Part I: Organisation and management
introduction

Tuberculosis bacteriology is one of the fundamental aspects of a national tuberculosis control programme and a key component of the DOTS strategy, yet the tuberculosis laboratory service is often the most neglected component of these programmes.

Diagnosis of tuberculosis and monitoring of treatment progress rely heavily on bacteriological examination of clinical specimens. The usefulness, priority and scope of the various techniques used in tuberculosis bacteriology depend on the epidemiological situation prevailing in individual countries and on the resources available.

Microscopy
Despite recent advances in mycobacteriology, early laboratory diagnosis of tuberculosis still relies on the examination of stained smears. Microscopy of sputum smears makes a particularly important contribution since the technique is simple, inexpensive and detects those cases of pulmonary tuberculosis who are infectious, ie. those responsible for maintaining the tuberculosis epidemic. Currently no other diagnostic tool is available which could be implemented affordably.

Smear sensitivity*: Direct smear microscopy using acid-fast stains is generally considered to be a relatively insensitive diagnostic procedure, with the reported sensitivity ranging from 25% to 65% when compared to culture. What is not generally appreciated, however, is that smear sensitivity varies with the type of lesion, the type and number of specimens, the mycobacterial species, staining technique and the alertness and persistence of the microscopist. In addition, the above sensitivity refers to bacteriological diagnosis of pulmonary TB, but smear examination identifies the cases which are sources of infection to the community, with a sensitivity of approximately 90%.

*Sensitivity =Capacity of a diagnostic test to distinguish correctly in a population those individuals who have the disease, ie. the true positives

The minimum number of acid-fast bacilli necessary to produce a positive smear result has been estimated to be between 5 000 and 10 000 per millilitre (Table 1). Sputum specimens from patients with pulmonary tuberculosis - particularly those with cavitary disease - often contain sufficiently large numbers of acid-fast bacilli to be readily detected by direct microscopy. The sensitivity can further be improved by examination of more than one smear from a patient. Many studies have shown that examination of two smears will on average detect more than 90% of infectious tuberculosis cases, both in low- and high-prevalence countries. The incremental yield of acid-fast bacilli from serial smear examinations has been shown to be 80%-82% from the first, 10-14% from the second and 5-8% from the third examination.

Table 1. Numbers of acid-fast bacilli observed in smears, concentrations of culturable bacilli in sputum specimens and probability of positive results¹

¹ David HL. Bacteriology of mycobacterioses. US Department of Health, Education and Welfare, Public Health Service, Communicable Disease Centre, Atlanta, USA, 1996.

Considering the type of specimen, overnight sputum is better than sputum collected on the spot; for operational reasons, however, it is often necessary to depend on both. The recommended policy for sputum collection for case-finding by microscopy is, therefore, as follows:

• one spot specimen when the patient first presents to the health service
• one early morning specimen (preferably the next day)
• one spot specimen when the early morning specimen is submitted for examination

Smear sensitivity is very low in extrapulmonary and childhood tuberculosis and in most cases of disease caused by mycobacteria other than tubercle (MOTT) bacilli. In high-prevalence countries these conditions are, however, far less common than pulmonary tuberculosis and are not rapidly progressive or highly infectious. From a public health perspective, infectious pulmonary cases have the highest priority for detection and treatment until cure.

Smear specificity*: When discussing the specificity of acid-fast smears, a distinction must be made between those instances where the specimen truly does not contain tubercle bacilli (ie. false-positive smears) and those where tubercle bacilli present in the specimen fail to grow in culture (ie. false-negative cultures).

*Specificity=Capacity of a diagnostic test to distinguish correctly in a population those individuals who do not have the disease, ie. the true negatives

When false-negative cultures are excluded, a residual number of genuine false-positive smear remain. Direct smear microscopy for diagnosis of acid-fast bacilli has a specificity of more than 98%. The preventable causes of false-positive smears are multiple: Mycobacteria often contaminate the water or solutions used in staining, while transfer of acid-fast organisms from positive to negative smears may occur during the staining process or via microscope immersion oil or objective lenses. False positive smears have also been reported to result from failure to properly filter and regularly replace reagents, from staining of artifacts present in the smear, from insufficient de-staining of non-acid-fast organisms and from retention of acid-fast stain by nonmycobacterial organisms.

Auramine staining procedures may produce a higher false-positive rate than carbolfuchsin methods, presumably because auramine may stain inanimate objects. Studies conducted in laboratories in Africa showed that about 1% of false positive smears could be attributed to transfer of organisms and another 1% to various administrative errors.

Careful quality control of microscopy procedures and review of positive smears can markedly decrease the incidence of false-positive smears.

Predictive value*: Sensitivity and specificity are not the only determinants of the value of microscopy as a diagnostic test. Microscopy can produce results of varying accuracy in varying epidemiological situations, even though the sensitivity and specificity remain constant.

*Predictive value = Probability of having the disease among those classified as positive (positive predictive value) or probability of those not having the disease among those classified as negative (negative predictive value)

The prevalence of tuberculosis has a decisive influence on the value of microscopy as measured by the predictive value: a high predictive value (90% or more) can be achieved when the prevalence of tuberculosis in the population tested is 10% or more. With such prevalence rates, high predictive values can be obtained with a test of high specificity, even if its sensitivity is low (as is the case with acid-fast microscopy). The prevalence of tuberculosis among adult patients attending health centres with complaints of prolonged chest symptoms in high prevalence countries is usually in the order of 10%, which provides the basis for microscopy as a good diagnostic test for tuberculosis in these countries.

Most patients with infectious tuberculosis have respiratory symptoms and the use of smear microscopy in those presenting to health services with suggestive symptoms constitutes the most efficient means of case detection.

Culture
Examination by mycobacterial culture provides the only definitive diagnosis of tuberculosis. However, the usual microbiological techniques of plating clinical material on selective or differential culture media and subculturing to obtain pure cultures cannot be applied to tuberculosis bacteriology. Compared with other bacteria which typically reproduce within minutes, Mycobacterium tuberculosis proliferate extremely slowly (generation time 18-24 hours). Furthermore, growth requirements are such that it will not grow on primary isolation on simple chemically defined media. The only media which allow abundant growth of M. tuberculosis are egg-enriched media containing glycerol and asparagine (LJ), and agar/liquid media supplemented with serum or bovine albumin.

Depending on the type of culture medium and decontamination method used, as few as ten viable bacilli can be detected. Culture increases the number of tuberculosis cases found, often by 30-50% and detects cases earlier, often before they become infectious.

If specimens from non-sterile body sites are not decontaminated, tubercle bacilli will easily be overgrown by more rapidly dividing organisms, eg. bacteria and fungi. Selective decontamination of specimens is, therefore, required to destroy rapidly growing contaminants. If these procedures are not properly controlled they may adversely affect the viability of tubercle bacilli, resulting in false-negative cultures. In general, if less than 2% of a laboratory?s mycobacterial cultures become overgrown with bacterial or fungal contaminants, the decontamination procedure is overly harsh and may be inhibiting/preventing growth of tubercle bacilli. On the other hand, false-negative cultures may also result when inadequate decontamination procedures allow overgrowth of the medium by contaminating organisms.

False negative cultures may also occur if there are inordinate delays between specimen collection and processing that allow progressive dying-off of tubercle bacilli. Lastly, false-negative cultures may result when incubation is not done for a full eight weeks, since some tubercle bacilli (particularly some drug-resistant strains) may require extended periods of incubation to produce visible growth.

Culture is much more costly than microscopy, requiring facilities for media preparation as well as skilled staff. Culture should be used selectively, in the following order of priority:

Drug susceptibility testing
Drug susceptibility testing of M. tuberculosis isolates is of considerable value for epidemiological purposes. The prevalence of resistance in new tuberculosis patients (primary resistance) is a good measure of the efficiency of treatment services and guides the choice of regimens in treatment programmes. Trends in resistance among previously-treated patients (acquired resistance) point to failures in control programme management. Susceptibility tests may also be of value for individual patients if there is a failure during chemotherapy or a relapse after successful treatment. However, routine susceptibility testing of cultures from new patients imposes an unrealistic burden on laboratory services and with the documented success of four-drug treatment regimens (even in the presence of single-drug resistance), its benefit cannot be justified.

Species identification
Identification of mycobacterial species may have value in countries where tuberculosis incidence is low and where a substantial number of mycobacterial isolates may be of other species, particularly in HIV-infected patients. However, in developing countries where more than 85% of the disease burden is due to tubercle bacilli, species identification other than M. tuberculosis is of little value.