The representativeness of sentinel practice networks

Background The representativeness of practice networks is important when using the information obtained to guide health policy. Aim To develop a model for examining the representativeness of practice networks. Methods Comparison of surveyed population, practice structure and prescribing characteristics with the national data using the Weekly Returns Service (WRS) for 2006 as an example of practice network. The population monitored was compared with the national PCT population. The practice postcode was linked to the Index of Multiple Deprivation and the distribution compared with the national equivalents. Doctor and practice-specific structural data (obtained by questionnaire) and practice-prescribing data were compared with the national equivalents. The significance of differences was evaluated using non-parametric tests. Results The WRS population was closely matched with the national data by age, gender and deprivation index. Compared with the national equivalents, WRS practices, included more younger GPs, had a larger average list per GP and fewer practices with a list of less than 1499 per GP. Prescribing patterns were similar to their PCT equivalents excepting for small reductions of antibacterial prescribing (items 7% and cost 5%). Conclusion We demonstrate a low-cost model methodology for examining the representativeness of practice networks using independent data with minimum practice input.