Weight, haemoglobin, malnutrition predict risk of death in Zambian children without ART

A low weight-for-age, low haemoglobin and malnutrition are all strong predictors of poor survival in HIV-positive Zambian children not receiving antiretroviral therapy, and may prove useful markers to enable more intensive monitoring or priority antiretroviral treatment (ART) in settings where CD4 cell counts are not available, Zambian and British researchers report in the August 15^{th/sup> edition of the Journal of Acquired Immune Deficiency Syndromes.}

Around two million children are living with HIV or AIDS in sub-Saharan Africa. Despite national and international efforts to increase access to ART only a small proportion of these children are on antiretroviral drugs. In some rural settings, life-saving cotrimoxazole prophylaxis (CTX) might not be available to these patients for various logistic reasons.

Children who are not on ART and might not be on CTX are prey to a legion of opportunistic infections. Providing these children with quality healthcare including palliative treatment requires knowledge about predictors of mortality. However, high-tech laboratory equipment such as flow cytometers for CD4 cell counts are largely unavailable outside major hospitals in most countries. They are too expensive, require expensive reagents, and need highly-trained personnel.

Alternative cheaper and more sustainable low-tech laboratory assays and clinical guidelines for following up CLWA under resource-poor settings are required. Simple clinical or laboratory predictors, such as total lymphocyte counts (TLC), haemoglobin and/or albumin levels, and weight-for-age that could be used in rural and semi-urban settings need to be validated.

A team of Zambian and British (MRC Clinical Trials Unit) investigators addressed this issue using data from the CHAP study, a randomised study of cotrimoxazole prophylaxis in HIV-positive children aged one year to 14 years.

The study took place at the University Teaching Hospital (UTH) in Lusaka, Zambia between March 2001 and January 2003. Five hundred and fourteen children (median 5.5 years of age) were randomised to receive daily CTX or matching placebo. At baseline (trial entry), demographic, clinical, and laboratory data as well as the history of prior hospitalisations were obtained.

The demographic information obtained included age, sex, and primary carer. The comprehensive clinical data set collected included weight, height, and prior and current signs and symptoms of a range of diseases including oral candidiasis, malnutrition, diarrhoea, severe bacterial infections, fever, tuberculosis, malaria, etc. Laboratory data included haemoglobin and T-cell subsets. Clinical malnutrition was assessed using the WHO/Wellcome classification.

The children were followed up and assessed 4-weekly for 16 weeks and every eight weeks thereafter. A total of 164 children died. The usefulness of the baseline data as predictors of survival was investigated using a statistical model for survival.

Poor survival was predicted by low weight for age, CD4 percentage, haemoglobin, mother as primary carer, current malnutrition, oral candidiasis, previous hospital admissions for respiratory tract infections, or recurrent bacterial infections. The predictive value of oral candidiasis disappeared after correcting for CD4 percentage. This observation underscores the fact that oral candidiasis is a clinical proxy for low CD4 percentage.

Nutrition is of paramount importance in the health care of children with HIV and this study provided additional insights. First, prior clinical malnutrition with low current weight-for-age had a worse prognosis than weight-for-age alone. Second, malnutrition predicted mortality independently of CD4 percentage. Finally, children whose malnutrition had been successfully treated during a previous hospital admission were not at a higher risk.

Weight-for-age was a better predictor than height-for-age or weight-for-height. This is significant because weight is easier to measure and has been correlated with virologic response in HIV-positive children in Europe and the USA.

Of the two low-tech laboratory assays, haemoglobin was a better a predictor of mortality even after adjusting for CD4 percentage. By contrast, total lymphocyte count (TLC) was confounded by CD4 and, contrary to earlier studies, was a less reliable predictor of mortality. A published study previously demonstrated that an algorithm combining TLC, hemoglobin, and albumin was useful for clinical decision-making for providing health care to HIV/AIDS patients.

Such algorithms would be invaluable in resource poor settings in Zambia and other African countries. The authors point out the need for further studies to investigate the usefulness of both TLC and haemoglobin for patient follow up.