Methods
Study Participants
We conducted a cross-sectional diagnostic study at the Mulago National Referral Hospital in Kampala, Uganda on both paediatric in-patients and out-patients aged 2 months to 12 years with suspected pulmonary tuberculosis between January 2011 and January 2012. Eligibility for enrolment was based on the WHO case definition for a TB suspect and included having a persistent cough of 2 weeks or more and one of the following: household TB contact, unexplained weight loss or failure to gain weight, and unexplained fever for 2 weeks or more. We excluded children who were on anti- tuberculous therapy and those in whom sputum could not be obtained or sputum induction was contraindicated. The contraindications to sputum induction were adapted from the WHO guidelines for sputum induction and included: severe respiratory distress (including rapid breathing, wheezing, hypoxia); intubated patient; bleeding: low platelet count, bleeding tendency, severe nosebleeds (symptomatic or platelet count <50/ml blood); reduced level of consciousness and history of significant asthma (diagnosed and treated by a clinician). The exclusion criterion "children in whom sputum could not be obtained" was defined after systematic attempt of sputum induction.
Study Procedure
The study was approved by the School of Medicine Research and Ethics committee of Makerere University College of Health Sciences (# REC REF 2011–005). The children were enrolled upon provision of informed consent from the primary caregivers and assent for those older than 8 years. We conducted consecutive enrolment until the desired sample size was achieved. Upon enrolment, a clinical history and physical examination were conducted. The investigations performed included the TST, Complete Blood Count, HIV testing in children whose HIV status was unknown, and chest radiography. Each child's nutritional status was assessed using the WHO child growth standards. Weight for length/weight for height was used for children < 5 years while Body Mass Index (BMI) for Age was used for children ≥ 5 years. Severe wasting was defined as weight for length or weight for height or BMI for age Z score which was less than – 3SD.
Chest Radiography
The chest radiographs (Antero-Posterior views) were evaluated for pulmonary tuberculosis by two independent reviewers who used a standardised reporting form and were blinded to the results of the other investigations. A third independent reviewer evaluated a radiograph in the event that there was discordant reporting between the two reviewers. The reviewers classified the chest radiographs into probable TB (findings consistent with TB), possible TB (atypical findings), and TB unlikely (findings not consistent with TB or normal chest radiograph) based on the radiological features present. Findings consistent with TB included lymphadenopathy for children 5 years and below while, a typical adult type of chest radiograph appearance with predominance of upper lobe infiltrates with or without cavitation was used for older children. A miliary pattern was categorized as probable TB.
Tuberculin Skin Test
A qualified trained study nurse administered 0.1 ml of purified protein derivative (PPD; 2TU, PPD RT23, Staten Serum Institute, Denmark, Copenhagen) intradermally on the volar aspect of the less dominant forearm. The induration at the site of administration was measured in millimetres after 48–72 hours of administration. A positive skin test was defined as > 5 mm of the transverse induration in children with HIV infection and > 10 mm in children without HIV infection.
Sputum Sample Collection
A qualified trained study nurse performed sputum induction using nebulised 3% hypertonic saline after a minimum of 3 hours fast. The study participants underwent premedication with nebulised salbutamol prior to the administration of nebulised 3% hypertonic saline. A maximum of two attempts were conducted to obtain an adequate sample of at least 3 mls of sputum. However whatever volume of samples obtained was sent for analysis. The study participants had pulse oximetry monitoring during and for one hour after the procedure. The sputum sample was collected in a sterile container which was placed on an ice pack in a cool box, taken to an accredited mycobacteriology laboratory (Makerere University Department of Microbiology – Mycobacteriology Laboratory) and processed within two hours of collection. Digestion-Decontamination of sputum by N- acetyl L- cysteine (NALC)-Sodium Hydroxide method was performed prior to making smears for fluorochrome staining with Auramine O Phenol and examined according to standards. Samples were inoculated into the Mycobacteria Growth Indicator Tube (MGIT) (Becton Dickson, Franklin Lakes, NJ) and Lowenstein-Jensen (LJ) slants and incubated for 6 weeks on MGIT and 8 weeks on LJ culture. At least 0.5 ml of the sediment sample was stored at – 20°C for the Xpert MTB/RIF test. Due to the delay in delivery of the Xpert MTB/RIF cartridges and sample reagents, we were unable to run the Xpert MTB/RIF test on the freshly collected sputum samples. A culture was considered positive if mycobacterium tuberculosis growth was confirmed on either LJ or MGIT media. A culture was considered negative if no growth was confirmed on both LJ and MGIT media, if one culture result was negative and the other was contaminated. A culture was considered contaminated if both LJ and MGIT demonstrated contamination. Mycobacteria Other Than Tuberculosis (MOTT) was considered if both LJ and MGIT culture media grew MOTT. Mycobacterial identification was done using Capilia TB Neo™ (TAUN, Numazu, Japan) assay. The frozen samples were thawed to room temperature prior to addition of the Xpert MTB/RIF sample reagent. Three parts (1.5 ml) of the sample reagent were added to one part (0.5 ml) of the sediment sample as per the manufacturer's guide. The mixture was shaken and incubated at room temperature for 15 minutes. The mixture was then transferred to a pre labeled cartridge and subsequently loaded into the Xpert MTB/RIF machine. The results were printed out after a 2 hours cycle was complete. An Xpert MTB/RIF test was considered positive if MTB was detected and negative if MTB was not detected. We were unable to run a repeat test for the indeterminate MTB/RIF test result as the stored sputum sample was insufficient. Drug susceptibility testing was not performed in children identified with rifampicin resistance based on Xpert test due to cost limitations. The indeterminate test results were reported as such and were excluded from analysis. Quality control measures for the Xpert MTB/RIF test included negative and positive control tests for every new batch of Xpert MTB/RIF cartridges; monthly negative and positive control tests; monthly room swabs for mycobacteria tuberculosis contamination; and automated quality control for PCR. The laboratory technologists processing the sputum samples for culture and Xpert MTB/RIF were blinded to the results of the other test.
Blood Sample Collection
Using aseptic technique, 3mls of venous blood were drawn for CBC and HIV testing under aseptic conditions. The CBC assays were performed using Beckman Dickinson Coulter Counter (5 Part-Differential). HIV rapid testing was conducted on all children without a documented HIV status. HIV exposed infants aged 18 months or younger had an HIV DNA PCR done to confirm the diagnosis. The HIV rapid test algorithm used was that recommended by the Ministry of Health in Uganda.
Statistical Considerations
Sample Size At the time we started data collection, there were no published data on evaluation of the Xpert MTB/RIF among children. Sensitivity and specificity normograms with a fixed false positive rate of 5% were used to calculate the sample size for this study. Using a prevalence of culture confirmed pulmonary tuberculosis of 30% in a Ugandan paediatric population, sensitivity of the Xpert MTB/RIF test in the adult population of 98.2% and 95% confidence interval, we obtained a sample size 250 study participants.
Data Analysis The data were captured using EPI INFO version 3.5.3, analysed using STATA version 10 and OpenEpi. We described data for all enrolled study participants but excluded data of children with indeterminate MTB/RIF test results and contaminated sputum cultures from the diagnostic accuracy analysis. Study population characteristics were described using proportions with 95% CI, means with standard deviations, and medians with interquartile ranges (IQR). We excluded data of children with indeterminate MTB/RIF test results and contaminated sputum culture results from the diagnostic accuracy analysis. Sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios with their 95% CI for the Xpert MTB/RIF test were calculated using sputum culture as the gold standard. The chi square statistical test was used to compare the proportions. We explored the clinical characteristics associated with a positive Xpert MTB/RIF test result using multivariate logistic regression analysis. The clinical characteristics with a p value of < 0.2 on bivariate analysis and those associated with PTB were entered into the logistic regression model using the backward stepwise method. P values less than 0.05 were considered statistically significant.