Tuberculosis and War. Группа авторов

probabilities for progression from one step to the next in the chain of events are determined by various risk factors, both exogenous (such as exposure and acquisition of infection) and endogenous (such as progression from latent infection to overt clinical disease) [4].

      This model will guide the following discussion on the challenges and difficulties in obtaining an accurate picture of the epidemiologic situation of TB during warfare using selected examples to illustrate attempts in ascertaining transmission of M. tuberculosis, TB, and death from M. tuberculosis, addressing both numerator and denominator issues that are always present in epidemiology, but become accentuated during wartime.

      Transmission of Tubercle Bacilli

The extent to which M. tuberculosis is transmitted during a given period of time – the incidence of infection with M. tuberculosis – determines the subsequent incidence of TB (the disease). The latter determines TB fatality (the magnitude modified by chemotherapy) in the individual and thus TB mortality in the population. Because TB has no defined incubation period, the extent of transmission during a specified period of time also has long-term consequences as latent infection may progress to TB many years, even decades, after acquisition of infection. This is demonstrated in studies among untreated contacts (placebo-recipients in a preventive therapy trial) of newly diagnosed TB patients in the United States [5, 6]. The risk of progression was highest during the first 2 years following infection, and dropped to about one tenth of the initial annual risk in the subsequent years, but it never became zero as is the case with diseases that are characterized by a well-defined incubation period. Thus, the extent of transmission that occurs has both immediate and long-term implications for expected TB morbidity subsequent to acquisition of infection with M. tuberculosis.

Fig. 1. Model used in the discussion of the epidemiology of tuberculosis, based on the classification of tuberculosis by the American Thoracic Association and the Centers for Disease Control and Prevention. Adapted from [2, 3].

      Tuberculin Skin Test

The tool to measure infection with M. tuberculosis has long been the tuberculin skin test, but currently is gradually being replaced by the introduction of interferon-gamma release assays [7]. Various methods of applying tuberculin have been developed, mostly in the first decade of the 20th century, chiefly percutaneous [8] and intracutaneous testing [9, 10]. Infection with M. tuberculosis results in a cell-mediated immunologic response that can be elicited by a tuberculin skin test. Like any test, the tuberculin skin test has the intrinsic operating characteristics of sensitivity and specificity. Lack of sensitivity (i.e., the proportion with a false-negative result) is determined by the immune status of the person tested and is also reduced by faulty tuberculin reagents and technical errors [11–13]. The most prominent causes for lack of specificity (i.e., the proportion with a false-positive result) are cross-reactions attributable to infection with other mycobacteria, notably several types of environmental mycobacteria and the Bacille Calmette-Guérin (BCG) vaccine strain M. bovis BCG. The predictive value of a positive test diminishes rapidly with decreasing prevalence of the sought-after condition if the operating characteristics remain unchanged.

The estimated annual incidence of M. tuberculosis infection has dropped below 10% in industrialized countries by about 1910, and became even substantially lower by the time of World War II (WWII) [3]. In other words, a relatively small and increasingly smaller problem would require at least 2 sequential tuberculin surveys 1 year apart to determine the annual incidence of transmission of infection by M. tuberculosis. During World War I (WWI), BCG was not yet in use anywhere, but by WWII, most European countries – with the notable exceptions of Germany [14], the Netherlands [15], and the United Kingdom [16] – had begun introducing BCG vaccination nationally in a systematic manner. Thus, prior BCG vaccination adversely affected the tuberculin test specificity, and thus the predictive value of a positive test result in a large number of countries in Europe. Of lesser importance among younger population groups, repeat tuberculin testing can boost the response and falsely suggest incident infection [17]. To solve the problem of extracting incidence information from serial prevalence measurements [18, 19], the incidence of infection was approximated by the calculation of an average annual risk of infection, algebraically derived from single or repeat tuberculin skin test prevalence surveys [20, 21]. Whereas this approach ingeniously circumvents these difficulties, it has an intrinsic problem that it calculates an average over the lifetime of a person (Fig. 2). A tuberculin skin test survey is conducted among persons born in year b at calendar time b + a, where a is the age of the birth cohort under consideration. The annual risk of the observed prevalence of infection with M. tuberculosis is the algebraically derived average (using a compound interest calculation approach) of the probability of becoming infected (or reinfected) during the course of 1 year. By convention [22], it is put on the calendar time in between the years of birth and survey conducted as a pragmatic approximation. The actual incidence of infection, however, may have decreased or increased over the life span of the tested individual, or even taken a more complex course, and there is no way to actually know its precise current level. It follows that the younger the tested population, the closer the calculated annual risk of infection approximates to the actual incidence of infection, while conversely, the older the population tested, the less certain one can be about the actual magnitude of current transmission. Thus, a recent increase in infection incidence will show up only with a substantial delay and is averaged out to a much-diluted value.

Fig. 2. Difference between calculated average annual risk of infection and actual incidence of infection with M. tuberculosis.

Fig. 3. Secular trend in the average annual risk of infection with M. tuberculosis in selected European countries. Adapted from data given in [20, 23–27]. A straight line (The Netherlands, Norway) is only