Statistics from Altmetric.com
- EPTB, extrapulmonary tuberculosis
- HIV, human immunodeficiency virus
- PEM, protein energy malnutrition
- PTB, pulmonary tuberculosis
- TB, tuberculosis
- TST, tuberculin skin test
Human immunodeficiency virus (HIV) infected children are at increased risk for tuberculosis (TB).1 Previous reports suggest that HIV infected children respond poorly to antituberculosis therapy compared with HIV uninfected children.2–5 We describe outcome during antituberculosis therapy and overall mortality in HIV infected children with culture confirmed TB in an area with endemic TB and HIV.
This retrospective cohort study assessed all children ⩽13 years of age with confirmed HIV infection and culture confirmed tuberculosis, hospitalised from January 1992 to December 2000 and from January 1998 to December 2000, respectively, at two referral centres in the Western Cape Province, South Africa. Antiretroviral therapy was unavailable in the public health sector.
TB disease was confirmed by isolation of Mycobacterium tuberculosis from any source in the presence of symptoms, and HIV infection by enzyme linked immunosorbent assay (ELISA) and polymerase chain reaction or detection of p24 antigen in children ⩽15 months of age. HIV disease was staged according to Centers for Disease Control criteria.6
TB treatment included six month regimens of three (isoniazid, rifampicin, and pyrazinamide) or four drugs (ethambutol or ethionamide added), according to national guidelines. On discharge, children completed clinic based directly observed therapy.
Clinical information was collected through patient chart review. Severe protein energy malnutrition (PEM) was defined according to the Wellcome classification (marasmus and/or kwashiorkor): a negative tuberculin skin test (TST) as a Mantoux test of ≤5 mm or Tine test with papules indurated ⩽2 mm. Chest radiographs were read according to a standardised method by two independent readers.7
Outcome, including cure and death during TB therapy, and total mortality at the end of the study period, were assessed. Cure was defined as culture negativity at therapy completion and/or radiological improvement of chest radiographic features. All children were followed through July 2002 or until lost to follow up, after deliberate attempts at tracing them at this point through local clinics, home visits or telephone calls. If lost to follow up, the date last seen was recorded.
Data were analysed using SPSS (version 11.5). Descriptive analyses for clinical characteristics were done for n = 93 episodes by HIV stage, and outcomes for n = 87 patients. Kaplan-Meier and Cox proportional hazard modelling compared long term survival by HIV stage; patients lost to follow up were censored. The study was approved by the ethics committee of the Faculty of Health Sciences, Stellenbosch University.
Eighty seven children were eligible for the study; all were included in analysis, representing 93 TB episodes: 2 children with HIV stage A, 30 stage B, and 55 stage C disease. Two children died before initiation of TB therapy. No children were lost to follow up during therapy. The median treatment duration was 6 months (mean 7.2 months; p = 0.352) and mean number of antituberculosis drugs 3.3; three children had undocumented adherence or defaulted on TB therapy.
Age at TB diagnosis ranged from 1 to 158 months (median 24); age at HIV diagnosis ranged from 1 to 98 months (median 17) (table 1). Drug resistance was detected in 9 (15.5%) of 58 TB episodes (62.3% tested); 5 (8.6%) had multidrug resistance. A TST, performed and read in 78 episodes (83.9%), was positive in 44 (56.4%).
The dominant findings on chest radiography were mediastinal adenopathy in 56 episodes (60.2%), and lobar/segmental opacification in 51 (54.8%). There was severe pulmonary disease, defined as ⩾1 of the following: miliary disease, cavities, bronchopneumonic opacification, or lobar disease involving ⩾2 lobes, in 55 episodes (59.2%, p = 0.169). CD4+ T lymphocyte count at diagnosis ⩽15% was significantly associated with miliary disease (p = 0.05); however, CD4+ T lymphocyte count data was incomplete (missing data in 45.2%).
Outcome during antituberculosis therapy
Bacteriologic cure was confirmed in 44/56 episodes (78.6%) where cultures were repeated and in 44/93 total episodes (47.3%) (table 2). There was radiological improvement of TB features in 30 (32.3%), partial or no improvement in 40 (43%), and missing data in 23 (24.7%) episodes. There was cure based on a combination of bacteriological and/or radiological criteria in 54 episodes (58.1%; missing data in 13; 14%). Mycobacterium tuberculosis was isolated in 12 episodes (13%) ⩾6 months after treatment initiation: in eight episodes with HIV stage B and in four with stage C disease (p = 0.565). These children were all treatment adherent; none showed chest radiographic improvement at ⩾6 months after initiation of therapy and none had drug resistance.
There were 18 deaths during TB therapy: four in children with HIV stage B and 14 with stage C disease (p = 0.107); median duration until death 116 days. Four deaths (22.2%) occurred within two months following treatment initiation. Two children died before therapy was initiated, one due to Pneumocystic jivoveci pneumonia (PCP) and the other to bacterial septicaemia. In univariate analysis, age ⩽1 year (p = 0.002), severe PEM at diagnosis (p = 0.009), and a negative TST (p = 0.009) were significantly associated with mortality during therapy.
Outcome after completion of antituberculosis therapy
Sixteen deaths occurred after antituberculosis therapy completion. There were 34 (39.1%) total deaths: 10 in children with HIV stage B and 24 in stage C disease (p = 0.181). Thirteen children (14.9%) were lost to follow up after treatment completion. TB accounted for 17.6% of deaths, including multidrug resistant disease (3; 8.8%), TB meningitis (2; 5.8%) and miliary disease (1; 2.9%). Non-TB deaths were attributed to bacterial pneumonia and/or septicaemia in 13 (38.2%), severe AIDS wasting in two (5.9%), severe gastroenteritis in two (5.9%), PCP in two (5.9%), HIV encephalopathy in one (2.9%), disseminated varicella in one (2.9%), disseminated BCG disease in one (3.6%), and undocumented causes in six (17.6%).
Kaplan-Meier analysis of all deaths over total follow up in days (range 6–3730, median 3602 days) showed no significant differences between HIV disease stages (−2 log likelihood ratio, p = 0.1327) (see fig 1). In multivariate Cox survival analysis, HIV disease category, severe PEM at diagnosis, and lack of cure at end of TB therapy were significantly associated with overall mortality (table 3).
Fourteen children (16.1%) were retreated after treatment completion; in six (42.9%) there was bacterial confirmation by culture.
The most important findings of this study are the poor outcomes of HIV infected children with culture confirmed TB on antituberculosis therapy in the absence of antiretroviral therapy and the high mortality during antituberculosis therapy and at follow up. Most deaths were not directly tuberculosis related but due to acute opportunistic infections. Children with advanced HIV disease, severe malnutrition, and incomplete cure at the end of TB therapy were at greatest risk. There was a higher proportion of extrapulmonary TB noted in children with more advanced HIV disease; however cause and effect are difficult to distinguish.
What is already known on this topic
HIV infected children have a poor response to standard antituberculosis therapy
HIV infected children with culture confirmed tuberculosis have a high mortality
What this study adds
HIV infected children with advanced HIV disease, severe malnutrition, and incomplete cure at the end of antituberculosis therapy are at greatest risk of death; death is mainly due to acute opportunistic infections
HIV infected children with culture confirmed tuberculosis have a high risk of recurrent tuberculosis
The confirmation of cure in childhood paucibacillary disease is difficult without bacteriologic proof. In children where repeat cultures were not available at the end of therapy, we relied on improvement of chest radiographic features. The high proportion where no substantial improvement was noted may indicate inadequate treatment response, and/or the co-prevalence of other chronic lung disease. However, in children where failure of bacteriological cure was documented, chest radiographic lesions indicative of TB had similarly not improved. The high proportion of children with persistent positive cultures six months or more after initiation of therapy while adherent, suggests suboptimal treatment response.
This study has limitations: the confounding effects of HIV on TB presentation and TST, the limitations of chest radiograph interpretation in HIV infected children, the study’s retrospective nature, incomplete data on CD4+ T lymphocyte counts, the lack of postmortem examination data, and possible inadequate control for adherence to antituberculosis therapy. However, in this large cohort, despite limited resources, diagnosis was confirmed by culture in all children, and follow up was achieved with chest radiograph and/or culture in the majority.
In summary, there is an urgent need to assess therapeutic and preventative strategies, including antiretroviral and preventative therapy aimed at TB and other opportunistic infections, to improve treatment outcome, and reduce mortality in this vulnerable group.
We acknowledge the Department of Science and Technology through Technology and Human Resources for Industry Programme, South Africa (THRIP) for funding to ACH, and the Nestle Community Research Fund for financial support.
Published Online First 17 June 2005
Competing interests: none
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.