Analytical approach:
A Markov model was developed to simulate the treatment of disease under the two options based on different pain levels. The time horizon of the analysis was 1 year. The authors stated that the perspective of the health care payer was adopted in the study.
Effectiveness data:
The clinical data were derived from a large, randomised double-blind clinical trial that enrolled 368 patients. This study compared usual care (placebo) against three different dosages of pregabalin (150, 300 and 600 mg/day). The duration of follow-up was 13 weeks. The four different patient groups were similar at baseline. Assumptions were necessary to extrapolate 12-week results to a 1-year time horizon. The key clinical estimates were pain scores, which were transformed in transition probabilities to be fitted in the model.
Monetary benefit and utility valuations:
Utility valuations were derived from a Belgian prospective observational study of 88 patients who completed the Short-Form 36 (SF-36). The results of the SF-36 were converted to utility weights using the SF-6 algorithm.
Measure of benefit:
The summary benefit measure was the quality-adjusted life-years (QALYs). These were estimated using the decision model.
Cost data:
A breakdown of the cost items was not provided. Thus, it was not clear which cost categories, other than drug costs, were included. In general, costs were associated with each health state of the Markov model. The cost data were derived from a Belgian observational study of 88 patients, whose medical charts were prospectively reviewed over a 1-month period. Drug dosages were based on the data from the RCT. The costs were in euros (EUR). The price year was 2003.
Note: since this abstract was written the author has confirmed that the cost data included all costs (drugs, tests, consults, hospital stay) from the perspective of the payer.
Analysis of uncertainty:
A probabilistic sensitivity analysis was undertaken to address the issue of variability in the cost estimates, the range of which was very wide because of a small proportion of patients requiring hospitalisation. The sensitivity analysis focused also on variability in utility valuations. Both the costs and utilities were assigned probabilistic distributions (beta and normal, respectively).