The use of propensity score methods in psychiatric research

The use of propensity score methods has become increasing widespread in public health, in medicine and in many other settings involving research with observational studies. The methods provide a useful and straightforward way of simultaneously controlling for many variables that potentially confound the relationship between a specific outcome and the treatment or exposure under study. The methods assist the researcher in avoiding certain dangers in more traditional regression modelling and, in the case of binary outcomes, allow for the control of a greater number of potential confounders. The use of propensity score techniques has been fairly limited in psychiatry, largely to studies of health care delivery and use. An introduction to propensity score methods is given and a series of examples illustrating their use in psychiatric research is presented.