Analytical approach:
A Markov model, with a time horizon of one year, was used to combine the published data. The authors reported that the perspective was that of the health care system.
Effectiveness data:
The effectiveness data were from systematic reviews of pressure ulcer prevention, a survey of pressure ulcer prevalence in Canadian hospitals, a large study of home care in Ontario, and other literature searches. The main effectiveness parameter was the relative risk of pressure ulcers with or without the pressure-redistribution mattresses. These estimates were from a randomised trial of emergency department patients, that was identified by two published systematic reviews of pressure ulcer prevention.
Monetary benefit and utility valuations:
The utility estimates were from studies that used the Health Utilities Index (HUI). These estimates were identified by the searches performed for the effectiveness data.
Measure of benefit:
The benefit measure was quality-adjusted life-years (QALYs) gained.
Cost data:
The direct costs included the mattresses and the treatment of pressure ulcers, which included ward care, drugs, overheads, capital depreciation, and home care upon discharge from hospital. The costs for the mattresses were from a telephone survey of three mattress manufacturers. The cost of treating pressure ulcers in hospital was from 47 teaching and community hospitals in Ontario; the costs for patients with pressure ulcers were compared with those without them. The home care costs were similarly calculated from a large database of home care in Ontario. The price year was 2009. All costs were reported in Canadian dollars (CAD).
Analysis of uncertainty:
One-way sensitivity analyses were undertaken by varying the key model inputs over plausible ranges. The impact of including effectiveness estimates, for pressure-relieving mattresses, in health care settings other than the emergency department was assessed, and a hospital perspective was analysed. A probabilistic sensitivity analysis, with distributions applied to all the model parameters, and value-of-information analyses were undertaken.