Analytical approach:
A Markov model was constructed with time horizons of 10 and 15 years and yearly cycles. The authors stated that a payer’s perspective was used in accordance with the National Institute for Clinical Excellence guidance.
Effectiveness data:
The clinical data were obtained from systematic reviews of the literature and expert opinion. MEDLINE, EMBASE, Ovid, and Cochrane database searches were performed (1980 to 2006), without language restrictions, and a manual search of the article references was also completed. The authors searched for comparative and non-comparative studies. These studies were critically appraised by two authors and some were selected on the basis of published recommendations regarding their inclusion in decision analytic models. Four experts were also consulted. A weighted mean was obtained for each parameter, and ranges from experts were used in the sensitivity analysis. The main clinical outcome was failure rates. These were extracted from 12 studies and two systematic reviews including 880 patients in total (103 received PS, and 777 received DS). No other details of the study designs were reported.
Monetary benefit and utility valuations:
Quality-adjusted life-years (QALYs) were derived from SF-36, Gastrointestinal Quality of Life (GIQL), and Wexner incontinence score data using “validated conversion mathematics”. The data were obtained from the selected literature and expert opinion.
Measure of benefit:
The measure of benefit was QALYs. Although, a net health benefits (QALYs) versus willingness to pay (threshold at £10 000) line graph and incremental cost-effectiveness ratio (ICER) scatter plots were also reported.
Cost data:
The costs included those of the surgical procedure, anorectal physiology, endoanal ultrasound, and outpatient appointments. All costs were based on the British National Health Service reference costs for 2004 to 2005, and were expressed in UK pounds sterling (£).
Analysis of uncertainty:
A one-way sensitivity analysis was performed for all variables. The following scenarios were also run: 10- and 15-year time horizons; a subgroup of patients with obstetric anal sphincter injury (OASI) only, with the lowest plausible range of failure probabilities used for the DS group; and a group of patients under 40 years old, with the lowest range of probability failures used for both the PS and DS groups. Each scenario was run for 10,000 iterations and the ranges were reported, but the probability distributions were not.