Interventions:
The treatments for overactive bladder were included in the model were clearly defined. However, the authors acknowledged that there were formulations of oxybutynin and tolterodine (and additional potential agents) that were not compared in the model. It was not completely clear which drug would normally be prescribed in the study setting; standard practice should be an essential comparator for local decision-making. Given the large number of alternative treatments available, the assumption of no further treatment options after the failure of second-line therapy may not be realistic.
Effectiveness/benefits:
The authors chose the VECTOR trial to inform transitions in the model because it directly compared oxybutynin and solifenacin. A mixed-treatment meta-analysis, including both direct and indirect evidence for all potential treatments would have been more appropriate for deriving treatment effectiveness. The authors stated that there were several transitions within the model that were assumed to have a zero probability due to no events occurring in the VECTOR trial; it was unclear whether these assumptions were realistic, as a lack of events may be a factor of the VECTOR trial's small sample and short duration. The model also assumed that patients could not transition between health states in second-line therapy and after treatment discontinuation; the basis for this was unclear.
The utility measurement tool was not stated. Several key parameters in the model were based on assumptions. It was unclear whether the assumptions were justified and whether all assumptions were tested in sensitivity analyses.
Costs:
Costs appeared to have been generally derived from appropriate sources. The price year was clearly stated and methods of cost inflation were referenced. Resource use for general practitioner and outpatient visits were assumed; the validity of these assumptions may not be applicable to all settings.
Analysis and results:
The analysis of uncertainty was limited. Many parameters in the model were not varied, and the one-way and scenario sensitivity analyses gave no clue as to the likelihood of any result. A full probabilistic sensitivity analysis using Monte Carlo simulation would have to evaluate uncertainty would have been useful. The results were clearly reported.
The authors conducted a thorough discussion, comparing their work to that of others and acknowledging the limitations of their model and its remaining uncertainty.
Concluding remarks:
The study excluded most of the potential treatments for overactive bladder. There remained significant uncertainty in the model parameters, methodology and structure; the authors' conclusions appropriately reflected this.