Objectives: The purpose of this study was to develop, implement, and assess estimation procedures to prevent adverse drug reactions (ADRs), especially drug-induced metabolic disorders, based on the subjective symptoms (complaints) of patients.
Methods: Our own database called CARPIS (Case Reports of Adverse Drug Reaction and Poisoning Information System) was started in 1987 and now contains ca. 23,000 case reports of ADRs. We extracted 264 cases of drug-induced metabolic disorders from CARPIS, such as hyperglycemia, hypoglycemia, and hypokalemia. Evaluation scores were created based on the subjective symptoms and the backgrounds of these patients. The scores were applied to the 264 cases to demonstrate their efficiency in estimation of ADRs.
Results: The evaluation scores estimated ADRs as follows: 39.6% of the 96 hyperglycemia cases, 53.6% of the 84 hypoglycemia cases, and 59.5% of the 84 hypokalemia cases. The validity measures of the evaluation scores were estimated to be as follows: for hyperglycemia sensitivity (SE) = 39.6%, specificity (SP) = 99.0%, predictive value of positive test (PVP) = 97.4%, positive likelihood ratio (PLR) = 39.6, and negative likelihood ratio (NLR) = 0.61; for hypoglycemia, SE = 53.6%, SP = 93.0%, PVP = 86.5%, PLR = 7.7 and NLR = 0.50; and for hypokalemia, SE = 59.5%, SP = 99.0%, PVP = 98.0%, PLR = 59.5, and NLR = 0.41.
Conclusions: The proposed evaluation scores are a reliable estimation method to detect ADRs related to drug-induced metabolic disorders. The scores should be incorporated into the integrated ADR estimation system available at a Web site our institution is developing, along with other evaluation scores of drug-induced liver disorders, extrapyramidal symptoms, and leakopenia, and druy eruptions.