A healthy disposition? The use and limitations of the characteristics approach to general practice research.

A range of easily identifiable characteristics is often used by researchers and general practitioners to categorise primary care practices. In the United Kingdom, for example, practices can be defined as dispensing, single-handed or training. The availability of routinely collected data has led to a growing research literature that links practice characteristics to their workload, performance and costs. This paper examines the use and limitations of this 'characteristics approach' and argues that this type of research is often undertaken because it is easy to perform rather than because it is the most appropriate way to study primary care. Using this approach may lead to failure to do the following: to account for the environmental factors that determine the effects particular characteristics manifest; to identify the true relationships between the observed characteristics; to control for changes in the effects of characteristics over time; to differentiate between the behaviour of individual members of a group with the same characteristic and that of the group as a whole; to assign the correct causality to relationships between practice characteristics, workloads, performance, and costs. The characteristics approach should be used with great caution by general practice researchers.