Distinct Risk Factors for Clinical and Bacteriologically Confirmed Tuberculosis among Child Household Contacts in a High-Burden Setting

INTRODUCTION

In 2018, there were approximately 10 million new cases of tuberculosis globally; about 11% of these were child tuberculosis (TB) cases.¹ Therefore, the identification, diagnosis, and treatment of child TB cases are paramount to the global TB control strategy. In recent years, active case finding and contact investigations in high-burden settings have taken the place of passive self-presentation to clinics as preferred strategies for identifying children at high risk of developing TB disease.^2–8 Contact tracing typically incorporates symptom screening followed by a clinical evaluation and specimen collection in symptom-positive contacts; contact investigations may include differentiated care strategies to reduce barriers to screening, testing, treatment, and prevention.^9–11 Tests for TB are often not available in high-burden settings, and symptom screening of child TB contacts can additionally determine eligibility for TB preventive therapy (TPT) in low-resource areas.^12–15 Microbiological testing for child TB is complicated by barriers to specimen collection and poor sensitivity in detecting paucibacillary disease, so child TB cases in high-burden areas are predominately diagnosed via clinical determination.^16,17 Herein, we describe factors associated with active child TB among child contacts evaluated as part of our community-based contact-tracing project, BUTIMBA.

METHODS

The kingdom of Eswatini (formerly Swaziland) bears one of the world’s highest TB burdens, compounded by a high HIV prevalence. BUTIMBA, “The Royal Hunt” in siSwati, was an active case finding and household contact-tracing project implemented between 2013 and 2015 in three of Eswatini’s four regions.^18,19 Seven TB basic management units (BMUs), including national and district hospitals, referral clinics, and a rural health center, participated in the program. Study staff at each BMU were trained to identify index cases beginning anti-TB treatment (ATT) and perform household contact tracing and evaluation using a novel family mapping tool.

Eligible index cases provided demographic and clinical information for household contacts including age, gender, HIV status, sleep location, relationship to index case, and TB symptoms. Although study staff enumerated all household contacts, child household contacts (< 15 years of age) were prioritized for screening and evaluation. Index cases were asked to identify symptom screen–positive household contacts and return to the BMUs for further evaluation; index cases who failed to return to the BMU and who gave permission received visits at their homes by screening officers. Only contacts identified as symptom positive by index cases or by clinicians and screening officers during clinical evaluations were further evaluated for TB. Screening officers attempted to collect expectorated sputum in the community, whereas clinicians administered alternative sputum collection methods (induction or gastric or nasopharyngeal aspiration) and chest radiography in the BMUs. No individual participant consent was required for program evaluation as the data were analyzed anonymously. Approval was obtained from all necessary ethical bodies including the Baylor College of Medicine Children’s Foundation-Swaziland and the Swaziland Ethics Committee (24047712/24045469), and the Baylor College of Medicine Institutional Review Board (H-35028).

Samples were processed via acid-fast bacillus smear, culture, and GeneXpert cartridge-based PCR (Cepheid, Sunnyvale, CA) according to the Ministry of Health standard diagnostic algorithms. Clinicians at each BMU evaluated and diagnosed contacts for TB which was reported as clinically diagnosed or bacteriologically confirmed. Eligible, symptom-negative child contacts were offered isoniazid preventive therapy administered through the BMUs in accordance with national guidelines.

This cross-sectional analysis focuses on child household contacts of adult index cases; results considering all age-groups are reported elsewhere.^18,19 Univariate and multivariate regression analyses were conducted to determine risk factors associated with any child TB diagnosis and bacteriologically confirmed child TB. Final models were determined using a purposive variable selection strategy; P-values less than 0.05 were considered significant. Generalized estimating equations were used to compensate for correlation within households. All analyses were conducted using Stata 16.0 (StataCorp, College Station, TX).

RESULTS

BUTIMBA enrolled 2,589 TB index cases at ATT initiation, across seven BMUs, in three regions of Eswatini (Figure 1). Of these, 1,568 index cases (median age 32.2 years, interquartile range (IQR): 23.5–41.7) reported information on 5,413 child household contacts (median age 7.2 years, IQR: 3.6–10.7). Upon evaluation by study clinicians, 82 child household contacts (1.5%) were diagnosed with TB disease, 57 clinically determined and 25 bacteriologically confirmed. Of the 82 cases of child TB disease, 68 reported having one or more TB symptoms.

View Full Size

Study population.

Citation: The American Journal of Tropical Medicine and Hygiene 103, 6; 10.4269/ajtmh.20-0522

• Download Figure
• Download figure as PowerPoint slide

Table 1 contains demographic information for BUTIMBA’s study population. Enumerated child household contacts of TB index cases were well distributed across age and gender and representative of the target population.²⁰ Additional measures of socioeconomic status show that 73.2% of households included a miner and 27.4% of households had experienced a death within the past 2 years. Relationship with the index case was correlated with sleep location relative to the index case and was not included in model building to prevent multicollinearity bias. Furthermore, 38.9% of child household contacts of adult TB index cases were reported to have at least one TB symptom. Although self-reported (caretaker-reported) child HIV prevalence was low (3.4%), 53.9% of child contacts reported “unknown” HIV status.

Univariate and multivariate clustered logistic regression analyses of risk factors for any child TB diagnosis are provided in Table 2. Gender, presence of miner in household, and death in household within past 2 years were excluded from the final model based on lack of significance in univariate analyses. Sleeping in the same bed as the index case, compared with sleeping in a separate structure within the homestead/household, was significant in the univariate analysis, but not in the multivariate model. Children younger than 5 years were approximately 2.2 times more likely to have TB than children aged 5–14 years. In addition, children with positive HIV status (caregiver reported) were approximately five times more likely to have TB than those with negative HIV status. Children who reported one or more symptoms of TB were approximately 9.5 times more likely to have TB than those who did not report symptoms. Of note, children with “unknown” HIV status were less likely to have TB.

Univariate and multivariate clustered logistic regression analyses of risk factors for bacteriologically confirmed child TB were also conducted. Children living with HIV or who reported symptoms of TB were slightly less likely to have bacteriologically confirmed TB (odds ratio [OR]: 0.982, P-value: < 0.001 and OR: 0.995, P-value: < 0.001, respectively).

DISCUSSION

Child TB is among the top 10 causes of child death globally; clinicians’ ability to control and prevent child TB is restricted by patient access to health services, limited clinical resources and technology, and inadequate diagnostic techniques.⁶ The aim of BUTIMBA was to arm healthcare providers in a resource-limited, highly HIV/TB burden African country with a greater arsenal of tools to identify, screen, diagnose, treat, and prevent TB in children. By integrating active case finding with differentiated facility- and community-based contact management, BUTIMBA reached a previously inaccessible population of children at high risk for TB. In addition to providing support for essential public health programming, BUTIMBA characterized the population of children exposed to TB in their homes and identified risk factors associated with TB.

Our findings reinforce established knowledge of child TB while providing new insights into disease dynamics in an endemic region. Young children and children living with HIV are at highest risk for TB disease. “Unknown” HIV status appeared to have a protective effect. We hypothesize that caregiver knowledge and recall of HIV status could be a marker of previous interaction with the healthcare system; whereas all exposed children should have received HIV testing at 6 weeks of age in Eswatini, caregivers may not be informed of negative results or may not recall receiving negative results. HIV/AIDS-related stigma may also have contributed to caregiver fear in learning or disclosing their child’s HIV status. Children with one or more TB symptoms were more likely to be evaluated and diagnosed with clinical TB, but less likely to have bacteriologically confirmed TB, highlighting the degree of subjectivity in clinical determination of child TB.

Our findings align with global consensus on the importance of household contact evaluations for active TB disease and reinforce the importance of risk factors such as sleep location, age, and HIV status. The proportion of bacteriologically confirmed TB among cases reflects the enormous difficulty of confirming TB in children, and the reduction in bacteriological confirmation in children living with HIV may result from clinicians’ reduced threshold for diagnosing TB in this group. In high-burden low-resource settings, providers are especially driven to capture and treat any case of child TB, which may result in overdiagnosis.

The ability to identify, diagnose, treat, and prevent child TB cases is an essential but challenging facet of global TB control strategy. Child contacts of adult TB index cases represent an important population at risk for the development of TB disease. Tuberculosis prevention and control programs should integrate screening and TB evaluation with initiation of antituberculosis therapy or TPT. Public health practitioners in high-burden areas must therefore implement innovative, sustainable, and culturally appropriate interventions to capture and retain child contacts of TB cases in comprehensive contact management programs.