Abstract
Introduction:
This study investigates drivers of childhood pulmonary tuberculosis (PTB) using a childhood ecosystem approach in South Africa. An ecosystem approach toward identifying risk factors for PTB may identify targeted interventions.
Methods:
Data were collected as part of a prospective cohort study of children presenting at a primary care facility or tertiary hospital with possible TB. Characterization of the childhood ecosystem included proximal, medial, and distal determinants. Proximal determinants included child characteristics that could impact PTB outcomes. Medial determinants included relational factors, such as caregiver health, which might impact interactions with the child. Distal determinants included macro-level determinants of disease, such as socioeconomic status and food insecurity. Children who started on TB treatment were followed for up to 6 months. Multivariate regression models tested independent associations between factors associated with PTB in children.
Results:
Of 1202 children enrolled, 242 (20%) of children had confirmed PTB, 756 (63%) were started on TB treatment, and 444 (37%) had respiratory conditions other than TB. In univariate analyses, childhood malnutrition and caregiver smoking were associated with treated or confirmed PTB. In multivariate analyses, proximal factors, such as male gender and hospitalization, as well as low socioeconomic status as a distal factor, were associated with PTB.
Conclusions:
Interventions may need to target subgroups of children and families with elevated proximal, medial, and distal risk factors for PTB. Screening for risk factors, such as caregiver’s health, may guide targeting. The provision of social protection programs to bolster economic security may be an important intervention for attenuating childhood exposure to risk factors.
Keywords: conceptual hierarchical frameworks, TB determinants
1 |. INTRODUCTION
Childhood tuberculosis (TB) is a major global health problem. Among the 10 million individuals estimated to have TB disease in 2019, 12% were children. In 2019, the estimated total of HIV-positive deaths for children 14 years old or less was 36,000.1 Tuberculosis ranks among the top 10 global causes of death for children, including children under 5 years.2 The prevalence of TB is especially high among children and adults from low- and middle-income countries, such as South Africa.3,4 South Africa has one of the highest country burdens of TB globally, with over 300,000 people falling ill with TB in 2019.1
Conceptual hierarchical frameworks are useful to characterize the etiology of chronic health conditions, such as TB. Previous studies have utilized frameworks that organize determinants of chronic disease into proximal, medial, and distal levels of the ecosystem.5,6 Consideration of these ecosystem dimensions of health can guide efforts to reduce disease burden for priority populations in the TB epidemic, such as children. Attention toward the ecological aspects of TB health for children is vital for several reasons. Diagnosis and appropriate treatment of TB among children are dependent on having an effective, healthy caregiver. Individuals and families affected by TB are often faced with challenging economic environments. These adverse economic environments can contribute to psychological distress, reduce access to protective resources, and increase individual vulnerability to poor TB outcomes.7,8 Such environments can predispose children to malnutrition, which can increase the risk of childhood TB.9–11 More directly, caregiver’s health, including psychological distress and substance abuse, has been linked to TB nonadherence among adults; these caregiver factors may also affect TB outcomes among children. Therefore, the overarching child ecosystem may affect childhood TB incidence and outcomes. The child ecosystem, including caregiver health and the broader social and economic environment, are vital factors to address.
The aim of this study was to investigate factors associated with pulmonary tuberculosis (PTB) in children and explore health markers that characterize the ecosystem of childcare to better understand the links between the childhood ecosystem and PTB.
2 |. METHODS
2.1 |. Study design and setting
Data were derived from a prospective study of children enrolled with possible TB disease in a TB diagnostic study at two sites in Cape Town, South Africa: (1) Nolungile Clinic, a primary healthcare clinic in an urban low socioeconomic area, and (2) Red Cross War Memorial Children’s Hospital, a tertiary care referral hospital.12,13 Children with possible PTB were followed at 1, 3, and 6 months for a response to treatment.
2.2 |. Enrollment and participant consent
Children were enrolled in the study between September 2010 and July 2017. Inclusion criteria were (1) children less than 15 years old and (2) clinically possible pulmonary TB with or without extrapulmonary TB. Possible TB was based on cough of any duration and one of the following: household TB contact, recent weight loss, positive tuberculin skin test or a chest radiograph12,13; (3) known HIV status or consent for HIV testing; and (4) informed consent from the parent or legal guardian and verbal assent from children over 7 years of age. TB therapy was initiated at the discretion of the treating doctor. Follow-up visits were done at 1, 3, and 6 months for children on TB therapy and at 1 and 3 months for those not treated. Children were recorded as lost to follow-up (LTFU) if they were enrolled in the study and completed all study activities, including sampling and tests but did not return for follow-up visits despite three attempts by the clinical staff to find the child and caregiver.
Ethics approval for the study was obtained from the Human Research Ethics Committee of the Faculty of Health Sciences at the University of Cape Town (UCT HREC:045/2008). Written informed consent was obtained from a parent or legal guardian, and verbal assent was obtained from children 7 years or older.
2.3 |. Outcome measures
Outcome measures used in our child–ecosystem framework include TB diagnostic category and TB treatment. To look for patterns among TB-related determinants using the childhood ecosystem approach, malnutrition was selected as an outcome measure for an additional analysis given the known risk factors between malnutrition and childhood TB. PTB diagnostic categorization was based on clinical and microbiological investigations, in line with National Institute of Health (NIH) consensus definitions as “confirmed PTB” (microbiologically confirmed), “unconfirmed PTB” (microbiologically negative, clinically diagnosed), and “unlikely PTB” (lower respiratory disease not due to TB with improvement on follow-up in the absence of TB treatment).14 We compared two groups of children with TB (those treated for TB, defined as those clinically diagnosed with PTB and those microbiologically confirmed) to children in the category “unlikely” TB.
2.4 |. Ecosystem measures
Determinants of TB diagnostic category and TB treatment were organized into proximal, medial, and distal levels as part of our hierarchical framework for understanding the childhood ecosystem for TB in South Africa. Proximal determinants were understood as factors that were specific to the individual child characteristics, including demographic factors that were hypothesized to impact child TB outcomes, such as gender and TB treatment. Medial determinants included interpersonal factors that might impact childhood TB. These factors included aspects of caregiver health that might impact interactions with the child (psychological health, substance use, stress) and other interpersonal measures, such as social support. Distal determinants included macro-level determinants of disease, such as socioeconomic status (SES) and food insecurity.
2.5 |. Child-focused measures
Clinical diagnostics for children other than TB diagnosis and treatment were operationalized as proximal measures in PTB consistent with how they were operationalized in the chronic disease context by Egger and Dixon.5 Measures include gender and previous hospitalization for TB.
2.6 |. Caregiver questionnaire
To capture the medial and distal measures of the child ecosystem, caregivers completed a self-administered questionnaire at enrollment to assess the risk factors for TB. The questionnaires included assessment of SES,7 substance abuse, psychological distress, perceived stress, social support, and on caregiver’s health, including substance use. Six measures were used to contextualize the child ecosystem measuring health status for caregivers, including the following.
2.6.1 |. Kessler’s psychological distress scale
Kessler’s psychological distress scale (K10)15 was used to measure caregiver’s psychological distress. For K10, a total score was calculated; a score ≥ 30 was indicative of severe psychological distress. The K10 scale has been previously validated among adults in South Africa, ages 18 or older, with moderate discriminate validity.16
2.6.2 |. Perceived stress
The Cohen Perceived Stress Scale (PSS)17 was used to measure caregiver’s perceived stress level. For the PSS, a total score was calculated; a score ≥ 20 was indicative of high-stress levels.
2.6.3 |. Perceived social support
A modified Multidimensional Scale of Perceived Social Support18 (MSPSS) was used to assess caregiver’s access to social support, comprising 12 questions with a four-point scaling a total score was calculated; three levels of support were generated, <28 high, 28–38 moderate, and >38 low.
2.6.4 |. Substance use
The Alcohol Use Disorders Identification Test-C (AUDIT-C)19 was used to collect information about alcohol use, and the Fagerström Test for Nicotine Dependence (FTND) information about cigarette smoking.20 The AUDIT-C is a 3-item version of the longer AUDIT scale.19 Studies have found high comparability between the AUDIT-C and the full AUDIT.21 To measure other substance use, the Drug Use Disorders Identification Test (DUDIT) was administered. The 11-item questionnaire did not have an adequate response rate for items 4–11. Due to the low response rate for scale, the measure was unusable.
The FTND is an instrument used to assess the intensity of physical addiction to nicotine.20 The measure consists of six questions with scores of 0–1 for yes/no items and scores of 0–3 for multiple-choice items. The thresholds for dependence are the following: scores 1–2: low, 3–4: low to moderate, 5–7: moderate, >8: high. For the FTND, only classification of caregivers as smokers or nonsmokers was possible.
2.6.5 |. Nutritional status
To capture the distal factors of the child ecosystem, we examined food insecurity. Specifically, we examined child malnutrition as a proxy for food insecurity. We examined childhood nutrition by calculating height-for-age (HAZ), weight-for-age (WAZ), and body mass index z scores using World Health Organization (WHO) Child Growth Standards.22,23 Under-weight-for-age was defined as a weight-for-age z score <−2, malnutrition as weight-for-age < −3, stunting as height-for-age z score < −2, and severe stunting as a height-for-age z score of < −3.
2.6.6 |. Socioeconomic status and educational attainment
To capture distal factors, SES was assessed using a list of 14 common household items and amenities, adapted from an SES scoring system used in social epidemiology in South Africa.7 The median number of assets reported by the caregivers was seven assets and a cut point was established. Lower SES was defined as a household with seven or fewer assets and higher SES defined as a household with eight or more. An additional measure of household monthly income was assessed to further understand caregiver resources; this included earnings and social grant support. Household monthly income was categorized based on the three following categories <$68, $68–340, and >$340 USD. South African Rand monthly income amounts were converted to USD for this analysis.24 In addition, caregiver education was explored as a risk factor for childhood TB. Education was classified as <10 or ≥10 years of schooling.
2.7 |. Data analyses
Data analyses were performed using STATA 15.1.25 Exploratory data analysis of categorical and continuous variables included frequency tables and histograms of continuous variables to determine distribution. Simple descriptive statistics were used to characterize the study population. Normally distributed continuous data were summarized by mean and standard deviations (SD); non-normally distributed continuous data by the median and interquartile range (IQR). Categorical data were summarized as number and proportion. Statistical tests included the χ2 test of equal proportions and the Kruskal–Wallis comparison of medians. Separate multivariate regression models were developed to further investigate the collective relationship between independent variables in the childhood ecosystem model, PTB in children, TB treatment, and malnourished children. Only variables that were significant in the univariate models in addition to age and gender were included in multivariate models. Results of the regression models are reported as odds ratios (ORs) with 95% confidence intervals (CIs). Separate regression analyses looked at factors related to nonadherence to follow-up visits. Statistical tests were two-sided at α = .05.
3 |. RESULTS
Over a 7-year period, 1738 children were enrolled at the two study sites (Figure 1). During this period, 1202 (69%) caregivers completed questionnaires. Of these, 959 (80%) were completed by the mother, 57 (5%) by the father, and 117 (10%) by another family member. Six percent of caregivers were not family members or were not specified. Overall, 1135 (94%) of caregivers were female, with a median (IQR) age of 30 (25–37) years; 7 (0.6%) caregivers were younger than 18 years; and 371 (34%) of caregivers reported being HIV-infected (Table 1). Two hundred and ninety (24%) of the caregivers were previously treated for TB, and 880 (73%) caregivers completed at least 10 years of schooling.
FIGURE 1.
Flow diagram of childhood tuberculosis diagnosis across study sites
TABLE 1.
Baseline characteristics of caregivers, by child’s TB category
| Overall 1202 | Confirmed TB 242 (20) | Unconfirmed TB 560 (47) | Unlikely TB 400 (33) | p value | |
|---|---|---|---|---|---|
| Socio-demographic | |||||
| Age, median in years (IQR) | 30 (25, 37) | 32 (26, 38) | 30 (25, 37) | 29 (25, 35) | .005 |
| Age < 18 years | 7 (0.6) | 1 (0.4) | 6 (1.1) | 0 (0.0) | .092 |
| Gender, female, n (%) | 1135 (94) | 225 (93) | 534 (95) | 376 (94) | .363 |
| HIV infected, n (%) | 371 (34) | 78 (35) | 162 (32) | 131 (35) | .445 |
| Previous TB treatment, n (%) | 290 (24) | 50 (21) | 126 (23) | 114 (29) | .037 |
| Low socioeconomic status, n (%) | 648 (62) | 132 (69) | 294 (62) | 222 (58) | .055 |
| Monthly income category, n (%) | |||||
| <$68 | 275 (59) | 68 (68) | 136 (63) | 71 (47) | |
| $68–340 | 177 (38) | 29 (29) | 71 (33) | 77 (51) | .003 |
| >$340 | 15 (3) | 3 (3) | 8 (4) | 4 (3) | |
|
Psychological Kessler’s psychological distress scale category (K10), n (%) | |||||
| Well | 348 (40) | 67 (40) | 173 (43) | 108 (37) | |
| Mild disorder | 128 (15) | 24 (14) | 48 (12) | 56 (19) | |
| Moderate disorder | 118 (14) | 23 (14) | 56 (14) | 39 (13) | .249 |
| Severe disorder | 269 (31) | 52 (31) | 130 (32) | 87 (30) | |
| Cohen’s perceived stress category | |||||
| High stress, n (%) | 308 (45) | 68 (49) | 140 (45) | 100 (44) | .594 |
| Perceived social support category n (%) | |||||
| High | 538 (61) | 106 (61) | 256 (61) | 176 (60) | |
| Moderate | 281 (32) | 56 (32) | 133 (32) | 92 (32) | .991 |
| Low | 67 (8) | 13 (7) | 30 (7) | 24 (8) | |
| School category | |||||
| <10 years of schooling | 321 (27) | 78 (32) | 156 (28) | 87 (22) | |
| ≥10 years of schooling | 880 (73) | 164 (68) | 403 (72) | 313 (78) | .010 |
| Substance use | |||||
| Fagerström, any category of smoking, n (%) | 233 (23) | 55 (27) | 116 (25) | 62 (18) | .031 |
| Active alcohol use, n (%) | 166 (22) | 25 (18) | 82 (24) | 59 (22) | .415 |
Abbreviations: IQR, interquartile range; TB, tuberculosis.
3.1 |. Characteristics of children
Most children, 833 (69%) were hospitalized. In addition, 369 (31%) presented as outpatients at a clinic not needing hospitalization. The median (IQR) age of the sample was 2.9 (1.5–5.5) years; 604 (50%) were male and 167 (14%) were HIV-infected. The median (IQR) z score for weight-for-age was −0.7 (−1.6 to 0.2) and height-for-age −1.0 (−2.0 to 0.2). In all, 265 (26%) were classified as stunted, 190 (17%) as under-weight-for age, 126 (12%) as severely stunted, and 72 (7%) as malnourished (Table 2).
TABLE 2.
Baseline characteristics of children by TB category
| Overall 1202 | Confirmed TB 242 (20) | Unconfirmed TB 560 (47) | Unlikely TB 400 (33) | p value | |
|---|---|---|---|---|---|
| Age, median in years (IQR) | 2.9 (1.5, 5.5) | 3.9 (1.7, 7.0) | 2.6 (1.3, 5.2) | 2.8 (1.4, 5.2) | <.001 |
| Gender, n (%) male | 604 (50) | 132 (55) | 299 (53) | 173 (43) | .003 |
| HIV, n (%) infected | 167 (14) | 38 (16) | 79 (14) | 50 (13) | .499 |
| Previous TB treatment, n (%) | 120 (10) | 21 (9) | 48 (9) | 51 (13) | .085 |
| Started on TB treatment, n (%) | 746 (63) | 239 (99) | 507 (93) | 0 (0) | <.001 |
| Hospitalized, n (%) | 833 (69) | 216 (89) | 417 (74) | 200 (50) | <.001 |
| Ambulatory, n (%) | 369 (31) | 26 (11) | 143 (26) | 200 (50) | |
| Weight-for-age, median (IQR) | −0.7 (−1.6, 0.2) | −0.9 (−2.0, −0.1) | −0.7 (−1.6, 0.1) | −0.5 (−1.3, 0.4) | <.001 |
| Height-for-age, median (IQR) | −1.0 (−2.0, 0.2) | −1.3 (−2.3, −0.4) | −0.9 (−1.9, 0.2) | −0.9 (−2.0, 0.8) | <.001 |
| BMI-for-age, median (IQR) | −0.3 (−1.5, 0.9) | −0.3 (−1.7, 0.9) | −0.4 (−1.6, 0.8) | −0.2 (−1.5, 1.1) | .295 |
| Underweight-for-age, n (%) | 190 (17) | 53 (25) | 90 (17) | 47 (13) | <.001 |
| Malnourished, n (%) | 72 (7) | 21 (10) | 32 (6) | 19 (5) | .059 |
| Stunting, n (%) | 265 (26) | 66 (31) | 112 (24) | 87 (25) | .104 |
| Severe stunting, n (%) | 126 (12) | 33 (16) | 55 (12) | 38 (11) | .230 |
| 3-month visit status, n (%) | |||||
| Completed | 1095 (91) | 229 (95) | 477 (85) | 389 (97) | <.001 |
| LTFU | 74 (6) | 7 (3) | 64 (11) | 3 (1) | |
| Withdrawn | 27 (2) | 6 (2) | 13 (2) | 8 (2) | |
| Died | 6 (0.05) | 0 (0) | 6 (1) | 0 (0) | |
Abbreviations: BMI, body mass index; IQR, interquartile range; LTFU, lost to follow-up; TB, tuberculosis.
3.2 |. Proximal factors
3.2.1 |. Child TB disease
There were 242 (20%) children with confirmed PTB, 560 (47%) with unconfirmed TB, and 400 (33%) with unlikely TB (Table 2). Overall, 746 (63%) children were started on TB treatment (Table 3). There were 444 children who were not started on TB treatment, of whom 44 were LTFU, and were classified in the unconfirmed category as improvement without TB treatment could not be ascertained. Treated PTB was associated with the following child characteristics in univariate analyses: male (p = .004), previous hospitalization (p < .001), lower weight-for-age z score (p = .001), lower height-for-age z score (p = .003) (Table 3). Previous TB treatment was found to be protective against being started on TB treatment (p = .028). In the multivariate analysis, male gender (p = .004) and previous hospitalization (p < .001) remained significantly associated with treated PTB as proximal factors.
TABLE 3.
Unadjusted and adjusted baseline associates of tuberculosis disease in children
| Treated for TB N = 746 (63) | Respiratory illness not TB N = 444 (37) | Univariate analysis |
Multivariate analysis |
|||
|---|---|---|---|---|---|---|
| OR (95% CI) | p value | OR (95% CI) | p value | |||
| Children’s characteristics | ||||||
| Age, median in years (IQR) | 3.0 (1.5, 5.6) | 2.7 (1.3, 5.3) | 1.0 (1.0–1.1) | .058 | ||
| Gender, n (%) male | 398 (53) | 199 (45) | 1.4 (1.1–1.8) | .004 | 1.5 (1.1–2.0) | .004 |
| HIV, n (%) infected | 106 (14) | 59 (13) | 1.1 (0.8–1.5) | .655 | ||
| Previous TB treatment, n (%) | 64 (9) | 56 (13) | 0.7 (0.4–1.0) | .028 | 0.7 (0.5–1.1) | .151 |
| Hospitalized, n (%) | 588 (79) | 237 (53) | 3.3 (2.5–4.2) | <.001 | 2.8 (1.2–3.8) | <.001 |
| Ambulatory, n (%) | 158 (21) | 207 (47) | ||||
| Weight-for-age, median (IQR) | −0.8 (−1.7, −0.1) | −0.5 (−1.4, 0.4) | 0.9 (0.8–0.9) | .001 | 0.9 (0.8–1.0) | .132 |
| Height-for-age, median (IQR) | −1.0 (−2.1, −0.1) | −0.9 (−2.0, 0.6) | 0.9 (0.8–1.0) | .003 | ||
| BMI-for-age, median (IQR) | −0.4 (−1.6, 0.8) | −0.2 (−1.5, 1.0) | 1.0 (0.9–1.0) | .359 | ||
| Malnutrition, n (%) | 50 (7) | 22 (5) | 1.4 (0.8–2.4) | .179 | ||
| Caregiver’s characteristics | ||||||
| Age, median in years (IQR) | 31 (25, 37) | 29 (25, 36) | 1.0 (1.0–1.0) | .478 | ||
| Gender, n (%) female | 707 (95) | 418 (94) | 1.1 (0.7–1.9) | .645 | ||
| HIV, n (%) infected | 220 (32) | 146 (35) | 0.9 (0.7–1.1) | .271 | ||
| Previous TB treatment, n (%) | 164 (22) | 120 (27) | 0.8 (0.6–1.0) | .050 | 1.0 (0.7–1.3) | .774 |
| Low socioeconomic status, n (%) | 396 (64) | 248 (59) | 1.2 (1.0–1.6) | .109 | ||
| Monthly income < $68, n (%) | 193 (64) | 78 (48) | 1.9 (1.3–2.9) | .001 | ||
| Education | ||||||
| <10 years of schooling, n (%) | 217 (29) | 97 (22) | 1.5 (1.1–1.9) | .006 | 1.3 (1.0–1.9) | .090 |
| Substance use | ||||||
| Active smoker | 157 (25) | 70 (19) | 1.5 (1.1–2.0) | .019 | 1.0 (0.7–1.4) | .801 |
| Active alcohol use, n (%) | 99 (22) | 67 (23) | 1.0 (0.7–1.4) | .844 | ||
| Severe psychosocial distress | 168 (32) | 95 (29) | 1.1 (0.8–1.5) | .472 | ||
Note: 12 children are not included in this table, as it is not known if they started on TB treatment due to lost-to-follow-up.
Abbreviations: 95% CI, 95% confidence interval; BMI, body mass index; IQR, interquartile range; OR, odds ratio; TB, tuberculosis.
3.3 |. Adherence to the 3-month follow-up visit
Of the 1202 children whose caregivers completed the psychosocial questionnaire at baseline, 6 (0.5%) children died before the 3-month visit, and 27 (2%) were disenrolled due to relocation. Overall, adherence in this study was high, approaching 94%. Of the remaining 1169, 74 (6%) did not attend the 3-month visit and were considered LTFU (Table 2). LTFU was associated with children with severe stunting (p = .017), hospitalization (p < .001), caregiver smoking (p = .009), or severe psychological distress as assessed based on the Kesler 10 questionnaire administered to the caregiver questionnaire (p = .019) in univariate analyses. In a multivariate logistic regression model, severe stunting was independently associated with nonadherence to the 3-month follow-up visit (Table S1).
3.4 |. Medial factors
3.4.1 |. Caregiver characteristics
Overall, 233 (23%) caregivers reported smoking while 166 (22%) consumed alcohol; 308 (45%) of caregivers were classified as experiencing high stress (Table 1). Of all caregivers, 538 (61%) perceived that they had high social support, 281 (32%) reported moderate social support, and 67 (8%) reported low social support (Table 1). Active smoking (p = .019) was a caregiver characteristic associated with childhood TB in univariate analyses. Only a limited number of substance abuse questions were completed by caregivers, resulting in missing data. Previous TB treatment of the caregiver was protective in the univariate analysis (p = .050) (Table 3). A similar association was found when analyzed by microbiologically confirmed PTB (Table 4).
Table 4.
Unadjusted and adjusted baseline associates of mycobacteriologically confirmed TB versus unlikely TB
| Confirmed TB N = 242 (38) | Unlikely TB N = 400 (62) | Univariate analysis |
Multivariate analysis |
|||
|---|---|---|---|---|---|---|
| OR (95% CI) | p value | OR (95% CI) | p value | |||
| Children’s characteristics | ||||||
| Age, median in years (IQR) | 3.9 (1.7, 7.0) | 2.8 (1.4, 5.2) | 1.1 (1.1–1.2) | <.001 | 1.1 (1.0–1.3) | .012 |
| Gender, n (%) male | 132 (55) | 173 (43) | 1.6 (1.2–2.2) | .006 | 1.9 (1.2–3.0) | .005 |
| HIV, n (%) infected | 38 (16) | 50 (13) | 1.3 (0.8–2.1) | .242 | ||
| Previous TB treatment, n (%) | 21 (9) | 51 (13) | 0.7 (0.4–1.1) | .130 | ||
| Hospitalized, n (%) | 216 (89) | 200 (50) | 8.3 (5.3–13.0) | <.001 | 6.7 (3.9–11.8) | <.001 |
| Weight-for-age, median (IQR) | −0.9 (−2.0, −0.1) | −0.5 (−1.3, 0.4) | 0.8 (0.7–0.9) | <.001 | 0.9 (0.8–1.1) | .201 |
| Height-for-age, median (IQR) | −1.3 (−2.3, −0.4) | −0.9 (−2.0, 0.8) | 0.8 (0.7–0.9) | <.001 | ||
| BMI-for-age, median (IQR) | −0.3 (−1.7, 0.9) | −0.2 (−1.5, 1.1) | 1.0 (0.9–1.1) | .960 | ||
| Malnutrition, n (%) | 21 (10) | 19 (5) | 2.1 (1.1–4.0) | .024 | ||
| Caregiver’s characteristics | ||||||
| Age, median in years (IQR) | 32 (26, 38) | 29 (25, 35) | 1.0 (1.0–1.0) | .011 | 1.0 (1.0–1.0) | .237 |
| Gender, n (%) female | 225 (93) | 376 (94) | 0.8 (0.4–1.6) | .607 | ||
| HIV, n (%) infected | 78 (35) | 131 (35) | 1.0 (0.7–1.4) | .916 | ||
| Previous TB treatment, n (%) | 50 (21) | 114 (29) | 0.7 (0.4–1.0) | .027 | 0.6 (0.4–1.1) | .118 |
| Low socioeconomic status, n (%) | 132 (69) | 222 (58) | 1.6 (1.1–2.3) | .017 | 1.7 (1.0–2.7) | .036 |
| Monthly income < $68, n (%) | 68 (68) | 71 (47) | 2.4 (1.4–4.1) | .001 | ||
| Education | ||||||
| <10 years of schooling, n (%) | 78 (32) | 87 (22) | 1.7 (1.2–2.5) | .003 | 1.3 (0.8–2.2) | .337 |
| Substance use | ||||||
| Caregiver active smoker | 55 (27) | 62 (18) | 1.7 (1.1–2.5) | .016 | 1.0 (0.6–1.8) | .939 |
| Active alcohol use, n (%) | 25 (18) | 59 (22) | 0.8 (0.5–1.3) | .376 | ||
| Severe psychosocial distress | 52 (31) | 87 (30) | 1.1 (0.7–1.6) | .767 | ||
Note: Monthly income was not included in the multivariate model as less than 50% of caregivers responded.
Abbreviations: 95% CI, 95% confidence interval; BMI, body mass index; IQR, interquartile range; OR, odds ratio; TB, tuberculosis.
3.4.2 |. Psychological distress
Of all caregivers, 863 (72%) completed all questions on the K-10 questionnaire. The median score (IQR) was 23 (15–30); 348 (40%) were classified as being well, 128 (15%) as having mild distress, 118 (14%) as having moderate distress, and 269 (31%) as having severe distress (K-10 score ≥ 30). There was no difference in the proportion of caregivers with severe psychological distress across the three categories of children with PTB (p = .249) (Table 1). For the outcome of malnutrition, severe psychological distress was associated with malnutrition among children in both the univariate and multivariate analyses (p = .028) and (p = .015), respectively (Table 5).
Table 5.
Unadjusted and adjusted baseline associates of malnutrition at baseline
| Malnourished (WAZ ≤−3) N = 72 (7) | Not malnourished N = 1028 (94) | Univariate analysis |
Multivariate analysis |
|||
|---|---|---|---|---|---|---|
| OR (95% CI) | p value | OR (95% CI) | p value | |||
| Children’s characteristics | ||||||
| Age, median in years (IQR) | 1.7 (0.9, 3.9) | 2.7 (1.4, 5.0) | 0.9 (0.8–1.0) | .016 | 0.9 (0.8–1.1) | .289 |
| Gender, n (%) male | 42 (58) | 518 (50) | 1.4 (0.8–2.2) | .194 | ||
| HIV, n (%) infected | 19 (26) | 126 (12) | 2.6 (1.5–4.5) | .001 | 4.4 (2.1–9.0) | <.001 |
| Previous TB treatment, n (%) | 10 (14) | 96 (10) | 1.6 (0.8–3.2) | .193 | ||
| TB treatment started, n (%) | 50 (69) | 625 (61) | 1.4 (0.8–2.4) | .179 | ||
| Hospitalized, n (%) | 66 (92) | 671 (65) | 5.9 (2.5–13.6) | <.001 | 3.7 (1.2–10.7) | .018 |
| Ambulatory, n (%) | 6 (8.3) | 357 (34.7) | ||||
| Caregiver’s characteristics | ||||||
| Age, median in years (IQR) | 30 (24, 37) | 30 (25, 36) | 1.0 (1.0–1.0) | .967 | ||
| Gender, female n (%) | 69 (96) | 974 (95) | 1.3 (0.4–4.2) | .688 | ||
| HIV, n (%) infected | 32 (49) | 315 (33) | 2.0 (1.2–3.3) | .008 | ||
| Previous TB treatment, n (%) | 23 (32) | 244 (24) | 1.5 (0.9–2.6) | .113 | ||
| Low socioeconomic status | 38 (61) | 477 (53) | 1.5 (0.9–2.7 | .144 | ||
| Monthly income <$68, n (%) | 18 (75) | 230 (57) | 2.2 (0.9–5.7) | .096 | ||
| Educational | ||||||
| <10 years of schooling, n (%) | 29 (40) | 256 (25) | 2.0 (1.2–3.3) | .005 | 1.7 (0.8–3.2) | .139 |
| Substance use | ||||||
| Active smoker | 33 (55) | 177 (20) | 4.8 (2.8–8.1) | <.001 | 3.5 (1.8–6.7) | <.001 |
| Active alcohol use, n (%) | 12 (27) | 142 (22) | 1.3 (0.7–2.6) | .438 | ||
| Severe psychosocial distress | 24 (44.4) | 220 (30.0) | 1.9 (1.1–3.3) | .028 | 2.2 (1.2–4.2) | .015 |
Abbreviations: 95% CI, 95% confidence interval; IQR, interquartile range; OR, odds ratio; WAZ, weight-for-age z score.
3.5 |. Distal factors
3.5.1 |. Socioeconomic status
Overall, 648 (62%) of households were classified as of low SES. When examined by monthly income, 275 (59%) of caregivers earned less than $68, 177 (38%) earned between $68 and 340, and 15 (3%) earned more than $340 per month (Table 1). Caregiver monthly income was associated with differences in childhood TB status (p = .003); as was caregiver educational attainment of less than 10 years (p = .010) (Table 1). Both relationships remained significant in the unadjusted model for PTB in children. In the univariate analyses, monthly income below $68 USD and educational attainment below 10 years were also associated with microbiologically confirmed TB (Table 4). Low SES remained statistically significant in the multivariate analyses for children with microbiologically confirmed PTB (p = .036), further highlighting the importance of this distal risk factor.
3.5.2 |. Nutritional status
Nutritional status was investigated in two ways. First, it was used as a proxy for food insecurity to explore the impact of this distal factor on PTB. Second, it was used as an outcome variable to look at patterns between TB determinants and risk factors related to TB. While all measures of nutrition were explored, WAZ was the best predictor for malnutrition (Table 2); models, therefore, reflect the use of WAZ < −3.
Malnutrition was associated with the following child characteristics: younger age (p = .016), HIV infection (p = .001), hospitalization (p < .001), and the following caregiver characteristics: less than 10 years of schooling (p = .005), HIV infection (p = .008), smoking (p < .001), and severe psychological distress, (p = .028). In a multivariate logistic regression model, child HIV-infection (p < .001), hospitalization (p = .018), caregiver active smoking (p < .001), and caregiver with severe psychological distress (p = .015) were independently associated with childhood malnutrition (Table 5).
4 |. DISCUSSION
Our findings indicate that several proximal, medial, and distal factors in a child’s ecosystem may impact the risk of childhood PTB and nutritional status. Proximal risk factors for childhood TB included sex and previous hospitalization. Consistent with this, young male South African children have been reported to be at higher risk for tuberculin skin test conversion, TB disease, and lower respiratory tract infection (LRTI), which may reflect a biological susceptibility due to hormonal or other factors.26 Medial risk factors included caregiver tobacco use. While caregiver tobacco use was not associated with childhood PTB beyond univariate analyses, caregiver tobacco use is a known risk factor for PTB. Tuberculosis and second-hand smoke exposure have a severe negative impact on the respiratory system.27 A meta-analysis conducted on second-hand smoke (SHS) exposure and TB risk has shown that children have more than a threefold increased risk of SHS-associated active TB compared to those not exposed to SHS. Similarly, caregiver psychological distress was not associated with childhood PTB. However, severe psychological distress among caregivers was common. This may reflect the burden of caring for a child with illness and environmental stressors inherent in living in poverty.1,2 The high number of caregivers surviving on a subsistence budget in our study may compound the elevated levels of psychological distress. This is consistent with the literature, which has found an association between caregiver psychological distress and socioeconomic status.7
Finally, we examined the distal factors linked to childhood TB. Caregivers with less than 10 years of schooling, a monthly income of less than $68 per month were distal determinants associated with childhood TB. As evidenced through the outcome of mycobacteriologically confirmed TB, limited monetary resources and educational attainment underscore the structural barriers some caregivers experience when caring for a child with TB.
Malnutrition is a known risk factor for childhood TB4; however, this link was only evidenced by our findings that low-weight or low-height for age were associated with childhood TB in the univariate analyses. This may also reflect that our control group had children sick with non-TB respiratory illnesses, for which malnutrition may also be a risk factor. Medial factors observed for the outcome of TB diagnostic category and TB treatment related to caregiver characteristics were also associated with childhood malnutrition at baseline. Caregiver smoking was an important risk factor for childhood malnutrition with a threefold increased risk for childhood malnutrition. Malnutrition was more than twice as likely to be present in children of caregivers who suffered from severe psychological distress. Given the nuanced links between childhood TB and proximal, medial, and distal risk factors in our data, child malnutrition appears to be a distal risk factor for childhood TB that needs further exploration. Additional studies using a child ecosystem approach are needed to confirm our findings in relation to child nutrition.
The proximal, medial, and distal factors identified in our ecosystem analysis help to provide an important context to the high prevalence of TB among children in this population. Promisingly, adherence in this study was extremely high. Despite the high adherence rates, rates of LTFU shed light on a subpopulation in our study that could benefit from family ecosystem-oriented interventions. Of the 6% of children that did not attend the 3-month visit, LTFU was associated with proximal, medial, and distal factors associated with childhood TB and childhood malnutrition that we identified using a childhood ecosystem framework. These findings indicate that caregivers with less schooling and suboptimal mental health may need additional support from healthcare systems to ensure that adherence for their child’s TB treatment stays high.
The high rates of psychological distress among caregivers are concerning. These data are consistent with reported rates of psychological distress in studies of adult TB patients in South Africa,8–10 and how this may relate to childhood TB, including adherence to follow-up visits. This is a research avenue that warrants further exploration.
Limitations include the questionnaires that were self-completed by caregivers. As such, the prevalence of smoking, alcohol use, or psychological distress may have been underestimated. That notwithstanding, a strong association was shown between these risk factors as organized through the childhood ecosystem approach and TB diagnosis, TB treatment, and childhood malnutrition. Second, only 69% of the caregivers of children consented to complete the questionnaires. Caregivers who elected not to participate in this study may have been less healthy (Table S2). Our reported prevalence rates represent the minimum prevalence for psychological distress or substance use. Reasons for not consenting were not recorded. No differences in age, level of schooling, or gender between those that consented and those that did not consent were found. We were not powered to investigate differences between ambulatory or hospitalized cases; however, categorization for TB disease, treatment, and follow-up was identical in both cohorts, enabling us to reliably pool data.
Despite these limitations, the findings emerging from this study provide a clearer picture of risks for childhood TB. For example, this study confirms poverty as a driver for TB. The results highlight the need for multisectoral interventions to reduce the burden of childhood TB, including job creation, education, and strengthened social support services. Specific, structural interventions to provide appropriate support protection strategies to reduce this burden include labor market support like job training or temporary subsidies to families in need. Additional research is required to explore the impact of the child ecosystem during transitional points in the development trajectory. For example, the role of family, peers, and individuals in the educational ecosystem may have varying points of prominence depending on what stage of development the child is in. Interventions should consider developmental milestones that may be leveraged for maximum intervention effect. Such studies may provide a more nuanced understanding and approach to risk reduction of transmission, diagnosis, or management of TB.
5 |. CONCLUSION
Our findings indicate the importance of proximal factors (e.g., male gender, sicker children e.g. those hospitalized), medial factors (e.g., caregiver tobacco use), and distal factors (e.g., educational attainment, economic stability, nutritional status) in childhood PTB. Our results underscore important entry points to leverage in future interventions. Current childhood TB interventions are primarily focused on the continuum of care in terms of prevention, adherence, and treatment. While these are critical components in achieving TB prevention, our findings suggest that other entry points include a more holistic family approach, such as the impact of distal factors on childhood TB. Future interventions may need to target subgroups of children and families who are at elevated risk for TB diagnosis and treatment. Within TB endemic settings, screening for risk factors, such as psychological distress or tobacco counseling among caregivers with children who are at high risk for TB or who have TB, as well as the provision of social protection programs to bolster economic security, may be critical in limiting children’s exposure to TB risk factors.
Supplementary Material
ACKNOWLEDGMENTS
We gratefully acknowledge the contributions of the NHLS diagnostic microbiology laboratory at Groote Schuur Hospital, the study laboratory and clinical staff at Red Cross Children’s Hospital and the Division of Medical Microbiology, and the children and their caregivers. The study was funded by the National Institute of Health, USA (1R01HD058971-01), the Medical Research Council of South Africa, the National Research Foundation (NRF) South Africa, and the European and Developing Countries Clinical Trials Partnership (EDCTP) (TB-NEAT; IP.2009.32040.009). Professor Dheda acknowledges funding from the SA MRC (RFA-EMU-02-2017), EDCTP (TMA-2015SF-1043, TMA-1051-TESAII, TMA-CDF2015), UK Medical Research Council (MR/S03563X/1), and the Wellcome Trust (MR/S027777/1). The writing of this publication was supported by the following National Institute on Drug Abuse (NIDA) predoctoral grant (F31DA049460).
Funding information
National Institutes of Health, Grant/Award Numbers: 1R01HD058971-01, F31DA049460; UK Medical Research Council, Grant/Award Number: MR/S03563X/1; South African Medical Research Council, Grant/Award Number: RFA-EMU-02-2017; Wellcome Trust, Grant/Award Number: MR/S027777/1; Medical Research Council of South Africa, the National Research Foundation (NRF) South Africa and the European and Developing Countries Clinical Trials Partnership (EDCTP), Grant/Award Number: TB-NEAT; IP.2009.32040.009; European and Developing Countries Clinical Trials Partnership, Grant/Award Numbers: TMA-2015SF-1043, TMA-1051-TESAII, TMA-CDF2015
Footnotes
ETHICS STATEMENT
The authors assert that all procedures contributing to this study comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Written informed consent was obtained from a parent or legal guardian and assent from children older than 7 years. The study was approved by the Human Research Ethics Committee of the Faculty of Health Sciences, University of Cape Town.
SUPPORTING INFORMATION
Additional Supporting Information may be found online in the supporting information tab for this article.
REFERENCES
- 1.World Health Organization. Global tuberculosis report 2020; 2019. https://apps.who.int/iris/bitstream/handle/10665/336069/9789240013131-eng.pdf.Accessed December 22, 2020.
- 2.Dodd PJ, Yuen CM, Sismanidis C, Seddon JA, Jenkins HE. The global burden of tuberculosis mortality in children: a mathematical modelling study. Lancet Glob Health. 2017;5(9):e898–e906. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lönnroth K, Glaziou P, Weil D, Floyd K, Uplekar M, Raviglione M. Beyond UHC: monitoring health and social protection coverage in the context of tuberculosis care and prevention. PLOS Med 2014; 11(9):e1001693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Graham SM, Sismanidis C, Menzies HJ, Marais BJ, Detjen AK, Black RE. Importance of tuberculosis control to address child survival. Lancet (London, England). 2014;383(9928):1605–1607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Egger G, Dixon J. Beyond obesity and lifestyle: a review of 21st century chronic disease determinants. BioMed Res Int 2014;2014:731685. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wamani H, Astrøm AN, Peterson S, Tumwine JK, Tylleskär T. Predictors of poor anthropometric status among children under 2 years of age in rural Uganda. Public Health Nutr 2006;9(3):320–326. [DOI] [PubMed] [Google Scholar]
- 7.Myer L, Stein DJ, Grimsrud A, Seedat S, Williams DR. Social determinants of psychological distress in a nationally–representative sample of South African adults. Soc Sci Med 2008;66(8):1828–1840. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Myer L, Ehrlich RI, Susser ES. Social epidemiology in South Africa. Epidemiol Rev 2004;26:112–123. [DOI] [PubMed] [Google Scholar]
- 9.Marais BJ. Childhood tuberculosis: epidemiology and natural history of disease. Indian J Pediatr 2011;78(3):321–327. [DOI] [PubMed] [Google Scholar]
- 10.Jaganath D, Mupere E. Childhood tuberculosis and malnutrition. J Infect Dis 2012;206(12):1809–1815. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Martin SJ, Sabina EP. Malnutrition and associated disorders in tuberculosis and its therapy. J Diet Suppl 2019;16(5):602–610. [DOI] [PubMed] [Google Scholar]
- 12.Zar HJ, Workman L, Isaacs W, et al. Rapid molecular diagnosis of pulmonary tuberculosis in children using nasopharyngeal specimens. Clin Infect Dis 2012;55(8):1088–1095. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zar HJ, Workman L, Isaacs W, Dheda K, Zemanay W, Nicol MP. Rapid diagnosis of pulmonary tuberculosis in African children in a primary care setting by use of Xpert MTB/RIF on respiratory specimens: a prospective study. Lancet. Glob Health. 2013;1(2):e97–e104. [DOI] [PubMed] [Google Scholar]
- 14.Graham SM, Cuevas LE, Jean–Philippe P, et al. Clinical case definitions for classification of intrathoracic tuberculosis in children: an update. Clin Infect Dis 2015;61(suppl 3):S179–S187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kessler RC, Andrews G, Colpe LJ, et al. Short screening scales to monitor population prevalences and trends in non–specific psychological distress. Psychol Med 2002;32(6):959–976. [DOI] [PubMed] [Google Scholar]
- 16.Andersen LS, Grimsrud A, Myer L, Williams DR, Stein DJ, Seedat S. The psychometric properties of the K10 and K6 scales in screening for mood and anxiety disorders in the South African stress and health study. Int J Methods Psychiatr Res 2011;20(4):215–223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24(4):385–396. [PubMed] [Google Scholar]
- 18.Zimet GD, Powell SS, Farley GK, Werkman S, Berkoff KA. Psychometric characteristics of the multidimensional scale of perceived social support. J Pers Assess. 1990;55(3–4):610–617. [DOI] [PubMed] [Google Scholar]
- 19.Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA. The AUDIT alcohol consumption questions (AUDIT–C): An effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Arch Intern Med 1998;158(16):1789–1795. [DOI] [PubMed] [Google Scholar]
- 20.Heatherton TF, Kozlowski LT, Frecker RC, Fagerström KO. The Fagerström test for nicotine dependence: a revision of the Fagerström tolerance questionnaire. Br J Addict 1991;86:1119–1127. [DOI] [PubMed] [Google Scholar]
- 21.Reinert DF, Allen JP. The alcohol use disorders identification test: an update of research findings. Alcohol Clin Exp Res 2007;31(2): 185–199. [DOI] [PubMed] [Google Scholar]
- 22.World Health Organization. WHO child growth standards, methods and development, Geneva, Switzerland; 2006. https://www.who.int/childgrowth/standards/Technical_report.pdf.Accessed November 11, 2020. [Google Scholar]
- 23.World Health Organization. WHO growth reference data for 5–19 years. Geneva, Switzerland; 2007. https://www.who.int/growthref/en/.Accessed November 11, 2020. [Google Scholar]
- 24.X–Rates Currency Calculator, 2020. https://www.x-rates.com/.Accessed November 11, 2020.
- 25.StataCorp. Stata statistical software: Release 15.1. StataCorp LLC. [Google Scholar]
- 26.Martinez L, leRoux DM, Barnett W, Stadler A, Nicol MP, Zar HJ. Tuberculin skin test conversion and primary progressive tuberculosis disease in the first 5 years of life: a birth cohort study from Cape Town, South Africa. Lancet Child Adolesc Health. 2018;2(1):46–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Patra J, Bhatia M, Suraweera W, et al. Exposure to second–hand smoke and the risk of tuberculosis in children and adults: a systematic review and meta–analysis of 18 observational studies. PLOS Med 2015;12(6):e1001835. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.