Association of copayments with healthcare utilization and expenditures among Medicaid enrollees with a substance use disorder.

BACKGROUND: The purpose of this study was to examine the association between copayments and healthcare utilization and expenditures among Medicaid enrollees with substance use disorders. METHODS: This study used claims data (2020-2021) from a private insurer participating in Arkansas's Medicaid expansion. We compared service utilization and expenditures for enrollees in different Medicaid program structures with varying copayments. Enrollees with incomes above 100 % FPL (N = 10,240) had copayments for substance use treatment services while enrollees below 100 % FPL (N = 2478) did not. Demographic, diagnostic, utilization, and cost information came from claims and enrollment information. The study identified substance use and clinical comorbidities using claims from July through December 2020 and evaluated utilization and costs in 2021. Generalized linear models (GLM) estimated outcomes using single equation and two-part modeling. A gamma distribution and log link were used to model expenditures, and negative binomial models were used to model utilization. A falsification test comparing behavioral health telemedicine utilization, which had no cost sharing in either group, assessed whether differences in the groups may be responsible for observed findings. RESULTS: Substance use enrollees with copayments were less likely to have a substance use or behavioral health outpatient (-0.04 PP adjusted; p = 0.001) or inpatient visit (-0.04 PP; p = 0.001) relative to their counterparts without copayments, equal to a 17 % reduction in substance use or behavioral health outpatient services and a nearly 50 % reduction in inpatient visits. The reduced utilization among enrollees with a copayment was associated with a significant reduction in total expenses ($954; p = 0.001) and expenses related to substance use or behavioral health services ($532; p = 0.001). For enrollees with at least one behavioral health visit, there were no differences in outpatient or inpatient utilization or expenditures between enrollees with and without copayments. Copayments had no association with non-behavioral health or telemedicine services where neither group had cost sharing. CONCLUSION: Copayments serve as an initial barrier to substance use treatment, but are not associated with the amount of healthcare utilization conditional on using services. Policy makers and insurers should consider the role of copayments for treatment services among enrollees with substance use disorders in Medicaid programs.

The Increasing Adoption of Out-of-Pocket Cost Caps: Benefits, Unintended Consequences, and Policy Opportunities.

This Viewpoint discusses potential benefits and unintended consequences of out-of-pocket cost caps in Medicare and the employer-sponsored health insurance market and provides suggested policy opportunities to address shortcomings.

Insulin Fills by Medicare Enrollees and Out-of-Pocket Caps Under the Inflation Reduction Act.

This study uses data from IQVIA's National Prescription Audit to assess the association of the Inflation Reduction Act's cap on cost sharing with insulin fills by Medicare beneficiaries.

The costs and inequities of precision medicine for patients with prostate cancer: A call to action.

Financial toxicity is a growing problem in the delivery of cancer care and contributes to inequities in outcomes across the cancer care continuum. Racial/ethnic inequities in prostate cancer, the most common cancer diagnosed in men, are well described, and threaten to widen in the era of precision oncology given the numerous structural barriers to accessing novel diagnostic studies and treatments, particularly for Black men. Gaps in insurance coverage and cost sharing are 2 such structural barriers that can perpetuate inequities in screening, diagnostic workup, guideline-concordant treatment, symptom management, survivorship, and access to clinical trials. Mitigating these barriers will be key to achieving equity in prostate cancer care, and will require a multi-pronged approach from policymakers, health systems, and individual providers. This narrative review will describe the current state of financial toxicity in prostate cancer care and its role in perpetuating racial inequities in the era of precision oncology.

Does Insulin Price Reduction Suggest a Path Forward for Lowering Drug Costs?

This Viewpoint investigates whether political pressure such as that used to decrease insulin pricing is a feasible method of lowering the cost of other drugs.

Cost sharing: implications of a well-intended benefits strategy.

DISCLOSURES: The thoughts and opinions expressed in this article are those of the author only and are not the thoughts and opinions of any current or former employer of the author. Nor is this publication made by, on behalf of, or endorsed or approved by any current or former employer of the author.

Paying patients to use lower-priced providers.

OBJECTIVE: Many employers have introduced rewards programs as a new benefit design in which employees are paid $25-$500 if they receive care from lower-priced providers. Our goal was to assess the impact of the rewards program on procedure prices and choice of provider and how these outcomes vary by length of exposure to the program and patient population. STUDY SETTING: A total of 87 employers from across the nation with 563,000 employees and dependents who have introduced the rewards program in 2017 and 2018. STUDY DESIGN: Difference-in-difference analysis comparing changes in average prices and market share of lower-priced providers among employers who introduced the reward program to those that did not. DATA COLLECTION METHODS: We used claims data for 3.9 million enrollees of a large health plan. PRINCIPAL FINDINGS: Introduction of the program was associated with a 1.3% reduction in prices during the first year and a 3.7% reduction in the second year of access. Use of the program and price reductions are concentrated among magnetic resonance imaging (MRI) services, for which 30% of patients engaged with the program, 5.6% of patients received an incentive payment in the first year, and 7.8% received an incentive payment in the second year. MRI prices were 3.7% and 6.5% lower in the first and second years, respectively. We did not observe differential impacts related to enrollment in a consumer-directed health plan or the degree of market-level price variation. We also did not observe a change in utilization. CONCLUSIONS: The introduction of financial incentives to reward patients from receiving care from lower-priced providers is associated with modest price reductions, and savings are concentrated among MRI services.

Accuracy of Physician Estimates of Out-of-Pocket Costs for Medication Filling.

Importance: One-third of US residents have trouble paying their medical bills. They often turn to their physicians for help navigating health costs and insurance coverage. Objective: To determine whether physicians can accurately estimate out-of-pocket expenses when they are given all of the necessary information about a drug's price and a patient's insurance plan. Design, Setting, and Participants: This national mail-in survey used a random sample of US physicians. The survey was sent to 900 outpatient physicians (300 each of primary care, gastroenterology, and rheumatology). Physicians were excluded if they were in training, worked primarily for the Veterans Administration or Indian Health Service, were retired, or reported 0% outpatient clinical effort. Analyses were performed from July to December 2020. Main Outcomes and Measures: In a hypothetical vignette, a patient was prescribed a new drug costing $1000/month without insurance. A summary of her private insurance information was provided, including the plan's deductible, coinsurance rates, copays, and out-of-pocket maximum. Physicians were asked to estimate the drug's out-of-pocket cost at 4 time points between January and December, using the plan's 4 types of cost-sharing: (1) deductibles, (2) coinsurance, (3) copays, and (4) out-of-pocket maximums. Multivariate linear regression was used to assess differences in performance by specialty, adjusting for attitudes toward cost conversations, demographics, and clinical characteristics. Results: The response rate was 45% (405 of 900) and 371 respondents met inclusion criteria. Among the respondents included in this study, 59% (n = 220) identified as male, 23% (n = 84) as Asian, 3% (n = 12) as Black, 6% (n = 24) as Hispanic, and 58% (n = 216) as White; 30% (n = 112) were primary care physicians, 35% (n = 128) were gastroenterologists, and 35% (n = 131) were rheumatologists; and the mean (SD) age was 49 (10) years. Overall, 52% of physicians (n = 192) accurately estimated costs before the deductible was met, 62% (n = 228) accurately used coinsurance information, 61% (n = 224) accurately used copay information, and 57% (n = 210) accurately estimated costs once the out-of-pocket maximum was met. Only 21% (n = 78) of physicians answered all 4 questions correctly. Ability to estimate out-of-pocket costs was not associated with specialty, attitudes toward cost conversations, or clinic characteristics. Conclusions and Relevance: This survey study found that many US physicians have difficulty estimating out-of-pocket costs, even when they have access to their patients' insurance plans. The mechanics involved in calculating real-time out-of-pocket costs are complex. These findings suggest that increased price transparency and simpler insurance cost-sharing mechanisms are needed to enable informed cost conversations at the point of prescribing.

Patient administrative burden in the US health care system.

OBJECTIVE: To assess the prevalence of patient administrative tasks and whether they are associated with delayed and/or foregone care. DATA SOURCE: March 2019 Health Reform Monitoring Survey. STUDY DESIGN: We assess the prevalence of five common patient administrative tasks-scheduling, obtaining information, prior authorizations, resolving billing issues, and resolving premium problems-and associated administrative burden, defined as delayed and/or foregone care. Using multivariate logistic models, we examined the association of demographic characteristics with odds of doing tasks and experiencing burdens. Our outcome variables were five common types of administrative tasks as well as composite measures of any task, any delayed care, any foregone care, and any burden (combined delayed/foregone), respectively. DATA COLLECTION: We developed and administered survey questions to a nationally representative sample of insured, nonelderly adults (n = 4155). PRINCIPAL FINDINGS: The survey completion rate was 62%. Seventy-three percent of respondents reported performing at least one administrative task in the past year. About one in three task-doers, or 24.4% of respondents overall, reported delayed or foregone care due to an administrative task: Adjusted for demographics, disability status had the strongest association with administrative tasks (adjusted odds ratio [OR] 2.91, p < 0.001) and burden (adjusted OR 1.66, p < 0.001). Being a woman was associated with doing administrative tasks (adjusted OR 2.19, p < 0.001). Being a college graduate was associated with performing an administrative task (adjusted OR 2.79, p < 0.001), while higher income was associated with fewer subsequent burdens (adjusted OR 0.55, p < 0.01). CONCLUSIONS: Patients frequently do administrative tasks that can create burdens resulting in delayed/foregone care. The prevalence of delayed/foregone care due to administrative tasks is comparable to similar estimates of cost-related barriers to care. Demographic disparities in burden warrant further attention. Enhancing measurement of patient administrative work and associated burdens may identify opportunities for assessing quality, value, and patient experience.

Cost-sharing With Medicaid Expansion in Michigan: Obligations and Propensity to Pay.

BACKGROUND: Six states expanding Medicaid under the Affordable Care Act have obtained waivers to incorporate cost-sharing. OBJECTIVE: We describe the magnitude and distribution of cost-sharing imposed by the Healthy Michigan Plan and enrollees' propensity to pay. RESEARCH DESIGN: Enrollees are followed for at least 18 months (6-mo baseline period for utilization and spending before receipt of first cost-sharing statement; ≥12 mo follow-up thereafter to ascertain obligations and payments). Analyses stratified by income, comparing enrollees with income less than Federal Poverty Level (FPL) who faced only utilization-based copayments and those greater than or equal to FPL who also faced premium contributions. SUBJECTS: A total of 158,322 enrollees aged 22-62 who initially enrolled during the first year of the program and remained continuously enrolled ≥18 months. RESULTS: Among those enrolled ≥18 months, 51.0% faced cost-sharing. Average quarterly invoices were $4.85 ($11.11 for those with positive invoices) for income less than FPL and $26.71 ($30.93 for those with positive invoices) for incomes greater than or equal to FPL. About half of enrollees with obligations made at least partial payments, with payments being more likely among those >100% FPL. Payment of the full obligation was highest in the initial 6 months. CONCLUSIONS: Many payment obligations go uncollected, suggesting that in a system without the threat of disenrollment, the impacts of cost-sharing may be muted. Similarly, the ability of cost-sharing to defray the program's budgetary impact may also be less than anticipated.

Effects of inadequate coverage on healthcare utilization: A regression discontinuity analysis.

ABSTRACT: To examine the impact of inadequate health insurance coverage on physician utilization among older adults using a novel quasi-experimental design in the time period following the elimination of cost sharing for most preventative services under the US Affordable Care Act of 2010.The Medical Expenditure Panel Survey full year consolidated data files for the period 2010 to 2017 were used to construct a pooled cross-sectional dataset of adults aged 60 to 70. Regression discontinuity design was used to estimate the impact of transitioning between non-Medicare and Medicare plans on use of routine office-based physician visits and emergency room visits.For the overall population, gaining access to Medicare at age 65 is associated with a higher propensity to make routine office-based visits (2.94 percentage points [pp]; P < .01) and lower out-of-pocket costs (-23.86 pp; P < .01) Similarly, disenrollment from non-Medicare insurance plans at age 66 was associated with more routine office-based visits (3.01 pp; P < .01) and less out-of-pocket costs (-8.09 pp; P < .10). However, some minority groups reported no changes in visits and out-of-pocket costs or reported an increased propensity to make emergency department visits.Enrollment into Medicare from non-Medicare insurance plans was associated with increased use of routine office-based services and lower out-of-pocket costs. However, some subgroups reported no changes in routine visits or costs or an increased propensity to make emergency department visits. These findings suggest other nonfinancial, structural barriers may exist that limit patient's ability to access routine services.

Out-of-Pocket Spending for Influenza Hospitalizations in Medicare Advantage.

INTRODUCTION: Although many Medicare Advantage plans have waived cost sharing for COVID-19 hospitalizations, these waivers are voluntary and may be temporary. To estimate the magnitude of potential patient cost sharing if waivers are not implemented or are allowed to expire, this study assesses the level and predictors of out-of-pocket spending for influenza hospitalizations in 2018 among elderly Medicare Advantage patients. METHODS: Using the Optum De-Identified Clinformatics DataMart, investigators identified Medicare Advantage patients aged ≥65 years hospitalized for influenza in 2018. For each hospitalization, out-of-pocket spending was calculated by summing deductibles, coinsurance, and copays. The mean out-of-pocket spending and the proportion of hospitalizations with out-of-pocket spending exceeding $2,500 were calculated. A 1-part generalized linear model with a log link and Poisson variance function was fitted to model out-of-pocket spending as a function of patient demographic characteristics, plan type, and hospitalization characteristics. Coefficients were converted to absolute changes in out-of-pocket spending by calculating average marginal effects. RESULTS: Among 14,278 influenza hospitalizations, the mean out-of-pocket spending was $987 (SD=$799). Out-of-pocket spending exceeded $2,500 for 3.0% of hospitalizations. The factors associated with higher out-of-pocket spending included intensive care use, greater length of stay, and enrollment in a preferred provider organization plan (average marginal effect=$634, 95% CI=$631, $636) compared with enrollment in an HMO plan. CONCLUSIONS: In this analysis of elderly Medicare Advantage patients, the mean out-of-pocket spending for influenza hospitalizations was almost $1,000. Federal policymakers should consider passing legislation mandating insurers to eliminate cost sharing for COVID-19 hospitalizations. Insurers with existing cost-sharing waivers should consider extending them indefinitely, and those without such waivers should consider implementing them immediately.

Impact of the Affordable Care Act Insurance Marketplaces on Out-of-Pocket Spending Among Surgical Patients.

OBJECTIVE: To evaluate the association between the introduction of the Affordable Care Act (ACA) Health Insurance Marketplaces ("Marketplaces") and financial protection for patients undergoing surgery. BACKGROUND: The ACA established Marketplaces through which individuals could purchase subsidized insurance coverage. However, the effect of these Marketplaces on surgical patients' healthcare spending remains largely unknown. METHODS: We analyzed a nationally representative sample of adults aged 19-64 who underwent surgery in 2010-2017, using the Medical Expenditure Panel Survey. Low-income patients eligible for cost-sharing and premium subsidies in the Marketplaces [income 139%-250% federal poverty level (FPL)] and middle-income patients eligible only for premium subsidies (251%-400% FPL) were compared to high-income controls ineligible for subsidies (>400% FPL) using a quasi-experimental difference-in-differences approach. We evaluated 3 main outcomes: (1) out-of-pocket spending, (2) premium contributions, and (3) likelihood of experiencing catastrophic expenditures, defined as out-of-pocket plus premium spending exceeding 19.5% of family income. RESULTS: Our sample included 5450 patients undergoing surgery, representing approximately 69 million US adults. Among low-income patients, Marketplace implementation was associated with $601 lower [95% confidence interval (CI): -$1169 to -$33; P = 0.04) out-of-pocket spending; $968 lower (95% CI: -$1652 to -$285; P = 0.006) premium spending; and 34.6% lower probability (absolute change: -8.3 percentage points; 95% CI: -14.9 to -1.7; P = 0.01) of catastrophic expenditures. We found no evidence that health expenditures changed for middle-income surgical patients. CONCLUSIONS: The ACA's insurance Marketplaces were associated with improved financial protection among low-income surgical patients eligible for both cost-sharing and premium subsidies, but not in middle-income patients eligible for only premium subsidies.

The costs of treating and not treating patients with chronic myeloid leukemia with tyrosine kinase inhibitors among Medicare patients in the United States.

BACKGROUND: Patients with high cost-sharing of tyrosine kinase inhibitors (TKIs) experience delays in treatment for chronic myeloid leukemia (CML). To the authors' knowledge, the clinical outcomes among and costs for patients not receiving TKIs are not well defined. METHODS: Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, the authors evaluated differences in TKI initiation, health care use, cost, and survival among patients with CML with continuous Medicare Parts A and B and Part D coverage who were diagnosed between 2007 and 2015. RESULTS: A total of 941 patients were included. Approximately 29% of all patients did not initiate treatment with TKIs within 6 months (non-TKI users), and had lower rates of BCR-ABL testing and more hospitalizations compared with TKI users. Approximately 21% were not found to have any TKI claims at any time. TKI initiation rates within 6 months of diagnosis increased for all patients over time (61% to 85%), with greater improvements observed in patients receiving subsidies (55% to 90%). Total Medicare costs were greater in patients treated with TKIs, with approximately 50% because of TKI costs. Non-TKI users had more inpatient costs compared with TKI users. Trends in cost remained significant when adjusting for age and comorbidities. The median overall survival was 40 months (95% confidence interval [95% CI], 34-48 months) compared with 86 months (95% CI, 73 months to not reached), respectively, for non-TKI users versus TKI users, a finding that remained consistent when adjusting for age, comorbidities, and subsidy status (hazard ratio, 2.23; 95% CI, 1.77-2.81). CONCLUSIONS: Approximately 21% of all patients with CML did not receive TKIs at any time. Cost-sharing subsidies consistently are found to be associated with higher initiation rates. Non-TKI users had higher inpatient costs and poorer survival outcomes. Interventions to lower TKI costs for all patients are desirable.

Impact of co-payment level increase of antidiabetic medications on glycaemic control: an interrupted time-series study among Finnish patients with type 2 diabetes.

BACKGROUND: A new special reimbursement scheme (SRS) for non-insulin medications used for treatment of hyperglycaemia in type 2 diabetes (T2D) was implemented in Finland on January 1, 2017. The new SRS affected all community-dwelling Finnish T2D patients as all community-dwelling residents are eligible for reimbursement for prescription medications. The aim of the study was to evaluate the impact of this co-payment increase on glycaemic control among Finnish T2D patients. METHODS: Data on glycaemic control were collected with HbA1c measures from electronic health records from primary health care and specialized care in the North Karelia region, Finland, from patients with a confirmed T2D diagnosis in 2012 who were alive on January 1, 2017 (n = 8436). Average HbA1c levels were measured monthly 36 months before and 33 months after the policy change. Consumption of diabetes medications was measured with defined daily doses (DDDs) based on reimbursed medication purchases. Interrupted time series design analysed with segmented regression model was applied to examine the effect of the policy change on average HbA1c levels. RESULTS: Eight thousand one hundred forty-three T2D patients had at least one HbA1c measurement within 01/2014-9/2019. Mean age of the patients was 68.1 (SD 11.3) years and 53.0% were women. Average time since T2D diagnosis was 11.5 (SD 6.1) years. An estimated increase of 0.81 (95% confidence interval, CI, 0.04-1.58) mmol/mol in average HbA1c levels was detected at the time of the policy change. In subgroup analyses, strongest effects were detected among patients who used only other diabetes medications than insulin or metformin in 2016 (3.56 mmol/mol, 95% CI 2.50-4.62). Meanwhile, yearly consumption of diabetes medications decreased slightly from 618.9 (SD 487.8) DDDs/patient in 2016 to 602.9 (SD 475.6) DDDs/patient in 2017 (p = 0.048). CONCLUSIONS: Simultaneously with the increase of the co-payment level, the average HbA1c level increased among T2D patients from the North Karelia region, Finland. This may be explained by the decreased consumption of diabetes medications between 2016 and 2017. Special attention should be allocated to glycaemic control of patients utilizing only other antidiabetic medications than metformin or insulin.

Carve-out plan financial requirements associated with national behavioral health parity.

OBJECTIVES: To examine changes in carve-out financial requirements (copayments, coinsurance, use of deductibles, and out-of-pocket maxima) following the Mental Health Parity and Addiction Equity Act (MHPAEA). DATA SOURCE/STUDY SETTING: Specialty mental health benefit design information for employer-sponsored carve-out plans from a national managed behavioral health organization's claims processing engine (2008-2013). STUDY DESIGN: This pre-post study reports linear and logistic regression as the main analysis. DATA COLLECTION/EXTRACTION METHODS: NA. PRINCIPAL FINDINGS: Copayments for in-network emergency room (-$44.9, 95% CI: -78.3, -11.5; preparity mean: $56.2), outpatient services (eg, individual psychotherapy: -$7.4, 95% CI: -10.5, -4.2; preparity mean: $17.8), and out-of-network coinsurance for emergency room (-11 percentage points, 95% CI: -16.7, -5.4; preparity mean: 38.8 percent) and outpatient (eg, individual psychotherapy: -5.8 percentage points, 95% CI: -10.0, -1.6; preparity mean 41.0 percent) decreased. Probability of family OOP maxima use (29 percentage points, 95% CI: 19.3, 38.6; preparity mean: 36 percent) increased. In-network outpatient coinsurance increased (eg, individual psychotherapy: 4.5 percentage points, 95% CI: 1.1, 7.9; preparity mean: 2.7 percent), as did probability of use of family deductibles (15 percentage points, 95% CI: 6.1, 23.3; preparity mean: 38 percent). CONCLUSIONS: MHPAEA was associated with increased generosity in most financial requirements observed here. However, increased use of deductibles may have reduced generosity for some patients.

Patient and Plan Spending after State Specialty-Drug Out-of-Pocket Spending Caps.

BACKGROUND: Specialty drugs are used to treat complex or life-threatening conditions, often at high financial costs to both patients and health plans. Three states - Delaware, Louisiana, and Maryland - passed legislation to cap out-of-pocket payments for specialty drugs at $150 per prescription. A concern is that these caps could shift costs to health plans, increasing insurance premiums. Estimates of the effect of the caps on patient and health-plan spending could inform future policies. METHODS: We analyzed a sample that included 27,161 persons under 65 years of age who had rheumatoid arthritis, multiple sclerosis, hepatitis C, psoriasis, psoriatic arthritis, Crohn's disease, or ulcerative colitis and who were in commercial health plans from 2011 through 2016 that were administered by three large nationwide insurers. The primary outcome was the change in out-of-pocket spending among specialty-drug users who were in the 95th percentile for spending on specialty drugs. Other outcomes were changes in mean out-of-pocket and health-plan spending for specialty drugs, nonspecialty drugs, and nondrug health care and utilization of specialty drugs. We compared outcomes in the three states that enacted caps with neighboring control states that did not, 3 years before and up to 3 years after enactment of the spending cap. RESULTS: Caps were associated with an adjusted change in out-of-pocket costs of -$351 (95% confidence interval, -554 to -148) per specialty-drug user per month, representing a 32% reduction in spending, among users in the 95th percentile of spending on specialty drugs. This finding was supported by multiple sensitivity analyses. Caps were not associated with changes in other outcomes. CONCLUSIONS: Caps for spending on specialty drugs were associated with substantial reductions in spending on specialty drugs among patients with the highest out-of-pocket costs, without detectable increases in health-plan spending, a proxy for future insurance premiums. (Funded by the Robert Wood Johnson Foundation Health Data for Action Program.).

Savings or Selection? Initial Spending Reductions in the Medicare Shared Savings Program and Considerations for Reform.

Policy Points Concerns have been raised about risk selection in the Medicare Shared Savings Program (MSSP). Specifically, turnover in accountable care organization (ACO) physicians and patient panels has led to concerns that ACOs may be earning shared-savings bonuses by selecting lower-risk patients or providers with lower-risk panels. We find no evidence that changes in ACO patient populations explain savings estimates from previous evaluations through 2015. We also find no evidence that ACOs systematically manipulated provider composition or billing to earn bonuses. The modest savings and lack of risk selection in the original MSSP design suggest opportunities to build on early progress. Recent program changes provide ACOs with more opportunity to select providers with lower-risk patients. Understanding the effect of these changes will be important for guiding future payment policy. CONTEXT: The Medicare Shared Savings Program (MSSP) establishes incentives for participating accountable care organizations (ACOs) to lower spending for their attributed fee-for-service Medicare patients. Turnover in ACO physicians and patient panels has raised concerns that ACOs may be earning shared-savings bonuses by selecting lower-risk patients or providers with lower-risk panels. METHODS: We conducted three sets of analyses of Medicare claims data. First, we estimated overall MSSP savings through 2015 using a difference-in-differences approach and methods that eliminated selection bias from ACO program exit or changes in the practices or physicians included in ACO contracts. We then checked for residual risk selection at the patient level. Second, we reestimated savings with methods that address undetected risk selection but could introduce bias from other sources. These included patient fixed effects, baseline or prospective assignment, and area-level MSSP exposure to hold patient populations constant. Third, we tested for changes in provider composition or provider billing that may have contributed to bonuses, even if they were eliminated as sources of bias in the evaluation analyses. FINDINGS: MSSP participation was associated with modest and increasing annual gross savings in the 2012-2013 entry cohorts of ACOs that reached $139 to $302 per patient by 2015. Savings in the 2014 entry cohort were small and not statistically significant. Robustness checks revealed no evidence of residual risk selection. Alternative methods to address risk selection produced results that were substantively consistent with our primary analysis but varied somewhat and were more sensitive to adjustment for patient characteristics, suggesting the introduction of bias from within-patient changes in time-varying characteristics. We found no evidence of ACO manipulation of provider composition or billing to inflate savings. Finally, larger savings for physician group ACOs were robust to consideration of differential changes in organizational structure among non-ACO providers (eg, from consolidation). CONCLUSIONS: Participation in the original MSSP program was associated with modest savings and not with favorable risk selection. These findings suggest an opportunity to build on early progress. Understanding the effect of new opportunities and incentives for risk selection in the revamped MSSP will be important for guiding future program reforms.