Sensitivity and specificity of three methods of detecting adverse drug reactions.

The sensitivity and specificity of three methods of detecting adverse drug reactions (ADRs) were determined. Minimal use of a voluntary ADR reporting program prompted this investigation of three ADR detection methods, as follows: screening of laboratory reports, pharmacist screening of medication orders, and voluntary reporting. A total of 98 patients who were receiving oral or i.v. digoxin therapy, oral or i.v. theophylline therapy, or i.v. gentamicin therapy were randomly selected and monitored for possible ADRs. A physician reviewed the charts of patients with suspected ADRs using the Naranjo algorithm to assess causality. The chart review served as the reference method to which the other three methods were compared. Thirteen "true" (i.e., confirmed by the Naranjo algorithm) ADRs were identified in 11 different patient charts, resulting in a 13.3% ADR incidence rate for the 98 sampled patients. For the three ADR detection methods, the decreasing order for level of sensitivity was screening of laboratory reports, pharmacist screening of medication orders, and voluntary reports; however, only the difference between laboratory reports and voluntary reports was significant. For level of specificity, the decreasing order for the three methods was voluntary reports, pharmacist screening of medication orders, and pharmacist screening of laboratory reports; the differences among all three methods were significant. Screening of laboratory reports and pharmacist screening of medication orders are two detection methods that appear to exhibit an appropriate combination of sensitivity and specificity for identifying ADRs; trials with larger sample sizes are needed to confirm the results of this study.