Rising prescription drug cost sharing increases mortality among Medicare beneficiaries

What are the health consequences when patients reduce their consumption of prescribed drugs in response to higher out-of-pocket expenses? In The health costs of cost sharing (NBER Working Paper 28439), researchers Amitabh Chandra, Evan Flack, and Ziad Obermeyer use the distinctive cost-sharing features of Medicare Part D to demonstrate that such reductions can increase mortality.

Their analysis uses the fact that two 65-year-old Medicare enrollees with identical prescriptions, but different birth months, could face dramatically different charges at the end of the year. Several unique features of Medicare Part D during the 2007-2012 study period contribute to this fact. The patient’s out-of-pocket share of prescription drug costs (coinsurance rate) in the standard Part D plan was typically 25% when total drug expenses were less than approximately $2,500. But the coinsurance rate jumped to 100% in the ‘coverage gap’ after total drug spending exceeded the $2,500 threshold, before dropping to 5% above the ‘catastrophic’ spending limit. (about $5,700). Beneficiaries tend to enroll in Medicare the month they turn 65, but the spending thresholds that determine coinsurance rates aren’t prorated for those who enroll in the middle of the year. year. Because of these characteristics, beneficiaries whose birthdates fall earlier in the year are more likely to enter the coverage gap or exceed the catastrophic limit in their first year of coverage.

The researchers use a random sample of 20% of newly enrolled Medicare beneficiaries aged 65, focusing on December results among those who enrolled between February and September. They report that 11.8% of February enrollees fell into the coverage gap in December, while only 1.5% of September enrollees did. Similarly, 1.7% of enrollees in February were in the catastrophic coverage category in December, compared to 0.2% of enrollees in September.

Of course, beneficiaries with low prescription drug use are unlikely to fall into the coverage gap, regardless of their birthday. To better identify beneficiaries who might fit into different levels of coverage, researchers are using machine learning tools to predict annual prescription drug spending for each Medicare beneficiary. The algorithm predicts what expenses would be in the absence of cost sharing, based on the expenses of similar Medicare beneficiaries who face no out-of-pocket due to Medicaid eligibility.

The figure shows the results for Medicare beneficiaries with projected annual prescription costs high enough for them to enter the coverage gap, based on their date of birth, but unlikely to exceed the catastrophic threshold. Among this group, the first panel shows that December co-insurance rates are highest for those born in February and drop by an average of 2.3 percentage points in each subsequent month of enrollment.

The second panel of the figure shows a similar trend for death rates in this group. On average, each subsequent month of enrollment reduces December mortality by 0.0113 percentage points, a decrease of 9%.

The opposite patterns emerge for beneficiaries with planned expenses in the upper 3 percentiles, who could exceed the catastrophic limit and end the year with a 5% coinsurance rate. For this group, which is not shown in the figure, each subsequent month of enrollment is associated with a higher coinsurance rate in December and a higher mortality rate.

Assuming that all mortality differences are due to coinsurance differentials, the researchers estimate that a 1 percentage point increase in the coinsurance rate results in a 0.00435 percentage point increase in the mortality rate. They attribute these mortality differences to the reduction in prescription drug use induced by higher out-of-pocket expenses. The same 1 percentage point increase in coinsurance is associated with 0.031 fewer prescriptions filled during the month and a $5.54 reduction in total prescription drug expenditure between payers. Interestingly, some recipients respond to increased out-of-pocket expenses by choosing not to fill any prescriptions, regardless of the number of medications they were previously taking or their individual health risks.

In response to higher out-of-pocket expenses, patients have reduced their consumption of a wide range of medications, many of which have beneficial effects on mortality that have been established by clinical trials, such as those that control cholesterol, blood pressure and blood sugar. These reductions are not limited to patients who are at relatively low risk of adverse health events. Rather, the researchers find similar price sensitivity in patients who, based on machine learning predictions, are most at risk for adverse events such as heart attacks and strokes.

To rationalize the decisions observed in this study, one would have to assume that the Medicare beneficiaries in the sample valued an additional year of life at $6,628, a value that is at least an order of magnitude below the conventional range. estimates. Another plausible explanation is that patients make significant errors in evaluating the relative costs and benefits of purchasing prescription drugs.

The standard economic rationale for increased cost sharing is that it reduces the overuse of medical care. But the findings of this study indicate that cost-sharing also leads to reckless treatment cuts that put recipients’ lives at risk. The researchers conclude that redesigning prescription drug coverage to better accommodate these behavioral patterns would be a cost-effective way to improve health outcomes.

The researchers acknowledge support from National Institute on Aging grant P01AG005842.

Comments are closed.