3121 — An approach to detect novel adverse drug events using medical and pharmacy claims databases.

Nebeker J (SLC GRECC ), Shen S (University Utah), Rupper R (SLC GRECC), Samore M (SLC IDEAS Center)

Objectives:
Present a framework for detection of possible adverse drug events using administrative databases. We demonstrate our approach using cholinesterase inhibitors (ChEIs) for the treatment of dementia using Utah Medicaid data. Four classes of adverse drug events: death, expected (gastro-intestinal and psychological disturbance), suspected (respiratory disturbance), and idiosyncratic reactions (hepatic and hematological disturbance) were evaluated.

Methods:
We conducted an open cohort study that linked data on drug utilization and medical claims from Utah Medicaid beneficiaries between 1/01/2003 and 12/31/2005. The study was limited to patients aged 50 and older who had a dementia diagnosis. Incident ChEI users were compared to non-users. Propensity score matched cohorts were established for each adverse outcome to evaluate cumulative effects of drug exposure.

Results:
Of the expected reactions, ChEI exposure was significantly associated with gastrointestinal episodes (Hazard Ratio [HR]:2.02; 95%CI:1.28-3.2). ChEI exposure was not significantly associated with respiratory episodes (HR:1.21; 95%CI:0.81-1.79). Of the idiosyncratic reactions, ChEI exposure was significantly associated with hematological episodes (HR:2.32; 95%CI:1.47-3.67).

Implications:
Administrative data can be used to examine specific drug classes and individual drugs for known and unknown adverse drug events. The ADE framework of initially examining mortality, expected, suspected events, and then novel idiosyncratic reactions will foster understanding of drug safety and generate hypotheses for future investigations. Further evaluation is needed to determine whether or not the increased rate of hematological episodes is causally related to ChEI exposure.

Impacts:
The Veteran’s Affairs Healthcare system is a rich source of clinical and administrative data and can be used to identify severe adverse drug events that may not have been identified during clinical trials.