Interventions:
The interventions compared were reasonably simple and were adequately described, although it was unclear whether there any comparators had been omitted or whether there was only one available type of each needle.
Effectiveness/benefits:
The methods used to identify and select the data were not reported, which meant it was not possible to make any assessment on validity or whether the best available data were used. The selection of data values was not discussed; for example, it was not clear why the probability of post-lumbar puncture headache was taken from two different sources when the two interventions were compared in the single study which was used for cutting needle probabilities alone. The focus of the analyses presented was costs, even though the cost differences were driven by clinical effectiveness differences. A more comprehensive discussion surrounding the validity of these estimates would have reduced the high levels of uncertainty.
Costs:
The costs included in the analysis were relevant to the stated study perspective and were well reported. The sources of resource use data were provided and appeared relevant to the study setting. As with the effectiveness estimates, it was not always clear why certain data had been selected; there appeared to be a mixture of Medicare costs and Medicare reimbursement rate. Whether these were actual differences or just terminology, and whether this would have had an impact on the conclusions reached was not clear. The method of inflation was appropriate and reported. Discounting was not appropriate, so not performed. Overall, the level of reporting surrounding the costs was adequate.
Analysis and results:
The analysis was undertaken using a decision model, which was adequately reported. The focus was clearly on cost outcomes, so it was unclear whether patient preference or differential impact on health-related quality of life during the procedure existed, which could be incorporated into the model. However, with no clinical utility of the tests evaluated, the results could only be seen as partial. The selection of model inputs appeared ad-hoc, so the level of uncertainty may be high. Uncertainty analysis was addressed using appropriate methods. The base-case results and results of sensitivity analysis were adequately reported. However, given the lack of reporting surrounding the derivation of clinical data, the results were uncertain.
Concluding remarks:
The focus of the model was cost outcomes and, as a consequence, the derivation of the clinical estimates driving the difference in costs was not well reported, which makes the results presented uncertain.