Ep16-258-22 Cost-effectiveness analysis of 9-month MDR-TB treatment regimen, as evaluated within the STREAM trial. further presentation of differences in food-supplement spending and working-hours disaggregated by sex

Background: The STREAM Stage 1 trial demonstrated that the Short-regimen reduced health system costs by 20-25%, and was associated with an earlier return to work and a reduction in the purchase of supplementary food. The health economic data collected permit a num- ber of secondary analyses, including differences patient costs by gender, and probabilistic estimates of the likeli- hood that the Short-regimen is cost-effective. Design/Methods: Patient data were collected at 12- week intervals from the start of treatment up to week 132. Mean spending and mean working hours at each week were estimated by gender. Probabilistic sensitiv- ity analyses were conducted via boostrapping trial data for both Ethiopia and South Africa. The probability of cost-effectiveness was estimated for each country, at a range of willingness-to-pay thresholds for the trial pri- mary outcome. Results: Men spent $221 (Short-regimen) or $249 (Long- regimen) more than women on supplements. Although men reported working longer hours than women, the difference was not substantial. From a health system perspective, the probability that the Short-regimen is cost-effective is above 95% if the willingness to pay threshold for each additional favourable outcome is less than US$19,000 in Ethiopia and US$14,500 in South Africa. Conclusions: Our results suggest that expenditure on supplementary food is a more important driver of gen- der differences in MDR-TB patient costs than hours worked, irrespective of regimen. They also suggest that the Short-regimen is highly likely to be cost-effective, al- though the trial primary outcome was a composite for which a willingness-to-pay threshold was challenging to determine precisely. Further research is required to de- termine why more was spent by men than women on supplementary food, and to establish relevant willing- ness-to-pay thresholds for MDR-TB favourable treat- ment outcomes.