Analytical approach:
A decision-analytic model was used to synthesise the data from key published studies and epidemiology reports. These data were primarily from the Clinical Study on Tacrolimus Ointment Over the Long-Term (CONTROL Study; Thaci, et al. 2008 and Wollenberg, et al. 2008, see ‘Other Publications of Related Interest’ below for bibliographic details). The time horizon was 12 months and the authors stated that they took a UK NHS perspective.
Effectiveness data:
The key clinical outcomes were the days in disease exacerbation, the discontinuation rate, and the time until discontinuation. These data were from the CONTROL Study for maintenance treatment, manufacturer trial reports (Data on file at Astellas Pharma Europe R&D. Study references FG-506-06-40 and FG506-06-41, 2007), and a relevant publication (Zuberbier, et al. 2006, see ‘Other Publications of Related Interest' below for bibliographic details) that included patients with moderate-to-severe atopic dermatitis. Adverse events were assumed to be equivalent for the two strategies.
Monetary benefit and utility valuations:
Utility data were used to adjust the survival days and applied to the period of disease exacerbation. These values were from the adult participants in the CONTROL Study and a 2005 published health technology assessment for children.
Measure of benefit:
The measure of benefit was quality-adjusted life-years (QALYs).
Cost data:
The direct medical costs included the tacrolimus ointment, consultant dermatology and out-patient visits, phototherapy, topical steroids, oral medications, and in-patient stays. Published sources were used to measure and value these resources. These included MIMS 2009, NHS National Reference Costs, and the Personal Social Services Research Unit 2008. The costs of discontinuing tacrolimus were from 16 clinical experts. Costs were reported in UK pounds sterling (£).
Analysis of uncertainty:
The model parameters were examined in one-way and probabilistic sensitivity analyses. The one-way analyses on all parameters used ranges of values that were the 95% confidence interval limits in the literature. The results were reported in a table and the text.