The diagnosis of paediatric pulmonary tuberculosis is difficult, especially in young infants who cannot expectorate sputum spontaneously. Breath testing has shown promise in diagnosing respiratory tract infections, but data on paediatric tuberculosis are limited. We performed a prospective cross-sectional study in Kenya in children younger than five years with symptoms of tuberculosis. We analysed exhaled breath with a hand-held battery-powered nose device. For data analysis, machine learning was applied using samples classified as positive (microbiologically confirmed) or negative (unlikely tuberculosis) to assess diagnostic accuracy. Breath analysis was performed in 118 children. The area under the curve of the optimal model was 0.73. At a sensitivity of 86 % (CI 62-96 %), this resulted in a specificity of 42 % (95 % CI 30-55 %). Exhaled breath analysis shows promise as a triage test for TB in young children, although the WHO target product characteristics were not met.

For microbiological confirmation of pediatric pulmonary tuberculosis (PTB), gastric aspirates (GA) are often operationally unfeasible without hospitalization, and the encapsulated orogastric string test is not easily swallowed in young children. The Combined-NasoGastric-Tube-and-String-Test (CNGTST) enables dual collection of GA and string specimens. In a prospective cohort study in Kenya, we examined its feasibility in children under five with presumptive PTB and compared the bacteriological yield of string to GA. Paired GA and string samples were successfully collected in 95.6 % (281/294) of children. Mycobacterium tuberculosis was isolated from 7.0 % (38/541) of GA and 4.3 % (23/541) of string samples, diagnosing 8.2 % (23/281) of children using GA and 5.3 % (15/281) using string. The CNGTST was feasible in nearly all children. Yield from string was two-thirds that of GA despite a half-hour median dwelling time. In settings where the feasibility of hospitalisation for GA is uncertain, the string component can be used to confirm PTB.

Diagnosis of tuberculosis (TB) among young children (<5 years) is challenging due to the paucibacillary nature of clinical disease and clinical similarities to other childhood diseases. We used machine learning to develop accurate prediction models of microbial confirmation with simply defined and easily obtainable clinical, demographic, and radiologic factors. We evaluated eleven supervised machine learning models (using stepwise regression, regularized regression, decision tree, and support vector machine approaches) to predict microbial confirmation in young children (<5 years) using samples from invasive (reference-standard) or noninvasive procedure. Models were trained and tested using data from a large prospective cohort of young children with symptoms suggestive of TB in Kenya. Model performance was evaluated using areas under the receiver operating curve (AUROC) and precision-recall curve (AUPRC), accuracy metrics. (i.e., sensitivity, specificity), F-beta scores, Cohen's Kappa, and Matthew's Correlation Coefficient. Among 262 included children, 29 (11%) were microbially confirmed using any sampling technique. Models were accurate at predicting microbial confirmation in samples obtained from invasive procedures (AUROC range: 0.84-0.90) and from noninvasive procedures (AUROC range: 0.83-0.89). History of household contact with a confirmed case of TB, immunological evidence of TB infection, and a chest x-ray consistent with TB disease were consistently influential across models. Our results suggest machine learning can accurately predict microbial confirmation of M. tuberculosis in young children using simply defined features and increase the bacteriologic yield in diagnostic cohorts. These findings may facilitate clinical decision making and guide clinical research into novel biomarkers of TB disease in young children.

BACKGROUND: Pediatric tuberculosis (TB) remains a critical public health concern, yet bacteriologic confirmation of TB in children is challenging. Clinical, demographic, and radiological factors associated with a positive Mycobacterium tuberculosis specimen in young children (5years) are poorly understood. METHODS: We conducted a prospective cohort study of young children with presumptive TB and examined clinical, demographic, and radiologic factors associated with invasive and noninvasive specimen collection techniques (gastric aspirate, induced sputum, nasopharyngeal aspirate, stool, and string test); up to 2 samples were taken per child, per technique. We estimated associations between these factors and a positive specimen for each technique using generalized estimating equations (GEEs) and logistic regression. RESULTS: A median (range) of 544 (507-566) samples were obtained for each specimen collection technique from 300 enrolled children; bacteriologic yield was low across all collection techniques (range, 1%-7% from Xpert MTB/RIF or culture), except for lymph node fine needle aspiration (29%) taken for children with cervical lymphadenopathy. Factors associated with positive M. tuberculosis samples across all techniques included prolonged lethargy (median [range] adjusted odds ratio [aOR], 8.1 [3.9-10.1]), history of exposure with a TB case (median [range] aOR, 6.1 [2.9-9.0]), immunologic evidence of M. tuberculosis infection (median [range] aOR, 4.6 [3.7-9.2]), large airway compression (median [range] aOR, 6.7 [4.7-9.5]), and hilar/mediastinal density (median [range] aOR, 2.9 [1.7-3.2]). CONCLUSIONS: Identifying factors that lead to a positive M. tuberculosis specimen in very young children can inform clinical management and increase the efficiency of diagnostic testing in children being assessed for TB.

BACKGROUND: Tuberculosis (TB) is a leading cause of illness and death in children globally. Improved bacteriologic and clinical diagnostic approaches in children are urgently needed. METHODS: In a prospective cohort study, a consecutive series of young (<5 years) children presenting with symptoms suggestive of TB and parenchymal abnormality on chest radiograph in inpatient and outpatient settings in Kisumu County, Kenya from October 2013 to August 2015 were evaluated at baseline and over 6 months. Up to 14 specimens per child were tested for the Mycobacterium tuberculosis complex by fluorescence microscopy, Xpert MTB/RIF and mycobacterial culture. Using detailed clinical characterization, cases were retrospectively classified according to standardized research case definitions and the sensitivity and specificity of microbiological tests on different specimen types were determined. RESULTS: Among 300 young children enrolled, 266 had sufficient information to be classified according to the research clinical case definition. Of these, 36% (96/266) had TB disease; 32% (31/96) with bacteriologically confirmed intrathoracic TB. Compared to culture, the sensitivity of a single Xpert test ranged from 60 to 67% and specificity from 97.5 to 100% for different specimen types. CONCLUSIONS: Despite extensive specimen collection and laboratory testing, TB could not be bacteriologically confirmed in almost two-thirds of children with intrathoracic TB classified by research clinical case definitions. Improved diagnostic tests are needed to identify children with TB and to exclude other potential causes of illness.

BACKGROUND: HIV testing efficiency could be improved by focusing on high yield populations and identifying types of health facilities where people with undiagnosed HIV infection are more likely to attend. METHODS: A retrospective cohort analysis of data collected during an integrated TB/HIV active case-finding intervention in Western Kenya. Data were analyzed from health facilities' registers on individuals who reported TB-suggestive symptoms between 1 July and 31 December 2018 and who had an HIV test result within one month following symptom screening. We used logistic regression with general estimating equations adjusting for sub-county level data to identify health facility-level predictors of new HIV diagnoses. RESULTS: Of 11,376 adults with presumptive TB identified in 143 health facilities, 1038 (9%) tested HIV positive. The median HIV positivity per health facility was 6% (IQR = 2-15%). Patients with TB symptoms were over three times as likely to have a new HIV diagnosis in private not-for-profit facilities compared to those in government facilities (adjusted odds ratio (aOR) 3.40; 95% CI = 1.96-5.90). Patients tested in hospitals were over two times as likely to have a new HIV diagnosis as those tested in smaller facilities (i.e., health centers and dispensaries) (aOR 2.26; 95% CI = 1.60-3.21). CONCLUSION: Individuals with presumptive TB who attended larger health facilities and private not-for-profit facilities had a higher likelihood of being newly diagnosed with HIV. Strengthening HIV services at these facilities and outreach to populations that use them could help to close the HIV diagnosis gap.

BACKGROUND: Recommended by the World Health Organization as an initial diagnostic test for TB in children, Xpert(®) MTB/RIF is widely implemented in many countries, including Kenya.METHODS: Three hundred HIV-positive and negative children (<5 years) were enrolled in Kisumu County, Kenya, from October 2013 to August 2015. Multiple specimen types were collected from each child and tested using Xpert, liquid culture, and phenotypic drug susceptibility testing (DST). Samples positive for rifampin (RIF) resistance on Xpert were tested using line-probe assay and sequencing.RESULTS: Of 32 children with bacteriologically confirmed TB, 27 had positive Xpert results. Of these, 3/27 (11%, 95% CI 4-28) had RIF resistance detected on Xpert, but not by phenotypic DST, line-probe assay, or sequencing. For these three children, five Xpert tests showed RIF resistance; all five tests had semi-quantitative "very low" results and delay or absence of probe D signal, whereas no Xpert results with higher semi-quantitative results showed RIF resistance. All three children responded well to standard TB treatment.CONCLUSIONS: False RIF resistance may be detected in pediatric specimens. Further study is needed to determine if false RIF resistance is associated with low bacterial load.

IMPORTANCE: Criterion-standard specimens for tuberculosis diagnosis in young children, gastric aspirate (GA) and induced sputum, are invasive and rarely collected in resource-limited settings. A far less invasive approach to tuberculosis diagnostic testing in children younger than 5 years as sensitive as current reference standards is important to identify. OBJECTIVE: To characterize the sensitivity of preferably minimally invasive specimen and assay combinations relative to maximum observed yield from all specimens and assays combined. DESIGN, SETTING, AND PARTICIPANTS: In this prospective cross-sectional diagnostic study, the reference standard was a panel of up to 2 samples of each of 6 specimen types tested for Mycobacterium tuberculosis complex by Xpert MTB/RIF assay and mycobacteria growth indicator tube culture. Multiple different combinations of specimens and tests were evaluated as index tests. A consecutive series of children was recruited from inpatient and outpatient settings in Kisumu County, Kenya, between October 2013 and August 2015. Participants were children younger than 5 years who had symptoms of tuberculosis (unexplained cough, fever, malnutrition) and parenchymal abnormality on chest radiography or who had cervical lymphadenopathy. Children with 1 or more evaluable specimen for 4 or more primary study specimen types were included in the analysis. Data were analyzed from February 2015 to October 2020. MAIN OUTCOMES AND MEASURES: Cumulative and incremental diagnostic yield of combinations of specimen types and tests relative to the maximum observed yield. RESULTS: Of the 300 enrolled children, the median (interquartile range) age was 2.0 (1.0-3.6) years, and 151 (50.3%) were female. A total of 294 met criteria for analysis. Of 31 participants with confirmed tuberculosis (maximum observed yield), 24 (sensitivity, 77%; interdecile range, 68%-87%) had positive results on up to 2 GA samples and 20 (sensitivity, 64%; interdecile range, 53%-76%) had positive test results on up to 2 induced sputum samples. The yields of 2 nasopharyngeal aspirate (NPA) samples (23 of 31 [sensitivity, 74%; interdecile range, 64%-84%]), of 1 NPA sample and 1 stool sample (22 of 31 [sensitivity, 71%; interdecile range, 60%-81%]), or of 1 NPA sample and 1 urine sample (21.5 of 31 [sensitivity, 69%; interdecile range, 58%-80%]) were similar to reference-standard specimens. Combining up to 2 each of GA and NPA samples had an average yield of 90% (28 of 31). CONCLUSIONS AND RELEVANCE: NPA, in duplicate or in combination with stool or urine specimens, was readily obtainable and had diagnostic yield comparable with reference-standard specimens. This combination could improve tuberculosis diagnosis among children in resource-limited settings. Combining GA and NPA had greater yield than that of the current reference standards and may be useful in certain clinical and research settings.

OBJECTIVE: To evaluate the utility of a broad and non-specific symptom screen for identifying people with undiagnosed HIV infection. DESIGN: Secondary analysis of operational data collected during implementation of a cluster-randomized trial for tuberculosis case detection. METHODS: As part of the trial, adults reporting cough, fever, night sweats, weight loss, or difficulty breathing of any duration in the past month were identified in health facilities and community-based mobile screening units in western Kenya. Adults reporting any symptom were offered HIV testing. We analysed the HIV testing data from this study, using modified Poisson regression to identify predictors of new HIV diagnoses among adults with symptoms and initially unknown HIV status. RESULTS: We identified 3,818 symptomatic adults, referred 1424 (37%) for testing, of whom 1065 (75%) accepted, and 107 (10%) were newly diagnosed with HIV. The prevalence of new HIV diagnoses was 21% (95% CI: 17-25%) among those tested in health facilities and 5% (95% CI 4-7%) among those tested in mobile units. More men were diagnosed with HIV than women despite fewer men being screened. People who reported 4-5 symptoms were over twice as likely to be diagnosed with HIV compared to those reporting 1-3 symptoms (adjusted prevalence ratio [aPR] in health facilities = 2.58, 95% CI, 1.65-4.05; aPR in mobile units = 2.63, 95% CI, 1.37-5.03). CONCLUSION: We observed a high yield of new HIV diagnoses among adults identified by active application of a broad symptom screen. Integrated tuberculosis and HIV screening using could help close the detection gap for both conditions.

SETTING: Efficient tuberculosis (TB) active case-finding strategies are important in settings with high TB burdens and limited resources, such as those in western Kenya. OBJECTIVE: To guide efforts to optimize screening efficiency, we identified the predictors of TB among people screened in health facilities and communities. DESIGN: During February 2015-June 2016, adults aged >/=15 years reporting any TB symptom were identified in health facilities and community mobile screening units, and evaluated for TB. We assessed the predictors of TB using a modified Poisson regression with generalized estimating equations to account for clustering according to screening site. RESULTS: TB was diagnosed in 484 (20.3%) of 2394 symptomatic adults in health facilities and 39 (3.4%) of 1424 in communities. In health facilities, >10% of symptomatic adults in all demographic groups had TB, and no predictors were associated with a >/=2-fold increased risk. In communities, the independent predictors of TB were male sex (adjusted prevalence ratio [aPR] = 4.26, 95%CI 2.43-7.45), HIV infection (aPR 2.37, 95%CI 1.18-4.77), and household TB contact in the last 2 years (aPR 2.84, 95%CI 1.62-4.96). CONCLUSION: Our findings support the notion of general TB screening in health facilities and evaluation of the adult household contacts of TB patients.

OBJECTIVE: To evaluate the utility of a broad and non-specific symptom screen for identifying people with undiagnosed HIV infection. DESIGN: Secondary analysis of operational data collected during implementation of a cluster-randomized trial for tuberculosis case detection. METHODS: As part of the trial, adults reporting cough, fever, night sweats, weight loss, or difficulty breathing of any duration in the past month were identified in health facilities and community-based mobile screening units in western Kenya. Adults reporting any symptom were offered HIV testing. We analysed the HIV testing data from this study, using modified Poisson regression to identify predictors of new HIV diagnoses among adults with symptoms and initially unknown HIV status. RESULTS: We identified 3,818 symptomatic adults, referred 1424 (37%) for testing, of whom 1065 (75%) accepted, and 107 (10%) were newly diagnosed with HIV. The prevalence of new HIV diagnoses was 21% (95% CI: 17-25%) among those tested in health facilities and 5% (95% CI 4-7%) among those tested in mobile units. More men were diagnosed with HIV than women despite fewer men being screened. People who reported 4-5 symptoms were over twice as likely to be diagnosed with HIV compared to those reporting 1-3 symptoms (adjusted prevalence ratio [aPR] in health facilities = 2.58, 95% CI, 1.65-4.05; aPR in mobile units = 2.63, 95% CI, 1.37-5.03). CONCLUSION: We observed a high yield of new HIV diagnoses among adults identified by active application of a broad symptom screen. Integrated tuberculosis and HIV screening using could help close the detection gap for both conditions.

Setting: Although Kenya has a high burden of tuberculosis (TB), only 46% of cases were diagnosed in 2016. Objective: To identify strategies for increasing attendance at community-based mobile screening units. Design: We analysed operational data from a cluster-randomised trial, which included community-based mobile screening implemented during February 2015-April 2016. Community health volunteers (CHVs) recruited individuals with symptoms from the community, who were offered testing for human immunodeficiency virus (HIV) and sputum collection for Xpert((R)) MTB/RIF testing. We compared attendance across different mobile unit sites using Wilcoxon rank-sum test. Results: A total of 1424 adults with symptoms were screened at 25 mobile unit sites. The median total attendance among sites was 54 (range 6-134, interquartile range [IQR] 24-84). The median yields of TB diagnoses and new HIV diagnoses were respectively 2.4% (range 0.0-16.7, IQR 0.0-5.3) and 2.5% (range 0.0-33.3, IQR 1.2-4.2). Attendance at urban sites was variable; attendance at rural sites where CHVs were paid a daily minimum wage was significantly higher than at rural sites where CHVs were paid a nominal monthly stipend (P < 0.001). Conclusion: Mobile units were most effective and efficient when implemented as a single event with community health workers who are paid a daily wage.