Denominator estimation: Approaches in the Hamburg paediatric sentinel network

Study objective - The aims were to develop an estimator for the size of paediatric practices to be used as a denominator for purposes of comparison; to analyse the age structure of the patients attending paediatric practices and to check the necessity for an age specific denominator; and to validate the denominator information by other available data. Design - This was an observational study. Setting/participants - A sentinel network was set up comprising 26 self selected paediatric practices. Weekly patient contacts in relation to age and sex were counted three times during the study period of two years. In addition, accounting data, including the total number of children treated in a given three month period (quarter), were available. Main results - Weekly patient contact counts were stable over time, not in terms of the absolute number of contacts but in the rank positions of the practices (r ^s = 0.86) and in their age structure. The age distribution of weekly patient contacts differed significantly between the practices. Cross validation of the weekly contact count by means of the quarterly accounting data resulted in a rank correlation of r^s = 0-90. Conclusions - Sentinel networks with paediatric practices should use age specific denominator information. Weekly contact group, estimated by counts in a sample of weeks, is a stable and easily available denominator for sentinel practices in the context of the German health care system.