It Pays to Reward Doctors for Risk-Stratifying Patients, Early Data Hint

“There's 330 organizations of all different types all across the country enrolled, so I think it's promising in that it has clearly improved the care processes for things like risk stratification,” lead author G. Greg Peterson, PhD (Mathematica, Washington, DC), told TCTMD.

The incentivized model was launched in 2017 by the US Centers for Medicare & Medicaid Services (CMS) with a goal of reducing the rate of first-time MI and stroke in high-risk patients. Provider organizations that enrolled were allocated to intervention or usual care, with the intervention group required to stratify patients by 10-year CVD risk and provide prevention management services, including discussion about risk scores, individualized risk-reduction plans, as well as annual in-person risk assessments and follow-up.

While the model is still ongoing and has not yet reported its primary outcome, the secondary analysis, published online in JAMA Cardiology by Peterson and colleagues, shows a statistically significant, though numerically small, improvement in the intervention group with regard to initiation or intensification of statins and antihypertensives within a year of enrollment.

The enrolled organizations include primary care and cardiology practices, healthcare centers, and  hospital-based outpatient departments. Of 125,436 Medicare beneficiaries included from these organizations, high-risk patients in the intervention group had a mean age of 74 years at enrollment and 91% of them were above threshold systolic blood pressure level (> 130 mm Hg), LDL-cholesterol level (> 70mg/dL), or both. Additionally, 69% of those in the intervention group were already on statins, 90% were on an antihypertensive drug, and the average number of office visits in the previous year was 10.

At 1 year, initiation or intensification of statins or antihypertensives was 37.3% in the intervention group versus 32.4% in the control group (P < 0.001). Although providers were only paid for assessment and management of high-risk patients, who had a 30% or higher 10-year risk of MI or stroke, there also were higher rates of initiation or intensification of statins or antihypertensives among medium-risk patients in the intervention group, who had a 15% to 30% risk (27.9% vs 24.8; P < 0.001). In both the intervention and control groups, the number of medium-risk patients was more than double that of high-risk patients.

Along with the difference in medication use, the model also showed an impact on clinical endpoints, with a 1.2% lower mean blood pressure among high-risk patients in the intervention group at 1 year compared with controls (P = 0.003), as well as a 2% lower mean LDL-cholesterol level (P = 0.003).

Systematic Improvement ‘Spillover’

“I think the spillover to the medium-risk group is important,” Peterson noted, adding that it may be a result of more systematic use of risk scores in daily practice among enrollees, a key takeaway that the model is meant to encourage. “The impact on blood pressure and cholesterol are certainly modest to small, however you want to think about it. But when you spread those impacts over a large population, they can still have a meaningful effect population-wide,” he added.

Million Hearts pays participating organizations $10 for each eligible beneficiary they risk-stratify. In the first year, cardiovascular management fees are fixed at $10 per beneficiary per month for each high-risk enrollee. In year 2 and later, the cardiovascular management fees are replaced with risk-reduction payments scaled to performance in reducing 10-year predicted risk among those who were high-risk at initial enrollment (up to a maximum of $10 per beneficiary per month). Control organizations also get paid to collect and report clinical data, but they are not asked to calculate CVD risk scores or to make changes to their usual clinical care.

Peterson said it’s too early to know if the pay-for-performance model will have a significant impact on the primary outcome of first incidence of MI or stroke over 5 years. The model is scheduled to be ongoing through 2022. However, despite showing only a modest difference so far between the intervention and control groups on medication management, preliminary data to 2.5 years suggest that the difference is sustained, he said.

“If the benefits of medication accumulate over time, then the impact on risk of an actual event could get larger over time, even if the impact on medications is just a sustained difference,” Peterson added.