Preventive Cardiovascular Care Delivery and Medicaid Accountable Care Organizations

INTRODUCTION

Cardiovascular mortality is rising among low-income working-age adults in the United States.¹ Statins are highly effective for preventing cardiovascular events, yet remain underutilized in low-income groups despite their elevated risk.²

Accountable Care Organizations (ACOs) are payment models that hold clinicians and health systems accountable for the quality of care and total costs of an assigned patient population. ACOs aim to improve high-value preventive care delivery – including statin prescribing – by incentivizing care coordination, population health management, and performance-based accountability.³ While ACOs have led to modest gains in cardiovascular care quality in Medicare, little is known about their impact on younger, low-income adults in Medicaid, where the opportunity to narrow treatment gaps may be greatest.⁴

As of 2023, only 14 states have implemented Medicaid ACOs.³ Understanding whether these models improve cardiovascular preventive care could inform decisions about their broader adoption nationwide. We leveraged state-level variation in the implementation of Medicaid ACOs to evaluate their impact on statin utilization.

METHODS

Using Medicaid claims (2016–2019), we identified beneficiaries aged ≥26 years in Massachusetts, Rhode Island, and New York. Massachusetts launched statewide voluntary Medicaid ACOs in 2018. Providers in 17 ACOs shared accountability with managed care organizations under capitated payments or directly with Medicaid under fee-for-service. By 2019, ~75% of the eligible population was ACO-attributed.⁵ Rhode Island also implemented Medicaid ACOs in 2018, requiring managed care plans to contract with seven Accountable Entities, capturing ~70% of beneficiaries.⁶ New York served as the control state because of its proximity, large Medicaid population, and high-quality claims data. New York had established a small number of regional Medicaid ACOs in 2012 – providing time for policy effects to stabilize – but did not launch new statewide ACOs during the study period, making it a more appropriate comparator than states that had never pursued similar reforms.⁷

We included beneficiaries with hyperlipidemia, diabetes, coronary artery disease (CAD), peripheral arterial disease (PAD), or ischemic stroke/transient ischemic attack (TIA) using International Classification of Diseases codes. The pre-period was 2016 and the post-period was 2018–2019 due to the COVID-19 pandemic; 2017 was blanked due to ACO pilots. Beneficiaries in long-term care and those not continuously enrolled within each period were excluded.

The primary outcome was statin utilization, defined as ≥1 annual statin fill among beneficiaries with statin-eligible conditions, consistent with established quality metrics.⁸ To assess whether ACO implementation was associated with broader improvements in cardiovascular preventive care, secondary outcomes included ≥1 annual antihypertensive medication fill among beneficiaries with hypertension and ≥1 annual lipid test among beneficiaries with diabetes or CAD. Medications were identified using National Drug Codes and lipid testing using Current Procedural Terminology codes.

Using a difference-in-differences design, we compared changes in each outcome before (2016) and after (2018–2019) ACO implementation in Massachusetts and Rhode Island (intervention) versus New York (control). We fit beneficiary-level linear regression models adjusted for age and sex; the coefficients of group-by-period interaction terms represented the difference-in-differences estimates. Parallel trends were assessed visually using quarterly data. Analyses were conducted in SAS. Beth Israel Deaconess Medical Center provided ethical approval.

RESULTS

The study included 919,794 unique beneficiaries in the intervention states (mean age 49.2 years; 57.8% female) and 2,470,455 in the control state (mean age 50.8 years; 59.3% female). Statin use was low across both groups, with fewer than half receiving statin therapy (Figure).

Figure: Changes in cardiovascular preventive care delivery before and after Medicaid Accountable Care Organization implementation in Massachusetts and Rhode Island (intervention) versus New York (control).

The plots demonstrate the quarterly prevalence of outcomes in the intervention states (Massachusetts and Rhode Island) which implemented Medicaid ACOs statewide in 2018 as well as the control (New York). 2017 was excluded due to pilot ACO programs in the intervention states. The corresponding tables provide the annualized outcome prevalence in the pre-implementation period (2016) and the post-implementation period (2018–2019) in each group, as well as the difference-in-differences estimates, adjusted for age and sex, with 95% confidence intervals.

Hyperlipidemia was defined by at least one claim with ICD-10 codes E78.0-E78.5.

Diabetes was defined as one inpatient claim for ICD-10 diagnosis code E11.xx (any position), or two outpatient claims for ICD-10 diagnosis code E11.xx (any position) at least 30 days apart but no more than 180 days apart.

Coronary artery disease was defined by at least one claim with ICD-10 codes I20-I25.

Ischemic stroke or TIA was defined as at least one claim with ICD-10 codes I63, G45, I69.3x, or Z86.73.

Hypertension was defined as at least one claim with ICD-10 codes I10-I15.8.

Beneficiaries could contribute to more than one condition-specific cohort if they had multiple qualifying conditions. Each outcome was assessed among all beneficiaries eligible for that specific measure, regardless of eligibility for other measures.

Statin use was defined as a pharmacy claim for at least one of the following per year, including any combination medications: atorvastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, simvastatin.

Hypertension medication use was defined as a pharmacy claim for at least one of the following per year, including any combination medications: acebutolol, aliskiren, amiloride, amlodipine, atenolol, azilsartan, benazepril, betaxolol, bisoprolol, bumetanide, candesartan, captopril, carvedilol, chlorothiazide, chlorthalidone, clonidine, diltiazem, doxazosin, enalapril, eplerenone, felodipine, fosinopril, furosemide, guanfacine, hydralazine, hydrochlorothiazide, indapamide, irbesartan, isosorbide dinitrate, isradipine, labetalol, levamlodipine, lisinopril, losartan, methyldopa, metolazone, metoprolol, minoxidil, moexipril, nadolol, nebivolol, nicardipine, nifedipine, nisoldipine, olmesartan, perindopril, pindolol, prazosin, propranolol, quinapril, ramipril, sacubitril, spironolactone, telmisartan, terazosin, torsemide, trandolapril, triamterene, valsartan, verapamil.

Lipid testing was defined as at least one claim with CPT code 80061, 82465, 83718, or 84478 per year.

ACO: Accountable Care Organization; CI: confidence interval; CPT: current procedural terminology; DiD: difference-in-differences estimate; ICD: international classification of diseases; MA: Massachusetts; NY: New York; RI: Rhode Island; TIA: transient ischemic attack.

ACO implementation was associated with an increase in statin use across all conditions. For hyperlipidemia, diabetes, and CAD, statin use increased in the intervention states while declining in the control state, yielding an adjusted difference-in-differences of +12.2 (95% CI: 11.9, 12.5), +12.0 (11.6, 12.4), and +10.8 (10.3, 11.4) percentage points (pp), respectively. For ischemic stroke/TIA and PAD, statin use increased in both groups, but gains were greater in the intervention states (adjusted difference-in-differences +8.8 pp [7.9, 9.7] and +9.9 pp [9.1, 10.7]) (Figure).

ACO implementation was also associated with improvements in secondary outcomes: antihypertensive use (adjusted difference-in-differences +10.1 pp [9.8, 10.4]), lipid testing in diabetes (adjusted difference-in-differences +7.7 pp [7.3, 8.1]), and lipid testing in CAD (adjusted difference-in-differences +6.6 pp [6.0, 7.2]). Pre-implementation trends were parallel across all outcomes (Figure).

DISCUSSION

After Massachusetts and Rhode Island adopted Medicaid ACOs, statin use for hyperlipidemia, diabetes, CAD, PAD, and stroke/TIA increased relative to the control state. Medicaid ACOs were also associated with improvements in antihypertensive use and lipid testing. These findings suggest that Medicaid ACOs may strengthen preventive cardiovascular care delivery for low-income adults.

ACOs were associated with an increase in statin utilization even though statin prescribing was not an explicit quality benchmark. Together with gains in antihypertensive use and lipid testing, these findings suggest ACOs may have broadly strengthened preventive care delivery beyond narrow, measure-specific metrics.³ As out-of-pocket statin costs are minimal in Medicaid, the low baseline use likely reflects non-financial barriers such as fragmented care, competing priorities, and unmet social needs.² By supporting care navigation, population health management, and efforts to address social risk factors, ACOs may help providers reach patients who might otherwise go untreated.

The potential benefits of ACOs are not necessarily limited to their attributed patients. By 2019, over 70% of eligible beneficiaries in both states were ACO-attributed, indicating that a substantial proportion of the Medicaid primary care infrastructure operated under ACO incentives.^5,6 We therefore evaluated the broader Medicaid population to capture potential spillover effects and observed improvements across all outcomes. These findings align with evidence suggesting that ACO participation can influence provider practice patterns regardless of an individual patient’s attribution.^9,10

This study has limitations. There was a drop in quarterly statin fills and lipid testing in the control state for some, but not all, conditions between 2016 and 2018, though annualized values reflect smaller changes. Differential coding changes could bias results, although condition prevalence trends were similar between groups. Claims cannot capture clinical details or medications filled through retail discount programs. Generalizability may be limited to states with similar policy environments.

Our findings indicate that broader adoption of Medicaid ACOs could reduce persistent gaps in preventive cardiovascular care for low-income populations.² Future studies should evaluate their impact on clinical outcomes.

What is the clinical question being addressed?

Did Medicaid Accountable Care Organizations (ACOs) improve preventive cardiovascular care delivery?

What is the main finding?

Medicaid ACO implementation was associated with significant increases in statin use, antihypertensive use, and lipid testing, suggesting broader adoption could narrow gaps in preventive cardiovascular care for low-income populations.

Access to Data and Data Analysis:

Dr. Marinacci, Dr. Wadhera, and Mr. Zheng had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.