Diagnostic Algorithms for Cardiovascular Death in Administrative Claims Databases: A Systematic Review

Area of Science:

• Health Informatics
• Epidemiology
• Regulatory Science

Background:

• Accurate identification of cardiovascular (CV) deaths is crucial for assessing medication safety, public health, and regulatory science.
• Administrative health plan claims databases are valuable resources for epidemiological research.

Purpose of the Study:

• To systematically review algorithms used to identify cardiovascular death in administrative health plan claims databases.
• To evaluate the accuracy of existing algorithms for identifying sudden cardiac death (SCD), myocardial infarction (MI)-related death, and stroke-related death.

Main Methods:

• Systematic literature search of MEDLINE, EMBASE, and Cochrane Library (2012-2017).
• Inclusion of observational studies evaluating algorithm performance (sensitivity, specificity, PPV, NPV) for CV death using electronic health care data.
• Risk of bias assessment using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) instrument.

Main Results:

• Five studies met the inclusion criteria, focusing on SCD (4 studies) and MI/stroke-related death (1 study).
• Positive predictive values (PPVs) were reported for all studies; other accuracy parameters were incompletely reported.
• Algorithms combining medical claims and death certificates showed high PPVs for SCD (86.8% and 92.3%). Algorithms using death certificates alone had lower PPVs (27% and 32%).

Conclusions:

• Two algorithms utilizing medical claims data with or without death certificates demonstrate accuracy for identifying sudden cardiac death (SCD) in pharmacoepidemiologic studies.
• Development of valid algorithms for identifying myocardial infarction (MI)- and stroke-related deaths is a critical research priority.