To our knowledge, this is the first observational study to examine the impact of Medicaid states with sobriety restrictions on HCRU and costs among patients initiating HCV DAAs. Our analysis revealed that the cohort of DAA initiators in states with sobriety restriction versus states with no sobriety restrictions had a greater proportion of patients with HCRU for the different types of healthcare settings in this analysis; a higher number of all-cause hospitalizations, ED visits, and OP visits; and greater all-cause healthcare costs for the different types of healthcare settings at both the patient level and visit level. Specifically, patients residing in sobriety restriction states had nearly a twofold and 1.5-fold greater likelihood of IP and OP visits, respectively, compared to those in no sobriety restriction states. Similarly, patients in sobriety restriction states had nearly 2.5 times, 1.8 times, and 1.5 times greater risk for higher rates of hospitalization, OP visits, and ED visits, respectively, compared to those in no sobriety restriction states. Additionally, mean per-patient healthcare costs were three times higher for medical costs, and approximately 1.4 times higher for pharmacy and overall costs for patients insobriety restriction states compared to those in no sobriety restriction states. The two study cohorts appear to be balanced overall, despite the absence of matching techniques. Age, gender, opioid use disorder, substance-related disorder, CCI (not including less mild/moderate liver disease), smoking and tobacco use, screening and counseling, and prescriber restrictions that were hypothesized to potentially act as confounders in the causal pathway were included as covariates in the regression analysis. While the CCI variables were included to account for disease severity, they may not be adequate to fully capture HCV severity. It is possible that severity may not be fully accounted for, given the absence of clinical variables that can fully capture liver severity.

While the authors are unaware of other real-world studies that have examined the cost consequences of states with sobriety restrictions regarding HCV DAA access among patients with HCV, results of this study are consistent with findings of a published Markov modeling analysis that predicted the potential for significantly lower healthcare expenditures due to expanded access and improved health outcomes with the removal of access restrictions to HCV treatment, including disease severity restrictions [17]. Historically, several state Medicaid programs have experimented with various market access and formulary restrictions to manage drug therapy for patients with chronic conditions. However, these restrictions have often had the unintended consequences of cost shifting or cost increases rather than reduction in costs. These cost shifts often may occur due to poorer clinical outcomes and higher healthcare utilization and costs across various settings [18]. In fact, patients denied access to HCV therapies may develop hepatic fibrosis, progress to cirrhosis, and have a greater risk for developing end-stage liver disease (ESLD) and extra hepatic manifestations that can significantly increase HCRU and cost burden. Indeed, the literature suggests that deferring HCV therapy increases the risk of liver-related complications and death [19].

We believe that there may be several reasons for the higher adjusted HCRU and costs in SR versus NSR cohorts. First, it is plausible that patients’ HCV disease contributing in the wake of delayed treatment initiation may be a contributing factor to higher HCRU/costs in the SR cohort. Indeed, we observed that a higher percentage of patients in SR cohort may have moderate to severe liver disease, suggesting that they may be at higher risk of developing hepatic fibrosis, progress to cirrhosis, and ESLD requiring intensive use of HCRU. While our analysis showed that sobriety restriction states and no sobriety restriction states experienced delays from HCV diagnosis to DAA initiation, the SR cohort had longer delays to DAA access vis-à-vis NSR cohort. Second, while the length of restriction delays may not play a significant role, the higher percentage of patients in SR cohort with moderate-severe liver disease at baseline may be a contributing factor. However, the HCRU/cost differences persisted even after adjusting for their baseline differences. Third, the mandatory requirement to show evidence of sobriety in sobriety restriction states (potentially by completing substance use treatment). It is possible that HCRU and costs of sobriety treatment and rehabilitation contributed to the differences between states with sobriety restriction and states with no sobriety restriction. While we did not adjust for baseline HCRU/costs given that they are in the causal pathway, we examined the interaction between baseline SUD and restriction status. Specifically, the aORs of models with and without the interaction term suggest that the overall patterns and statistical significance of the associations remain consistent for all HCRU outcomes other than professional visits. While the models without the SUD interaction term showed that IP, OP, and ED are significantly higher among SR (states with restrictions), the models with the interaction between SUD*SR suggest that all HCRU (IP, OP, ED) including PV visits are also significantly higher among patients in the SR group. Given that the interaction between SUD and SR group does not substantially modify the observed associations for overall HCRU, sobriety restrictions may be directly associated with higher HCRU in our analysis.

Our analysis is consistent with the extensive literature that has examined the impact of restrictions imposed by state Medicaid programs in several therapeutic areas. The literature consistently suggests that restrictions imposed by state Medicaid programs almost always shift the costs within the system rather than reduce it, and may have the unintended consequences of increasing HCRU/costs, due to potential deleterious clinical outcomes. Since this study was intended to study the differences in HCRU/cost differences between “any” states with sobriety restriction versus states with no sobriety restriction on access to DAAs, the study did not examine the association between length of access delays and HCRU/cost outcomes. Interestingly, this analysis indicated that patients residing in sobriety restriction states and no sobriety restriction states had access delays, with > 50% of patients in both sobriety restriction states and no sobriety restriction states receiving a first prescription for DAA after 90 days of HCV diagnosis. While patients in sobriety restriction states experienced a 2-week longer delay in access to DAAs than those in no sobriety restriction states regarding receiving their first prescription of DAA, it is unclear whether it was clinically meaningful or whether it resulted in any differences in HCRU outcomes. Therefore, future investigations may be needed to examine the association between the length of access delays and HCRU/cost outcomes. Results of this research suggest that Medicaid market access restriction policies such as sobriety restriction may not be consistently aligned with clinical evidence or current HCV treatment guidelines. It should be noted that in March 2023, the White House and the National Institutes of Health announced a National Program for HCV Elimination, aiming to address HCV through enhanced outreach, screening, and treatment. Our findings suggest that existing Medicaid coverage restrictions will continue to impede the achievement of this national goal as well as the WHO 2030 hepatitis elimination goals. While several states have been working to lift sobriety restriction for DAAs in recent years, our analysis points to important considerations regarding sobriety as well as other DAA-related Medicaid restrictions and provides opportunities for improving outcomes by addressing access barriers to elimination.

While being the first real-world study to examine the consequences of sobriety restrictions, these findings have several implications. First, by assessing the impact of sobriety restriction across different states in a case–control fashion, by comparing patients in states that implemented sobriety restriction (13 states) to those that did not (16 states), it provided robust insights into the potential impact of “sobriety restriction” and HCRU and costs. Second, our findings are generalizable to Medicaid program members, who constitute a significant proportion of patients with HCV in the US, demonstrating the impact of sobriety restriction across different states with variations in their sociodemographic composition. Third, these findings underscore important policy implications for the future about the lack of usefulness of sobriety restriction on effective medications, such as DAAs. While medications are often targeted by cost-containment programs, existing research does not support the idea that arbitrary drug access limits effectively reduce overall costs while maintaining essential care for low-income, chronically ill populations [20]. Finally, our findings also challenge the belief that changes in one area of care can have simple and isolated effects without any adverse impact on other parts of the healthcare system. In our study, these market access disruptions related to delayed treatment initiation led to expensive increases in healthcare utilization, including more IP, ED visits, OP visits, and professional consultations [20]. Moreover, sobriety restriction on HCV treatment create barriers to care, potentially accelerating disease progression, increasing HCRU, and undermining public health goals by perpetuating HCV transmission, especially among people who inject drugs [21]. Eliminating sobriety restriction would align state Medicaid policies with clinical guidelines from the American Association for the Study of Liver Diseases/Infectious Diseases Society of America, which recommend DAA therapy for all patients with chronic HCV, regardless of substance use history. As emphasized by the National Academies of Sciences, Engineering, and Medicine [22], universal DAA access is crucial for eliminating HCV as a public health threat. Therefore, state Medicaid programs should prioritize policies ensuring treatment for all HCV-diagnosed patients, as sobriety restriction and other restrictions like fibrosis levels force patients to delay treatment until their health declines, ultimately resulting in higher long-term costs and worse outcomes. Considering these recommendations, our study found that healthcare costs were higher across all settings in states that enforced sobriety restrictions. In fact, the findings from this study confirm that costs are only shifted from one part of the healthcare system to another. Thus, these market access restrictions are called into question and suggest the potential futility of implementing sobriety restriction to allow DAA access to patients. Further research could investigate the impact of DAA access restrictions beyond sobriety restriction on healthcare expenditures related to clinical outcomes (e.g., fibrosis, cirrhosis). Understanding these impacts is essential for evaluating the broader economic and clinical implications of any restrictions on the use of DAAs and sobriety restriction and guiding policy decisions to improve healthcare outcomes for individuals with chronic HCV.

While it is possible that the potential misalignment of the post-index period may result in higher costs in SR group, we believe that it is less likely for two reasons. First, we discovered that SUD treatment costs were a top driver of costs in both states with restrictions (SR) and states without restrictions (NSR). Second, we found that costs related to “cirrhosis” or “liver disease” treatment-related services were higher among patients residing in SR states versus NSR states. Put together, these results suggests that SUD costs do not drive the differences, and costs of services related to “cirrhosis” and “liver disease-related treatment” may be one of the reasons that costs were different between the cohorts. Additionally, our recent re-analysis examining the interaction between SUD and restriction status (SR) indicates that the interaction term SUD*SR is significantly associated with HCRU, especially for professional visits. While patients in states with sobriety restrictions (SR) are required to complete SUD treatment and show evidence of sobriety to gain access to DAAs, patients in states with no sobriety restriction did not have such restrictions. The high costs associated with SUD treatment in the NSR cohort also suggest that patients in no sobriety restriction states may be treated appropriately with DAAs while providing them with SUD treatment to ensure sobriety. Overall, these results may, indeed, suggest that preventing or delaying access to DAAs may be futile and can result in potentially unintended adverse clinical, HCRU, and cost consequences.

As is the case with claims-based analysis, this study is subject to limitations related to potential miscoding, under-coding, or other issues that are characteristic of claims data used for billing and reimbursement purposes. While administrative claims are excellent measures to assess HCRU/cost outcomes, they may not be able to account for patient characteristics that are unobservable in claims data such as this. Despite adjusting for key confounders (age, gender, comorbidities, and SUD), it is possible for residual confounding to exist given the absence of unobservable clinical and socioeconomic variables. To the extent that all patients in the study included the indigent Medicaid beneficiaries with similarities in their economic status, residual confounding may be minimized. While this analysis suggests that SR versus NSR cohorts had higher adjusted HCRU and costs, findings of this analysis should be interpreted with caution as they are not intended to infer causation.

The present study did not evaluate other access barriers (e.g., specialist counseling, fibrosis requirement) beyond sobriety restriction, which could also impact healthcare costs. This omission suggests that other factors influencing DAA access and healthcare utilization were not accounted for, potentially skewing the results. Additionally, for states that lifted sobriety restriction during the study period (e.g., Arizona and Texas), patients initiating DAAs after the policy changes were excluded from the SR cohort to avoid confounding. However, it is plausible that residual effects from these transitional periods may still exist. Furthermore, it is important to recognize that there may be underlying differences between the analytic samples in the states, which could impact the study outcomes. These differences might include demographic, socioeconomic, or healthcare system variations that were not controlled for, potentially confounding the results. States with screening and counseling mandates or recommendations, but without delaying access to DAA prescriptions, were not included in this study. This exclusion means that our findings may not represent the full spectrum of access barriers faced by patients with HCV across different states.

Since our patient identification period year was 2021, the first full year in the post-coronavirus disease (post-COVID) pandemic era, many states experienced different exposures and responses to COVID, potentially resulting in dramatic effects on HCRU, and consequently costs. Although state-specific COVID incidence rates by the month of index date are available from Johns Hopkins COVID dashboard, the inclusion of state-specific monthly COVID incidence was outside the scope of the current analysis. It should be noted that this analysis did not account for states that have removed sobriety restriction since 2021. Future analysis may consider the impact of differential COVID incidence rates as an additional covariate in the analysis.

Finally, the claims analysis design limited the ability to assess outcomes beyond medical care usage. Consequently, we could not directly determine if the higher HCRU and costs in SR versus NSR cohorts were driven by worsened clinical outcomes (i.e., liver functional status, fibrosis, cirrhosis) possibly due to delays in HCV DAA access in states with sobriety restriction. Nonetheless, our data provide some indication that worsened clinical outcomes among patients in states with sobriety restriction may be the reason for higher HCRU and costs.