Analytical approach:
: A semi-Markov model was used to compare the cost-effectiveness of the two options to prevent hospital-acquired pressure ulcers among patients admitted to hospital. The time horizon for costs was until patient discharge, up to a maximum of one year. The time horizon for outcomes was the lifetime of the patient. The authors reported that a societal perspective was adopted.
Effectiveness data:
The effectiveness data were extracted from the literature indexed in the MEDLINE database. The main effectiveness estimate was that of the intervention in reducing the incidence of hospital-acquired pressure ulcers. These data were from a published study.
Monetary benefit and utility valuations:
The utilities for the health states were based on European Quality of life (EQ-5D) questionnaire scores. All utilities were divided by 365 to produce quality-adjusted life-days to match the cycle length. The data were from published studies and other sources.
Measure of benefit:
Quality-adjusted life-years (QALYs) were the measure of benefit and they were discounted at an annual rate of 3%.
Cost data:
The direct costs included those of the interventions (including risk assessments, support surfaces, chair cushions, nutrition, repositioning, dealing with moisture and incontinence, and unforeseen costs); the treatment of deep tissue injuries; the treatment of stage I or II hospital-acquired pressure ulcers; and the treatment of stage III or IV hospital-acquired pressure ulcers. The costs of preventing and treating of stage I or II ulcers were from a micro-costing study performed by the authors. Those of stage III or IV ulcers were from published studies. The costs of a deep tissue injury were assumed to equal the costs of daily prevention. The price year was 2009 and all costs were reported in US dollars ($).
Analysis of uncertainty:
One-way sensitivity analyses were conducted to examine the impact of variations in the assumptions on the outcomes. The base-case estimates were varied by ±15%. Threshold analyses were performed for some inputs by changing their values by more than ±15%. A probabilistic sensitivity analysis was undertaken, by applying a distribution to each variable, to assess the uncertainty in all the model parameters simultaneously.