Poisoning and epidemiology: 'toxicoepidemiology'

Abstract

1. There is little hypothesis-testing clinical research performed in toxicology. Randomized clinical trials are rare and most observational studies are performed on highly selected patients and are subject to marked bias. Thus, for many poisonings, our approach has been based almost entirely on deduction from known pharmacological/toxicological effects, generalizations from drugs within the same therapeutic class, animal data and case reports. This is also far from satisfactory, as many toxicological mechanisms are poorly understood and not related to the therapeutic class. 2. Although we need much better data to address the clinical and public health aspects of poisoning, there are many practical and ethical reasons why randomized clinical trials are difficult in this field. However, the scope for observational research, in particular population-based clinical epidemiology, is almost unlimited. The collection of data on human poisoning is facilitated because most non-fatal overdoses are admitted to hospital and by legal requirements to report to the coroner deaths that are due to poisoning. In the present article I argue that 'toxicoepidemiology', meaning the application of epidemiological methods to the problem of acute poisoning, is the best means we have of addressing deficiencies in our knowledge of poisoning. 3. Examples are given of a variety of observational research strategies, ranging from audit to meta-analysis, that may be applied to clinical toxicology. From coronial and clinical data obtained from reasonably well-defined populations, it has been possible to identify a number of previously unrecognized differences in the severity and spectrum of toxicity between and within drug classes. Also, the demographic risk factors for poisoning and the reproducibility, validity and optimal use of diagnostic and therapeutic interventions can be assessed. 4. The major limitations to the range of associations and interventions that may be studied are the need to achieve adequate power to study uncommon outcomes or poisonings and the ability to replicate findings at other centres using similar methodology. The expansion of data collection to other centres has the potential largely to overcome these obstacles.