Interventions:
The authors justified their selection of the comparators. Trimethoprim-sulphamethoxazole was the preferred antibiotic, given its efficacy and low cost, but increasing resistance had discouraged its use. Fluoroquinolone was an alternative treatment, but was not recommended for uncomplicated UTIs to avoid the development of uropathogen resistance. Nitrofurantoin had been used as an alternative treatment, without developing high resistance.
Effectiveness/benefits:
A systematic review was performed to identify the inputs for the cost-minimisation analysis, but the methods and other details of the studies used were not reported. The mean values were not weighted by features, such as sample size, and the authors did not discuss the homogeneity of the studies. These issues should be considered when assessing the validity of the clinical data.
Costs:
The cost categories appear to have reflected the stated perspective of the third-party payer. It appears that all the relevant cost items were included and only rare events were not considered. Some unit costs together with ranges found in the literature were reported, while fewer resource use data were provided. The authors did not describe in detail the sources for the cost data, but it appears that standard US sources were generally used. The cost data were varied in the sensitivity analysis and this showed the impact of the drug costs. The price year was not explicitly reported, but might have been 2010.
Analysis and results:
The results were selectively reported, without the total costs and giving only the findings of the threshold analyses. The uncertainty was investigated in a deterministic analysis and the results were selectively presented. The authors stated that a limitation of their analysis was the high variability in values found in the literature and the lack of some data for the model. It was stated that this was the first model comparing the three antibiotics for UTIs. The findings appear to be specific to the authors’ context and cannot be generalised to other settings.
Concluding remarks:
The analysis used a cost-minimisation approach, despite differences in treatment efficacy. Within the limitations of this approach, the authors’ conclusions appear to be valid.