Interventions:
Comprehensive details of the intervention and comparators were clearly reported. The most appropriate comparator, standard care, was included in the analysis. No other relevant alternative interventions were discussed.
Effectiveness/benefits:
The effectiveness estimates, and the methods used to derive them, were clearly reported. Future effects were not discounted, which was reasonable given the short time horizon. Given the likely impact on patients' quality of life, inclusion of a quality of life measure would have provided a more comprehensive assessment of the benefit of the intervention. The centres participating in the trial were selected based on their need to improve quality and safety indicators; these factors may make the results obtained less generalisable to other settings. The selection and randomisation of patients was sufficient, but it was not clear if the lack of blinding could have introduced bias.
Costs:
The included costs were appropriate for the adopted perspective, but the cost of training was not included in the analysis. Since PUPP included a unique training component, the lack of inclusion of this cost may have underestimated the cost of the intervention. The costs and their sources were clearly reported. The costs were specific to the setting. Future costs were not discounted, which was reasonable for the short time horizon. Costs were appropriately inflated to 2010 prices.
The authors stated that the cost of hospitalisation was not included because it was not applicable to the nursing facilities; this narrow perspective assumption may limit the usefulness of the results. The authors recommended that future studies should assess the cost impact to the hospital and payer, which the authors hypothesised would be very large.
Analysis and results:
The decision model was clearly described including a diagram. Results of the analysis were clearly reported. Key details of the clinical trial, including the study design, patient characteristics and inclusion criteria, were clearly reported. An appropriate method of randomisation was used to assign participants to treatment groups. A possible limitation of the study was the short time horizon and the selection of centres. If significant differences in costs and health effects could be expected beyond six months, the time horizon may have been inadequate to assess the cost-effectiveness of the intervention.
Only a limited threshold analysis was completed. As such, it was not possible to make any definitive conclusions on the robustness of the results. Ideally, a multivariate deterministic or probabilistic sensitivity analysis that assessed the effect of single or joint parameter uncertainty on the results should have been completed.
Concluding remarks:
The study methodology and reporting was adequate. The authors' conclusions appear to be appropriate, but the robustness of the results is unclear.