We took a closer look at one of the case studies from Millions Saved - following references to make sure that the account given by the primary sources matched the picture given by the case study. The case study we examined was Case 3: Controlling Tuberculosis in China.
What was done?
Did it work?
Data collection: China conducted randomized national surveys of TB prevalence in both 1990 and 2000.5
Analysis: A study by the China Tuberculosis Control Collaboration (2004) used regression analysis to compare provinces with DOTS to provinces without DOTS. It used regression analysis to attempt to control for a variety of differences in these provinces not related to changes brought about by the program.6 The dissent (more about this below) summarizes the finding of this study:7
Dissent and possible alternative explanations: A comment on the study by the China Tuberculosis Control Collaboration (2004) notes that the 1990 and 2000 surveys were different in many ways, including smaller sample size for the 2000 survey, thus making the comparison less than ideal. It also observes,8
Its observation of the available data shows that the DOTS provinces were indeed higher-income, although not dramatically so:9
Despite these reservations, the authors conclude,10
Bottom line: The impact demonstration didn't have the rigor of a randomized controlled trial, but such an expectation would be unreasonable for a "macro" story such as this one. The evidence is fairly compelling and Levine's (2007) conclusion about this program's impact is, on the whole, reasonable.
What did it cost?
$130 million, roughly half of which was covered by the government and half of which was covered by a grant from the IDA arm of the World Bank.11
This matches the figure in the Levine (2007) case study.12
- Chen, Xianyi, et al. 2002. The DOTS strategy in China: Results and lessons after 10 years (PDF). Bulletin of the World Health Organization 80(6):430-436.
- Levine, Ruth. 2007. Case 3: Controlling tuberculosis in China (PDF). In Case studies in global health: Millions saved. Sudbury, MA: Jones and Bartlett.
- China Tuberculosis Control Collaboration. 2004. The effect of tuberculosis control in China. Lancet 364(9432):417-422. Available at http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(04)16764-0/fulltext#article_upsell (free registration required; accessed July 14, 2009). Archived by WebCite® at http://www.webcitation.org/5uYku5mws.
- Squire, S. Bertel, and Shenglan Tang. 2004. How much of China's success in tuberculosis control is really due to DOTS? Lancet 364(9432):391-392. Available at http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(04)16777-9/fulltext?_eventId=login#bib1 (free registration required; accessed July 14, 2009). Archived by WebCite® at http://www.webcitation.org/5uYl0TBj1.
- The third nationwide random survey for the epidemiology of tuberculosis in 1990. 1992. Zhonghua Jie He He Hu Xi Za Zhi 15(2):69-71, 125. Abstract available at http://www.ncbi.nlm.nih.gov/pubmed/1394592 (accessed July, 14 2009). Archived by WebCite® at http://www.webcitation.org/5uYlAkoyI.
- World Bank. 2002. China - Infectious and endemic disease control project: Implementation completion and results report (PDF). Washington, DC: World Bank.
"The Bank was the only financier of the project but development, implementation, and supervision of the project would also require effective cooperation with the World Health Organization (WHO)." World Bank 2002, Pg 3.
World Bank 2002, Pg 4.
Levine 2007, Pg 1.
China Tuberculosis Control Collaboration 2004 claims 13, citing Chen 2002, Pg 2, which says "In 1996, Sichuan province split off a municipality called Chongqing. Thereafter, there were 12 provinces and one municipality in the project. To simplify the presentation of data, however, we have reported project results using the old undivided province of Sichuan."
- See "The third nationwide random survey for the epidemiology of tuberculosis in 1990" 1992 for 1990 study. China Tuberculosis Control Collaboration 2004, Pg 422, Reference 8 also cites the 1990 study but we haven't found it online.
- "Diagnosis" section of China Tuberculosis Control Collaboration 2004, Pg 418 describes the medical methods of assessing TB.
- "The 1990 survey used stratified random sampling as in 2000." China Tuberculosis Control Collaboration 2004, Pg 419.
"In multivariate regression, we fitted a separate model for each year, with age (in 5-year classes), sex, urbanisation, region, and project included as categorical variables, together with terms for the interactions between project and sex, project and urbanisation, and project and region." China Tuberculosis Control Collaboration 2004.
Squire and Tang 2004, Pg 391.
Squire and Tang 2004, Pg 391.
Squire and Tang 2004, Pg 391.
Squire and Tang 2004, Pg 392.
"The total project cost at appraisal was US$271 million equivalent. This was to be financed by an IDA credit of SDR 95.9 million, equal at the time to US$129.6 million. The provincial governments were to contribute US$139.2 million equivalent to financing the provincial components of the project. The central government was to finance US$2.2 million equivalent for the central component. The final total project cost of US$264.9 million is 97.7% of the total cost that was estimated at appraisal. This small change in total project cost was due to the exchange rate fluctuation between the SDR and US$ from appraisal to project closing, although 100% of the credit (SDR 95.5 million) was expended. The final financing for the project was US$126.7 million from IDA, US$137.5 million from the provincial governments, and US$0.7 million from the central government. ... The actual expenditure by component was US$129.7 million for TB, compared to US$102.5 million at appraisal, US$139.7 million for schistosomiasis, compared to US$145.4 million at appraisal, and for the central component, US$2.7 million equivalent, compared to US$4 million equivalent at appraisal. As noted earlier, much of the increased expenditure on TB was used for improving case detection and management, and was financed largely by savings on the cost of drugs. Those savings stemmed from the reduced need for drugs to the delayed start up of TB efforts, rapid decline in endemicity of schistosomiasis, and much lower than expected cost of drugs that were procured domestically." World Bank 2002, Pgs 12-13.
"The total cost amounted to nearly $130 million in the end." Levine 2007, Pg 6.