Chronic condition mortality in the Medicare population

The census and proportion of the US population that is age 65 years and older has never been as high. Medicare data files are a valuable source of data on medical conditions and care that can be used to study the older age population. We obtained access to The CMS Medicare data files including a 5% sample of annual Beneficiary Annual Summary (BASF) files for the years 1999-2009, and the most current Vital Status file available at the time of our request (2011). The Vital Status file enabled us to assess longitudinal follow-up for survival analysis. Data from over 3 million beneficiaries were available. The BASF files include summarized data pertaining to condition categories, defined by the Chronic Conditions Data Warehouse (CCW), which was of primary interest for this analysis. Cox regression models were used to assess the mortality risk associated with a set of 15 chronic conditions, as well as severity factors based on summary claims data. We were able to confirm a number of expectations, such as the high level of mortality risk with lung cancer, congestive heart failure, and in the oldest ages, Alzheimer's disease and related dementias. We were also able to identify chronic conditions that behave more as chronic conditions individuals may "live with" rather than "die of." Depression, diabetes, prostate and breast cancer are present for longer durations and/or are associated with low or no increased mortality risk in the Medicare population. Inpatient confinement or skilled nursing facility utilization were markers for increased risk, as expected. Unexpectedly, frequent physician visits (>10/year) was a marker for more favorable mortality, perhaps indicating that close supervision of chronic conditions lead to improved survival.