Bacteriological measures for the detection of cases of pulmonary tuberculosis*

Bull. Org. mond. Sante 1971, 45, 551-558 Bull. Wld Hith Org. Bacteriological Measures for the Detection of Cases of Pulmonary Tuberculosis* WILLIAM CHAN,1 MICHAEL CHIA,2 LO KWEE LEE,3 & DAVID M. MACFADYEN4 There is a need for more detailed injormation on the relative merits of the culture and direct sputum smear examination methods, separately and in various combinations, using one or more sputum specimens. The efficiency and relative cost of different bacteriological measuresfor the detection ofcases ofpulmonary tuberculosis, in the conditions of the Singa- pore tuberculosis control programme, are demonstrated. The findings provide guidance for deciding how, and in what order, these bacteriological measures should be used, and may assist planners and executives of national tuberculosis control programmes in the rational commissioning of laboratory facilities that already exist or are to be established. INTRODUCTION In a prevalence survey in India (Raj Narain et al., 1968), problems were encountered in defining a " case " of pulmonary tuberculosis. It was suggested that the feasibility of making a bacteriological classi- fication on the basis of smear and culture examina- tion of two sputum specimens should be explored. In the tuberculosis services in Singapore, two speci- mens are routinely collected from each patient re- gistered for treatment. The results obtained by smear and culture examination of each specimen have been used to group patients into various categories in order to compare the relative merits of different examinations in different combinations, in terms of efficiency and cost. MATERIALS AND METHODS Study population During the first half of 1969, 1178 persons were registered with the government health service, on the basis of a radiological examination, to be treated * The findings reported in this study are derived from a nation-wide investigation of treatment efficacy being con- ducted by the Tuberculosis Research Committee, Singapore. This committee consists of representatives of the Ministry of Health, the Tuberculosis Control Unit, Tan Tock Seng Hospital, the Singapore Anti-Tuberculosis Association, and the World Health Organization. The collection of data for the present report was facilitated by a WHO research grant and the processing of data was undertaken by the electronic data processing unit of the Ministry of Finance, Singapore. for tuberculosis. At the time of registration a small proportion of them, 226 (19 %), had already been treated for more than 1 month. Sixteen patients failed to provide the two sputum specimens re- quired for the study. The remaining 1 162 constitute the study population. Bacteriological procedures Two specimens were collected from each patient, one on each of 2 consecutive clinic days, in the pre- sence of a trained supervisor. When the specimen was inadequate, coughing and expectoration were induced by tickling the patient's larynx with a sterile laryngeal swab. A patient was considered to have failed to produce the requisite specimens if the second one was collected more than 7 days after the first. All specimens were examined by direct microscopy in a single laboratory and by culture in an adjacent laboratory. Smears were prepared direct from the sputum specimens without homogenization. A puru- lent or other suitable portion of sputum was smeared on to a slide with a sterile orange-stick and was dried and fixed by heat. The smears were stained with auramine and examined by fluorescence microscopy. After the results had been read and recorded, the negative slides were chemically cleansed for re-use 1 Chest Physician, Tan Tock Seng Hospital, Singapore. 2 Senior Registrar, Statistical Research Unit, Ministry of Health, Singapore. 3 Bacteriologist, Tuberculosis Control Unit, Singapore. 4WHO Senior Tuberculosis Adviser, Singapore. 2749 551 W. CHAN AND OTHERS and the positive ones were discarded. Specimens were cultured either on the day of collection or on the fol- lowing working day. Those not processed immedi- ately were stored in a refrigerator before culture. De- contamination was effected with an equal volume of 4% sodium hydroxide. After having been homoge- nized by shaking, the specimen was incubated at 370C for 20 minutes. It was then centrifuged at 3 000 to 4 000 rev/min for 15 minutes. The supernatant fluid was poured off, and the deposit was washed with distilled water and centrifuged again. The final depo- sit was inoculated, after mixing, on to each of two slopes of Lowenstein-Jensen medium without potato starch (Cruickshank, 1965). The cultures were in- cubated at 370C and were inspected for growth weekly, the results being reported as soon as growth appeared or, if there was no growth, at the end of 8 weeks. RESULTS The bacteriological results for the 1162 patients included in the study have been distributed accord- ing to various categories derived by dichotomous analysis (Table 1). Two results are considered for each smear examination-namely, positive and negative-but 3 results are considered for each cul- ture examination-positive, negative, and contami- nated. Thus, for the 2 specimens, there were (2 x 2) x (3 x 3) = 36 categories. All the data in this re- port are derived from the regrouping of these cate- gories. Correlation between the results of smear examination of the first and secoiid specimens The first sputum specimen was positive on direct smear examination in 428 (37%) and negative in 734 (63%) of the 1162 patients (Table 2). Smear examination of the second specimen in the 734 pa- tients initially found to be smear-negative yielded a further 72 positive results. Thus altogether 500 (43 %) of the 1162 patients had a positive smear and 662 (57 %) a negative smear, as assessed by examination of the 2 sputum specimens. The results of the cul- ture examination were then correlated to those of the smear examination in order to determine whether or not bacteriological classification on this basis would be reliable. Correlation between the results of smear examination anid the results of culture examination of the first and second specimens Among the 500 smear-positive patients (Table 3), the first culture was positive in 399 (80 %), negative in 44 (9 %), and contaminated in 57 (11 %). Among the 44 patients with a positive smear and a negative culture, the culture of the second sputum specimen was positive in 32 (73%), negative in 9 (20%), and Table 1. Bacteriological results from 2 sputum specimens from 1 162 patients examined by smear and culture First Second I First culture I Second culture Totalsmear smear result result result resulIt positive negative positive positive negative contaminated positive negative negative positive contaminated positive negative contaminated positive negative negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated positive negative contaminated 277 7 37 15 2 2 35 4 6 24 6 0 5 2 0 3 1 2 42 0 6 12 5 1 3 0 3 81 46 9 63 372 29 6 39 17 total 1 162 552 BACTERIOLOGICAL DIAGNOSIS OF TUBERCULOSIS Table 2. Correlation between the results of examining first and second smears in 1 162 patients with pulmonary tuberculosis Result Result of examining first smear of examining second smear positive negative total positive 385 72 457 negative 43 662 705 total 428 734 1 162 contaminated in 3 (7 %). As for the 57 patients with a positive smear and contaminated culture, the cul- ture of the second specimen was positive in 41 (72 %), negative in 5 (9%.) and contaminated in 11 (19%). Thus, in cases where the smear was positive and the initial culture was negative or contaminated, the re- sults of culture examination of the second specimen indicated that all but a small proportion of such patients were in fact excreting tubercle bacilli. There- fore, for practical purposes, it would be safe to assume that a patient who is smear-positive is bac- teriologically positive. The results of culture in the 662 smear-negative patients (Table 4) were then considered. The first culture was positive in 136 (20%) of these patients, negative in 464 (70 %), and contaminated in 62 (10 %). Thus culture examination of the first sputum speci- men in the smear-negative patients led to the identi- fication of an additional 136 cases. If these are added to the 500 who were classified as positive on the basis of 2 smear examinations, the number of bac- Table 3. Correlation between the results of culture of first and second specimens in 500 patients in whom 1 of 2 specimens was smear-positive Result 0 Result of first culture of second culture positive negative con- totalItaminated toa positive 343 32 41 416 negative 13 9 5 27 contaminated 43 3 11 57 total 399 44 57 500 Table 4. Correlation between the results of culture of first and second specimens in 662 patients in whom 2 specimens were smear-negative Result Result of first culture of second culture positive negative tained total positive 81 63 6 150 negative 46 372 39 457 contaminated 9 29 17 55 total 136 464 62 662 teriologically positive cases increases to 636. How- ever, the remaining patients were not all bacterio- logically negative: culture of a second specimen showed a further 63 (14%) of the 464 in whom the result of culture of the first specimen was reported to be negative, and 6 (10%) of the 62 in whom the first culture was contaminated, to be excreting tubercle bacilli, bringing the total number of bac- teriologically positive cases to 705 (61 %). Therefore, in almost one-quarter of the cases in which both smear examinations had given negative results, it proved incorrect to classify the patient as bacterio- logically negative. The same was true, though to a lesser degree, when diagnosis was based on smear examination of two specimens plus culture examina- tion of the initial specimen only. Alternative classifications of bacteriological status If the bacteriological status had depended on the result of the first culture, independently of the re- sult of smear examination, 535 (46%) would have been counted as positive, 508 (44 %) as negative, and, for 119 (10%), no classification would have been possible owing to contamination (summation of totals, Tables 3 and 4). The proportion of undeter- mined culture results might have been reduced by taking into account the results from a second specimen when the first was contaminated. By thus taking in each case the first culture result available (Table 1), the number classified as positive and negative would have been 582 (50 %) and 552 (48 %), respectively, and the number with contamination would have been only 28 (2%O). On the other hand, if a positive bacteriological status had been found on the basis of the first specimen only, taking any 553 554 W. CHAN AND OTHERS Table 5. Eight categories of bacteriological positivity in 1 162 newly registered tuberculosis cases: yield and relative cost Number Relative cost Specimens Diagnostic method of bacteriologically of bacteriological positive cases measures a smear examination only 428 1.0 initial specimen culture examination only 535 3.5 only smear examination plus culture examination, routinely in all patients 612 4.5 smear examination, with culture examination only in smear-negative patients 612 3.2 smear examination in all patients, with smear examination of a subsequent specimen in patients initially smear-negative 500 1.6 initial specimen plus a culture examination in all patients, with culture examination of a subsequent subsequent specimen in patients yielding a contaminated result on the initial culture 582 4.0 specimen smear examination in all patients, smear examination of a subsequent specimen in patients initially smear-negative plus culture examination of an initial specimen in patients smear-negative on both specimens 636 3.5 smear and culture examination routinely in both specimens in all patients 705 9.1 a According to Piot & Sundaresan, the estimated unit cost of smear examination alone is 3.20, and that of smear and culture exami- nation, 14.50. positive result whether by smear or by culture (Table 1), then the number of bacteriologically posi- tive cases would have amounted to 612 (53 %). Yield and relative cost of 8 alternative diagnostic methods The number of " bacillary " cases detected among the 1162 patients varied between 428 and 705 de- pending upon the number of specimens and the combination of diagnostic methods considered. The number of bacteriologically positive cases that would be detected by each of 8 different diagnostic methods is given in Table 5. The cost of each method has been calculated 1 and relative costs are tabulated in Table 5 together with the corresponding yields. Results of examining the first specimen only: (l) smear examination enabled 428 bacteriologi- cally positive cases to be detected; (2) culture examination led to the discovery of 535; (3) smear and culture examination, carried out routinely in all patients, revealed 612 bacteriologi- cally positive cases; and (4) the same number (612) would have been de- ' Piot, M. & Sundaresan, T. K. (1967) Unpublished docu- ment WHO/TB/Techn. Information/67.55. tected by routine smear examination in all cases, with culture examination only in smear-negative patients. Results of examining both specimens: (a) initial smear examination, performed routinely in all patients, with smear examination of a subse- quent specimen in patients initially smear-negative, would have yielded 500 cases; (b) initial culture examination in all patients, with culture examination of a subsequent specimen in patients yielding a contaminated result on the initial culture, would have enabled 582 cases to be identi- fied; (c) initial smear examination, done routinely in all patients, with smear examination of a subsequent specimen in patients initially smear-negative, plus culture examination of an initial specimen in patients negative on repeated smear examination, would have allowed the detection of 636 bacteriologically positive cases; and (d) smear and culture examination performed routinely in both specimens in all patients enabled 705 bacteriologically positive cases to be found. If the cost of examining one specimen from each patient by smear alone is taken as 1.0, then the re- lative cost of routine smear plus culture in a single specimen from all patients is 4.5 times as much. The BACTERIOLOGICAL DIAGNOSIS OF TUBERCULOSIS Table 6. Probability of obtaining positive, negative, or contaminated culture results on examination of a first and second specimen in 385 patients positive on smear examination of both specimens, in 115 positive on smear examination of one specimen, and in 662 negative on smear examination of both specimens Positive smear Negative smear both smears positive one smear positive, one smear negative both smears negative probability of probability of probability of probability of probability of probability of obtaining a given obtaining a given obtaining a given obtaining a given obtaining a given obtaining a given result on a single result on a subsequent result on a single result on a subsequent result on a single result on a subsequent culture examination culture examination culture examination culture examination culture examination culture examination positive 0.8549 positive 0.7904 positive 0.5664 positive 0.8416 negative 0.0348 positive 0.7261 negative 0.1377 positive 0.2160 negative 0.3811 contami- contami- contami- _ __nated 0.1111 nated 0.0719 nated 0.0524 positive 0.6875 positive 0.5897 positive 0.1183 negative 0.0416 negative 0.1250 negative 0.1696 negative 0.3590 negative 0.6956 negative 0.8078 contami- contami- contami- nated 0.1875 nated 0.0513 nated 0.0738 _~~~ positive 0.8000 positive 0.5000 positive 0.1282 conta- conta- conta- minated 0.1169 negative 0.0667 minated 0.1043 negative 0.0833 minated 0.0884 negative 0.5812 contami- contami- contami- nated 0.1333 nated 0.4167 nated 0.2906 estimated cost of a policy of examining a specimen by culture only in patients who are smear-negative on examination of 2 specimens is 3.5 times as much. The cost of routine smear and culture examination of both specimens in all patients is estimated to be 9.1 times as much as that of routine smear exami- nation of a single specimen. Thus a policy of placing full reliance upon the smear findings and ofexamining a specimen by culture only in those patients who are negative on repeated smear examination would have yielded 636 bacteriologically positive cases in the present series, whereas routine smear and culture of one specimen from each of the 1162 patients-a more costly policy-would have detected only 612 cases. Probability of obtaining positive or negative culture results in patients with a positive or negative smear In order to assess whether a policy of full reliance upon positive smear findings is justified, the proba- bility of obtaining a given culture result has been calculated, from the data given in Table 1, for the 500 smear-positive patients and for the 662 who were smear-negative. In making these calculations it has been assumed that the time sequence in collecting the two specimens did not influence the results ob- tained. The probability that smear-positive patients will yield a positive culture should ideally be unity and the probability of a negative culture should be zero. The actual findings when both specimens were smear-positive approximate closely to the ideal (Table 6), the probability of obtaining a positive re- sult from culture of a single specimen being 0.8416 and that of obtaining a negative result, 0.0416. Even in patients in whom one specimen was smear-posi- tive and the other smear-negative, the results were close to the ideal, the corresponding probabilities being 0.7261 and 0.1696, respectively. In contrast, in patients with both specimens smear- negative, the probability of obtaining a positive cul- ture result from examination of a single specimen was 0.2160 and that of a negative result, 0.6956. Thus culture examination of a single specimen sel- dom failed to confirm a positive smear finding, even when only 1 of the 2 specimens yielded a positive smear result, whereas, with patients who were nega- tive on smear examination of 2 specimens, there was 1 chance in 5 of obtaining a positive culture result. Ideally, the probability that patients who were culture-negative on examination of a single speci- 2 555 W. CHAN AND OTHERS men will yield a positive culture on examination of a subsequent specimen should be zero. The actual findings (Table 6) were quite different, at least for smear-positive patients, since the probability that pa- tients who were negative on a single culture examin- ation would yield a positive culture on subsequent examination was 0.6875 for those in whom both spe- cimens were smear-positive and 0.5897 for those in whom one specimen was smear-positive. Indeed, even in patients with both smears negative, the probability of obtaining a subsequent positive culture follow- ing a negative culture result on an initial specimen was 0.1183, or 1 chance in 8. Thus if culture exa- mination is used to assess the reliability of smear findings, results from more than one specimen must be available. DISCUSSION In a painstaking review of the global epidemiology of tuberculosis (Nagoya University, Department of Preventive Medicine, 1967), the incidence rates report- ed from 136 countries were compared. These ranged from 15 per 100 000 for Cyprus to 8 312 per 100 000 for French Somaliland. A major factor in the wide variation in the rates reported from country to country and from year to year is undoubtedly the lack of an accepted bacteriological definition of this disease. In only one country was the annual rate of disease incidence given for " bacillary " cases. However, even in such purely bacteriological terms, the definition of a case would vary with the number ofspecimens and with the bacteriological examination used. For instance, in the present series the number of bacillary cases detected by smear examination of a single specimen was 428 and the number detected in the same series by smear and culture examina- tion of 2 specimens was 705. Clearly it is essential to adopt a consistent definition of a " case " if com- parisons of incidence rates are to serve any purpose. Definition on the basis of smear examination is the simplest and cheapest. However, even with such a simple measure, it would be necessary to standardize the type of specimen examined and the microscopy system employed in order to be really precise. In the Singapore programme, sputum specimens are collected under supervision from persons attending the clinic, as this type of specimen is considered to be the most suitable for a community tuberculosis control programme. Again for practical reasons, fluorescence microscopy is used as this system is less costly than Ziehl-Neelsen microscopy when more than 30 specimens are examined daily (Mitchison, 1968). In any case-finding and treatment programme a consistent working definition is all that is required, the objective being to identify persons who are ex- creting tubercle bacilli and to ensure that they be- come non-infectious. In surveys of disease pre- valence an objective definition has been sought. Raj Narain et al. (1968) considered definition of the disease on the basis of smear examination to be un- suitable, in such survey work, in view of the high frequency of results falsely reported as positive be- cause of technical artefacts. In the present series 44 patients were smear-positive on examination of one or both sputum specimens but negative on initial culture examination; 32 (73 %) of them proved to be culture-positive on culture examination of a second specimen. Thus, when tubercle bacilli were identified on smear examination, culture examination of two specimens confirmed the reliability of the finding in all but a very small proportion of cases. This finding in patients who attend treatment centres in the routine way has probably more epidemiological relevance than a contrary finding in survey work, since the incidence of new cases is almost universally reported, whereas comparative prevalence rates are available for very few countries. Many infectious cases would be incorrectly classi- fied as non-infectious with a diagnosis based on smear examination, even when two specimens were examined. However this would also be true of a diagnosis based on the culture of a single specimen. In 662 patients, the negative result was based on smear examination of 2 specimens and, in 464, on smear examination of 2 specimens plus culture of the initial specimen. A subsequent culture examina- tion in each group indicated that 136 (20'/) and 63 (14%o), respectively, had been incorrectly classi- fied. Thus, for the precise determination of infec- tious cases, it would be necessary to undertake cul- ture examination of more than one specimen from each patient. However, the cost of routine smear and culture examination of 2 specimens was estimated to be 9 times that of smear examination of a single speci- men; therefore diagnosis on this basis would be possible only in countries with ample economic re- sources. Moreover, these estimates of cost have been made for a series in which the proportion of posi- tive results is high as a result of initial radiographic screening. The cost of case detection based on culture would be relatively greater if the frequency of positivity were lower. 556 BACTERIOLOGICAL DIAGNOSIS OF TUBERCULOSIS The prime purpose in collecting bacteriological data in a routine service is to identify infectious sources so that they may be rendered non-infectious. Bacteriological definition for statistical comparison is a subsidiary aim. Smear examination might fulfil both purposes in countries where resources are li- niited or where faiilure to eliminate the infectious sources of tuberculosis is widespread. Indeed, it is remarkably efficient, since the proportions of positive results from smear examination of 2 specimens (43 %) and culture examiniation of the initial specimen (47 0) were closely similar in the present series. This is confirmatory evidence for the statement of Mitchison (1968) that " smear examination, especial- ly of several specimens from each patient, is almost as efficient as culture examination ... in developing countries. . . ". However, it may be expected that, as facilities for smear examination become uni- versally available and treatment results more satis- factory, countries will wish to expand their bac- teriological facilities to provide for routine culture. A guide to the economical use of such facilities is provided by the present report. Routine smear and culture examination of the first specimen yielded 612 cases, but a greater number of cases would have been detected at a lesser cost by relying fully UpOIn the positive findings from1 smear examination of 2 specimens and of undertaking a single culture ex- amination in smear-negative patients. Confirmation by culture of a positive smear finding did not appear to be necessary in the present series, since the pro- bability of obtaining a negative culture result from a single specimen was very low, both when each of 2 specimens was smear-positive (0.0416) and when one was smear-positive and the other smear-negative (0.1696). On the other hand, for patients who were smear-negative on each of 2 specimens, the probabili- ty of obtaining a positive result on culture exami- nation of a single specimen was relatively high (0.2160). Thus the most profitable way of using cul- ture facilities would be to reserve them for persons who show symptoms of tuberculosis but are negative on repeated smear examination. In this article, various bacteriological measures for the detection of cases of pulmonary tuberculosis have been described. Clearly, as more ample re- sources become available, planners and executives of tuberculosis control programmes will prefer exami- nation methods that give a higher yield. The pre- sent analysis provides a guide to the more profitable use of culture facilities that already exist or whose establishment is planned. ACKNOWLEDGEMENTS The authors are grateful for the willing assistance given by all staff participating in this investigation. RESUME LES TECHNIQUES BACTERIOLOGIQUES DANS LA DETECTION DES CAS DE TUBERCULOSE PULMONAIRE Les resultats des examens microscopiques directs et des cultures de deux echantillons de crachats recueillis chez 1162 malades atteints de tuberculose pulmonaire, di Singapour, ont ete repartis dichotomiquement en 36 categories. On les a ensuite analyses afin d'evaluer, en termes d'efficacite et de coCit, les avantages respectifs des techniques bactriologiques, employees isolement ou en associations diverses. Selon les combinaisons, le nombre des cas bacilliferes decouverts a varie de 428 a 705. Le premier echantillon de crachats s'est revel positif a l'examen direct chez 428 inalades sur 1162. L'examcn d'un second specimen, chez les 734 patients initialement negatifs, a fait decouvrir 72 autres cas positifs. Au total, 500 malades ont donc ete reconnus bacilliferes par l'examen microscopique direct de deux 6chantillons. Les cultures ont confirme la fiabilite des resultats de la bacil- oscopie: bien qu'une premiere culture soit restee nega- tive chez 44 malades a frottis positifs, la culture du second echantillon a confirme que la grande majorite de ces sujets (73 %) excretaient des bacilles. On a mis en regard le resultat de la culture d'un echantillon unique et trois combinaisons de resultats d'examens directs (deux echan- tillons positifs; un echantillon positif et un negatif; deux echantillons negatifs): les probabilites d'obtenir un resultat positif a la culture etaient respectivement de 0,8416; 0,7261 et 0,2160. La culture d'un echantillon unique confirrne donc, tres generalement, la positivite de l'examen direct, menme lorsqu'un seul des deux frottis montre la presence de bacilles. Par contre, si les deux frottis sont negatifs, une culture unique n'a qu'une chance sur cinq de donner un resultat positif. Si le cout de l'examen direct d'un echantillon de cra- chats est estim6 arbitrairement a 1,0, le cout relatif de l'examen direct et de la culture d'un echantillon unique 557 558 W. CHAN AND OTHERS est 4,5 fois plus eleve si ces examens sont effectu6s chez tous les malades et 3,2 fois plus eleve lorsque la culture n'est pratiquee que chez les sujets a frottis negatifs. En faisant entierement confiance a la bacilloscopie et en ne recourant a la culture d'un echantillon que pour les sujets negatifs aux deux examens directs, on aurait decouvert 636 cas bacilliferes dans la serie de malades etudiee. Par contre, en procedant chez chaque malade a un examen direct et a la culture d'un echantillon pratique plus couteuse - on n'aurait detecte que 612 cas bacilliferes. Quant a l'examen de routine de deux echantillons par frottis et par culture, pratique chez tous les malades - association qui donne les meil- leurs resultats (705 cas decouverts) -, son coOt est 9,1 fois plus eleve que celui de l'examen direct d'un seul echantillon. REFERENCES Cruickshank, R. (1965) Medical microbiology, 11th ed., Edinburgh & London, p. 753 Mitchison, D. A. (1968) Bull. int. Un. Tuberc., 41, 139 Nagoya University, Department of Preventive Medicine (1967) Global epidemiology of tuberculosis, Nagoya Raj Narain, Nair, S.S., Naganna, K., Chandrasekhar, P., Ramanatha Rao, G. & Pyare Lal (1968) Bull. Wld Hlth Org., 39, 701