Evaluation of Thai Algorithm Usage for Adverse Drug Reaction Monitoring - การประเมินการใช้ Thai Algorithm สำหรับการติดตามอาการไม่พึงประสงค์จากการใช้ยา

Abstract

          The WHO-UMC causality assessment criteria, the standard method for routine pharmacovigilance (PV) was rarely used in Thailand in comparison with the Naranjo’s algorithm. There were some questions that Naranjo’s algorithm is unable to comply with the current clinical practice and/or were interpreted differently among observers. To address this problem, Thai algorithm was developed. This study aimed to evaluate the validity and reliability of both the Thai algorithm and Naranjo’s algorithm for assessing a possible causal link between a drug and an adverse event in routine PV practice using WHO-UMC causality assessment criteria as the reference; as well as assessing the acceptability of the Thai algorithm. Six hundred individual case safety reports (ICSRs) during 2012 were randomly sampling from 30 hospitals of Ministry of Public Health, stratified by 3 levels of hospital size. A hospital pharmacist with PV qualification was selected to evaluate 200 reports (drug-event pair) collected from each level. Every report was randomly assessed by 3 independent observers and 3 methods. Results expressed as certain/ probable probability scores were defined as positive result and assessed forvalidity and reliability using sensitivity, specificity and Kappa statistics. The frequency was used for acceptability. A total of 600 ratings for each method were assessed. It was found that, comparing with the WHO-UMC methods, the Thai algorithm’s agreements of positive result was achieved with 361 agreements, with a sensitivity of 75.1% and a specificity of 66.4%. The assessment using Naranjo’s algorithm was similar to Thai algorithm, with 363 agreements of positive result, with a sensitivity of 75.5% and a specificity of 67.2%. The inter-observer reliability of three observers was fair. Cohen’s kappa value of the WHO-UMC method, Thai algorithm and Naranjo’s algorithm was 0.33, 0.22 and 0.21 respectively. In conclusion, the Thai and Naranjo’s algorithms had similar sensitivity and specificity for assessing ICSRs in daily practice when using the WHO-UMC causality assessment criteria as a reference. Majority of the respondents agreed to use the Thai algorithm; and recommended to develop a software so as to make it more convenient to use.

Key words: Thai algorithm, Naranjo’s algorithm, WHO-UMC causality assessment, validity, reliability