Tuberculosis and HIV interaction in Ethiopian children : aspects on epidemiology, diagnosis and clinical management

Abstract

This thesis investigates the influence of HIV on childhood tuberculosis (TB) in a high TB and HIV endemic setting. It is based on four studies involving the same study population of 522 tuberculous children and their 25 3 non-tuberculous controls, consecutively enrolled as out- and inpatients at the major paediatric hospital in Addis Ababa, Ethiopia.

The first study identifies HIV as a major risk factor for TB. HIV prevalence among tuberculous children in Addis Ababa was 12.5%, compared with 1.6% among the controls (AOR 12.7; 95% Cl 2.9,55).

The second study explores possible background features associated with dual infection. These were young age, high educational level of mothers, having lost one or both parents and being BCG vaccinated. The findings have implications for the National Tuberculosis Control Programme, suggesting higher demands on hospital beds for the youngest, active case- finding in connection to adult smear positive cases, and preventive strategies beyond BCG vaccination, since the protective effect against TB is lost among HIV-positive children.

Childhood TB is caused by recent transmission. Drug susceptibility testing and molecular typing of M. tuberculosis isolates from paediatric patients may thus add to the surveillance of adult strains of M. tuberculosis, since the latter tend to reflect a combination of newly acquired and older reactivated TB. In the third study, a positive culture of Mycobacterium tuberculosis was isolated from 191 out of 362 (53%) children with pulmonary manifestations. HIV infection was negatively associated with a positive culture (OR 0.3; 95% Cl 0.2, 0.6). Drug sensitivity testing was done in 167 and molecular strain typing, using the IS6110 (RFLP) and direct repeat (spoligotyping) genetic markers, in 163 of the isolates of Mycobacterium tuberculosis. Resistance to any of the standard anti-TB drugs was found in 13% of the isolates, isoniazid resistance (11 %) being the most important. No resistance to rifampicin was found. Two clusters included 29% of the bacterial isolates by RFLP. Spoligotyping reduced this figure to 26%. HIV was associated neither with drug resistance nor with any specific cluster within this sample frame.

The fourth study presents results from the diagnosis and follow-up of the same 517 tuberculous children included in the second study. HIV-positive patients were more symptomatic and the very young severely underweight compared with the HIV-negative TB patients. The tuberculin test was less sensitive and chest radiography less specific in dually infected children. The diagnostic delay, counted as number of visits to a health institution before the diagnosis of TB, was also higher in this group (OR 2.1; Cl 1.2, 3.7).

The children were followed until 6 months after treatment completion. Adherence to treatment was high (96%), only few side effects of TB drugs were noted and the cure rate was 58% for HIV-positive and 89% for HIV-negative patients. A fatal outcome was found in 40% of the HIVpositive compared to 7% in the HIV-negative TB patients (AOR 6.0; 95% Cl 3.0, 11.4) and HIV was the only identified predictor for death. Within the HIV-positive group, a low weight-for- age could be used to identify children at highest risk of a fatal outcome.

The need for routine HIV screening, including counselling, of all TB patients is essential in order to diminish the risk of over- as well as under-diagnosis of dually infected children and to secure an adequate, safe treatment of TB as well as of other opportunistic infections. Possibly, such measures might reduce the high mortality among HIV-positive tuberculous children encountered in this study.