Validation of claims‐based diagnostic and procedure codes for cardiovascular and gastrointestinal serious adverse events in a commercially‐insured population

• DOI: 10.1002/pds.1924
• Publication Date: 2010-02-05T09:54:07Z
• Authors: Peter M. Wahl; Keith Rodgers; Sebastian Schneew...; Brian F. Gage; Javed Butler; Charles Wilmer; Marshall Nash; Gregory Esper; Norman Gitlin; Neal Osborn; Louise J. Short; Rhonda L. Bohn
• Keywords: Cyclooxygenase 2 Inhibitors; Databases, Factual; Pharmacoepidemiology; Anti-Inflammatory Agents, Non-Steroidal; Myocardial Infarction; Nonprescription Drugs; Severity of Illness Index; United States; 3. Good health; Brain Ischemia; Stroke; 03 medical and health sciences; 0302 clinical medicine; International Classification of Diseases; Predictive Value of Tests; Humans; Epidemiologic Methods; Gastrointestinal Hemorrhage; Algorithms; Retrospective Studies

AbstractPurposeTo validate administrative claims codes with medical chart review for myocardial infarction (MI), ischemic stroke, and severe upper gastrointestinal (UGI) bleed events in a large, commercially‐insured US population.MethodsThese validation studies were part of a larger study examining the risk of MI, ischemic stroke, and severe UGI bleeds in patients receiving a new prescription of selective cyclooxygenase (COX)‐2 inhibitors (coxibs) and non‐over‐the‐counter (OTC) non‐steroidal anti‐inflammatory drugs (NSAIDs), between 1 July 2002 and 30 September 2004. Patients from the study cohort and other health plan members from the HealthCore Integrated Research DatabaseSM (HIRD) experiencing these events were selected for these studies. The positive predictive value (PPV) of each of the claims code algorithms, using medical chart review as the gold standard, was calculated.ResultsTwo hundred charts per event were abstracted. The PPV for MI was 88.4% (177/200; 95%CI, 83.2–92.5%); PPV for ischemic stroke was 87.4% (175/200; 95%CI, 82.0–91.7%); PPV for severe UGI bleed was 56.5% (109/193; 95%CI, 49.2–63.6%). Refining the ischemic stroke claims algorithm resulted in a PPV of 95.5% (95%CI, 91.0–98.2%); refining the claims algorithm for severe UGI bleed resulted in a PPV of 87.8% (95%CI, 78.7–94.0%).ConclusionThe results suggest that, for certain adverse events, claims data can serve as the basis for pharmacoepidemiology research and drug safety surveillance in the US. Copyright © 2010 John Wiley & Sons, Ltd.