Comparison of two recently published algorithms for assessing the probability of adverse drug reactions

Abstract

1 A simple valid and reliable method for estimating the probability of adverse drug reactions (adverse drug reactions probability scale, APS) has been recently described (Naranjo et al., 1981a). 2 The results using APS were compared to those obtained with another more detailed algorithm (adverse reactions scoring system, ASS) described by Kramer et al. (1979). 3 Sixty-three randomly selected adverse drug reactions (ADRs) were rated by two observers, using APS and ASS one year apart. The cases were ordered in a random sequence. Between-raters reliability using APS (R(est) = 0.96 and ASS (R(est) = 0.86), was very high. 4 ADR scores obtained with both methods were highly correlated (r = 0.82, P less than 0.001). However, time spent using ASS was significantly longer (paired t-test, t = 1.70, P less than 0.05). 5 These results suggest that while ASS is somewhat more complex than APS both are equally reliable and will give very similar conclusions regarding the probability of ADRs. Such algorithms must be used if the clinical assessment of ADRs is to become acceptably reliable.