© 2008 American Thoracic Society
Decline in Prevalence of Latent Tuberculosis Infection: Is the Waning of Tuberculin Reaction a Factor?From the Authors:We appreciate the opportunity to highlight the similarities and discuss the differences between our two articles on the epidemiology of tuberculosis (TB) infection in the United States population, which represent two independently conducted analyses (1, 2). Both our articles were based on data from two National Health and Nutrition Examination Surveys (NHANES). The first was conducted in 1971–1972; the second was designed and implemented by the authors of one of the articles (2) and was conducted in 1999–2000. First and foremost, it is reassuring that the prevalence estimates of TB infection reported in both analyses are nearly identical, with differences falling well within the surveys' margins of error. The most notable difference between the analyses relates to the projected number of persons infected with TB nationwide in the 1999–2000 survey. Dr. Khan and colleagues restricted their analysis to survey participants who had a tuberculin skin test (TST) placed and read. They applied the NHANES 1999–2000 Medical Examination Center 2-year sample weights, which adjust for unequal probability of selection and nonresponse, to the household interview and physical examination portions of the survey. This procedure resulted in a projected population size of 227 million people. The finding of a 4.2% infection prevalence led to a corresponding estimate of 9.5 million persons infected with TB. Dr. Bennett and colleagues further adjusted the Medical Examination Center 2-year sample weights for nonparticipation in the TST portion of the survey. TB infection rates were then applied to the respective U.S. 2000 Census population groups. This resulted in estimates based on the entire civilian, noninstitutionalized United States population, or 268 million people; the 4.2% prevalence led to an estimate of 11.2 million persons infected with TB. (Details on how sample weights and prevalence estimates were calculated by Bennett and coworkers can be found in the online supplement to their article [2].) Despite this difference, it is reassuring that each of our reports arrive at similar conclusions: (1) the United States has experienced substantial declines in the national reservoir of TB infection over the past 30 years; (2) the prevalence of infection among the nation's foreign-born population is now 10 times higher than that of the U.S.-born population; and (3) in addition to domestic TB control strategies, TB elimination in the United States will require further efforts to control and prevent TB in high-burden countries. In his letter, Dr. Rieder points out that, despite attention to training TST readers, our data show an apparent "digit preference" for 10-, 15-, and 20-mm TST readings. He suggests a mixture analysis with a Bayesian approach as a statistically "more solid" way to estimate infection prevalence. We acknowledge that other analytic approaches for modeling TB infection data could have been used. In our analyses, we chose to use an approach that could be directly compared with current and historical literature and control guidelines for TB infection in the United States. Our data on reactions to PPD-B, which reflect potential cross-reactivity with nontuberculous mycobacteria, did not result in a significantly different tuberculosis infection estimate, and this reduces the potential contribution of one major cause of concern. The entire data set can be accessed at http://www.cdc.gov/nchs/about/major/nhanes/exam99_00.htm. In their letter, Drs. Leung and Yew speculate that some of the observed U.S. decline in the prevalence of TB infection might be attributable to a diminishing TST response with advancing age. By comparing birth cohorts from the mid-1920s to mid-1940s (i.e., those 25–44 yr of age in 1971 and those 55–74 yr of age in 1999), Leung and Yew highlight a decline in the prevalence of TB infection (9.3 to 6.3%, respectively). This hypothesis is intriguing and could potentially be a contributing factor. Consistent with this hypothesis are studies demonstrating the limitations of one-step tuberculin skin testing (which was the method of testing in both NHANES surveys) among elderly persons (3). However, it is also important to consider that the compositions of these birth cohorts and their risks of TB infection are likely quite different due to a combination of effects, including shifting immigration patterns (e.g., in 1971–1972, 93% of the weighted population was born in the United States compared with 87% in 1999–2000), attrition from emigration and mortality, and changes in the incidence of active TB in the United States (4) and globally (5) over time. A competing explanation would be that these cohorts differ in significant ways that make it difficult to predict the magnitude of decline in TB infection over time.
Centre for Research on Inner City Health
Centers for Disease Control and Prevention
Centers for Disease Control and Prevention
Centre for Research on Inner City Health
Centers for Disease Control and Prevention
University Health Network FOOTNOTES Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
