OBJECTIVES: Clinical effectiveness of non-randomized studies can be influenced by confounding factors. One of the methods that can be used to get a precise estimate of a clinical effect from non-randomized clinical trials is a propensity score. The aim of the study was to evaluate the possibilities of using propensity score for assessing the efficiency of health technologies and its practical use in the cost effectiveness analysis.

METHODS: In the study we retrospectively reviewed the records of 615 patients with atrial fibrillation (228 patients were treated with conventional pharmacology therapy and 387 patients were treated with catheterization ablation). Two methods of propensity score matching were used: nearest neighbour and caliper matching. Balanced data sets were further used to evaluate treatment efficacy (percentage of patients with sinus rhythm at the end of follow-up). Costs were assessed from the perspective of health care payer and then used to calculate cost-effectiveness.

RESULTS: In the original data set of 615 patients, the sinus rhythm in the conventional treatment group was at 9.20% patients and in the ablation treatment group at 86.3% patients. The CEA was for the conventional group 4660 and for the ablation group 3001. Using nearest neighbour matching method, we created data set of 196 patients in both treatment groups. The CEA was for the conventional group 4660 and for the ablation group 3022. Using caliper matching method, we created data set of 124 patients in both treatment groups. The CEA was for the conventional group 4420 and for the ablation group 2973.

CONCLUSIONS: The results of the two balanced data sets were not significantly different from the results of the original data set but using propensity score helped to get a more precise estimate of the clinical effect on the basis of proved removal of bias from the original data.