Quality of CAD Care Varies by Type of Medicare Coverage

What’s being incentivized may need a rethink given the lack of a link between process measures and outcomes, the study author says.

Quality of CAD Care Varies by Type of Medicare Coverage

Patients with CAD with private insurance coverage through the Medicare Advantage (MA) program are more likely to receive medications for secondary prevention, but that does not necessarily translate into improved control of blood pressure or LDL cholesterol, a new analysis shows.

After accounting for potential confounders, enrollment in MA was associated with higher odds of receiving beta-blockers (OR 1.10), ACE inhibitors or angiotensin II receptor blockers (ARBs; OR 1.13), and statins plus those two types of medications (OR 1.23), according to researchers led by Jose Figueroa, MD (Brigham and Women’s Hospital and Harvard Global Health Institute, Cambridge, MA). There were no significant differences in use of statins alone or referral to cardiac rehabilitation.

The greater use of preventive drugs did not translate, however, into improvements in levels of blood pressure or LDL cholesterol, deemed “intermediate outcomes,” the investigators report in a study published online February 20, 2019, ahead of print in JAMA Cardiology.

“The one thing that is reassuring, though, is that it does seem to be that providers are much more likely to be prescribing these medications, when indicated, to MA patients than they are relative to fee-for-service patients,” Figueroa told TCTMD. “So for people who are interested in improving quality of care, at least you can think of MA as being a potential strategy.”

But, he added, it might be necessary to identify different metrics that are more closely aligned with patient outcomes to target for quality improvement.

Medicare Advantage Enrollment Growing

Figueroa said it’s important to examine these issues because more and more people have been enrolling in MA over the past decade. About one-third of Medicare beneficiaries are now using the program, which allows them to get coverage through private insurance plans. It remains unclear, however, whether quality of care is better or worse in MA-covered patients.

“There’s the thought that in fact it is likely that [MA] would be more willing to influence providers to prescribe the right medications or maybe potentially contract out with providers that are much more likely to provide higher quality of care,” Figueroa said.

To explore the issue, the investigators turned to the National Cardiovascular Data Registry’s PINNACLE registry, an outpatient-based cardiac quality improvement registry that contains more detailed clinical information than administrative data sets used in prior studies. The analysis included 35,563 adults diagnosed with CAD who were enrolled in MA and 172,732 who were enrolled in fee-for-service Medicare coverage. CAD was defined as a history of MI, PCI, or CABG.

On average, patients enrolled in MA were slightly younger, less likely to be white and to have atrial fibrillation or flutter, more likely to use tobacco, and more likely to be female and to have heart failure, diabetes, peripheral vascular disease, and chronic kidney disease.

This suggests that MA patients were as sick or sicker than those with fee-for-service coverage, which contrasts with prior studies showing a healthier patient population within MA. The investigators were “a little bit surprised” to see that, Figueroa said, who pointed out that the current analysis was better positioned than others to examine differences between patient populations because of the availability of more granular clinical data entered into electronic health records.

As for quality of care, prior to adjustment for potential confounders, use of secondary prevention treatments was higher in the MA versus fee-for-service group.

   Use of Secondary Prevention (Unadjusted Rates), MA Versus Fee-For-Service Medicare

 

Medicare Advantage

Fee-For-Service Medicare

P Value

Beta-Blockers

80.6%

78.8%

< 0.001

ACE Inhibitors/ARBs

70.7%

65.1%

< 0.001

Statins

68.4%

64.5%

< 0.001

All Three Drug Classes

48.9%

40.4%

< 0.001

Referral for Cardiac Rehab

6.4%

5.5%

< 0.001

            Abbreviation: ARBs, angiotensin II receptor blockers.

After adjustment, only the differences for beta-blockers, ACE inhibitors/ARBs, and all three drug classes remained significant.

Where Are the Improvements in Outcomes?

Asked why those advantages for the MA program did not translate into improvements in outcomes, Figueroa noted that the analysis was limited in that the registry did not include information on important patient outcomes like mortality, complications, readmissions, or patient experience.

He and his co-authors expand on possible reasons for lack of a difference in intermediate outcomes in their paper, pointing out that this study would not have picked up smaller effects that could be important on a population level; process measures are not always strongly related to outcomes; and older individuals have not been well represented in clinical trials of preventive therapies, so there may be differences in effectiveness.

Taken all of that into consideration, “this may indicate that the process measures currently being collected are inadequate to drive improvements in outcomes in an older Medicare population,” they say. “Therefore, policy makers might temper their expectations on using MA plans, especially as enrollment continues to grow, as a means for improving patient outcomes for Medicare-enrolled patients.”

Efforts to improve outcomes could be bolstered, Figueroa said, by rethinking which factors are tracked to assess quality of care. “The things that are being incentivized under Medicare Advantage may need some adjustment to measures that are actually much more correlated with better quality of care when it comes to patient outcomes,” he said.

Or, he added, it might make sense to incentivize improved patient outcomes directly, without looking at process measures as proxies for quality of care. Figueroa acknowledged, however, that there are concerns with using patient outcomes to assess quality because of challenges with accounting for things that are outside of a treating physician’s control, like social determinants of health. “We don’t have very good risk-adjustment models,” he said. But, he added, “it is potentially a strategy to just go right out and incentivize patient outcomes and not just processes of care.”

Noise Versus Signals

In an accompanying editorial, Paul Heidenreich, MD (VA Palo Alto Health Care System, CA), lays out some of the challenges with using outcomes to assess quality of care.

One of the appealing features of an outcome measure is that an outcome is influenced by all the relevant differences in the quality of care, not just those that are measurable,” he says. “However, outcomes, including observed to expected outcomes, are also influenced by case mix, the way data are measured (eg, overcoding or undercoding), and random variation. While one can attempt to adjust for case mix, it is difficult to adjust for the method of comorbidity measurement and impossible to exclude random variation.

“Thus, much of the difference in outcomes may be noise as opposed to signal,” he continues. “This signal-to-noise ratio is only going to worsen as the quality of care continues to improve, making the differences in outcomes between clinicians mostly noise.”

Heidenreich points out, too, that studies powered to detect differences in processes of care are rarely powered for outcomes as well.

“In summary, as tempting as it may be, we should not let the lack of an observational outcome supersede a benefit in a process of care that is shown to improve outcomes in randomized clinical trials,” he argues. “The low-quality signal-to-noise ratio for observational outcome measures, combined with the huge sample size that is needed to show an outcome difference given a small process of care difference, makes a null outcome finding difficult to interpret. We will be more accurate and helpful to health system users by combining process and outcome measures when assessing and labeling quality of care.”

Disclosures
  • Figueroa reports being partly funded by a grant from the National Center for Advancing Translational Sciences.
  • Heidenreich reports no relevant conflicts of interest.

We Recommend

Comments