Medical Cost Offsets from Prescription Drug Utilization among Medicare Beneficiaries

In November of 2012, the Congressional Budget Office (CBO) changed its methodology for estimating the financial impact of legislation affecting prescription drug utilization among Medicare beneficiaries.  CBO now assumes that a 1% increase in the number of prescriptions filled prompts a 0.20% decrease in spending on other medical services, such as emergency department visits and hospitalizations.  This new, literature-based assumption applies to the use of all prescription drugs by the general Medicare population.

In this paper, I extend earlier work on the value of medication adherence by estimating the impact of condition-specific prescription drug use on medical spending among seniors with chronic vascular disease.  I analyzed data on elderly individuals (age≥65) who were continuously enrolled in employer-sponsored insurance from 1/1/2005 through 6/30/2008, and diagnosed with and on pharmacotherapy for one or more of four chronic vascular diseases:  congestive heart failure (n=8,080), hypertension (n=53,859), diabetes (n=19,035), and dyslipidemia (n=22,813).  Prescription drug utilization was measured using the number of days of medication supply on hand for each condition.  To control for time invariant unobserved confounders, condition-specific linear fixed effects models of medical spending were estimated as a function of prescription drug utilization, health status as measured by the Charlson Comorbidity Index and year indicators.  To account for non-linear relationships, both days’ supply and days’ supply-squared were entered into the equations, and Ramsey’s RESET test supported this choice of functional form.  Finally, to be comparable to the CBO’s reported metric, marginal effects of prescription drug utilization in each of the chronic conditions were calculated as elasticities.

Results indicate that 1% increases in condition-specific prescription drug utilization were significantly (p<0.001) associated with reductions in seniors’ gross medical costs in the amounts of:  0.77% for congestive heart failure, 1.17% for hypertension, 0.83% for diabetes, and 0.63% for dyslipidemia.  These results demonstrate that medical cost offsets from prescription drug utilization likely vary by chronic condition, and that impacts for therapeutic classes used to treat these four conditions—which represent 40% of Medicare Part D utilization—may be between 3 and 6 times greater than CBO’s assumption.  In dollar terms, these relative impacts are not trivial.  For example, 53% of Medicare (fee-for-service) beneficiaries have the comorbidity combination of hypertension plus high cholesterol—with average annual medical costs of $13,825.  The current findings suggest that a 5% increase in use of antihypertensive medication by these patients may prompt reductions in medical (Part A and B) costs of more than $800 annually per beneficiary.

More research is needed to derive a comprehensive set of condition-specific medical cost offset estimates from prescription drug utilization.  Future studies should expand beyond Medicare to populations in Medicaid, Veteran Affairs, and private insurance—recognizing that use of pharmaceuticals will likely differ by demographic, socioeconomic, and health-related characteristics.  This work will better equip policymakers to design, evaluate, and implement programs that optimize prescription drug utilization, and minimize overall healthcare costs.