This was a prospective cross-sectional community-based study. Study participants were recruited from three high burden tuberculosis communities: Ravensmead, Uitsig (R/U) and Site C, Khayelitsha (Site C), in Cape Town, Western Cape Province, South Africa. The communities are demographically heterogeneous, and respectively represent ethnic groups that are predominantly of South African Mixed (R/U) and Xhosa African ancestry (Site C).
The tuberculosis notification rate in the province was 994 per 100 000 in adults and 671 per 100 000 in children aged 0 – 14
years in 2008 (Western Cape Department of Health, unpublished data). There is near universal infant bacille Calmette-Guérin (BCG) vaccination.
Prior surveys from school-going children in R/U found that 25% were stool positive for Ascaris lumbricoides
and 51% for Trichuris trichiura
]. Routine clinic and school-based deworming has since become standard of care in local health care centres. There are currently no published data on the prevalence of helminth infection in children in Khayelitsha.
Participants were recruited following identification of consenting adults (≥ 18
years of age), who routinely started tuberculosis treatment in the preceding 3
months (“source case”) at the local clinic. Participants were eligible for enrolment if they were a household child contact between 3
months and 15
years of age. Households were identified through TB source cases (“tuberculosis households”), and matched neighbouring households without known tuberculosis exposure (“neighbouring households”). The ratio between tuberculosis and neighbouring households was approximately 1:1.
We identified tuberculosis households within 2–4
weeks following the adult source case identification, then completed a home visit and collected data on M.tb
exposure and demographic characteristics. In addition, all adult household members were screened for symptoms of tuberculosis followed by sputum collection, if they were symptomatic.
Following written informed consent from the parents/caregivers, children aged 3
months – 15
years who were not currently on tuberculosis therapy, were enrolled in the study. Children were deferred for enrolment if there was a) prior TST administration in the preceding 3
months, b) live measles or polio vaccine within the past 6
weeks or c) recent acute illness requiring hospitalization (preceding 6
weeks). Children on isoniazid preventive therapy (IPT) were not excluded, whereas HIV-infected children were.
All children were investigated for M.tb
infection and disease using standard protocols [27
], including TST, a standard symptom questionnaire , blinded independent dual expert review of chest radiographs (postero-anterior and lateral), sputum or gastric washing for mycobacterial culture (Mycobacterial Growth Indicator Tubes; Becton Dickinson, Sparks, USA), a TST, and clinical examination.
The TST (Mantoux method; 2 tuberculin units purified protein derivative; Statens Serum Institute, Copenhagen) was administered intra-dermally on the volar aspect of the left forearm by trained standardized study nurses, and read within 48–72
hours using the ball-point and ruler method and callipers. The TST was regarded as positive if measuring ≥10
mm in HIV-negative children for clinical care. Unless already known to be HIV-infected, participants were tested for HIV (Abbott Determine Rapid HIV test), followed by a confirmatory laboratory based ELISA/PCR test if positive or indeterminate.
Data were collected to quantify the degree of M.tb
contact score) [29
]. In the absence of a gold standard of infection, we used this score to quantify the degree of exposure in children as a surrogate reference standard for M.tb
. The score includes measures of source case infectivity, the average reported duration of exposure to the source case, and the child’s reported proximity to the source case. TST results in children from the study communities were previously shown to be well correlated with this score [29
Table 1 Paediatric Mycobacterium tuberculosis contact score incorporating measures of infectivity, proximity and infectivity of documented exposure []
We also recorded information on BCG vaccination history and scar presence, previous and current isoniazid preventive therapy, previous tuberculosis treatment, nutritional status (z-scores), household composition, measures of socio-economic status (household assets), and reported history of deworming during the preceding 6
months. Z-scores were calculated with reference to the 2006 World Health Organization Child Growth Standards for children up to five years of age and the 2000 Centres for Disease Control and Prevention growth charts reference population for children that were older than 5
]. Reported socio-economic indicators were based on surveys in our study communities and were used in the absence of reliable data on reported income, given high unemployment rates in these impoverished urban communities.
Sensitization to Ascaris lumbricoides
was determined by the ImmunoCAP specific-IgE testing which measures circulating Ascaris-
specific IgE concentration in serum. This method was developed for clinical diagnosis and laboratory use where reagins belonging to the IgE antibody class, which develop shortly after exposure, are measured by a fluoroenzymeimmunoassay [33
]. Results are expressed as kU/L and are typically categorised in 6 groups (Table
) or interpreted as positive or negative based on a binary classification (25).
Data analysis was performed using STATA/SE version 12.0 (StataCorp LP, StataCorp, 4905 Lakeway Drive, College Station, Texas 77845 USA). An alpha level of less than or equal to 0.05 was regarded as statistically significant. The main outcome of interest was M.tb
infection, measured by a positive TST (≥10
mm). The main exposure of interest, Ascaris
IgE, was categorised according to RAST categories and dichotomized (positive if ≥0.35kU/L). This measure, consistent with previous work (25) was used as a measure of Ascaris
infection or sensitization.
We assessed the effect of predictors of M.tb infection in univariate analysis, with the primary determinant being Ascaris IgE status in 271 enrolled participants. Child baseline characteristics were described overall, and stratified according to Ascaris IgE status for baseline characteristics. Continuous data were described using mean and standard deviations and median and inter-quartile ranges, dependent on data distribution. Categorical variables were described using frequencies and percentages.
We developed logistic regression models that included Ascaris IgE status, age and the M.tb contact score. These measures were significant during univariate analysis or were clinically or epidemiologically relevant as covariates. Additionally, the developed model was examined for interactions between the explanatory and environmental variables in order to establish if possible effect modification is present. Results of the regression model are presented as adjusted odds ratios (aORs) and corresponding 95 percent confidence intervals (95% CIs).
We complied with the Helsinki Declaration and the study was approved by the Health Research Ethics Committee of the Stellenbosch University (reference number N05/07/129 and N08/08/207).