Summary rates.

The use of summary rates of reporting health data is discussed. Directly and indirectly adjusted rates are compared under two models of no interaction. A method of analysis is described whereby a model is first fitted to data in order to investigate assumptions, and subsequently the smoothed rates based on the model are summarized using an adjusted rate. We prefer directly adjusted rates when such a summarization is appropriate and the required data are available. An example is given which illustrates the method of estimation.