Multilevel assessment of immunisation uptake as a performance measure in general practice.

OBJECTIVE: To provide a measure of general practitioners' performance regarding uptake of immunisation against pertussis, taking account of the impact of patient characteristics on levels of uptake. DESIGN: Multilevel model of immunisation status against six measures of patient characteristics (level 1 predictor variables) with practice constraints as level 2 variables. SETTING: 126 practices in southern England. SUBJECTS: 2048 infants identified from infant surveillance and immunisation records. MAIN OUTCOME MEASURES: Top 10 practices with respect to uptake of pertussis immunisation according to a "null" model (model A) and according to a model that included six level 1 variables (model B); differences in ranking between the two models. RESULTS: Practices with low numbers of infants' requiring immunisation had imprecise and unstable uptake rates (range 0%-100%). With the multilevel procedure, after controlling for patient characteristics, practices in suburban catchment areas comprised largely of mature or young professionals performed best. Most improved performances when patient characteristics were taken into account were in practices in areas with a stable population and local authority housing--one such practice improved its ranking by 47 places. CONCLUSIONS: Crude uptake rates are inadequate performance indicators. Alternative approaches suggest that praiseworthy efforts to raise immunisation rates in unpromising areas are unrewarded by simple target based assessments.