above, the test characteristics change: sensitivity 94%, specificity 91%, positive predictive value 53%, negative predictive value 99%.

Table 1 summarizes test characteristics of the methods outlined.　

　

C.　Definitions and classification in tuberculosis

　

Standardized definitions and classification of tuberculosis were developed many years ago⁷⁵.　These have been revised from time to time, most recently by the World Health Organization in 1991⁷⁶.

　

1.　Tuberculosis deaths

The most important cause of death from tuberculosis is the failure to diagnose the disease before the patient has died³⁶.　This appears to be due to a simple failure of medical practitioners to consider tuberculosis in the differential diagnosis.

As an epidemiological endpoint, the nature of the death of tuberculosis patients must be precisely defined: tuberculosis may be the cause of the death (a minority of cases), may only contribute to the death, or may be unrelated to the cause of death³⁶.　Mortality statistics may comprise a mixture of these.　Prior to the introduction of chemotherapy for tuberculosis, mortality data were more reliable as an indicator of the level and trend of disease; since that time, mortality data have become unreliable.

　

2.　Tuberculosis disease

For purposes of reporting, any patient judged by a health worker to have tuberculosis and commenced on chemotherapy is to be reported as a case of tuberculosis⁷⁶.　However, cases should be reported separately according to the location of their disease (pulmonary or extrapulmonary) and the results of bacteriologic examination.　It is recommended that, to establish a diagnosis of tuberculosis, a case should have two separate types of evidence of disease from the following: smears; cultures; typical radiographic, pathologic or clinical evidence.

For purposes of international comparison, cases bacteriologically confirmed should be reported separately⁷¹, indicating whether only confirmed on direct microscopy or additionally confirmed on culture.

　

3.　Tuberculous infection

Tuberculous infection in man is evidenced by a significant degree of induration in response to the tuberculin skin test.　The strengths and limitations of the tuberculin skin test have already been enumerated.　Nevertheless, the presence of a reaction to tuberculin is associated with an increased probability of development of tuberculosis and the greater the size of induration in response to the tuberculin skin test, the greater the likelihood of subsequently developing clinical tuberculosis^{77 78 79}.　Determining the presence or absence of infection using the results of tuberculin skin testing is made difficult as the distribution of reaction size may or may not be normal and the likelihood of infection can only be inferred according to probability; the results of tuberculin skin testing are not categorical.　Nevertheless, various "cut-points" in the normal distribution have been selected for various purposes to indicate a sufficient probability of infection⁸⁰.

　

4.　Epidemiological categories

Table 2 indicates the epidemiological categories of tuberculosis.　These categories have been validated in a number of epidemiological studies^{39 81 82}, identifying groups according to differing prognoses, defined by: presence or absence of pathological changes due to, or clinical history of tuberculosis; presence or absence of significant response to tuberculin; and presence or absence of vaccination.

　

D.　Distribution of tuberculosis

　

1.　Person

　

Age

Tuberculosis fatality rises sharply with age³⁶, to a large degree, a function of death due to causes other than tuberculosis, although death due to tuberculosis also rises with age.

The association between age and development of tuberculosis disease varies according to the epidemiological situation.　Where current transmission of infection is high, disease is most common in young adults (aged 20-30)^{9 83}.　Where current transmission has been substantially reduced, disease is most frequent in older persons^{84 85}; this has been convincingly shown to be due to a cohort phenomenon related to the higher probability of having been previously infected among earlier birth cohorts⁸⁶.　Even when the disease is more frequent in older persons, the probability of developing disease among those already infected is greatest in young persons⁸⁷.

The relationship of age to the probability of becoming infected with Mycobacterium tuberculosis has been estimated by Styblo⁸⁸.　No matter whether the prevalence of tuberculosis is high or is declining, an individual who will become infected is most likely to become infected before the age of 20 years.　This is understandable when tuberculosis is extremely common and most members of the community become infected at some point in their lives.　When tuberculosis is at a very low level and has been so for a prolonged period, the majority of the population remains uninfected.　Under these circumstances, the likelihood of becoming infected may differ from this pattern, at least in subsets of the population, as has been the demonstrated in the poor, inner-city areas of large cities in North America^{43 89 90}, much as Heimbeck demonstrated for student nurses in Norway in the 1920s and 1930s⁸ when universal infection in childhood began to diminish.

前ページ　　　目次へ　　　次ページ