Application and validation of case-finding algorithms for identifying individuals with human immunodeficiency virus from administrative data in British Columbia, Canada

To define a population-level cohort of individuals infected with the human immunodeficiency virus (HIV) in the province of British Columbia from available registries and administrative datasets using a validated case-finding algorithm. Individuals were identified for possible cohort inclusion from the BC Centre for Excellence in HIV/AIDS (CfE) drug treatment program (antiretroviral therapy) and laboratory testing datasets (plasma viral load (pVL) and CD4 diagnostic test results), the BC Centre for Disease Control (CDC) provincial HIV surveillance database (positive HIV tests), as well as databases held by the BC Ministry of Health (MoH); the Discharge Abstract Database (hospitalizations), the Medical Services Plan (physician billing) and PharmaNet databases (additional HIV-related medications). A validated case-finding algorithm was applied to distinguish true HIV cases from those likely to have been misclassified. The sensitivity of the algorithms was assessed as the proportion of confirmed cases (those with records in the CfE, CDC and MoH databases) positively identified by each algorithm. A priori hypotheses were generated and tested to verify excluded cases. A total of 25,673 individuals were identified as having at least one HIV-related health record. Among 9,454 unconfirmed cases, the selected case-finding algorithm identified 849 individuals believed to be HIV-positive. The sensitivity of this algorithm among confirmed cases was 88%. Those excluded from the cohort were more likely to be female (44.4% vs. 22.5%; p