherent (MPR ≥ 0.80)||1,457 (60.16)||1,326 (59.46)||72 (46.20)||71 (46.71)||31 (36.47)|
|% continuing treatment beyond 24 weeks||1,676 (0.69)||1,461 (0.66)||91 (0.58)||91 (0.60)||44 (0.52)|
|MPR at first 48 weeks among patients continuing treatment beyond 24 weeks (standard deviation)||0.73 (0.22)||0.73 (0.22)||0.75 (0.22)||0.73 (0.22)||0.70 (0.26)|
The patients on PEG2A + R treatment had nearly eight PEG2A prescriptions on average during their treatment period and nearly the same number of ribavirin prescriptions. The average cost of PEG2A prescriptions was $12,656, while ribavirin prescriptions cost $9551, resulting in a total therapy cost of $22,207. The average cost of PEG2B + R prescriptions was slightly lower, at $21,681. The patients had fewer prescriptions and lower costs for IFN + R therapy compared with the PEG2A + R and PEG2B + R combination therapies. The patients on monotherapies incurred the lowest prescription costs because of lack of the ribavirin component. The patients received PEG2A + R therapy for approximately 224 days (32 weeks). The average overall duration for PEG2B + R was slightly lower at 214 days. The average duration of IFN + R was 187 days. The patients on PEG2A and PEG2B monotherapies received treatment for longer periods (233 and 283 days, respectively).
The mean MPR for the patients on PEG2A + R and PEG2B + R was similar at 0.76 and 0.74, respectively. MPR was the lowest for the patients on IFN + R (0.57). Approximately 60% of the patients on PEG2A + R and PEG2B + R were adherent to therapy. Adherence rates were lower for PEG2A and PEG2B monotherapies (46% and 47%, respectively) and for IFN + R combination therapy (36%). Among the patients who received treatment beyond 24 weeks, the average 48-week MPR for PEG2A + R and PEG2B + R was slightly lower, at 0.73.
Treatment Patterns and Adherence by Clinical Characteristics
Table 5 presents treatment patterns of the patients who initiated combination peginterferon regimens (PEG2A + R or PEG2B + R) stratified by genotype, SVR attainment, and disease severity. As anticipated, the patients infected with genotype 1 virus had longer treatment duration (237 days) compared with 161 days for the patients infected with genotype 2/3 virus. As a result, those with genotype 1 had twice the number of prescriptions on average and incurred higher prescription costs ($23,666 vs. $14,979). MPR at week 24 was slightly higher for the patients with genotype 1 compared with the patients with genotype 2/3. The percentage of the patients who were adherent was the same across both genotype categories.
Table 5. Peginterferon and ribavirin combination therapies, stratified by clinical characteristics
|Mean number of prescriptions obtained (standard deviation)|| || || || || || || || |
| Peginterferon prescriptions||7.84 (4.73)||8.42 (4.83)||4.82 (2.13)||8.05 (4.24)||7.93 (5.10)||8.64 (4.81)||8.44 (5.15)||8.08 (5.41)|
| Ribavirin prescriptions||7.55 (4.71)||7.86 (4.12)||4.61 (1.85)||8.08 (4.60)||8.71 (6.90)||8.58 (5.20)||8.23 (5.02)||7.70 (5.08)|
|Mean total cost in $ (standard deviation)|| || || || || || || || |
| Peginterferon costs||12,377 (7,171)||13,712 (6,902)||8,789 (4,412)||12,760 (6,128)||11,983 (7,694)||13,356 (7,142)||13,058 (7,522)||12,446 (8,214)|
| Ribavirin costs||9,572 (5,942)||9,954 (4,841)||6,190 (3,454)||10,121 (5,774)||10,849 (8,872)||10,464 (6,557)||10,081 (6,165)||9,390 (6,413)|
|Duration (days) of therapy|| || || || || || || || |
| Mean (standard deviation)||219.17 (125.02)||237.35 (130.00)||160.5 (74.62)||220.99 (115.98)||193.77 (127.00)||250.78 (131.78)||233.33 (143.95)||205.85 (134.49)|
| Minimum||1, 1,249||8, 511||26, 372||10, 708||1, 601||1, 1,249||28, 1,184||28, 541|
|Maximum|| || || || || || || || |
|Distribution|| || || || || || || || |
| Duration < 4 weeks||360 (7.74)||7 (9.86)||3 (10.71)||24 (7.52)||27 (12.22)||46 (6.21)||17 (8.29)||3 (7.50)|
| 4 < duration ≤ 16 weeks||692 (14.88)||8 (11.27)||2 (7.14)||35 (10.97)||37 (16.74)||88 (11.88)||25 (12.20)||9 (22.50)|
| 16 < duration ≤ 24 weeks||713 (15.33)||8 (11.27)||10 (35.71)||59 (18.50)||42 (19.00)||109 (14.71)||28 (13.66)||6 (15.00)|
| 24 < duration ≤ 40 weeks||1,157 (24.87)||13 (18.31)||11 (39.29)||78 (24.45)||53 (23.98)||168 (22.67)||53 (25.85)||7 (17.50)|
| 40 < duration ≤ 48 weeks||608 (13.07)||11 (15.49)||1 (3.57)||53 (16.61)||19 (8.60)||108 (14.57)||28 (13.66)||5 (12.50)|
| Duration > 48 weeks||1,122 (24.12)||24 (33.80)||1 (3.57)||70 (21.94)||43 (19.46)||222 (29.96)||54 (26.34)||10 (25.00)|
| MPR‡ at first 24 weeks (standard deviation)||0.75 (0.28)||0.77 (0.26)||0.74 (0.27)||0.79 (0.26)||0.69 (0.30)||0.82 (0.22)||0.76 (0.28)||0.59 (0.29)|
| % adherent at 24 weeks (MPR ≥ 0.80)||2,783 (59.82)||43 (60.56)||17 (60.71)||206 (64.58)||116 (52.49)||479 (64.64)||128 (62.44)||20 (50.00)|
| % continuing treatment beyond 24 weeks||3,137 (0.67)||52 (0.73)||13 (0.46)||229 (0.72)||131 (0.59)||528 (0.71)||146 (0.71)||26 (0.65)|
| MPR at first 48 weeks among patients continuing treatment beyond 24 weeks (standard deviation)||0.73 (0.22)||0.78 (0.19)||0.57 (0.20)||0.71 (0.24)||0.70 (0.22)||0.76 (0.21)||0.73 (0.22)||0.68 (0.26)|
When stratified by SVR attainment status, the number of prescriptions, treatment costs, and duration of treatment were slightly higher for those patients who attained SVR. The patients who attained SVR had a significantly higher MPR compared with those patients who did not attain SVR (0.79 vs. 0.69). Nearly 65% of the patients who attained SVR were adherent to therapy compared with only 52% of those who did not attain SVR. The patients who attained SVR were also more likely to continue treatment beyond 24 weeks compared with those who did not (72% vs. 59%). Among those patients who continued beyond 24 weeks, however, the rate of adherence was similar between the SVR attainers and the nonattainers.
The patients with severe disease had fewer prescriptions, lower costs, and shorter treatment duration compared with the mild and moderate patients. They also had much lower adherence rates, perhaps because of a higher rate of treatment-related adverse events. The average MPR for the patients with mild disease was 0.82, for moderate disease, MPR was 0.76, and for severe disease, MPR was 0.59. Adherence rates were 65%, 62%, and 50%, respectively. Fewer patients with severe disease received treatment beyond 24 weeks, compared with the patients with mild and moderate disease.
Adherence and Costs
Table 6 presents generalized linear model estimates of the impact of medication nonadherence on HCV-related and all-cause costs. The adherent patients had greater total HCV-related costs compared with the nonadherent patients ($20,132 vs. $12,259; P < 0.01) primarily because of higher pharmacy costs incurred from stricter refill compliance by the adherent patients ($18,963 vs. $10,232; P < 0.01). When pharmacy costs are excluded, the adherent patients incurred lower total HCV-related costs compared with the nonadherent patients ($1370 vs. $2463; P < 0.01). HCV-related inpatient costs incurred by the nonadherent patients were significantly greater than those costs incurred by the adherent patients ($13,162 vs. $8733; P < 0.01). This could be due to the fact that the nonadherent patients are more likely to experience disease progression, develop liver cirrhosis, and need expensive inpatient procedures such as liver transplants. This relationship could also be mediated by the observation of higher adherence rates among those who attained SVR: in a subanalysis conducted among the patients for whom SVR attainment was evaluated, monthly per patient costs after treatment completion in those who failed to attain SVR were roughly double ($1436) the costs incurred by the patients who achieved SVR ($717). Finally, the adherent patients incurred lower all-cause costs in all categories except physician office visits and prescription drugs. Total all-cause costs, excluding pharmacy, were $6320 for the nonadherent patients and $3693 for the adherent patients (P < 0.01).
Table 6. HCV-related and all-cause costs by 24-week adherence status
|Hospitalizations||8,733 (2,282)||13,612 (4,425)||<0.0001||10,920 (2,651)||17,377 (5,591)||<0.0001|
|ER visits||601 (177)||764 (213)||<0.0001||691 (113)||881 (305)||<0.0001|
|Office visits||435 (37)||320 (40)||<0.0001||1,115 (271)||1,082 (330)||<0.0001|
|Other OP/ancillary||772 (258)||835 (195)||<0.0001||1,705 (496)||2,508 (1,079)||<0.0001|
|Laboratory services||491 (78)||425 (77)||<0.0001||666 (118)||671 (207)||0.2979|
|Pharmacy||18,963 (439)||10,232 (265)||<0.0001||21,953 (812)||12,645 (1,356)||<0.0001|
|Nonpharmacy total||1,370 (420)||2,463 (1,371)||<0.0001||3,693 (1,205)||6,320 (3,545)||<0.0001|
|Total||20,132 (436)||12,259 (1,218)||<0.0001||25,651 (1,864)||18,910 (4,492)||<0.0001|
Our study is subject to several limitations inherent in most analyses of retrospective claims data. First, the patients were identified based on ICD-9-CM diagnosis codes that if recorded inaccurately, may have caused some of the patients to be misidentified as having HCV. In addition, the patients who were coded as having acute HCV, and therefore excluded from our analysis, could in fact be chronic HCV patients who were misclassified. The impact of misclassification bias stemming from analyses of ICD-9-CM codes in claims data has been described further in previous research [18,19]. The validity of our results therefore depends on the accuracy of record keeping among providers who submit claims in the Ingenix Impact database. Second, laboratory test results were available for only a small subset of the patient population. Better availability of laboratory results would provide genotype information for a larger number of patients, thereby allowing us to assess adherence over the entire treatment period of 48 weeks for the patients with genotype 1. Third, medication refill patterns in administrative data, although reflective of real-world prescription dispensing, may not reveal the true intent or directions of the prescribing physician. Failure to refill a prescription may occur for clinically appropriate reasons, such as treatment discontinuation because of medication side effects, adverse drug interactions, medication titration, or abnormal laboratory results . These factors may cause MPR-based compliance measures in claims data to understate actual adherence. Conversely, the observation of a prescription claim assumes complete use of the medication obtained. It is possible, however, that individuals may dispose of medication before refill or stockpile it for future use . These factors may cause the MPR to overstate actual adherence. In spite of these considerations, administrative prescription data remain a valid and widely accepted source for estimating adherence with chronic-use medications using validated measures such as the MPR [11,20]. Finally, our database captures only three ICD-9-CM codes (one primary code and two secondary codes) for each claim. Although we define HCV-related costs based on claims with an HCV diagnosis in the primary or either of the nonprimary positions, HCV diagnoses lower than the third position were not captured. Because HCV may still yield a contribution to the reason for and costs of a health-care encounter even at lower coding positions, our estimates of direct HCV-related costs are likely conservative.
Despite these limitations, this is the first study to document in detail real-world treatment patterns and adherence rates for key therapies for chronic HCV among a population of managed care enrollees in the United States. Our study found that peginterferon plus ribavirin combination regimens are the mainstay of HCV treatment. Combination therapy patients with genotype 1 received treatment for only 33 weeks on average compared with the expected treatment length of 48 weeks. Combination therapies are also costly, with managed care payers incurring over $20,000 per patient for treatment. Adherence to HCV treatments appears to be suboptimal, and with the exception of pharmacy costs, nonadherence has a significant negative impact on disease-specific and all-cause medical costs. Inpatient costs are the key drivers of increased costs among nonadherent patients. New therapies with greater efficacy or those that promote better adherence and impose a lower cost burden on patients and payers might help in improving SVR.
The authors would like to thank Letha Healey and Mani Subramanian of Human Genome Sciences Inc. for their assistance with and contributions to this article.
Source of financial support: This study and the writing and preparation of this manuscript were funded in full by Human Genome Sciences Inc. and Novartis Pharma AG. All authors had complete access to the data that support this manuscript. Debanjali Mitra and Keith L. Davis are employees of RTI Health Solutions. At the time this manuscript was written, Jasmina Medjedovic was an employee of Novartis Pharma AG. Cynthia Beam is an employee of Human Genome Sciences Inc. Vinod Rustgi is an employee of Georgetown University Medical Center and served as a consultant to Human Genome Sciences, Inc.