Analytical approach:
The cost-effectiveness analysis was based on a seven-state Markov model of the ongoing risk of melanoma, progression, and death, over a lifetime. The authors stated that they took a societal perspective.
Effectiveness data:
The key effectiveness measures were the hazard ratio for invasive melanoma and the relative risk of squamous cell carcinoma. These were from the Nambour Skin Cancer Prevention Trial (NSCPT); a randomised controlled trial. An intention-to-treat analysis, with Cox proportional regression, was used to generate the hazard ratio. Each patient received a dermatological evaluation at the start of the trial in 1992 and then again in 1994 and 1996. The events that were included in the analysis were the incidences of basal cell carcinoma, squamous-cell carcinoma, and melanoma. Follow-up data on potentially cancerous skin lesions was obtained from medical records up to 2006 and synthesised with the trial data.
Monetary benefit and utility valuations:
The utility weights for different stages of melanoma were from standard gamble surveys, time trade-off surveys, or expert opinion.
Measure of benefit:
Quality-adjusted life-years (QALYs) gained were the summary measure of benefit. The benefits were discounted at 5% annually.
Cost data:
The costs included those of the health care provider and the household. The health care provider costs included monitoring, sunscreen, and subsequent health care costs. The household costs included time, out-of-pocket payments for sun screen, and the discounted lifetime cost of a melanoma diagnosis. The costs were from an analysis of data from the NSCPT and from other published Australian studies. All costs were inflated to 2010 Australian dollars (AUD) and they were discounted at 5% annually.
Analysis of uncertainty:
A sensitivity analysis included the effects of treatment on squamous-cell carcinoma. One-way sensitivity analyses were conducted on each of the model parameters over a range of plausible values. Threshold analysis was conducted for the key parameters to determine the parameter value at which the programme became cost-effective. Probabilistic sensitivity analysis was performed to assess the overall uncertainty in all the model parameters and the results were displayed on a cost-effectiveness plane and a cost-effectiveness ellipse.