From 2012–2014, ART-naïve CLHIV 0–12 years of age receiving routine medical care at five health facilities in Eastern Cape Province were enrolled into a previously described¹⁰ prospective cohort from the time they were identified as ART-eligible based on South African national guidelines^11,12. Children enrolled in the study received routine medical care at the participating health facilities and CLHIV at two clinics attended referral facilities for TB diagnosis and treatment. Quarterly study visits were co-scheduled with routine HIV care visits and included caregiver interviews to collect data on the child’s health and ART adherence. Data from medical charts and records were also abstracted.

During study enrollment and follow-up visits, caregivers were asked about the child’s TB history or any new diagnosis of TB. TB diagnosis was also abstracted from the child’s medical record at the health facility. Children were followed for up to 24 months with the last study visits in 2015. Enrolled children had routine CD4 cell count (CD4) and HIV RNA viral load (VL) monitoring based on 2010 and 2013 South African guidelines for children on ART^11,12. ART was started based on recommended first-line regimens from these same guidelines (abacavir [ABC], lamivudine [3TC], lopinavir/ritonavir [LPV/r] for children under three years and ABC, 3TC, efavirenz [EFV] for children three years and older.

A child had prevalent TB at study enrollment if there was a recorded TB disease diagnosis date in the medical record 90 days before or after the enrollment date, or if TB was reported in the medical chart or by a caregiver prior to the first follow-up visit which generally occurred 3 months after enrollment. Incident TB was defined as TB disease reported more than 90 days after study enrollment or after the first follow-up visit among children without prevalent TB at enrollment. Use of TB preventive therapy (TPT) prior to enrollment was ascertained from medical charts and defined for children with a recorded isoniazid preventive therapy start date prior to the date of enrollment. TB was categorized as pulmonary or extrapulmonary based on the recorded TB category in the medical record. Moderate malnutrition was defined as weight-for-age standardized Z-score (WAZ) less than or equal to two standard deviations below the mean. We considered the presence of an inside tap, inside toilet, or availability of electricity as a proxy for socioeconomic status or household income; children from households with all three were classified as the highest income proxy category, those with one or two were considered to have an intermediate income and those with none were considered to have the lowest income. Enrollment laboratory values (CD4, VL, and hemoglobin) included those recorded in the medical chart closest to the enrollment date (within one year before or up to 30 days after enrollment). Hemoglobin less than 8 mg/dL was defined as anemia. The definitions for all variables are otherwise included in Table 1.

In this secondary analysis, characteristics of children with and without prevalent TB, TB treatment or TPT at enrollment were assessed using chi square, Cochran Mantel Haenszel, Wilcoxon rank sum and Kruskal-Wallis tests for categorical and continuous variables. Characteristics of children with and without TB were assessed using logistic regression with generalized estimating equations (GEE) to account for clustering within health facilities. Our multivariable regression model added variables with a p-value of less than 0.10 on bivariable GEE by forward selection (variables with missing values for more than 50% of children were excluded from the multivariable model). P-values of less than 0.05 (excluding missing levels) were considered to be significant. Analyses were conducted in SAS 9.4 (SAS Institute, Cary, USA) and R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria) along with the R survival package¹³.

The TB incidence rate was determined using the person-time contributed by children beginning from enrollment until the date of TB diagnosis, or end of study follow-up. We used univariable Cox proportional hazards with GEE and a robust long-rank test to assess enrollment characteristics associated with incident TB disease among CLHIV without prevalent TB at enrollment and with at least one study follow-up visit after enrollment.

Ethical and procedural reviews were obtained from Columbia University Medical Center Institutional Review Board, University of Cape Town Human Research Ethics Committee, East London Hospital Complex Research Ethics Committee, Walter Sisulu University Health Research Ethics Committee, and the Eastern Cape Department of Health. The protocol was also reviewed by the Centers for Disease Control and Prevention (CDC) and was determined to be research, but CDC investigators did not interact with participants or have access to identifiable data or specimens. Informed consent was not required.

Among 446 CLHIV identified as being eligible for ART, 401 caregivers gave consent, and 397 CLHIV enrolled in the cohort because 4 children had previous ART¹⁰. Enrolled children had a median age of 1.8 years (interquartile range [IQR] 0.4–7.2 years); 37.3% (n=148) were less than one year of age, and 20.2% (n=80) of children were either single or double orphans. More than half of children were male (52.4%; n=208) or were in the highest income proxy category (n=213; 53.7%). Most children had a CD4 at enrollment ≥ 350 cells/µL (60.5%, n=240/397 or 70.5% among 338 children with data) and a VL ≥ 10⁵ copies/mL (62.2%, n=247/397 or 75.5% among 327 children with data). (Table 1).

Of 397 enrolled children, 114 (28.7%) had prevalent TB including: 29 (25.4%) infants younger than 1 year old; 39 (34.2%) children 1–4 years old; and 46 (40.4%) children 5–12 years old (Table 1). Children with prevalent TB had lower CD4 compared to children without prevalent TB (odds ratio [OR] 1.7; 95% CI 1.2–2.4) (Table 1). In the multivariable model, enrollment CD4 less than 350 cells/µL (adjusted OR [aOR] 1.6; 95% CI 1.1–2.2), higher household income proxy (aOR 1.8; 95% CI 1.3–2.6 for the highest income proxy and aOR 1.6; 95% CI 1.6–1.7 for the intermediate income proxy both compared to the lowest income proxy), and moderate malnutrition at enrollment (aOR 1.6; 95% CI 1.1–2.6) were associated with prevalent TB (Table 1).

There were 246 children without enrollment TB who had at least one follow-up visit among the 397 enrolled children. Of the 246 children with follow-up (and no TB at enrollment), 20 (8.1%) had TB during study follow-up resulting in a TB incidence rate of 5.2 cases per 100 person-years. Time from enrollment to ART initiation differed significantly comparing children with and without incident TB: among CLHIV with incident TB, median time to ART was 52 days [interquartile range (IQR) 2–130] versus 9 days [IQR 0–21] in CLHIV without incident TB (p=0.03). CLHIV who initiated ART more than or equal to 60 days after ART eligibility had a nearly five-fold increased risk of incident TB (hazard ratio [HR] of 4.7, 95% CI 2.3–9.4) compared to those who started ART sooner (Table 2). Moderate malnutrition (HR 1.8, 95% CI 1.1–2.7) and lack of cotrimoxazole (HR 2.3, 95% CI 1.0–4.9) at study enrollment were also associated with higher hazards of incident TB (Table 2). Among the 19 children with incident TB who started ART, the median time from ART initiation to TB diagnosis was 253.0 days (IQR 140.5– 422.5).

Among all 134 CLHIV with prevalent or incident TB, the type of TB was listed for 80 children (59.7%). Of those with known TB type, most (83.8%; n=67) had pulmonary TB, while 10.0% (n=8) had extrapulmonary TB, and 6.3% (n=5) had both pulmonary and extrapulmonary TB. Most TB was diagnosed with a chest radiograph alone (68.6%, n=70), otherwise TB was diagnosed by chest radiograph and sputum (7.8%; n=8), sputum alone (5.9%; n=6), clinical diagnosis alone (4.9%; n=5), or other or unlisted methods (12.7%; n=13). A record of ever receiving TB treatment was documented for 97.8% (n=131) of 134 children with prevalent or incident TB.

TPT data were recorded for 362 of 397 children enrolled. Of these 362 children, 31 (8.6%) received TPT. Overall, TB occurred in 8 (25.8%) of 31 children with TPT. Among the 8 children with TPT who were diagnosed with TB, 7 (87.5%) had both a recorded TPT start date and TB diagnosis date. These 7 children were on TPT for a median of 14.0 days (IQR 6.0–279.0 days; range 3.0 to 529.0 days) before TB was diagnosed. There were 232 children without TB at enrollment who had both follow-up and TPT data. Among these 232 children, incident TB occurred in 1 out of 23 (4.4%) children given TPT and 18 out of 209 (8.6%) children who did not get TPT (p=0.7) (One child with incident TB was missing TPT data). Receipt of TPT did not differ by age (data not shown).

Of 134 children with enrollment or incident TB, TB episode outcome was documented for n=107, or 91.5% of those eligible for outcome data. For children with documentation in the HIV records, 105 (98.1%) completed TB treatment, and 2 children defaulted TB treatment, including one child (8.3%) among the 12 with incident TB and known TB outcome. Among all 397 enrolled children, there were 35 deaths recorded (10.2%) among 342 children with known outcome. Children who died had a median age of 0.4 years (IQR 0.2–2.0 years). Verbal autopsy information was available for 32 deaths (91.4%) with a known cause of death recorded for 18 (56.3%) of these children. Among deaths with known cause, 3 (16.7%) deaths were reported to be related to TB.