Building a composite score of general practitioners' intrinsic motivation: A comparison of methods

Objective: Pay-for-performance programmes have been widely implemented in primary care, but few studies have investigated their potential adverse effects on the intrinsic motivation of general practitioners (GPs) even though intrinsic motivation may be a key determinant of quality in health care. Our aim was to compare methods for developing a composite score of GPs' intrinsic motivation and to select one that is most consistent with self-reported data. Study design: A postal survey. Setting: French GPs practicing in private practice. Main Measures: Using a set of variables selected to characterize the dimensions of intrinsic motivation, three alternative composite scores were calculated based on a multiple correspondence analysis (MCA), a confirmatory factor analysis (CFA) and a two-parameter logistic model (2-PLM). Weighted kappa coefficients were used to evaluate variation in GPs' ranks according to each method. Results: The three methods produced similar results on both the estimation of the indicators' weights and the order of GP rank lists. All weighted kappa coefficients were >0.80. The CFA and 2-PLM produced the most similar results. Conclusions: There was little difference regarding the three methods' results, validating our measure of GPs' intrinsic motivation. The 2-PLM appeared theoretically and empirically more robust for establishing the intrinsic motivation score.