Interpretation of hematological and biochemical laboratory data in large-scale, multicenter clinical trials

The laboratory assessment of drug tolerability is central to a long-term trial. But because of its volume, its multicenter origin, and the importance of nondrug factors, the analysis of these data is complicated. Various methods, i.e., collection of investigators' opinion, comparison of before- and after-treatment means, and analysis of transitions, were found to be unsatisfactory. A fourth method, described in this paper, seems to be more promising. An initial computer screening of the laboratory data is conducted to identify all patients with the potentially clinically relevant laboratory abnormalities. Each laboratory abnormally is then examined by the pharmaceutical physician with regard to the patient's sex and age, the trial diagnosis, concomitant and intercurrent illnesses, concurrent medication, unwanted effects, and other laboratory results, and each result is assigned a probable etiology according to a pre-defined classification system. By this method it is then possible to compare the frequency and severity of possible or probable drug-related laboratory abnormalities occurring with the various trial drugs. Our opinion regarding the importance of using this method would appear justified by the fact that a possible or probable drug effect was considered to have been responsible for only 15 per cent (112/760) of the potentially clinically relevant abnormal tests reported.