Data analyses

Missing data (25.1%) were imputed five times using predictive mean matching in SPSS, version 22 (23) and the costs and effects were bootstrapped separately for each imputation dataset in Microsoft Excel 2010. Reported costs and effects after 12 months of intervention were pooled means of imputed datasets and confidence intervals were estimated with the percentile method (i.e., using 2.5th and 97.5th percentile as the lower and upper bound, respectively). Results of the bootstrap analyses are presented in incremental cost-effectiveness planes and cost-effectiveness acceptability curves (CEACs) (24). In the planes, the differences in costs are presented on the vertical axis and the differences in WC and QALYs on the horizontal axis. Dots in the lower right quadrant are most favorable for the intervention, indicating more effectiveness and less costs compared to the control condition. The CEACs indicate the probability that the intervention is cost-effective, given societal willingness to pay for 1 cm reduction in WC or one QALY gained. In sensitivity analyses, CEACs are explored for persons with complete data. Budget impact estimates were calculated by multiplying projected numbers of participants in the different uptake scenarios with per-participant costs and adding over the 5-year time horizon.