Xpert MTB/RIF assay did not improve diagnosis of pulmonary tuberculosis among child contacts in Rwanda

Introduction To report on the diagnostic yield using the Xpert MTB/RIF assay on gastric lavage samples from children (<15 years) who were household contacts of tuberculosis (TB) cases in Kigali, Rwanda. Methods A cross-sectional study was conducted among 216 child contacts of index cases with sputum smear-positive TB over a 7 month period, from 1st August 2015 to 29th February 2016. Child contacts with tuberculosis-related symptoms or abnormal chest X-ray had sputum collected by gastric lavage on two consecutive days and samples were examined by smear microscopy, Xpert MTB/RIF assay and solid culture. Results Of the 216 child contacts, 94 (44%) were less than 5 years of age. Most of them 84 (89%) were receiving isoniazid preventive therapy at the time of screening. Thirty seven out of 216 children had TB-related symptoms. Only 4 (10.8%) were clinically diagnosed with TB; and none had bacteriologically confirmed tuberculosis. Conclusion The use of Xpert MTB/RIF assay did not contribute to bacteriological confirmation of active TB in child contacts in this study. The low prevalence of tuberculosis in child contacts in this study may reflect the high coverage of preventive therapy in young (<5 years) child contacts. The low sensitivity of Xpert MTB/RIF assay in contacts may also suggest likely reflection of paucibacillary disease.


Introduction
Tuberculosis (TB) is a major cause of morbidity and mortality among children (0-14 years) in resource-limited countries [1]. The World Health Organisation (WHO) estimated that 10% of the 9 million TB incident cases occurred in children in 2015 and that there were 210,000 TB-related deaths in children, including 170,000 in Human Immunodeficiency Virus (HIV)-uninfected children [2]. The annual report of Rwanda's 2013-2014 National Tuberculosis Program (NTP) indicated that child TB cases represented 6% of all notified TB cases, below the national target of 12% [3]. Among these cases, 68% were pulmonary TB and 22% were bacteriologically confirmed. These data suggest under-detection of TB in children in Rwanda, especially of clinically diagnosed cases.
There are well recognised challenges with detection and diagnosis, particularly in young children (<5 years) with paucibacillary disease, difficulty in obtaining samples and clinical overlap of TB with other common diseases such as severe pneumonia and malnutrition [4][5][6][7].
Young children (< 5 years) who develop active TB subsequent to infection with Mycobacterium tuberculosis, usually do so within one year of infection [8]. Children who are close to a TB index case are at high risk of TB infection [9][10][11][12]. Without any intervention, 5-10% of infected children will develop active TB within one year, with the highest prevalence of TB at the time of screening being in young children (< 5 years) [8,13]. Screening of child contacts of TB cases, prioritising index cases with sputum smear-positive pulmonary TB, is almost universally recommended and plays two important roles which include identification and evaluation of symptomatic contacts of any age requiring further diagnostic assessment of TB for early treatment (i.e. active case finding), and the provision of preventive therapy to "high-risk" contacts that do not have active TB [14].
Since 2006, WHO has recommended a symptom-based screening approach that allows the initiation of contact management and the provision of preventive therapy for asymptomatic young child contacts at the household or primary care level [15,16]. However, symptomatic contacts need further evaluation for TB and this remains challenging at the primary or secondary care level given the widely recognized limitations of current diagnostic tools especially in young children.
In 2013, the WHO endorsed the Xpert MTB/RIF assay for use in children [15,17] Study population: Index cases diagnosed with sputum smearpositive PTB between August 2015 and February 2016 who had at least one child less than 15 years old, but who were not a member of a household to which a previous selected index case belonged and still living in Kigali city, were eligible for inclusion in the study.
Identified index cases were requested, either via telephone calls or through trained community health workers (CHWs), to bring children with whom they live with to their PHC on a specific day to coincide with visits to that PHC by data enumerators. Child contacts were defined as those contacts of less than 15 years who shared the same household with a selected index case within the 3 months prior to the diagnosis of the index case. Therefore, all child contacts born after the index cases had started treatment or were not leaving with the index cases prior to the diagnosis of the index cases were excluded. Eligible children were enrolled following written informed consent by the parent or caregiver and children of 7 years and older also signed an assent form.

Data collection and management:
A structured questionnaire adapted from screening guidelines [26,27] was pre-tested and modified during a pilot study in two selected sites. Twelve data enumerators were trained to conduct interviews with parents/caregivers of selected child contacts and to collect data from TB registers and index case folders, using standardised data collection forms. We also trained 20 CHWs to explain the study to the parents/caregivers, and sensitise them to bring child contacts for screening at the PHCs. Data of the index case included: result of smear microscopy, demographic data, address of residence and telephone number. The uptake of IPT among child contacts following diagnosis of the index case was also recorded. The recorded data were validated by the index case, parents or caregivers of selected children once they were identified in order to ensure the accuracy of the data. The demographics and medical history of index cases were recorded; and all eligible children underwent clinical screening including nutritional assessment and Chest X-ray (CXR). The clinical screening focussed on symptoms suggestive of TB: cough for ≥ 2 weeks, haemoptysis, fever, failure to gain weight, absence of appetite, fatigue, and the presence of lymphadenopathy. Anteroposterior and lateral CXR were also performed on all the 216 children; and read by two independent experienced general practitioners, trained in interpreting CXR and blinded to the clinical details of participants and proofread by an experienced radiologist. Children with symptoms suggestive of TB and/or CXR "consistent with active TB", as described in Table 1 and were based on clinical screening, X-ray and microbiological investigations.
Children were categorized as follows: bacteriologically confirmed TB, unconfirmed TB and unlikely TB (Table 1). Categorical data were interpreted through frequency  had symptoms suggestive of TB and/or CXR "consistent with active tuberculosis" at the time of screening. Table 2 and Table 3  Data of uptake and adherence to IPT will be presented separately once follow-up of the cohort is complete. The majority of child contacts selected in the study were asymptomatic at the time of screening: 179/216 or 83% (95% CI, 77%-87%). All symptomatic child contacts (100%) were exposed to air pollution (tobacco smoke or burning wood) and the majority (64%) had their parents as index cases with 81% in contact with index cases for more than 8 hours per day. In addition, the majorities (81%) of these children were living in the households with more than two people and 78% of those households had just one bedroom. Among the 37 symptomatic child contacts, 92% had at least one symptom suggestive of TB (

Competing interests
The authors declare no competing interests.

Acknowledgments
We would like to express our sincere gratitude to study participants  Tables and figure   Table 1: Operational definition used in this study     Unconfirmed TB: as our study is constituted by child contacts, we will consider in this category a child who will display at least one of the symptoms suggestive of TB, CXR consistent with active TB Unlikely TB: symptomatic child contacts suspected of TB whose symptoms and/or CXR consistent with active TB spontaneously improved after seven days of antibiotics without receiving any TB treatment Nutritional assessment using Weight-for-Height Normal : Z score ≥ -2 of the WHO median Moderate malnutrition: Z score -3 to < -2 of the WHO median Severe malnutrition: Z score < -3 Abbreviation WHO: World Health Organisation; CXR: Chest X-ray Page number not for citation purposes 10