Jain’s Innovators: Lower Medication Costs Lead to Better Health Outcomes
by Sachin Jain, MD
The RAND Health Insurance Experiment (HIE) is considered a seminal study because it helped clarify the relationship among health status, health services utilization, and access to health insurance. A recent study by Harvard researchers, in which free medications were given to patients after myocardial infarction (MI), promises to be similarly influential.
In the study, published last fall in The New England Journal of Medicine,1 patients discharged from hospital after MI were randomized to receive their medications with no out-of-pocket costs or with their usual co-payments. Remarkably, patients who received their medications for free had fewer fatal and nonfatal vascular events as a result of higher rates of adherence.
The study, conducted in partnership with health insurance giant Aetna, has already had a significant impact on patient care: In 2013, Aetna will begin providing all patients in the post-MI period with reduced co-payments on their medications; other insurers are expected to follow suit.
I spoke recently to Niteesh Choudhry, MD, PhD, and William Shrank, MD, MSHS, the lead authors of the study, to better understand the implications of their work.
Your study provided free drugs for patients after they had heart attacks. What was the inspiration for this study?
Non-adherence to prescription medications is a central focus of our research—on average, patients take about 50-60% of the medications that have been prescribed. There are a lot of reasons for non-adherence, and cost, even for patients with drug insurance, appears to be central.
Lowering co-payments, at least for some drugs, had been proposed as a strategy for improving adherence and clinical outcomes, which may also have favorable economic effects. We wanted to test this hypothesis more formally. In addition, we sought to evaluate the feasibility of doing a policy study that was truly randomized, and thus would provide the kind of evidence we consider to be the “gold standard” in clinical medicine, but which is remarkably rare in the health services research community.
The headline finding of the study was remarkable and has very important policy implications—lowering barriers to access for medications after heart attacks reduced overall mortality. Can you say more about this finding and highlight some of the other important findings from your study?
Our study included patients who had been discharged from hospital after suffering an MI and randomized them to their usual pharmacy benefits or to “full coverage” (no co-payments, coinsurance, or contributions to their deductibles) for any brand or generic statin, beta-blocker, or ACEI/ARB. This strategy increased adherence to the medications we studied by 4 to 6 percentage points or 30-40% in relative terms.
The rate of the study’s primary outcome, a composite of major vascular events or revascularization, was lower in the intervention group (17.6 vs. 18.8 per 100 person-years) but this difference was not significantly different (HR = 0.93; 95% CI 0.82 to 1.04; p = 0.21) because our intervention did not influence rates of revascularization. Focusing only on rates of major vascular events, which was a pre-specified secondary outcome, we found a 14% statistically significant reduction.
Average health spending was lower in the full-coverage group ($66,008 vs. $71,778) and patient spending on drugs and for other health services was reduced by 26%, or about $500 per patient, over the course of the study.
So, in summary, eliminating co-payments for evidence-based CV medications increased adherence, reduced rates of major vascular events, lowered patient out-of-pocket spending— and did so without increasing insurer spending. Of course, even in the group that received free medications, adherence rates were quite poor—highlighting the need for other strategies to complement the one we evaluated.
How might the findings from your study influence medical benefit design?
While the changes in adherence we observed were modest, they resulted in significant improvements in clinically meaningful outcomes and reductions in patient spending, without increasing insurer spending. This is somewhat of a rarity, since the vast majority of effective quality-improvement strategies have negative economic implications (i.e., they increase spending).
Equally important, the strategy we evaluated is simple and scalable and can be implemented by large insurers almost immediately. Aetna has announced that it will begin offering reduced co-payments for patients who suffer a heart attack beginning in January 2013 and we expect that other insurers will follow suit. As a result, our results should have an almost immediate impact on insurance design and the generosity of coverage in the United States.
Our trial might also have implications for other patient groups—perhaps most obviously, Medicare beneficiaries, who on average are at very high risk of adverse events after an MI and whose insurance coverage is less generous than those who receive usual employer-sponsored coverage. Our results may also have implications for other drugs and diseases—for example, CHF, diabetes, and hypertension. However, it is unclear whether improvements in adherence for drugs used to treat those conditions, even if they were of similar magnitude to the ones we observed, would necessarily translate into clinically meaningful results.
Your study was unique in that it was an academic study that was performed in partnership with a commercial health plan. How might this approach to research be expanded to improve healthcare delivery?
Our study was a relatively unusual example of a randomized policy study. We used a rigorous research design to examine a policy solution (instead of a clinical question), and in so doing we have demonstrated the ability to perform high-quality research within the context of a real-world healthcare system. We also created a very effective partnership between academics and insurers that will lead to a lot of other useful research.
This approach to the conduct of clinical trials is also substantially cheaper and answers different types of questions than more conventional trials, and thus could be used to evaluate a wide variety of innovative strategies to improve healthcare quality and reduce spending.
Can you contextualize your findings in terms of other studies about medication adherence and health plan benefit design such as the RAND HIE?
Until the MI FREEE (Post-Myocardial Infarction Free Rx Event and Economic Evaluation) trial, the RAND HIE was the only published randomized intervention of different levels of patient cost-sharing for prescription drugs. In the HIE, patients were randomly assigned to insurance plans that varied the amount of cost-sharing they faced. Prescription drugs were a covered expense in all plans, although the scope of the cost-sharing changes were much broader. Patients enrolled in a plan with no cost-sharing purchased 33% more prescriptions than patients with 95% cost-sharing; however, these changes were not associated with significant changes in health status.
It is important to note that the HIE was conducted in the 1970s, when the evidence base for many chronic medications was not clearly defined (i.e., before “value” could really be assigned to different therapies) and before many widely used drugs, such as statins, were available. Further, this study measure did not measure adherence.
Subsequent to this, there have been a handful of non-randomized studies evaluating the idea of selectively lowering co-payments for evidence-based medications, but MI FREEE is the first—and potentially last—randomized study of this strategy and the only one that had sufficient statistical power to evaluate its impact of clinically meaningful outcomes.
What are some of the limitations of the study and potential future research directions you and your colleagues are considering?
There are lots of challenges in doing any research study, especially in a real-world setting. One of the central difficulties we faced is that we recruited subjects with hospital discharge claims that take time to become available in administrative databases. During the resultant delay between actual discharge and our identifying patients as eligible for free drugs (which was about 1.5 months on average), some patients may have become non-adherent. While using health services claims to identify subjects increases the generalizability of our findings to other insurers who seek to institute similar plans, it may have diminished the intervention’s effect.
Despite its limitations, we think our study has a lot of important implications for patients and our healthcare system. One of the most important findings of MI FREEE was that even by giving away drugs for free, rates of adherence remained suboptimal. This highlights the concept that there are many other contributors to non-adherence, such as knowledge, attitudes, complexity of treatment regimens, difficulties accessing medications, side effects, and even simple forgetfulness. Our research going forward seeks to tackle each of these barriers and to use simple, scalable approaches that could be easily applied to help large numbers of patients.
What are the major take-aways from your study for practicing clinicians?
While we evaluated a policy strategy that will be difficult for individual clinicians to implement on their own, our results highlight the magnitude of the non-adherence problem and the potential benefits of even small improvements in long-term medication use.
There are a few lessons that we think practitioners should take away:
- Non-adherence is common and clinicians should be sure and ask their patients about difficulties they may have had taking their medications as prescribed.
- Cost matters. Therefore, prescribers should consider lower cost, equally effective options for patients.
- Many factors contribute to non-adherence, and clinicians should work with their patients to address side effects and simplify treatment regimens whenever possible.
- Choudhry N, et al. N Engl J Med. 2011;365:2088-97.
Sachin H. Jain, MD, MBA, is a physician at Brigham
and Women’s Hospital and Harvard Medical School and
Senior Institute Associate at Harvard Business School’s
Institute for Strategy and Competitiveness. He was previously
a key official in the Obama Administration, where
he helped launch the Center for Medicare and Medicaid
Innovation and the HITECH Act’s Meaningful Use provisions.
He is a leading national authority on healthcare
< Back to Listings