Test characteristics
The technique has been used for only a few years and has only recently been standardized. In a blinded study utilizing RFLP to evaluate strains at the time of relapse in a material in Hong Kong60, 4%,of strains evaluated (27% of those noted to be different) were erroneously read. Use of this tool is currently evolving and test characteristics are difficult to evaluate. In addition, a "gold standard" for evaluation has not yet emerged.
Chest Radiography
Screening with chest radiography has been widely used in the past (and continues in many industrialized countries) for the purpose of detecting cases of tuberculosis. Only cases of tuberculosis in the chest (comprising 85 to 90% of cases in most locations) can be detected in this fashion. In addition, chest radiography has limited ability to distinguish those cases clnically active from those which are inactive (previously active cases which have now been cured). It is also limited in distinguishing tuberculosis from a small number of other conditions which might have similar radiographic features.
Test characteristics
Various studies have evaluated the use of chest radiography as a diagnostic or screening tool61 62. The definitive study was an international collaborative study undertaken by the International Union Against Tuberculosis63 64 65 based on chest radiographs selected from the community-based mass radiography survey in Norway.
In a study of 15,000 students, the sensitivity of chest radiography in detecting active tuberculosis was 96%, specificity 99%; positive predictive value 61%; negative predictive value greater than 99%. The low positive predictive value was a function of the low prevalence of disease. The IUAT study determined an "index of disagreement" among 90 readers. This varied from 28% disagreement on the presence of a cavity to 60% disagreement concerning the presence of abnormal lymph nodes. Surprisingly, there was 31% disagreement on the need for medical action.
A comparison of the level of disagreement, using the same methods of analysis,between chest radiography and sputum smear microscopy showed strikingly better performance of the smear microscopy66.
Clinical characteristics
Some patients are designated tuberculosis cases and treated for the disease, even in the absence of bacteriological confirmation. Criteria for the establishment of a diagnosis of pulmonary tuberculosis under such circumstances have been established many years ago67. To be designated a "case" of tuberculosis ("active" case in the classification), a patient with pulmonary disease whose bacteriological examinations were negative requires "serial postero-anterior chest radiographs which indicate progression without, or regression with, treatment at intervals of six months or less", in addition to having a significant reaction to the tuberculin test. An evaluation of such cases68 indicated that only 35% of those in whom there was no bacteriological confirmation of disease were eventually classified as having active tuberculosis based on regression of clinical and radiographic features under treatment, as compared with regression of radiographic features in 76% in whom cultures were positive. In this study, there were no clinical features which could adequately predict whether a patient would be ultimately judged to have active tuberculosis. Another study in East Africa69 identified prolonged (greater than 3 weeks) cough, chest pain (more than 15 days), absence of phlegm, and absence of breathlessness as key predictors of an eventual diagnosis of tuberculosis in those who were sputum smear negative. Unfortunately, this study included,in addition to pulmonary tuberculosis, other forms of respiratory tuberculosis and is thus not entirely comparable.
Test characteristics
In those judged to have tubereulosis based on radiographic criteria, in the absence of bacteriolological confirmation, the sensitivity of the diagnosis was 69%; specificity 76%; positive predictive value 69% and negative predictive value 76%. In those judged to have tuberculosis from among suspects, based on clinical symptoms, estimated sensitivity was 85%, specificity 67%; positive predictive value 43% and negative predictive value 94%.
The inaccuracy of clinical diagnosis based on radiographic features results in a serious inaccuracy in designating cases70. Moreover, even among those cases actually diagnosed, many are never included in the official statistics71.
3. Measuring infection
The tuberculin test is used to identify infection with Mycobacterium tuberculosis72. The recommended method is that developed by Mantoux which consists of intradermal injection of purified protein derivative of a standard type (either Tuberculin RT23 or Tuberculin-Seibert). Some 48 to 72 hours after injection of the material, the largest transverse diameter of induration is measured which is the result of the test which is to be recorded. This induration is reflective of the delayed type hypersensitivity due to infection with Mycobacteria and is caused by an influx of T-cells into the dermis where the material has been injected. Induration in reaction to the intradermal test may be the result of a number Mycobacteria species, notably M bovis as a result of vaccination, and, more significantly, of a variety of environmental mycobacteria which cause reactions of a smaller diameter. Reactions due to vaccination may wane over time and often have only a small effect on the estimation of prevalence of tuberculous infection. Reactions due to environmental mycobacteria, alternatively, may have a major effect. In addition, there are a number of methodological errors which may occur in the administration (injecting the material subcutaneously) or reading (terminal digit preference) of the test.
Test characteristics
Characteristics of the tuberculin test have been summarized73. In a large material obtained from recruits to the United States Navy74, the following test characteristics were determined: sensitivity 94%, specificity 99%. At a prevalence of 10%, Positive predictive value was 88% and negative predictive value 94%. However, where reactions from environmental mycobacteria are common, as noted in the review
