Conclusions
Regarding the cost-effectiveness of STI control, the position this chapter takes is "it depends." The health benefit in terms of numbers of disability-adjusted, discounted, healthy life years saved by curing or preventing a case of syphilis varies from 3 years in a person who has ceased all sexual activity to as many as 161 years in a sex worker with two partners a day. The cost of treating that prostitute for syphilis varies from US$5 to US$100. Thus, the cost per DALY of syphilis treatment can range from 100/3 or US$33 per DALY to 5/161 or less than US$0.05 per DALY.
As we learn more about the complexities of delivering STI treatment services and take into account the diversity of risk behavior, the ease with which STI interventions can be ascribed a simple cost-effectiveness ratio has declined. If no easy way to summarize experience to date with a simple cost-effectiveness ratio is available, how should we analyze economic investments in STI treatment? We believe that the way forward is a better understanding of why STI treatment and other health services vary so much in terms of their efficiency and effectiveness from one setting to another. By studying the determinants of this variation, we should gain an improved understanding of the full costs of high-quality STI service delivery and its place in the health sector investment picture.
Notes
1. The Gini coefficient is a measure of inequality that here we apply to income. If income is distributed equally in the population, the coefficient is equal to 0, and if a few individuals hold almost all the wealth, the coefficient is close to 1.
2. Given that WHO has expended enormous efforts to estimate discounted years of life lost, disability years lost, and DALYs lost as a result of STIs, the simplest and most direct approach for computing the DALY benefits of preventing or curing an STI in a single patient would be to use WHO's years of life lost, disability years lost, or DALY per case assumptions. Alternatively, one could simply divide WHO's aggregate values of these indicators by the incidence rate of each disease in each region to obtain the estimated burden per incident case. Unfortunately, neither the case-specific burden numbers nor the incidence rates that correspond to the DALY aggregates are available from WHO.
3. In contemporaneous data, STIs can either affect or be affected by HIV prevalence. To focus on the effects of an STI on HIV infection, we lag infection by an STI by seven years. Though partially correcting for simultaneity bias, this strategy does not allow us to identify whether lagged STI prevalence is directly affecting HIV infection or only serving as a proxy for the risky sexual behavior that drives both epidemics.
4. The samples include two measures of HIV prevalence (low- and high-risk groups) for some countries. These measures enable us to expand the sample used in the column (1) regression from 40 countries to 56 separate observations. Equations are estimated with Stata's cluster option to correct the standard errors of the coefficients for the correlation between the errors on separate observations from the same country. The variable urban high-risk dummy is used to shift the intercept coefficient for the high-risk sample in comparison with the low-risk one.
