1Relationship between acid-fast bacilli (5.0 mL) and culture for diagnosis of pulmonary tuberculosis (N = 330). http://phcfm.org/index.php/phcfm/article/downloadSuppFile/240/639
2Relationship between acid-fast bacilli (2.0 mL) and culture for diagnosis of pulmonary tuberculosis (N = 330). http://phcfm.org/index.php/phcfm/article/downloadSuppFile/240/640
ResultsThe majority of the patients, 110 (32.1%), were aged 18–29, 95 (28.8%) were aged 30–40, 68 (20.6%) were 41–51 years old, 38 (11.5%) were aged 52–62, 21 (6.4%) were 63–73 years old and 2 (0.6%) were aged 74–94. Of the 330 patients, 146 (44.2%) were male and 184 (55.8%) were female. The prevalence of AFB positive sputum results in the 2.0 mL samples was 17.6% (58/330) and 17.9% (59/330) in the 5.0 mL samples. Sixty patients (18.2%) were positive and 270 (81.8%) were negative for the AFB test in the 5.0 mL category, whilst 58 (17.6%) were positive and 272 (82.4%) were negative for the AFB test in the 2.0 mL category. The gold standard AFB culture results revealed 77 (23.3%) as positive and 53 (76.7%) as negative sputa. In the 5.0 mL sputum samples, out of the total of 77 specimens that were positive according to the gold standard culture test, 59 tested positive on AFB microscopy, resulting in a sensitivity of 76.6% (59/77), (95% CI 66.0% – 84.7%), (Table 1). Meanwhile, according to the culture test, 253 specimens tested negative whilst 252 tested negative by microscopy, missing one specimen (false negative), resulting in a specificity of 99.6% (252/253), (95% CI 97.8% – 99.9%). The positive predictive value (PV+) was 98.3% (59/60), (95% CI 91.1% – 99.7%) whilst the negative predictive value (PV-) was 93.3% (252/270), (95% CI 89.7% – 95.7%). The likelihood ratio (LR) for a positive microscopy in the 5.0 mL volume sputa was 192 [76.6 ÷ (100 - 99.6)], whilst the LR for a negative test was 0.23 [(100 - 76.6) ÷ 99.6].Table 2 demonstrates that out of the 77 sputum smears (from the 2.0 mL specimens) which tested positive by culture, 58 were picked up by microscopy, resulting in a sensitivity of 75.3% (58/77), (95% CI 64.6% – 83.6%). Meanwhile, out of the total of 253 specimens that were negative according to the culture test, 251 tested negative by microscopy and two were missed (false positives), resulting in a specificity of 99.2% (251 253), (95% CI 97.1% – 99.8%). The positive predictive value (PV+) was 96.7% (58/60), (95% CI 88.6% – 99.1%) whilst the negative predictive value (PV-) was 93.0% (251/270), (95% CI 89.3% – 95.4%). The likelihood ration (LR) for a positive microscopy in the 2.0 mL specimens was 94 [75.3 ÷ (100 - 99.2)], whilst the LR for a negative microscopy was 0.25 [(100 - 75.3) ÷ 99.2].The 5.0 mL sputum smear test gave false negative results in 23.4% (18/77), (95% CI 15.3% – 34.0%) patients whilst the 2.0 mL sputum smear test gave a false negative value of 24.7% (19/77), (95% CI 16.4% – 35.4%). There were 0.4% (1/253), (95% CI 0.07% – 2.2%) false positive results encountered for 5.0 mL and 0.8% (2/253), (95% CI 0.2% – 2.8%) for the 2.0 mL sample. DiscussionThe majority of the patients (60.9%) were between the ages of 18 and 40 years, of whom 53.7% were 18–29 years old. The incidence in the 18–40 age group in this study was slightly higher than a similar study conducted in Mexico (52%). 11We were focussing on PTB suspects on clinical grounds, hence a wider scope, whereas their focus was mainly on radiological changes. The fact that TB suspects are in the young adulthood age-group (hence breadwinners), has economic implications for the society. This places the onus on the clinician to properly diagnose each TB suspect by using the sputum quantity with a better yield, to enable early therapeutic intervention. The prevalence rate of the AFB positive sputa by microscopy was virtually similar in the 5.0 mL versus the 2.0 mL sputum specimens (17.9% versus 17.6%, respectively). The smear sensitivity rate of 76.6% obtained using the 5.0 mL sputum specimen, and 75.3% obtained using the 2.0 mL specimen, from patients suspected of having TB in this study, was in keeping with the upper limit of the range of sensitivity for smear microscopy reported by Foulds and O’Brien to vary from 30% to more than 70%.12 Matee et al., in a study conducted in the Muhimbili University in Dar es Salaam, Tanzania, sought to determine the diagnostic accuracy of sputum microscopy for active case findings of HIV-associated PTB using TB culture as the reference standard. No mention is made of the quantity of sputum used. However, their overall sensitivity and specificity of sputum microscopy was 61.8% and 99.7%, respectively, 13which were lower than found in this study. This could be ascribable to the relatively smaller sample of culture positive sputum results in this study compared to their sample (77 versus 212). In a study conducted by Warren et al. over two periods, 1992–1993 and 1996–1999, the yield using a minimum of 5.0 mL sputum specimen for AFB diagnosis was found to increase sensitivity for Mycobacterium tuberculosis (MTB), thus accelerating treatment of TB. In the first period, the acid-fast staining was carried out using sputum specimens regardless of volume. In that way, a sensitivity of 72.5% was obtained, compared to 92.0% for sputum specimens of volumes ≥ 5.0 mL during the second period. 14 The better yield of the sputum specimens with the minimum of 5.0 mL tallied with our study even though the methodologies used in the two studies differed. The relatively low sensitivity in our study could be attributable to the comparatively smaller sample size and the high prevalence of HIV seropositive patients (65.2%) in our Level One wards, because there is an association between decreased host immunity and reduced sputum smear positivity leading to an increase in false negative smears. 15In contrast with the likelihood ratio for a negative microscopy test that was almost similar in both the 5.0 mL and 2.0 mL sputum samples, the likelihood ratio for a positive microscopy in the 5.0 mL sample was almost double that in the 2.0 mL samples (192 versus 94, respectively), clearly in favour of the 5.0 mL sputum sample for AFB microscopy. Clinically, this finding indicates that by using a 5.0 mL specimen, the probability of finding a false positive in a truly negative result is almost 50% lower than by using the 2.0 mL specimen. This finding was corroborated by the relatively higher positive and negative predictive values in the 5.0 mL specimen category, compared to the 2.0 mL category. The difference is of clinical importance because the diagnostic value of a test depends on its positive and negative predictive values, which vary with the prevalence of the disease in a given community. 16As mentioned above, the smear microscopy used in this study was based on fluorescence microscopy of concentrated sputum sediments following centrifugation and is commonly considered to have superior sensitivity compared with conventional light microscopy. 17The false negative results were slightly lower in the 5.0 mL specimens compared to the 2.0 mL specimens (23.4% versus 24.7%, respectively). Furthermore, the false positive results in the 5.0 mL specimens were 50% lower than those in the 2.0 mL samples (0.4% versus 0.8%, respectively). An increase in false positive results reduces the specificity of a test by increasing the denominator of specificity. 18The clinical implication of an increase in false positives is the over-diagnosis of a condition with unnecessary commitment of resources on subjects who do not need the resources. Supervised sputum collection by a health care professional during sputum collection is an important factor in obtaining quality sputum samples. Improving sputum quality has been shown to improve the yield of sputum microscopy. 14Therefore, the best use of limited resources for the detection of smear-positive PTB cases would be to improve the quality of self-expectorated sputum collection and microscopy. Only supervised sputum collection was used in our study, we did not use augmented and invasive sputum collection techniques as recommended in other studies. 19,20 However, PTB suspects who are unable to produce sputum, should be followed-up regularly or even admitted to a health-care facility if possible. The inability to produce sputum is common in immunocompromised patients. 21There was a high prevalence of sero-positive patients in our Level One wards which led the researchers to the conclusion that every effort should be attempted to obtain adequate sputum from every suspected PTB sufferer. To enhance sputum collection, various methods are available, from sputum induction and physiotherapy, to more invasive procedures like bronchoalveolar lavage. 11,22 It has been established from the literature that patient education, allied health worker training and involvement of the community are also of value in a TB programme. 18Sputum collection should be done preferably in the early morning, in a secluded and well ventilated environment so that it does not pose a threat of infection to others. Supervised sputum collection combined with patient instruction, is of great value in obtaining a good quality sputum specimen (not mere saliva), as demonstrated by Sakundarno et al. in a study to determine factors associated with insufficient quality of sputum collected for PTB diagnosis in Indonesia. 23They deduced that, in their district, only a few PTB suspects (45%) provided samples of good colour, viscosity and volume (3.0 mL). One of the reasons identified for this result was poor patient education by health care workers on how to collect sputum. In our study patients were guided on the correct way to provide sputum. Sputum samples that were collected for this study were kept less than 1 hour after collection (en route to the laboratory), to avoid contamination with other micro-organisms from possible excessive specimen handling in the ward. Studies have shown that sputum specimens for AFB detection could be preserved safely in the refrigerator (4 ºC) for up to 8 weeks without affecting the sputum-smear positivity result. 24,25 It has also been shown that there is always the element of specimen container contamination in the wards which could lead to specimen contamination before the specimen reaches the laboratory. 26,27 Nevertheless, in spite of the precautions, 8.5% of the specimens in our study tested positive for mycobacteria other than tuberculosis (MOTT) by culture. Direct microscopy results tend to display variability depending on the microscopic field used, and the number of bacilli in a specific field as well as the level of expertise of the microscopist. Comparing direct sputum microscopy and the NALC-NaOH (N-acetyl-L-cysteine-sodium hydroxide) method, Farnia et al. established that the NSLC-NaOH method was more reliable. They found the sensitivity and specificity of NALC-NaOH method to be 83% and 97% respectively, which was higher than that of direct sputum microscopy which stood respectively at 46% and 90%. 28The NALC-NaOH method was used in this study. This method, however, is not performed in countries with limited resources. Errors by the microscopists in examining sputum smear slides could also have affected our study results, thereby increasing the number of false negative smears. A study has shown that implementation of proficiency testing of microscopists and the introduction of a rechecking system for external quality assurance could help in evaluating diagnosis. 29The process is labour intensive, however, especially in countries with a high TB burden. The limitation of this study was the relatively small sample size. Conclusion This study has indicated a tendency towards superiority in the AFB microscopy results yield of the 5.0 mL specimen category compared to the 2.0 mL category as indicated by the relatively higher sensitivity, specificity and predictive values. With regard to the likelihood ratio (LR), given a positive microscopy with a 5.0 mL specimen, the ratio of a truly positive to a truly negative ratio is 192:1, that is, it is 192 times more likely to be truly positive than truly negative. With a 2.0 mL specimen, the LR is 94:1. Therefore with a 5.0 mL specimen there is a greater chance (more than twice better) of finding a truly positive result. From this point of view, the 5.0 mL specimen is to be recommended. Acknowledgements This study was conducted in partial fulfilment of the requirements for the award of the Master of Medicine (Family Medicine) degree at the University of Limpopo, Medunsa campus. The researchers would like to thank Professor HS Schoeman and Ms AM Managa for guidance with statistical analyses. Competing interests The authors declare that they have no competing interests. Authors’ contribution I.R, M.L.H. and G.I. contributed to conception, design and data collection. All authors were involved in drafting the manuscript. P.D. read and approved the final manuscript. 1.EnarsonDAMurrayJFGlobal epidemiology of tuberculosisWN Rom and SM Garay199657752.10.1001/jama.282.7.6773.Global Tuberculosis Control: WHO Report 201020104. Tuberculosis Strategic Plan for South Africa; 2007–2011. [homepage on the internet][cited 2011 Mar 21]5.TodarKTodar’s online book of bacteriology. [homepage on the internet][cited 2010 Jan 12]6.WarrenJRBhattacharyaMDe AlmeidaKNFTrakasKPetersonLRA minimum 5.0 mL of sputum improves the sensitivity of acid-fast smear for mycobacterium tuberculosis200016115591562108061547.10.1016/0962-8479(95)90583-9 8.GuptaKBGargSUse of sputum induction for establishing diagnosis in suspected pulmonary tuberculosis2005521431469.Dr George Mukhari Hospital Records (Level One wards) 2009/1010.Perez-GuzmanCTorres-CruzAVillarreal-VelardeHVargaMHProgressive Age-related changes in Pulmonary Tuberculosis Images and the Effect of Diabetes2000162173817401106980511.10.1183/09031936.001229071844850312.FouldsJO’BrienRNew tools for the diagnosis of tuberculosis: the perspective of developing countries19982778783978352113.10.1186/1471-2458-8-6818289392226527214.SaglamLAkgunMAktasEUsefulness of induced sputum and fiberoptic bronchoscopy specimens in the diagnosis of pulmonary tuberculosis20053322602651579013915.HarriesADMaherDNunnPAn approach to the problems of diagnosing and treating adult smear-negative pulmonary tuberculosis in high-HIV-prevalence settings in sub-Saharan Africa199876665166210191561231248516.WuHPShiehWBHsienFKHuaCCThe Significance of Mycobacterium tuberculosis Antibody, Antigen 60 IgG in Patients with Abnormal Chest Radiography2004278698761575477617.CattamanchiADavisJLWorodriaWetalSensitivity and Specificity of Fluorescence Microscopy for Diagnosing Pulmonary Tuberculosis in a High HIV Prevalence Setting20091391130113619723403275458418.BellDJDacombeRGrahamSMetalSimple measures are as effective as invasive techniques in the diagnosis of pulmonary tuberculosis in Malawi20091319910419105886287367419.YüksekolIBalSOzkanMetalThe value of fiberoptic bronchoscopy in the diagnosis of smear negative pulmonary tuberculosis20035144054091514338920.10.1093/clinids/6.2.21421.XavierR GSavegnagoF LDamianFetalEvaluating the diagnosis of tuberculosis in induced sputum collection and bronchoalveolar lavage specimens200522.HadleyMMaherDCommunity involvement in tuberculosis control: lessons from other health care programmes2000454014081081573223.10.1186/1471-2466-9-1619426477268916524.BandaHTHarriesADBoereeMJetalViability of stored specimens for smear microscopy and culture2000432722741075107625.10.1016/0041-3879(83)90036-326.10.1590/S1806-3713200900050001227.10.1136/jcp.36.4.479640359849825128.0.1128/JCM.40.2.508-511.20021182596415341629.Martinez-GuarnerosABalandrano-CamposSSolano-CehMALipmanHRidderhofJCFlisserAImplementation of proficiency testing in conjunction with a rechecking system for external quality assurance in tuberculosis laboratories in Mexico20037651652112797692