Association of Direct-Acting Antiviral Therapy for Hepatitis C With After-Treatment Costs Among Medicare Beneficiaries

This cohort study analyzes the cost implications for Medicare of direct-acting antiviral medications used by beneficiaries with or without hepatitis C–related liver disease.


Introduction
Hepatitis C is an important public health problem in the United States for several reasons. It is a leading cause of serious and costly liver diseases, such as cirrhosis and liver cancer. 1 The financial burden it imposes is estimated to exceed $10 billion annually. 2 Hepatitis C was associated with more deaths than any other infectious disease in 2014 and 2015. 3,4 The introduction of highly effective direct-acting antiviral (DAA) therapy has provided an unprecedented opportunity to address hepatitis C. [5][6][7][8] However, high prices of DAA drugs have led third-party payers to restrict coverage for these medications while demanding more information on cost implications and the value of new treatments. 9-12 Although the prices of DAA drugs have decreased as more new drugs have entered the market, they are still high, and coverage by some payers remains restrictive. 12,13 Previous studies have evaluated the cost-effectiveness of DAA therapy compared with traditional peginterferon therapy and generally reported that DAA drugs would be an optimal treatment despite their high prices because they could reduce downstream medical costs and/or extend lives with their high cure rates. [14][15][16][17] However, in calculating posttreatment medical costs, those studies considered only the estimated costs associated with the expected hepatitis C progression (eg, costs of cirrhosis and hepatocellular carcinoma) probably because they lacked data on actual posttherapy total medical costs. 14,18,19 Patients with hepatitis C infection have several comorbidities because of the extrahepatic manifestation of hepatitis C, [20][21][22][23] and patients need medical care to manage those conditions even after the infection is cured. Information on total medical costs is crucial for assessing the cost-effectiveness or value of a new treatment, but evidence on actual cost implications after DAA therapy is scarce.
A few recent studies examined total medical costs 1 year after DAA therapy using data from patients with Medicaid coverage or private insurance. 24,25 One study found that medical costs 1 year after DAA treatment decreased from $14 014 to $12 327 among Medicaid beneficiaries 24 ; however, this work did not have a control or untreated group. Another study of commercial enrollees with cirrhosis found that total medical cost 1 year after DAA treatment was $19 300 compared with $33 039 in the untreated group; however, that study did not account for the difference in pretreatment costs, which were $17 537 in patients who used DAA therapy and $23 798 in those who did not use it. 25 To our knowledge, no research has been conducted to analyze the follow-up costs more than 1 year after DAA treatment. In addition, no study has examined the cost implications for Medicare, in which many baby boomers (the group with the highest prevalence of hepatitis C) are enrolled. 1,26 Medicare also covers younger adults with disabilities and low income, the other group with a high prevalence of hepatitis C. 9 Medicare is thus among the largest payers of DAA therapy. 27,28 In the present study, we examined how DAA treatment was associated with medical costs by following up patients with Medicare for up to 30 months after they completed DAA treatment. The goal of this study was to offer the first evidence of DAA cost implications for Medicare beyond a 1-year follow-up. administrative records. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Data were analyzed between September 1, 2019, and March 31, 2020.

Data Sources
The primary data were obtained from Medicare claims files for 2013 to 2017. To identify medical care use and costs, we analyzed data for these claim types: inpatient, skilled nursing facility, hospital outpatient, and carrier (noninstitutional, such as physician office) services. The 2013 files were used only to ensure no hepatitis C claims were made before 2014. We analyzed Medicare Part D data to identify DAA therapy use and costs. Master Beneficiary Summary Files provided data on demographics and health risk factors for each beneficiary.

Study Population
The study population comprised Medicare beneficiaries who had a recent (in 2014) diagnosis of hepatitis C after a 1-year washout period. We identified patients with hepatitis C using the standard algorithm of the Centers for Medicare and Medicaid Services (details in eAppendix 1 in the Supplement). 29 In addition, patients had to be enrolled in fee-for-service Medicare (ie, 0 months in Part C) and Part D during the study period (January 1, 2014, to December 31, 2017).
From the total population, we selected the study sample as shown in Figure 1. Patients who died within 6 months after diagnosis were excluded; DAA drugs were unlikely to be prescribed for those patients because a course of DAA therapy lasted 3 to 6 months. We identified the treatment group as patients who used 1 of the following DAA drugs: elbasvir and grazoprevir; ledipasvir and sofosbuvir; and ombitasvir, paritaprevir, and ritonavir plus dasabuvir, sofosbuvir, or sofosbuvir and velpatasvir. We assigned an index date to each patient as the first date of DAA prescription fills.
Patients with an index date on or after January 1, 2017, were excluded to allow all patients to have 6

JAMA Network Open | Health Policy
Association of Direct-Acting Antivirals for Hepatitis C With After-Treatment Medicare Costs months of DAA therapy and at least one 6-month follow-up period after treatment. We included patients who completed DAA therapy, which was defined as filling prescriptions for the expected duration of the drug, as identified in package inserts or randomized clinical trials (eAppendix 2 in the Supplement). We excluded patients with an index date within 6 months after diagnosis because those patients did not have pretreatment costs related to hepatitis C or liver disease-a key study variable.
The comparison (control) group comprised patients who did not use DAA therapy during the study period. Those who were treated with non-DAA hepatitis C treatments were excluded (N = 69; 0.02%). We selected comparison patients using time-to-treatment distribution matching. 30 We assigned a hypothetical treatment date to each DAA-untreated patient to make the distribution of time from diagnosis to treatment the same as that of the treatment group. Patients in the control group had to have 6 months before the assigned hypothetical treatment date and at least one 6-month follow-up after the hypothetical treatment.
We used propensity score matching to balance patient characteristics (demographics and health risks) during the treatment period and pretreatment spending. We matched patients with cirrhosis and those without cirrhosis separately. Matching in each group was performed with patients who had the same time-to-treatment range. The definitions and data sources of all study covariates are included in eTable 1 in the Supplement, and eTables 2 and 3 in the Supplement present results from the logistic regression to calculate propensity scores.

Follow-up Periods
We defined the pretreatment period as 6 months before the index date (eFigure in the Supplement).
The treatment period was the first 6 months after the index date. To assess medical costs after the completion of DAA therapy, we excluded records from the treatment period. We followed up patients up to 30 months after the treatment period. We required that all patients have follow-up in the first 6 months after treatment. Subsequent follow-up periods included only patients who were alive in each period to incur medical costs over the full 6 months. To perform a sensitivity analysis, we created a separate sample of patients who were alive throughout the study period.

Outcomes
We constructed 2 outcomes: hepatitis C or liver disease-related medical costs and total medical costs at 6-month intervals from the index date. We measured costs by Medicare-allowed payments, which included Medicare reimbursements, patient responsibilities, and any third-party payments (eTable 1 in the Supplement). We identified costs of inpatient, skilled nursing facility, emergency department, outpatient, and carrier services, but we did not include outpatient drug costs. All costs were measured in 2017 dollars, using the Consumer Price Index to adjust for inflation.
Total DAA costs were calculated as the sum of payer and patient responsibilities to enable their comparison with follow-up medical costs. The drug costs reported in the Medicare Part D data do not include manufacturer rebates or discounts. However, those rebates are accounted for in the Medicare payments to Part D plans. Thus, we discounted total DAA costs by 22%, the average discount on total drug costs for all brand-name Part D drugs in 2016. 31

Statistical Analysis
We used multivariate regression in a propensity score-matched cohort at the person-period level.
The regression model controlled for patient characteristics, which may change over follow-up periods after matching. We compared posttreatment changes in medical costs between the treatment group (used DAA therapy) and control group (did not use DAA therapy), a difference-indifferences approach. eAppendix 3 in the Supplement provides details of the regression and describes the variables used in the model. Among those variables were indicators that controlled for a common time trend over the follow-up periods and an indicator of DAA therapy use that controlled for permanent differences between the treatment and control groups. Interaction terms between

JAMA Network Open | Health Policy
Association of Direct-Acting Antivirals for Hepatitis C With After-Treatment Medicare Costs the time trend indicators and the DAA therapy use indicator were the variables of interest. Their coefficients measured the differences in changes in medical costs over time between the 2 groups.
The sum of these coefficients represented the cumulative cost change during the study period.
We estimated a generalized linear model with a log link and the gamma distribution to account for the skewed distribution of the cost variables. 32,33 We estimated the model separately for patients with and those without cirrhosis. This separate analysis approach allowed heterogeneity in the associations of all covariates with follow-up costs between the 2 groups. We clustered SEs at the patient level.
A 2-tailed P < .05 was considered statistically significant. We compared patient characteristics between the treatment and control groups using unpaired, 2-tailed t tests for binary and continuous variables and with χ 2 tests for categorical variables. We used SAS (version 9.4; SAS Institute Inc) and Stata (version 15; StataCorp LLC) for the analysis.

Additional Analyses
First, we performed a sensitivity analysis with patients who were alive throughout the study period to account for potential differences in mortality between the treatment and control groups. Second, we examined the association of DAA use with follow-up costs by time from hepatitis C diagnosis to treatment. We grouped patients who used DAA therapy into 2 categories of time to DAA therapy initiation: within 6 to 12 months and after 12 months of diagnosis. We compared changes in medical costs between patients in each of the 2 categories and patients in the control group. The model controlled for common time trends, permanent differences between patients with early and delayed treatments, and patient characteristics.

Descriptive Results
The primary analysis included a propensity score-matched cohort of 15 198   Among patients with cirrhosis, the mean 6-month hepatitis C or liver disease-related costs per patient who used DAA therapy decreased from $3422 before treatment to $2511 during the first 6 months after treatment (P < .001). Among patients without cirrhosis, the mean 6-month hepatitis C or liver disease-related costs per patient who used DAA therapy decreased from $879 before treatment to $311 after treatment (P < .001). Control patients showed a slight increase in hepatitis C or liver disease-related costs during the same period.
Among patients with cirrhosis, the mean 6-month total costs per patient who used DAA therapy decreased from $12 665 before treatment to $11 086 during 6 to 12 months after treatment (P < .001). Patients in the control group showed a slight increase in total costs during the 6-to 12-month follow-up period but a decrease from $15 200 to $13 135 (P < .001) during the 12-to 24-month follow-up period. Among patients without cirrhosis, the mean 6-month total costs per patient who used DAA therapy decreased from $7911 to $7881 (P < .001) during the 6-to 12-month followup period.

JAMA Network Open | Health Policy
Association of Direct-Acting Antivirals for Hepatitis C With After-Treatment Medicare Costs the first 6 months after treatment among those who used DAA therapy compared with nonusers and by $1654 (95% CI, −$2689 to −$620) during the 6 to 12 months after treatment (

Drug Costs
The mean (SD) DAA drug cost after 22% discounts were applied was $102 634 ($44 184) per patient with cirrhosis who used DAA therapy. The mean (SD) cost was $80 898 ($27 526) per patient without cirrhosis who used DAA therapy (eTable 5 in the Supplement).

Additional Analyses
The sensitivity analysis with patients who were alive throughout the study period used a propensity   Results from this sensitivity analysis (eTable 6 in the Supplement) were similar to the primary analysis, confirming that differences in mortality between the treatment and control groups did not affect the study findings.
For the time-to-treatment analysis, the mean (SD) time from hepatitis C diagnosis to DAA initiation was 16 (7.2) months, with 2924 patients (38%) treated within 6 to 12 months and 4675 patients (62%) after 12 months of diagnosis. Patients with delayed treatment incurred lower hepatitis C or liver disease-related costs at the time of diagnosis ($1380) but higher hepatitis C or liver disease-related costs for 6 months before treatment ($2274) compared with those with early initiation of DAA therapy (eTable 7 in the Supplement).
Time to DAA therapy initiation was associated with the degree of cost reduction (Figure 3 and

Discussion
In analyzing claims data from Medicare patients with hepatitis C, we found that DAA therapy was associated with decreases in posttreatment hepatitis C or liver disease-related costs and total medical costs. The reduction in hepatitis C or liver disease-related costs lasted for 30 months after treatment, but the reduction in total medical costs lasted for only 12 months.  The short duration of the reduction in total medical costs is an important extension of previous work, which was limited to the finding based on 1-year follow-up. 24,25 This finding suggests that the costs of other conditions eventually outweigh the reduction in hepatitis C or liver disease-related costs in patients who used DAA therapy and needed medical care after the hepatitis C infection was cured. Previous cost-effectiveness studies of DAA drugs lacked data on actual cost changes after DAA therapy and considered only estimated hepatitis C or liver disease-related costs. 14,18,19 Findings of the present study suggest that total medical costs, including costs of other conditions, should be considered in assessing cost implications of DAA medications.
The estimate of 1-year reduction in hepatitis C or liver disease-related costs in the present study (approximately $4984) is less than the $13 739 cost reduction for commercial insurance enrollees reported in previous work. 25 The reason for this difference may be that we compared changes in posttreatment costs between the treatment and control groups while controlling for differences in their pretreatment costs. The prior study compared costs in the posttreatment period only, even though the mean pretreatment costs in the treatment group were lower by $6261. 25 Another explanation may be that Medicare covers older adults compared with commercial insurance. Thus, the health risks in the present sample may have been different from the health risks of commercial enrollees.
The association of DAA therapy with hepatitis C or liver disease-related cost reductions was larger in patients with cirrhosis than in patients without cirrhosis during the 30 months of follow-up. This finding suggests that DAA therapy helps mitigate the immediate aggravation of the hepatitis C condition in patients with advanced disease progression. This understanding seems consistent with the approach by payers to give treatment priority to patients with advanced fibrosis, particularly in the early years of DAA therapy availability. 12

Limitations
This study has several limitations. First, the study did not use clinical information, such as viral load, genotype, and fibrosis stage, which are not recorded in Medicare claims. If those unobserved factors were associated with costs and their distribution was different in the treatment and control groups, selection bias (differences in patient characteristics between groups) may remain even with propensity score matching. Second, the study did not examine life-years saved after DAA therapy because analyzing mortality requires different approaches. However, the analysis with survivors suggested that the study results were not associated with mortality. Third, the data did not include

Conclusions
In this cohort study, DAA therapy appeared to be associated with decreases in hepatitis C or liver disease-related costs for 30 months after treatment, with a larger decrease found among patients with cirrhosis than among patients without cirrhosis. The decrease in total medical costs lasted for only 12 months after DAA therapy in both patients with cirrhosis and without cirrhosis. Studies with longer-term follow-up periods and with diverse outcomes are necessary to assess the value of DAA therapy.

JAMA Network Open | Health Policy
Association of Direct-Acting Antivirals for Hepatitis C With After-Treatment Medicare Costs