Tuberculosis in HIV-positive children often missed from incidence data: the influence of HAART on paediatric TB
12 May 2006. Related: Conference reports, Paediatric care, TB coinfection, CROI 13 (Retrovirus) 2006.
Polly Clayden, HIV i-Base
In the developing world, tuberculosis (TB) is a common opportunistic infection and cause of death in HIV positive children.
Neil Martinson from University of Witwatersrand presented findings from a study to compare the incidence of TB in HIV-positive children treated and not treated with HAART. He explained that in South Africa it is estimated that children aged 0-14 years make up 15% of the total TB caseload in high prevalence HIV/TB settings. Most are smear negative and therefore go uncounted by National Programmes or the WHO, so incidence among children appears low. However, higher incidence has been observed among studies of hospitalised children and autopsy studies.
In South Africa 10% of people receiving antiretrovirals are children. Adult cohort studies report 50 to 80% reductions in TB incidence with HAART treatment but equivalent data for children is scarce, so it is unclear whether or not this effect is sustained in children. It is also difficult to make TB diagnoses.
This study retrospectively reviewed notes from four South African antiretroviral clinics; three in Johannesburg and one in Cape Town. Dr Martinson noted that in parts of the Western Cape TB incidence has been reported as 1%.
Children <15 years with complete notes (recorded clinic visits at least once every 6 months), were eligible for the study. Follow-up was recorded from the first CD4 percentage until the last clinic visit.
Records were checked for a TB diagnosis, the basis for the diagnosis, and date(s) and duration of treatment. Children on TB treatment were temporally censored. If children received HAART, the date of initiation and duration of treatment was recorded.
There were a total 992 childrens records in the database (all children in the analysis) of these 719 had time off and on HAART and 262 had TB treatment for >1 month.
Baseline characteristics at study entry were: median age 6.8 years and median CD4 percentage 16.7 (10.0-24.4%). The girl to boy ratio was 0.98. 15.8% of children had TB at first visit and 39.8% had a history of TB prior to entering the study. The median length of follow up was 9 months for children not receiving HAART and 11.5 for children receiving HAART.
The investigators reported an incidence of TB per 100 child years among all children of 16.3 (14.1-18.8), and 6.3 (4.8-8.1) for those receiving HAART, representing a crude incidence rate ratio (IRR) 0.39 (95% CI: 0.3-0.5). Evaluating the children who had time on- and off-HAART they reported an incidence of 21.6 (18.5- 25.2) per 100 child years before they received HAART compared to 6.3 (4.9-8.2) after, (crude IRR 0.29 (95% CI: 0.2-0.4).
Overall few cases were confirmed (ie smear, culture or biopsy positive). However 34% of TB cases were confirmed in children receiving HAART, compared to 18% not receiving HAART. Dr Martinson suggested that this reflected the degree to which doctors were willing to investigate children not receiving HAART. The incidence of confirmed TB was 2.9 (2.0-4.0) per 100 child years among children not receiving HAART compared to 2.3 (1.3-3.2) per 100 child years in children receiving HAART (crude IRR 0.79 (95% CI: 0.4-1.3). For the children receiving HAART on and off, the incidence of confirmed TB was 3.7 (2.5-4.2) per 100 child years compared to 2.2 (1.4-2.9), (crude IRR 0.59 (95% CI: 0.3-1.09).
He noted that in all cases there was a marked reduction in TB prior to receiving HAART and that in his discussions with doctors they appeared to be more willing to give children TB treatment alone than with antiretrovirals.
In an analysis stratifying the risk by CD4 and viral load (see table 1) the rate reduction was sustained across all groups. He reported a possible dose response effect with viral load. He added that at low viral loads (<400 copies/mL) the TB incidence rate was very low, only 1.76 per 100 child years.
Table 1: Risk by CD4% and viral load
| CD4% | IRR |
|---|---|
| <15% | 0.35 (0.2-0.7) |
| 15-25% | 0.35 (0.1-1.2) |
| >25% | 0.23 (0.1-0.9) |
| Viral load (copies/mL) | |
| <400 | not applicable |
| 400-10,000 | 0.18 (0.03-1.1) |
| 10,000-100,000 | 0.31 (0.03-1.6) |
| >25% | 0.48 (0.2-1.2) |
Dr Martinson concluded that there are high rates of TB treatment in HIV positive children; that HAART reduces the incidence of TB treatment by 60% in children (and by 40% with confirmed TB); that few episodes of TB are confirmed microbiologically and the basis of decision to treat TB is influenced by HAART status. The investigators wrote: Overall, HAART protects HIV-infected children from TB but to a lesser extent than in adults.
However the study had a number of limitations compared to a clinical trial in that they did not use classic definitions of TB, there was no data on mortality, there was a bias towards older children (Dr Martinson asked what is happening to younger kids?), and that the review only included children that were hospital based. The results however were consistent with other previous studies of HAART and TB in children from Abidjan and Haiti.
Reference:
Martinson N, Moultrieh H, G Barry G et al. Incidence of tuberculosis in HIV-infected children: The influence of HAART. 13th CROI, Denver, 2006. Abstract 22.