THE LANCET • Vol 355 • January 29, 2000 369 Summary Background Necropsy studies from Africa have shown that Pneumocystis carinii pneumonia (PCP) is common in infants with HIV infection. We aimed to describe the rate, clinical presentation, and outcome of PCP in young Malawian children with acute severe pneumonia. Methods Children aged between 2 months and 5 years who were in hospital with a diagnosis of severe pneumonia were admitted to a study ward for clinical monitoring. We carried out blood culture, immunofluorescence on nasopharyngeal aspirate samples to test for PCP, polymerase chain reaction to detect HIV, and chest radiography. Findings 16 cases of PCP were identified among 150 children with radiologically confirmed severe pneumonia. All were HIV- positive and younger than 6 months. 21 children had bacterial pneumonia (including one who was also PCP positive) and 114 were not confirmed. The most common bacterial pathogens among children without PCP were Streptococcus pneumoniae (eight) and non-typhoidal salmonellae (seven). On admission, children with confirmed PCP had a lower mean age, body temperature, and oxygen saturation than children with bacterial pneumonia and were less likely to have a focal abnormality on auscultation. Oxygen requirements were much greater in children with PCP than those with bacterial pneumonias (96 of 105 hospital days vs 15 of 94, p<0·0001). Ten of 16 children with PCP and six of 21 with bacterial pneumonia died (relative risk 2·19 [95% CI 1·0–4·7]). The overall case-fatality rate of severe pneumonia was 22%. In addition to a strong association with PCP, a fatal outcome was significantly and independently associated with HIV infection (2·98 [1·1–7·9]) and with age under 6 months (2·76 [1·0–5·2]). Interpretation PCP is common and contributes to the high mortality from pneumonia in Malawian infants. Clinical features are helpful in diagnosis. The study highlights the impact of HIV infection and difficult issues of management in countries with few resources. Lancet 2000; 355: 369–73 Introduction Studies of vertical HIV transmission in many regions of Africa, including Malawi, show that childhood HIV infection is common. 1 Necropsy studies from the region have consistently shown that Pneumocystis carinii is common in children with HIV infection dying of pneumonia. 2–4 Though P carinii pneumonia (PCP) is commonly recognised clinically, a lack of diagnostic tests has meant that the number of cases of PCP among children presenting with acute pneumonia is unknown. A pilot study in Malawi diagnosed five cases of PCP by means of an immunofluorescence method that identified cysts on nasopharyngeal aspirate samples. 5 We did a prospective study of the causes and outcome of acute severe pneumonia in Malawian children. Methods Patients We did a prospective study at Queen Elizabeth Central Hospital, Blantyre, Malawi, during a 6-month period (July to December, 1996) when the rate of severe malaria is low. Clinical data were collected on all children aged between 2 months and 5 years with an admission diagnosis of severe pneumonia. This hospital serves Malawi’s largest city with a population of about 1 million. Children admitted with acute pneumonia have either presented directly to the hospital’s under-5 clinic or have been referred to the clinic from an urban or periurban health centre. Severe pneumonia was defined according to WHO criteria—cough or a difficulty breathing with chest indrawing. 6 Assessment on admission included a thick blood film to detect malaria parasites and measurement of packed-cell volume. Children from this group with severe pneumonia were admitted to the study ward, if beds were available, by one of the clinicians (SG, EM, MM). Children with a diagnosis on admission of severe pneumonia, who were not admitted to the study ward, were followed up to assess clinical course and outcome. Care was taken to avoid misdiagnoses due to other disorders such as severe malaria that also present with cough and respiratory distress. 7,8 Children with measles were followed up but not included in the study because they must be cared for in isolation. Children with severe malnutrition (eg, marasmus or kwashiorkor) were not included in our study because they must be transferred to the nutritional rehabilitation unit. Investigations The clinician enrolling patients collected clinical and demo- graphic data. Investigations included blood culture, immunofluorescence of nasopharyngeal aspirate samples, and chest radiography. Arterial oxygen saturation (SaO 2 ) was measured with pulse oximetry (Ohmeda, Louisville, KT, USA) on admission first with the child breathing air and then in oxygen if the child was hypoxic (SaO 2 <90%). Because most patients were infants, HIV infection was sought by nested PCR on DNA extracted from blood clots (with proteinase K and phenol-chloroform). Part of the HIV LTR gene was amplified with a primer (Neil Berry, University College London, London, UK) designed for detection of African HIV strains. The outer primer pair was Tar-3 (5'-ACCAGRTYTGAGCCTGGGA GCT) and N2607 (5'-CCTGTTCGGGCGCCACTGC TAGAGCTTTT) and the inner primer pair was prime Tar-2 (5'-TGAGCCTGGGAGCTCTCTGGCT) and N2609 (5'- CTGAGGGATCTCTAGDYACCAGAGT). Clinical presentation and outcome of Pneumocystis carinii pneumonia in Malawian children Stephen M Graham, Edward I Mtitimila, Henry S Kamanga, Amanda L Walsh, C Anthony Hart, Malcolm E Molyneux Department of Paediatrics (S M Graham FRACP, E I Mtitimila MB, H S Kamanga Dip Lab) and Wellcome Trust Research Laboratories (A L Walsh MIBiol, Prof M E Molyneux FRCP), College of Medicine, University of Malawi, Blantyre, Malawi; School of Tropical Medicine, Liverpool, UK (A L Walsh, Prof M E Molyneux); and Department of Medical Microbiology, University of Liverpool, Liverpool, UK (Prof C A Hart FRCPath) Correspondence to: Dr Stephen M Graham, Department of Medical Microbiology, University of Liverpool, 8th Floor, Duncan Building, Liverpool L69 3GA, UK (e-mail: smgraham@liverpool.ac.uk)