Doctors' self rating of skills in evidence based medicine

Editor—Young et al suggest that Australian general practitioners may be confused about certain basic epidemiological concepts.¹ Responsibility for some of this confusion presumably lies with whoever taught them epidemiology.

Part of the problem is suggested by the opacity and ambiguity of some of the epidemiological expert definitions against which general practitioners' knowledge was assessed. For example, for relative risk the expert definition was “Relative risk estimates the magnitude of an association between exposure and disease.” So it does, but how does this definition distinguish relative risk from absolute risk?

Furthermore, for relative risk reduction the “correct” definition was “Relative risk reduction is calculated as (control event rate−experimental event rate)/control event rate).” We suspect that many epidemiologists would also “fail” if judged against this criterion, since this is not the only way in which relative risk reduction can be calculated.

Similarly, although it is true that number needed to treat is the reciprocal of the absolute risk reduction, knowledge of this fact is, arguably, irrelevant to the practical application of the concept.

The perception that epidemiology is difficult is likely to be reinforced by definitions such as these. This may intimidate non-epidemiologists from engaging in critical appraisal of research evidence. The key issue is that it is not necessary to be able to define something to be able to understand it. Furthermore, the ability to memorise a definition does not necessarily indicate understanding of the underlying concept. The emphasis in teaching epidemiology should be not on mathematical formulas but on conveying understanding and relevance for health workers. Concentrating on the mathematical aspects of the definitions may be a hindrance rather than help.

If the knowledge gap identified by Young et al is real we need to examine how we are currently teaching epidemiology. If we accept their conclusions then current methods seem to be failing.