University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Effect of age, polymicrobial disease, and maternal HIV status on treatment response and cause of severe pneumonia in South African children: a prospective descriptive study
Global Health Sciences Literature Digest
Published June 4, 2007
Journal Article

McNally LM, Jeena PM, Gajee K, Thula SA, Sturm AW, Cassol S, Tomkins AM, Coovadia HM, Goldblatt D.Effect of age, polymicrobial disease, and maternal HIV status on treatment response and cause of severe pneumonia in South African children: a prospective descriptive study. Lancet 2007 Apr 28;369(9571):1440-51.


To measure predictors of treatment failure and the cause of non-responsive pneumonia in children admitted to hospital with severe pneumonia in Durban, South Africa.

Study Design

Prospective, descriptive study.


King Edward Hospital in Durban, KwaZulu-Natal, South Africa, a large hospital providing all levels of pediatric care. In 2002, the HIV prevalence among women in antenatal clinics in KZN was 36.5%, and antiretrovirals were not available. Haemophilus influenzae type b vaccination was introduced in South Africa in 1999.


358 children aged 1 month to <5 years and admitted to hospital with WHO-defined severe or very severe pneumonia from 2000-2002 were enrolled; 254 (71%) had WHO-defined very severe disease. Children who had been admitted in the preceding two weeks, had a chronic condition predisposing to pneumonia, a known allergy to study antibiotics, comorbid disease requiring different antibiotics, or wheezing responding to salbutamol, were excluded (n=31). Median age was 4.8 months (IQR 2.7-13.0).


A complete physical and laboratory evaluation was performed, with subsequent institution of WHO guidelines for treatment.

Evaluation: Lab investigations included a full blood count, blood culture, urine culture, and HIV testing. Evaluation of pneumonia included a chest radiograph, nasopharyngeal aspirate for viral culture, bronchio-alveolar lavage (NB-BAL), lung aspirates, induced sputum for Pneumocystis jirovecii pneumonia and tuberculosis (TB), and three gastric washings for TB. Pleural effusions were tapped and sent for microscopy and culture.

HIV testing: All children were initially tested by HIV EIA, followed by HIV viral load if antibody positive (including RNA and DNA PCR). Infants <1 yr old were classified as: 1. HIV-infected (EIA positive, RNA positive); 2. HIV-exposed, uninfected (EIA positive, RNA, and DNA negative 3. HIV-uninfected (EIA negative).

Treatment: The 1992 WHO guidelines for treatment of pneumonia were adapted. All children received high-dose benzylpenicillin and gentamicin irrespective of age; infants also received high-dose trimethoprim-sulfamethoxazole because of the increased risk of P. jirovecii. Intravenous antibiotics were given for a minimum of 48 hours and changed to oral medications as long as the child was not hypoxic (O2 saturation <92% on room air).

Primary Outcomes

The primary endpoint was treatment failure at 48 hours. Treatment failure was defined as: appearance of danger signs, low oxygen saturation, persistence of lower chest in-drawing, life-threatening or serious adverse drug reaction, received another antibiotic, newly diagnosed comorbid condition, parents or guardian withdrew consent for treatment, child left against medical advice, or death. Secondary endpoints were cumulative treatment failure at any time during hospital admission, or death. Only children with culture-proven M. tuberculosis were defined as having tuberculosis. A positive test for P. jirovecii was defined as the presence of five or more cysts by direct immunofluorescence.


Two hundred forty-two (68%) children were HIV-infected, 41 (11%) were HIV-exposed, uninfected, and 75 (21%) were HIV-uninfected. One hundred twenty-six (35%) children failed treatment by 48 hours, and 29 (8%) failed subsequently, mostly because of nosocomial or comorbid infections. Fifty-four (15%) children died during admission, 13 within 48 hours. Median hospital stay was eight days (IQR 4–13) and median duration of oxygen requirement was four days (IQR 2–8).

Failure to respond by 48 hours was independently predicted by the following: age <1 year (aOR =6.4, 95% CI 2.7–14.9), very severe disease (aOR= 2.5, 1.217–5.4), HIV status (HIV-infected 10.3, 3.3–32.1; HIV-exposed, uninfected 6.0, 1.6–2.4), and polymicrobial disease (two organisms, aOR=10.8, 4.4–26.4; p<0.0001). Among children ≥1 yr, there was no difference in outcome between HIV-infected and HIV-uninfected children. 106 (35%) children had no clinically significant organism identified, 141 (46%) had one, 53 (17%) had two, and four (1%) had three. All children with three organisms failed treatment. The prevalence of P. jirovecii among those who failed study therapy was 3/9 (33%) among HIV-exposed, uninfected infants and 29/74 (39%) HIV-infected infants, and none among HIV-uninfected infants. Streptococcus pneumoniae was the most common organism identified from admission blood culture.


The authors concluded that children who were HIV-infected, or who were HIV-exposed but uninfected, had worse outcomes than those who were HIV-uninfected. The reasons for poor outcome in uninfected, exposed children include reduced transplacental transfer of protective antibodies from HIV-infected mothers to their children, and altered CD4 immunity during the first year of life. These infants may also have had increased exposure to pathogenic organisms from their immunosuppressed mothers.

The authors state that in children >1 year old, the WHO guidelines are effective, irrespective of HIV-status. However, for those <1 year of age, the guidelines are inadequate, since 48% of infants failed therapy. Polymicrobial disease is a key predictor of treatment failure.

Quality Rating

There is no quality rating system for descriptive prospective studies. The high rates of HIV infection and very severe disease in children in this study are probably not representative of children routinely admitted to secondary level healthcare facilities.

In Context

Acute respiratory infections are the leading cause of childhood mortality worldwide. WHO’s guidelines for the management of acute respiratory infections in children younger than five years are based largely on studies published before the development of the HIV pandemic. Currently between 11%-45% of children admitted to hospital with severe pneumonia in southern African countries are co-infected with HIV, and HIV-related pneumonias are the leading cause of these admissions. In 2003 a WHO consultation recommended changing the guidelines to include provision of benzathine penicillin (or ampicillin) plus gentamicin in severe pneumonia, plus trimethoprim-sulfamethoxazole for infants <1 year or age; if there is inadequate treatment response by 48 hours, antibiotics are to be switched.

Programmatic Implications

The existing WHO guidelines (1) for treatment of pneumonia in young children, dating from 1990, and based on studies undertaken mostly in the pre-HIV era, need to be revised to reflect the high rates of HIV exposure and infection in southern Africa. The widespread rollout of prevention of mother-to-child transmission strategies and use of antiretrovirals will reduce the number of HIV-infected children admitted to hospital with pneumonia. However, the risk of pneumonia in HIV-exposed, uninfected infants will continue for some time. Because polymicrobial disease is a key predictor of treatment failure, there is a need for rapid low-cost diagnostic methods to assist clinicians identifying organisms and providing appropriate treatment. Further studies are urgently needed to address these issues.


  1. World Health Organization. WHO programme for the control of acute respiratory infections. Acute respiratory infections in children: case management in small hospitals in developing countries. Report no 5. Geneva: World Health Organization, 1990.