Background: Most tuberculosis (TB) control programs use positive sputum Ziehl-Neelsen (Z-N) smear microscopy results as a basis for TB case confirmation, valid enough to warrant commencement of treatment. Children do not only produce poor quality sputum a proportion cannot produce sputum, at all. Gastric aspirate (GA) smear microscopy, a recommended alternative, is the only available bacteriologic test in most health-care institutions. Objectives: This study was aimed to determine the usefulness of acid fast smears of GA for the diagnosis of pulmonary TB in children. Materials and Methods: A total of 263 GA samples from children with suspected pulmonary TB underwent Z-N staining and culture on Ogawa medium. Acid-fast isolates on Ogawa medium were sub-cultured on Ogawa medium containing para-nitrobenzoic acid (PNBA-Ogawa medium). Lack of growth on PNBA-Ogawa confirmed Mycobacterium tuberculosis complex (MTBC) and ruled out non-tuberculous mycobacteria. Validity parameters were calculated with GA culture as “Gold standard”. Results: GA samples from 33 (12.5%) of the 263 patients were smearing positive. However, GA culture on Ogawa medium was positive for MTBC in 97 (37%) of the 263 patients. Twenty-nine of the 97 culture-positive GA were also Z-N smear-positive, giving a GA Z-N smear sensitivity of 29.9%. Of the166 culture-negative cases, 162 were Z-N smear-negative, giving a GA Z-N smear specificity of 97.6%. Positive- and negative-predictive values were 88% and 70% respectively. Conclusion: GA smear microscopy is highly specific for diagnosis of pulmonary TB in children. The test can be useful in culture facility-limited settings as acceptable sensitivity can be achieved if specimen collection/handling protocols are strictly adhered to.
Keywords: Children, diagnosis, gastric aspirate, pulmonary tuberculosis
Tuberculosis (TB) is a world-wide pandemic that kills about 5,000 persons daily.  The impact of the disease is mainly in certain TB-endemic developing countries where resources for proper management are limited. ,, Sub-Saharan Africa bears about 30% of the world-wide TB burden and has the highest per capita incidence rate in the world. , Moreover, the highest number of deaths from TB in 2009 occurred in this sub-region.  In these endemic countries, childhood TB constitute more than 25% of the TB burden. 
Although the World Health Organization initiated response to the TB emergency is achieving positive effects in the adult population, the impact of the control programs on childhood TB burden is uncertain, especially in endemic areas. ,, This apparent failure of the national Tuberculosis control programs (NTPs) is due to both inaccuracy of diagnosis; and the focus of the “direct observed treatment short course (DOTS) programs on smear-positive cases. , Consequent on the fact that smear-positivity is very difficult to demonstrate in childhood cases, a vast majority of the cases could not access DOTS. , The diagnostic challenge is compounded by the fact that there is no satisfactory reference diagnostic standard for childhood TB diagnosis.  Further challenges specific to endemic areas include: Community-based rather than individual-based TB transmissibility; the high prevalence of childhood diseases that are clinically and radiologically similar to TB; and the high non-tuberculous mycobacteria (NTM) prevalence. ,, These challenges limit the discriminatory power of clinical, radiological, and immunodiagnostic methods in the endemic countries. ,, Thus, previously validated criteria such as positive history of exposure, Chest X-ray and tuberculin test became unreliable in endemic areas.  The non-specific nature of clinical criteria can lead to labeling 25% of children as TB patients.  It follows; therefore that the Nigerian National TB and Leprosy control program-recommended dependence on clinical parameters for diagnosis of TB in children who can’t produce sputum and sputum smear-negative cases should raise concern.  For any diagnostic method to be effective in endemic areas, it must be capable of discriminating between Mycobacterium tuberculosis complex (MTBC) infection and MTBC disease on one hand; and between MTBC disease and other common childhood diseases on the other hand. ,, This discriminatory capacity of TB diagnostic instruments is crucial because the NTPs in resource-limited settings mainly target disease eradication rather than infection eradication. ,, Considering its high discriminatory power, efforts should be made to make bacteriological diagnosis more relevant to childhood TB diagnosis in endemic areas.  For definitive bacteriologic diagnosis of pulmonary TB in young children, cultural isolation of MTBC from gastric aspirates (GAs) and demonstration of acid-fast bacilli by GA smear microscopy have been in use for decades.  GA smear microscopy, though it has lower sensitivity and specificity than GA culture, is readily available, affordable, and yields results so rapidly that it is primarily utilized by most health-care centers, including those with culture facilities. , Furthermore, there are potentials for improvement of sensitivity of GAs smear microscopy by following strict specimen collection and processing protocols; including prompt neutralization and centrifugation.  Such an affordable diagnostic tool, which can clearly discriminate between MTBC infection and MTBC disease, should be made more relevant, especially in endemic regions, once its current validity is ascertained. 
Several research efforts have been made on the validity of acid fast smear of GAs for the diagnosis of childhood MTBC disease. , A validation study of acid fast smear of GA specimens done more than 10 years ago in a non-endemic area, was prompted by an observed increase in NTM respiratory tract disease, which could potentially lead to false-positive acid-fast smears for TB.  Acid fast smears of GAs was then shown to be 19% sensitive and 100% specific for pulmonary TB diagnosis.  Consequent on the known high degree of mucosal colonization by NTM in this environment, there is a possibility of false-positive Ziehl-Neelsen (Z-N) test results, which raises questions on the specificity of the Z-N test. , There is a need to ascertain the current sensitivity, specificity, and predictive values of acid-fast smear of GA for the diagnosis of pulmonary tuberculosis (PTB) in this sub-region; by utilizing standard procedures and using the culture as “reference standard.” There is a paucity of published West African and Nigerian data addressing the validity of GA acid-fast smear as an indicator of pulmonary TB. Such validation based on local data is necessary to ascertain that the beneficiaries of DOTS are truly TB cases. False-positive GA acid-fast smear results will lead to waste of scarce resources and unnecessary toxicity. Furthermore, better understanding of the specificity and sensitivity of GA acid fast smear for pulmonary TB diagnosis could guide the physicians in planning the bacteriological work-up of suspected childhood TB cases.
The study was carried out in University of Benin Teaching Hospital, Benin, Nigeria. A total of 263 patients who presented and were registered in the pediatrics department of the hospital and who had features suggestive of TB were recruited. Approval was sought and obtained from the ethics and research committee of this hospital for this research work. Informed written consent was obtained from parents on behalf of their children. GA specimens were collected only from consenting patients. A semi-structured researcher-administered study questionnaire was used to assess demographic information, symptoms, duration of illness, history of TB contacts, clinical signs, previous TB diagnosis, current treatments, tuberculin sensitivity test, erythrocyte sedimentation rate (ESR), Human immunodeficiency virus status, and chest radiographic findings. GA specimens were tested for acid fast smear positivity and MTBC culture positivity.
For GA specimen collection, each mother was instructed not to feed the child again after supper and also, not to allow the child out of bed until the specimen was collected. By 6.00 am, each child was immobilized on a firm surface, the distance between the nose and the stomach was measured. After wiping the nostrils with dry swab, a naso-gastric tube was passed. As the tube reached the throat, a puff on the face of the child would stimulate swallowing reflex that enabled extension of the tube into the stomach. At least 5-10 ml of aspirate was collected each time in each case. Any reactive vomitus was also collected. Three consecutive specimens were collected from each patient every morning into properly labeled universal bottles. The specimens were tightly covered and delivered without delay to the laboratory where they were promptly neutralized with sodium bicarbonate. PH was first assessed before neutralization, which was effected by gradual addition of sodium bicarbonate and monitoring pH change with the pH meter. The neutralized specimens were promptly refrigerated, and temperature maintained at about 4°C until processing was carried out. Processing (microscopy and culture) of the neutralized specimens was carried out as early as possible but was never up to 24 h from time of collection. Specimen manipulations were carried out in a Biosafety cabinet type 2, and care was taken to avoid exposure to direct sunlight and heat.
A set of smears were made with sediments following decontamination with 4% NaOH and centrifugation at relative centrifugal force (RCF) of 3000× gravity (g) for 15 min. The smears of the sedements were prepared by placing 1-2 drops of homogenized sediment on a slide and letting it air-dry.
Positive control stained slides were prepared from skimmed milk suspension of known MTBC isolates. Negative control slides were made from Z-N staining of a smear of egg albumin. For each new batch of stains, new control Z-N staining slides were prepared.
Fixing the smear with heat was achieved by placing the air-dried smear on a hot plate set at 70°C for 1 h. Standard procedure and precautions were followed in the Z-N staining.
The stained slide was viewed with ×100 (oil-immersion) objective of a compound microscope. The acid-fast bacilli were seen as straight or bent pink-red rods, which could be banded owing to differential staining of different sections of the individual bacilli. They were seen also in clusters. Grading followed United States Centers for Disease Control and Prevention standards:  >9 bacilli/ Oil immersion field = 4+; 1-9 bacilli/Oil immersion field = 3+; 1-9 bacilli/10 Oil immersion fields = 2+; 1-9 bacilli/100 Oil immersion fields = 1+; 1-3 bacilli/300 Oil immersion fields = doubtful, repeat; 0 bacilli/300 Oil immersion fields = No acid fast bacilli (AFB) seen. Grades of smear-positive findings, 1+ to 4+, from at least one of the three patient’s samples were taken as positive results. A smear-positive finding of “doubtful” grade required at least another smear-positive sample from the same patient for an impression of a positive result to be made.
For GA culture, each sample was decontaminated by adding an equal volume of sterile 4% NaOH; homogenized with vortex mixer and left for 15 min without exposure to direct sunlight or heat. This was followed by centrifugation at RCF of 3000 g for 15 min. The sediment was homogenized with vortex mixer and 0.1 ml (3 drops) of the homogenate used to inoculate acidified Ogawa medium, which was incubated at 37°C in CO 2 enriched ambient air for the 1 st week and then in ambient air for the rest of 4-8 weeks. Culture bottles were placed horizontally for the first 24 h to enable absorption of innocula; after which they were placed upright. Sterile Ogawa media were used as negative controls. Cultures were examined every other day for the first 7 days within which rapidly growing mycobacteria and other contaminants were detected. Assessment of contamination was based on observation of growth before 2 weeks, observation of growth in association with media deterioration, Gram stain results, and Z-N staining results. Subsequently, cultures were examined weekly for the remaining 7 weeks until growth were detected. Cultures were reported as negative and discarded after 8 weeks of incubation.
Isolates appearing after 3 or more weeks of inoculation and showing rough, non-pigmented colonies and Z-N stain positivity were presumed to represent pathogenic species of MTBC. To further affirm that the isolate is MTBC and not NTM, suspension of the isolates in water was used to innoculate a medium containing 500 μg/ml para-nitrobenzoic acids (PNBA). Absence of growth in this medium was in keeping with MTBC. Only positive cultures were sub-cultured in PNBA-Ogawa media.
Microsoft Excel 2007 and SPSS version 16 software were used in the analysis of data generated by the study. Data generated by Excel filtering machine was used to prepare the tables used in computing the validity parameters: Specificity, sensitivity, negative-predictive value (NPV), and positive-predictive value (PPV).
Smear-positive cases were defined as cases whose GA smears were acid fast (contained acid-fast bacilli); although smear-negative cases were defined as those whose GA smears were not acid-fast.
Culture-positive cases were defined as cases whose GA cultures yielded acid-fast isolates on Ogawa medium and the isolates failed to grow in PNBA-Ogawa medium. Culture-negative cases were defined as cases whose GA specimens yielded “no growth” in Ogawa medium. “No growth” included AFB isolates on Ogawa medium, which also grew on PNBA-Ogawa. Sensitivity was defined as the percentage of smear-positive cases among culture-proven cases; although specificity was defined as the percentage of smear-negative cases among culture-negative cases.  PPV was defined as the percentage of true-positive results among all the positive results; although the NPV was defined as the proportion of truly negative results among all the negative results. 
False-positive smear results were defined as smear-positive cases that were not culture-positive; although false-negative smear results were defined as smear-negative results that were culture-positive. 
GA samples were collected from a total of 263 symptomatic pediatric patients who had features suggestive of PTB. One hundred and thirteen (43.0%) of the patients were males while 150 (57%) were females; giving a male-to-female ratio of 1:1.3 [Table 1]. The individual ages ranged from 3 months to 10 years; while their mean age was 2.8 ± 2.6 years. Majority (57.8%) of the patients were less than 2 years of age. Patients aged between 2 and 4 years were next in proportion (21.7%); followed by patients aged between 4 and 6 years (11.4%). Next in Proportion were patients aged 8 to 10 years (6.8%); although patients aged between 6 and 8 years constituted the least proportion (2.3%) [Table 1].
Two hundred and thirteen (80.6%) of the sampled patients had cough, fever, and weight loss/lack of weight gain of more than 2 weeks duration; 224 (84.8%) had radiological features suggestive of pulmonary TB; 158 (60.0%) had tuberculin sensitivity reading of >10 mm; although 69 (26.0%) presented as unresolved pneumonia of more than 2 weeks duration. Positive history of contact could be established in only 28 (10.6%) of the sampled patients; although 13 (5%) of them had ESR of ≥100 mm/h [Table 2].
Smears of specimens from 33 of the 263 respondents were Z-N-positive; giving a smear positive rate of 12.5% [Figure 1].
Of the 97 children with culture-positive samples, 29 also had positive Z-N smears; giving sensitivity score of 29.9% [Table 3].
As also shown in [Table 3], 162 of the 166 patients with culture-negative samples had Z-N smear-negative results; giving a specificity score of 97.6%.
[Table 3] Further shows that four of the smear-positive results, could not be confirmed by culture, and were therefore, referred to as false-positive smears; as opposed to the 29 positive smears that were confirmed by the culture, which were classed as true-positive smears.
Based on the above data, PPV of GA AFB smear was found to be equal to 87.9%.
[Table 3] also shows that 68 smear-negative cases could be detected by culture. These 68 smear-negative results were classed as false negatives; while the 162 negative smear results that were also culture-negative were classed as true-negatives. Based on the above data, negative predictive value (NPV) of GA AFB test was found to be equal to 70.4%.
In this study, the specificity or true-negative rate of GA acid-fast smear for MTBC was found to be 97.6%. The high specificity value found in this study suggests that the negative effect of NTM on the validity of the test results for diagnosis of childhood PTB is minimal. This high specificity finding agrees with the reports of Gómez Pastrana Durαn et al. and Bahammam et al. who got specificity values of 96.8% and 100% respectively. , An explanation for the high specificity in children in spite of the high prevalence of these NTM in our environment can be derived from a report, which showed that NTM colonization tend to affect older age groups; and that even when the NTM were present in the pediatric specimens, they were less likely to demonstrate smear positivity.  Furthermore, in the study by Bahammam, et al. None of the samples from the 15 children included in the study grew NTM; though there was a high rate of isolation of NTM in the older age groups. 
GA acid-fast smears have been known to have a low bacteriologic yield (sensitivity) among children, and this has led to several consequences, namely: Insignificant impact of TB control programs in children; loss of interest of physicians in the test; and the near-total dependence on clinical diagnostic criteria for TB detection in the National TB control program, as published by the Nigerian Federal Ministry of Health. , The sensitivity of 29.9% found in this study, though still low, is an improvement on the finding of 0% and 13% by Abadco and Steiner, and Gσmez Pastrana Durαn et al. respectively. , This improvement in sensitivity conforms to report that strict adherence to specimen handling protocol, including properly timed specimen collection and prompt neutralization can lead to some improvement in sensitivity.  The essence of a diagnostic tool is to make a diagnosis.  In addition, to ascertaining the sensitivity and specificity, which are measures of accuracy of the diagnostic instrument; there is still need to assess the proportion of correct diagnoses that the instrument is capable of. , Predictive values tell us the probability that patients with positive results (positive predictive value) or negative results (NPVs) have been correctly diagnosed. , The value of the two predictive validity parameters depend on the intrinsic sensitivity and specificity of the test instrument as well as on the prevalence of the disease in the study population. , The high prevalence of TB in endemic areas reduces the likelihood of false-positive results; and so can lead to high PPVs such as the 87.9% found in this study. The high PPV found in this study is comparable to the 100% reported by Bahammam.  In fact, within the under-5 age group in this study, the PPV was near 100%. This was because almost all the false-positive smear results in this study occurred in 10-year-old children; whereas the only false-positive result from under-5 patients was associated with culture contamination. These findings imply that a positive acid-fast smear result from a young child in Nigeria, and other TB-endemic areas, should warrant full course of treatment for PTB; although positive results from older children may require further corroboratory evidences, including cultural isolation of MTBC, and for the establishment of diagnosis.
The NPV of GA acid-fast smear was found to be 70.4% in this study. In this high TB prevalence area, false negative results will tend to be high, especially if the diagnostic tool is not very sensitive. However, the value of 70.4% is still reasonable; although, it indicates that additional or alternative assessments of negative microscopy results may be required. A higher NPV of 95% was found in a TB-non-endemic area.  NPV parameters were not calculated in the various studies carried out in Turkey and Nigeria, which are TB-endemic areas. 
The smear positive rate is not a validity parameter in the strict sense because it is not derived with reference to a “gold standard.”  It is however, an important value to consider in planning the laboratory work-up of a patient if GA smear microscopy is a consideration. With a highly specific test, every smear positive result will represent a bacteriologically confirmed case. The smear positive rate then becomes a better and more practical measure of diagnostic yield of the test instrument than the standard validity parameters, sensitivity and specificity, which are just measures of accuracy.  Moreover, some authors reported smear-positive rate in place of sensitivity.  The smear positive rate of 12.5% derived from this study is only slightly higher than the 10.8% reported by another study from Nigeria.  This rate of smear-positivity indicates that there should be considerations for prevention of transmission by these positive cases, in spite of the fact that children are not known to be effective transmitters of the disease. Moreover, there are several reports of TB-transmission by children. ,
Acid-fast smear of GAs is a highly specific test for the diagnosis of PTB in children. A positive acid-fast smear should warrant commencement of the full course of anti-TB therapy. The results of this study also show that the sensitivity of acid-fast smear of GAs can be increased by careful adherence to specimen handling protocol, especially prompt neutralization and proper timing of specimen collection.
Source of Support: None, Conflict of Interest: None
[Table 1], [Table 2], [Table 3]