Analytical approach:
The authors used a Markov model that allowed obesity and the absolute risk of obesity-related disease to vary by age. The nine obesity-related diseases were: stroke, ischaemic heart disease, hypertensive heart disease, diabetes mellitus, osteoarthritis, post-menopausal breast cancer, colon cancer, endometrial cancer, and kidney cancer. The time horizon was lifetime. The authors stated that they adopted a health sector perspective.
Effectiveness data:
The effectiveness of the interventions was measured by short-term weight loss, which lowered a patient's risk of disease. The estimates at one year were from a recently published meta-analysis. The change in disease risk was calculated using the Potential Impact Fraction, with relative risks from the World Health Organization (WHO) and a study on body mass index and the risk of developing diabetes. The effectiveness of usual care was based on the expected obesity-related disease trends in Australia. The drug effectiveness included the proportion of patients who did not take their medication. Weight regain data were from the Sibutramine Trial of Obesity Reduction and Maintenance (STORM) trial for sibutramine and were assumed to be the same for orlistat. It was assumed that none of the weight loss was permanent.
Monetary benefit and utility valuations:
Health-related quality of life utilities were calculated, for the nine disease states, using disability-adjusted life-year (DALY) weights from an Australian study.
Measure of benefit:
The DALYs averted were the summary measure of benefit and they were discounted at 3% annually.
Cost data:
The cost data for the nine diseases were from Australian government sources. The cost categories included the costs of medication, general practitioner visits, average annual health care, and patient time and travel. Patient time was valued at 25% of the average wage in Australia, and the results were presented both with and without travel and time costs. All costs were reported in 2003 Australian dollars (AUD), with adjustments made using Australian government indices. They were discounted at 3% annually.
Analysis of uncertainty:
One-way sensitivity analysis was conducted by adjusting the weight regain assumptions, including halving the rate of weight regain; halving the linear weight gain over three years, following the cessation of therapy; assuming 23% of weight gain was permanent; and adding a utility gain of 0.017 per unit reduction in body mass index, in a year. Uncertainty was quantified by calculating 95% confidence intervals around the model outputs, using Monte Carlo simulation.