The intervention appeared appropriate. A brief description of the intervention and comparator was supplied, and the subsequent treatment options were clearly reported. The comparator was appropriate and represented standard care in the absence of the intervention. The authors justified the choice of treatment options included, stating that they represented the current locally recommended treatments.
The effectiveness estimates were clearly reported. The sources used to derive estimates were clearly reported but the methods used to select sources were unclear. The authors justified their choice of paper used to derive uptake rates in the non-registry cohort, stating that it was chosen because it was one of a limited number of studies that measured both the uptake of all three treatment options considered. It was unclear why this paper was chosen over the mentioned alternatives, how those papers were initially selected, and why studies reporting on a selection of the treatment options were rejected. Given the lack of a systematic method of selection for sources used to derive effectiveness estimates and the exclusion of potentially relevant evidence, it was unlikely that the estimates represent best available evidence. Several strong assumptions were adopted regarding the effectiveness of the intervention and the subsequent treatments. In particular the authors did not justify their assumption that the registry would result in 100% uptake of treatment. This assumption seems particularly strong for chemoprophylaxis and education, which both had uptake rates significantly lower than 100% in the non-registry cohort. The authors highlighted that a limitation of their study was the lack of available data, and stated that there were no randomised controlled trials of the effectiveness of a registry, or of the interventions. The reliance on non-randomised comparative data means that the results were more likely to be susceptible to bias.
The costs included were appropriate to the perspective. The costs were appropriate to the population and setting, having been derived directly from data from the hospital in which the registry was conducted. Future costs were appropriately discounted. The unit costs applied in the model were clearly reported.
Analysis and results:
Only limited details of the model were reported. In particular the model structure was not reported, so the appropriateness of the model structure could not be assessed. The results of the analysis were clearly reported. No justification for the choice of parameter ranges used in the sensitivity analysis was given. It was unclear whether these ranges accurately reflected expected parameter uncertainty. Only a limited univariate deterministic sensitivity analysis was conducted, which was unlikely to adequately assess the effect of parameter uncertainty on the results. A probabilistic analysis (in which parameter values are drawn simultaneously from assigned probability distributions in order to assess the effect of joint uncertainty on the results) would have provided a more accurate assessment but the data may not have been available to undertake such an analysis. The interpretation of the results should account for the ranges selected for the sensitivity analyses. Given that the effectiveness and cost inputs were taken from a specific Australian hospital setting, the reader should carefully consider the applicability of the study (in particular the parameter estimates) to their situation before generalising the results.
There are some limitations in the study methodology which mean the authors conclusions should be used with caution. In particular the analysis suffered from a lack of data and strong assumptions regarding efficacy.