The use of claims data algorithms to recruit eligible participants into clinical trials

Abstract

Objectives: Recruitment strategies usually focused on a single International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code and rarely included exclusion criteria. The purpose of this study was to validate a claims-based algorithm to identify, from Veterans Affairs administrative data, eligible participants to be recruited into a hypertension trial.

Study design: Cross-sectional.

Methods: Subjects were labeled as eligible if they were 75 years or older, had a hypertension ICD-9-CM code (401.x-405.x, 437.2) and did not have a diabetes (250.xx) or stroke (430.x-436.x, 437.1, 437.9, 438.x) ICD-9-CM code. We compared the eligible subjects with the medical record-which was considered the gold standard-and we calculated the positive predictive value (PPV) of identifying a subject in the medical record.

Results: The algorithm identified 3591 elderly veterans with hypertension with no diabetes or stroke, and we reviewed the medical records of 76 randomly selected patients. In the sample of medical record review, the mean age in years was 83 ± 5.3, 48% had coronary artery disease, and the mean systolic blood pressure was 134 mm Hg ± 15.5. When compared with the medical record, the PPV for any hypertension code was 93% (95% CI, 85%-98%), and for the entire algorithm, including 75 years or older and the absence of both diabetes and stroke, the PPV was 83% (95% CI, 73%-91%).

Conclusions: The use of any ICD-9-CM code for hypertension is useful to identify elderly patients with hypertension. The algorithm to identify elderly patients with hypertension and without diabetes or stroke is a useful tool to also identify eligible patients for clinical trial participation.