Cardiovascular Outcomes in the Era of Claims Data
Has assessment of cardiovascular outcomes changed in when analysis of claims data is used?
Data from the Cardiovascular Health Study (CHS) were used for the present analysis. Event rates and risk factor associations were compared between adjudicated hospitalized cardiovascular events and claims-based methods of defining events. Specific outcomes examined included myocardial infarction (MI), stroke, and heart failure (HF), and were defined in three ways: 1) the CHS adjudicated event, 2) selected ICD9 diagnostic codes only in the primary position for Medicare claims data from the Center for Medicare and Medicaid Services, and 3) the same selected diagnostic codes in any position.
A total of 28,230 hospitalizations for the 5,888 CHS participants were identified in CHS. Of the 4,344 incident cardiovascular events, 59.0% were first identified by self-report, 57.0% for stroke, and 66.3% for angina. Claims-based methods for defining events (conventional method) had high positive predictive values, but with low sensitivities. The investigators provided an example of the positive predictive value of an ICD9 code of 410.x1 for a new acute MI in the first position, which was 90.6% but this code identified only 53.8% of the incident MIs in total. For CHS adjudicated events for MI, the incidence was 14.9 events per 1,000 person-years, and 8.6 for selected ICD9 diagnostic codes only in the primary position for Medicare claims data from the Center for Medicare and Medicaid Services, and 12.2 for the same selected diagnostic codes in any position. For cardiovascular disease risk factor associations, similar findings were observed for all three methods of defining events.
The investigators concluded that use of diagnostic codes from claims data as clinical events can lead to an underestimation of event rates. Additionally, claims-based events data represent a composite endpoint that includes the outcome of interest and selected (misclassified) nonevent hospitalizations.
In the era of big data, comparision of methods to assessment of outcomes is critical. Understanding how to assess outcomes with accuracy will allow investigators to improve quality of care; therefore, improving accuracy of accessment is imperative.
< Back to Listings