TRANSACTIONS OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (1996) 90,658~662

Clinical overlap between malaria and severe pneumonia in African children in hospital

Michael English1>2, Jeanine Puntl, Isaiah Mwangil, Kieran McHugh3 and Kevin Marsh1*2 ‘Clinical Research Centre, KEMRI Kilifi Unit, P.O. Box 428, Kilifi, Kenya; 2Nufield Department of Clinical Medicine and 3Department of Radiology John Radclzfle Hospital, Headington, Oxford, OX3 9D U, UK

Abstract Data collected from 200 children admitted to a hospital on the Kenyan coast who met a broad definition of severe acute respiratory infection (ARI) indicated that simple clinical signs alone are unable absolutely to distinguish severe AR1 and severe malaria. However, laboratory data showed that marked differences exist in the pathophysiology of unequivocal malaria and unequivocal ARI. Children in the former group had a higher mean oxygen saturation (97 vs. 94, P<O.OOl), mean blood urea level (5.3 vs. 1.9 mmol/L, P<O.OOl) and geometric mean lactate level (4.5 vs. 2.1 mmol/L, P<O.OOl), and lower mean haemoglobin level (5.3 vs. 9.0 g/dL, P<O.OOl) and base excess (-9.4 vs.-2.6, P<O.OOl) than those in the latter group. Using these dis- criminatory variables it was estimated that up to 45% of children admitted with respiratory signs indicative of severe AR1 probably had malaria as the primary diagnosis. Radiological examination supported this con- clusion, indicating that pneumonia characterized by consolidation was uncommon in children with respi- ratory signs and a high malarial parasitaemia (210 OOO/clL). There is no specific radiological sign of severe malaria. In practice, all children with respiratory signs warranting hospital admission in a malaria endemic area should be treated for both malaria and AR1 unless blood film examination excludes malaria. In those with malaria and clinical evidence of acidosis, but no crackles, antibiotics may be withheld while appropri- ate treatment for dehydration and anaemia is given. However, if clinical improvement is not rapid, antibi- otics should be started.

Keywords: malaria, Plasmodium falcipam, acute respiratory infection, pneumonia, children, Kenya

Introduction Several clinical studies have highlighted difficulties in

distinguishing mild malaria and acute respiratory infec- tion (ARI) in children seen at out-patients’ clinics and it has been recommended that treatment be given for both conditions in malaria endemic areas where blood film examination is not possible (O’DEMPSEY et al., 1993a; REDD et al., 1992). Recently, respiratory signs on admis- sion have been recognized as of major prognostic impor- tance in severe malaria (MARSH et al., 1995). We, and others (TAYLOR & MOLYNEUX, 1995), were concerned that overlap in clinical presentation of severe malaria and severe AR1 might result in inaccurate diagnosis and incorrect treatment. The possibility of extensive overlap is of particular importance since increasing emphasis is being; nlaced on the use of clinical algorithms for both out-gaiient and in-patient managemen? There are, how- ever, no published data describing the chest X-ray find- ings of children considered to ha;e malaria and r&pira- torv distress and no data on the abilitv of clinical or labbratory indices to differentiate seveie malaria and ARI.

Methods The study was performed in Kilifi District Hospital,

Kenya, from June to December 1994. The hospital serves an area of coastal Kenya with perennial malaria transmission including 2 seasonal peaks following the rains in May-July and November-December. The hospital out-patients’ clinic is attended by approxi- mately 30 000 children a year, of whom approximately 3000 are admitted (SNOW et al., 1994). Children studied were recruited from those admitted by the duty clinician (who was not part of the research team) to the general paediatric ward during the day. Any child meeting the following definition was considered eligible for the study: age 1 to 59 months, history of both cough and fe- ver for less than 14 d, and a respiratory rate of at least 50 in children aged l-l 1 months and at least 40 in children aged 12-59 months. In addition, one or more of the fol- lowing signs suggesting the presence of more severe ill- ness were required: indrawing (WHO, 1990), nasal flar- ing (LACKRITZ et al., 1992) or deep breathing (MARSH et

Address for correspondence: Dr Mike English, Molecular Parasitology Group, Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford, OX3 9DU, UK.

al., 1995), crackles, bronchial breathing (SPOONER et al., 1989), or respiratory rates of at least 70 in children aged l-11 months or 60 in those 12-59 months old CONY- ANGO et al., 1993). Children were excluded if there’was a clear history of asthma, congenital heart disease, or prior treatment for tuberculosis or if they were known to be seropositive for human immunodeficiency virus.

After obtaining informed consent, mothers were inter- viewed and children were examined bv a studv clinician with all results recorded on a standardproforma. Venous blood was obtained for full blood count (Coulter Elec- tronics. UK’I. thick and thin blood films which were Giems&sta&d and examined for asexual forms of Plas- modium falciparmm (enumerated per 100 white blood cells or 500 !red blood cells), sodium and potassium (CIBA- Corning, UK). urea and creatinine (Beckman Instru- ments, %SA),“venous blood gas (CIB-A-Corning, UK), and blood glucose and lactate (Analox Instruments, UK). Transcutaneous oxygen saturation was assessed by taking the average of 3 pulse oximetry measurements us- ing a reuseable probe (Nellcor, USA) applied to a digit from which a good pulse wave was obtained. Antero-pos- terior chest X-ray examination was done on all children. In the majority this was on admission but with critically ill children this was not possible as no portable X-ray fa- cility existed. With these children, X-rays were taken as soon as possible after admission.

All children with a clinical or radiological diagnosis of pneumonia were treated with either parenteral benzyl penicillin or chloramphenicol for a minimum of 24 h be- fore completing 7 d bf treatment with an oral formula- tion. Those with a blood slide containing P. falciaarum were treated with oral Fansidar@ - (suifadoxine- pyrimethamine) or parenteral quinine if prostrated or comatose (MARSH et al.. 19951. Intravenous fluids and whole blo;d transfusion’ were hsed to treat dehydration and severe anaemia respectively.

Chest X-rays, identified only by a study number, were interpreted by a paediatric radiologist who was unaware of the examination or laboratory findings or the clinical diagnosis. They were assigned to one or more of 5 de- scriptive categories: normal, consolidation (either lobar or patchy), peribronchial thickening, interstitial disease, or oedema/adult respiratory distress syndrome (ARDS). In addition cardiomegaly was assessed as absent, mild, moderate, or severe.

For the purposes of comparison, 2 polar reference

MALARIA AND PNEUMONIA IN AFRICAN CHILDREN 659

groups were defined: (i) unequivocal cases of severe pneumonia were considered to be those meeting the above clinical criteria who had consolidation on chest X-rav taken within 24 h of admission. a blood film neea- tive for malaria narasites, and a Blantvre coma score 53 (MOLYNEUX et al., 1989% (ii) unequivocal cases of ma- laria were those with a normal chest X-ray taken within 24 h of admission. l? fulciaarum uarasitaemia 3100

2 d 1 I

OOO/uL and Blantyre score 23. Thus children who were comatose (Blantyre score s2) were not included in either unequivocal group, to ensure that a primarily neu- rological disturbance did not account for respiratory ab- normalities.

To maximize the possibility of overlap, only data from the 175 children with an X-ray taken on admission were used when comparing clinical groups. This precaution was taken because of the theoretical possibility that a transient radiological abnormality, similar to consolida- tion but secondary to malaria, could occur. However, data from all 200 children were used when determining the ability of a clinical sign to identify radiological cony solidation. The differences between groups for continu- ous variables were examined by the; test and for cate- gorical variables by either the x2 test or Fisher’s exact test. All data were double-entered and verified using dBase IV@ and analysed using SPSS@.

Results Two hundred and seven children were initially re-

cruited to the study; however 7 were subsequently ex- cluded from analysis-2 were diagnosed as having con- genital heart disease, one with chronic liver disease, one with an encephalopathic illness clinically diagnosed as rabies, one who was not X-rayed until the sixth day after admission; 2 further children remained afebrile during admission despite a history of fever, had clinically re- versible airways obstruction, and were considered to have asthma. Of the 200 children remaining, with a mean age of 16 months (interquartile range 6-24 months), 15 children died (7.5%) and 137 (69%) had a blood slide with l? falciparum. Six (3%) of those re- cruited, all of whom entered the study because of tachypnoea, nasal flaring and deep breathing, had a clinical diagnosis of gastroenteritis complicated by severe dehydration. Each of these had a negative blood film, a normal chest X-ray and compensated metabolic acidosis. One hundred and seventy-five children (87%) were X-rayed on admission, a further 19 within 24 h, and the remaining 6 within 72 h. Clinical presentation of those X-raved on admission is summarized in the Figure. Thirteen children had a positive blood culture: in 3, whose chest X-ray examination was normal, Gram- negative pathogens were isolated (2 non-typhi Salmo- nella. one Escher-&a coli). Among the remaining 10 chil- dren who had consolidation in X-ray, Streptococcus pneumoniae was isolated from the blood of 8, Haemophi-

Figure. Clinical presentation of those children who had a chest X-ray (CXR) examination on admission.

Table 1. X-ray diagnoses of 200 children studied

Parasitaemia/ X-ray diagnosis FL Normal Consolidation Othera

None 30 (48%)b 27 (43%) l-9999 27 (56%) 18 (38%) Fi tz 00 10000-99999 42 (93%) 3 (7%) ~100000 41 (93%) 2 (5%) 1(2%)

al+ribronchial thickening or atelectasis. bPercentages of cases within each parasitaemia group.

lus influenzae from one, and Staphylococcus aureus from one.

X-ray findings The chest X-ray findings of all 200 children are pre-

sented in Table 1. Although radiological abnormalities were more common in children with-low parasitaemias, high parasitaemia together with consolidation did occur. Of the 2 children with parasitaemia 2100 000 u/L and consolidation, one was admitted with coma and respira- tory distress and thus aspiration pneumonia could not be excluded. In contrast, the other child was fully con- scious and the only indication for recruitment to the trial was a markedly raised respiratory rate. Amongst children with a blood slide showing I? falciparum. 23 (17%) also had consolidation on chest X-rat In addi- tion, on admission X-ray mild cardiomegaly was ob- served in 7 children and moderate cardiomegaly in one, all of whom had severe anaemia (haemoglobin G5 g/dL, n=64). Severe cardiomegaly was not observed.

Unequivocal severe pneumonia and malaria Clinical and laboratory data are presented in Table 2

Table 2. Admission characteristics of reference clinical groups

True True pneumoniaa malariaa P

No. of children Age (months) Duration of illness (d) Axillary temperature (“C) Respiratory rate (per min) Oxygen saturation (%) Liver size (cm) 2.2 (2.2) 2.6 (1.9) 0.4 l$emoglobm (g/dL) 9.0 (2.2) . <O.OOl

7.39d (0.08) ;;‘x’(“O~?) 0.6 Base excess -2.6d (6.1) -9.4 (6.8) <O.OOl Creatinine (ymol/L) Urea (mmol/L Lactate (mm0 1’ /L)e Glucose (mmol/L)

’ ’ 42 (is) 62 (29) 0.002 \9 (1.9) 5.3 (3.3) <O.OOl

2.1 (1.7,2.6) 4.5 (3.7,5.8) <O.OOl 5.6 (1.1) 5.1 (2.9) 0.5

‘Values are means (standard deviations in parentheses), except for lactate values.

bn=20. >=30. n=22.

eGeometric mean (95% confidence interval in parentheses).

for children meeting our ‘gold standard’ definitions of severe pneumonia (n = 26) and malaria (n = 38). Children with a diagnosis of malaria were more anaemic, had higher blood levels of lactate, creatinine, urea and oxy- gen saturation than those with a diagnosis of severe pneumonia. Respiratory signs were much more com- monly associated with compensated metabolic acidosis in malaria than in severe pneumonia (Table 3). Of the clinical signs evaluated, the presence of bronchial breathing, crackles, indrawing, severe indrawing, unilat- eral chest signs and wheeze and the absence of pallor, deep breathing or splenomegaly appear to be the most informative in discriminating severe pneumonia from

660 MICHAEL ENGLISH ETAL.

Table 3. Prevalence of signs in reference clinical groups

Prevalence (%)”

Pneumonia Malaria P

Bronchial breathing 21 3 0.04 Crackles 77 <O.OOl Deep breathing 15

ii 0.02

Grunting 16 0.15 Indrawing i: 26 0.001 Nasal flaring 62 53 0.5 Pallor 71 0.001 Severe indrawing 2;

7: 0.04

Splenomegaly 50 0.03 Unilateral signs 46 8 0.001 Wheeze Severe anaemiab

27 4:

0.006

Oxygen saturation <95% 4;’

3d 0.001 0.001

a26 children with pneumonia and 38 with malaria except where shown.

$I~30globin c5 g/dL.

Table 4. Comparison of admission groups using optimum discriminator variables

(P=O.Ol) but more likely to have crackles (P=O.OOl). In this group, 9 children had both crackles and low oxygen saturation while only one child had crackles and normal saturation.

Consolidation. Although some signs (bronchial breath- ing, unilateral signs, severe indrawing, wheeze and oxy- gen saturation ~95%) were specific (>80%) for consoli- dation they lacked sensitivity (< 50%). Only the presence of crackles had both moderate sensitivity (70%) and specificity (66%) for identifying consolidation in chil- dren with or without malaria parasitaemia (Table 5). However, even in this study area, where malaria was en- demic with reasonably low transmission, signs which are relatively specific for consolidation occurred just as com- monly in children with and without parasitaemia (Table 5). Therefore, as expected, no combination of signs al- lowed perfect discrimination of malaria, whether asymp- tomatic parasitaemia or clinical disease, and pneumonia. However, consolidation was rare in children with both deep breathing and pallor (6/60) and rarer still in those also without crackles (3/50), although the combination of deep breathing and pallor was present in only 30% of children overall (25% in those also without crackles).

No. of cases in which discriminator variable or value was present Comparison of proportions (P val~es)~

Groups Groups Groups Groups Group A GroupB Group C Group D Group F Avs. C Bvs. D Cvs.D Dvs. F

No. of children Base excess c-7.0 Lactate > 3.5 mmol/L Hbs6 g/dL Urea >4.0 mmol/L

6’(:06%) 3f( 12%)

1(4%)

Crackles 6 (20%)

21(70%)

10b(45%) 10’$3%) 38df:8%) 38 (55%) 0.6 0.2 1.0 0.008 0.1 0.07l 0.2 0.3

14 (56%) 47 (68%) 0.1 0.3 8 10b(45%) (32%) 36 2k(4%) (52%)

o.o$ ;‘; .

<O.OO\ 0.008 .

1 4:.:01

0.2 0.5l <O.OOl 0.08

17 (68%) 48 (70%) 0.007 0.9 10 (40%) 11 (16%) 00:; 0.03 0.07 0.01

aComparison of groups A vs. B for all dis- bn=22. k criminator variables, P>O.OS; comparison of

n=57.

groups A vs. F for all discriminator vari- )=23.

fn=27. Jn=15. ‘Fisher’s exact two-tailed test.

ables,P<O.OOl n=65.

malaria (Table 3). The degree of hepatomegaly, the pres- ence of a murmur, prostration (inability to sit unaided or to breast feed in children < 1 year old) or a history of seizures were not useful (data not shown).

The overlapping groups Malaria or ARI. Using variables that showed the most

marked differences in value or prevalence between un- equivocal malaria and severe pneumonia, differences in groups classified using admission data (Figure) were ex- plored (Table 4). All differences between groups A and F (see Table 4) were significant (P<O.OOl), reflecting the process used in choosing variables. There was no statisti- cally significant (i.e., P<O.Ol) difference between clini- cal groups A and B, but some marked differences be- tween groups A and B compared with group D (children with low parasitaemia). Thus children in group D were more likely to be anaemic and have raised blood lactate levels than those in group B (P<O.OOl and P=O.O08, re- spectively). However, children in group D were more likely (P<O.OOl) to have reduced oxygen saturation than children in group F (with a higher parasitaemia). Chil- dren in group B presented with crackles more com- monly than those in group D (P=O*O3), who in turn ap- peared to have crackles more often than those in group F (P=O.Ol). However, in view of the large number of com- parisons performed it is possible that these were chance findings. Within group D, children with oxygen satura- tion ~95% were less likely to have haemoglobin <6g/dL

Discussion AR1 and malaria have been said to account for ap-

proximately 30% of all childhood mortality in The Gam- bia (GREENWOOD et al., 1987), where levels of malaria transmission are similar to those found in many parts of sub-Saharan Africa. The true figure may, however, be considerably higher since the introduction of insecti- cide-impregnated bed nets, an intervention directed solely at malaria transmission, resulted in a massive re- duction in childhood mortality from all causes in the same area (rate ratio 0.37 for childhood mortality in in- tervention vs. control areas) (ALONSO et al., 1991).

While measures to prevent disease are therefore vital, effective case management of AR1 and malaria also re- mains essential. Many publications have addressed the issue of the most appropriate means to identify severe AR1 (CAMPBELL et al., 1989; SHANN et al., 1989; SPOONER et al., 1989; ONYANGO et al., 1993) and the World Health Organization (WHO) currently re- comends that chest indrawing should be used to distin- guish severe from non-severe pneumonia in children presenting to district hospitals (WHO, 1990). It is now well known that respiratory distress accompanying ma- laria, and sometimes associated with the presence of chest indrawing, carries a poor prognosis (MARSH et al., 1995; OLUMESE et al., 1995; WALLER et al., 1995). This raises the possibility that, in malaria endemic areas, se- vere malaria is misdiagnosed as severe pneumonia or that unrecognized severe pneumonia causes respiratory

MALARIA AND PNEUMONIA IN AFRICAN CHILDREN 661

Table 5. Prevalence and predictive ability of signs

No. of cases with sign and prevalence (%) within groups of children with and without I? fulciparum parasitaemia

Parasitaemic Not uarasitaemic *(n=63) (n= 137)

9 (6%)

Ability to predict consolidation on X-ray (n =200)

Sensitivitv Snecificitv Positive (%) ” - - (%) predictive value

Bronchial breathing Crackles Deep breathing Indrawing PRJ;;obflaring

Severe indrawing Unilateral signs Wheeze Severe anaemia Oxygen saturation 95%a

%=170.

13 (9%) 23 (17%) 14 (10%)

10 (16%) 44 (70%) 15 (24%j

15 (24%) 21 (33%) 19 (30%) 13 (21%) 21 (40%)

distress in children thought to have severe malaria. If in- adequate or inappropriate treatment results, this may in- crease morbidity and mortality.

Several problems make investigating the likely over- lap of severe malaria and severe pneumonia difficult, most obviouslv the lack of a ‘gold standard’ diagnosis for either condition. We attempyed to overcome this prob- lem by initially defining groups likely to be at either end of a spectrum of patients presenting with respiratory signs. Thus the presence of consolidation on X-ray in as- sociation with a history of fever, cough and clinical signs indicative of severe AR1 provided our ‘gold standard diagnosis of severe pneumonia. At the other end of the spectrum, children with a normal chest X-ray and P. falciparum parasitaemia 2100 OOO/uL were taken to rep- resent those whose respiratory signs were due solely to malaria. Such parasitaemias are more than 9 times higher than the 90th percentile of parasitaemias found amongst well children from the community in which the study took place (R. Snow, personal communication).

Laboratory data support the view that this approach defines 2 distinct pathophysiological entities. Children with malaria had higher urea, creatinine and lactate lev- els and evidence of compensated metabolic acidosis which could contribute to respiratory signs. Clinical and derived laboratory variables also indicated that, even amongst children with lower parasitaemias (< 10 OOO/uL) and a normal chest X-ray, malaria was often likely to be the cause of respiratory signs and not simply an inciden- tal findina. Thus 40% (10125) of such children X-raved on admiss?on had clinical and laboratory profiles consis- tent with unequivocal malaria and different from both children with unequivocal severe pneumonia and those with severe AR1 presumably due to non-bacterial respi- ratory pathogens (those with a negative blood slide and a normal chest X-ray). If these children and those with higher parasitaemia and a normal X-ray (group F) were combined, malaria might be considered the primary di- agnosis in as many as 45% (79/175) of children with se- vere resniratorv abnormalities who were X-raved on ad- mission: However, the true proportion of malaria- attributable disease is likelv to be sliahtlv lower. since in this study and in areas of Kenya not indemic for malaria (ONYANGO et al., 1993) almost half of the clinical cases of severe AR1 are associated with a normal chest X- ray-i.e., it is plausible that a small number of children with high parasitaemia and a normal chest X-ray also had ARI.

Although the presence of signs classically associated with severe pneumonia increases the likelihood that consolidation will be present on chest X-ray, they do not exclude the possibility that malaria accounts for, or con- tributes to, respiratory distress. In this study, if the cur- rent WHO guidelines for the management of pneumonia

80 0.48

0.53 0.41 0.13 0.32 0.26 0.13 0.32 0.59 0.33 0.09

in district hospitals (WHO, 1990) had been imple- mented without knowledge of the blood film result, and children with tachypnoea and indrawing had been given treatment only for severe pneumonia, up to 46% of chil- dren with high I? falciparum parasitaemia and a normal X-ray would not have received treatment for potentially life-threatening malaria. Our data therefore support the empirically derived practice of treating both malaria and AR1 in children in hospital with respiratory signs in ma- laria endemic areas. When blood film examination is possible, a film negative for l? .falciparum would allow treatment for AR1 alone but a positive smear does not exclude the uresence of ARI. If an X-rav facilitv exists. it remains a moot point whether those presenting with rks- piratory distress, falciparum parasitaemia above a locally defined threshold and a normal chest X-ray require treatment for ARI. One option with such children is to treat dehydration and anaemia which often underlie res- piratory abnormalities in severe malaria as appropriate. Failure to make a rapid clinical response can be taken to signify the presence of coexistent AR1 which requires prompt treatment.

The pathogenesis of respiratory signs in malaria is poorly understood. In severe disease hyperventilation to compensate for metabolic acidosis seems to be the most important cause of deep breathing, nasal flaring and in- drawing (MARSH et al., 1995). Fever, which alone may re- sult in raised respiratory rates (O’DEMPSEY et al., 1993 b), or parasite sequestration in the microvasculature of the lung (MACRHERSON et al., 1985; CORBETT et al., 1989) are alternative or additional causes. The latter has tenta- tively been suggested as the cause of interstitial changes on chest X-ray in some children with malaria (BYASS et al., 1991). However, even in children with high periph- eral l? falciparum parasitaemias, we found no evidence of pulmonary interstitial change. Severe anaemia with in- cipient cardiac failure has also been held to account for respiratory signs in children with malaria. Although we demonstrated mild or moderate cardiomegaly in a small number of children with severe anaemia (7/64), we did not find any clear radiological features of either pulmo- nary oedema or pulmonary congestion. In fact, although common in anecdote, the only formally documented cases of pulmonary oedema in African children have been at post-mortem in children admitted with cerebral malaria (WALKER et al., 1992). In these cases it is diffi- cult to exclude the possibility that pulmonary oedema was neurogenic in origin and simply a pre-terminal event. Furthermore, a more aggressive approach to re- suscitation of many anaemic children with malaria and severe respiratory distress appears to be well tolerated (ENGLISH et al., in press), and oxygen saturation in chil- dren with unequivocal malaria in this study, many of whom were anaemic, was normal.

662 MICHAEL ENGLISH ETAL.

In summary, signs of severe respiratory disease in Af- rican children in hospital with AR1 or malaria appeared largely to result from 2 distinct pathological processes with extensive overlap in clinical presentation. The de- gree of overlap was such that, without access to diagnos- tic facilities, immediate treatment for both conditions is the safest course of action. The results of blood film ex- amination could reduce the use of antimalarial drugs but antibiotics should nrobablv be witheld onlv from chil- dren with high parasitaemca and clinical evidence of se- vere anaemia and/or acidosis. If signs do not resolve rap- idly with appropriate treatment-in these particularly high-risk children (TAYLOR et al., 1993: MARSH et al., 1953, antibiotics should be given. ‘The inability of chest radiography to confirm all cases of respiratory infection and its limited availability in developing countries makes recommending the routine use of this relatively expensive investigation of less obvious value.

Acknowledgements This study is published with the permission of the director of

the Kenya Medical Research Institute (KEMRI) and was sup- ported by KEMRI and The Wellcome Trust (040313). Dr K. Marsh is a Wellcome Trust Senior Research Fellow in clinical science 1031342). We thank our collearrues at Kiliti District Hospital; in particular the staff of the X-ray department, and the KEMRI research unit in Kilifi for their invaluable assistance during this study.

References Alonso, I’., Lindsay, S., Armstrong, J., Conteh, M., Hill, A.,

David, I’., Fegan, G., De Francisco, A., Hall, A., Shenton, F., Cham, K. & Greenwood, B. M. (1991). The effect of insecti- cide treated bed nets on mortality of Gambian children. Lan- cet, 337, 1499-1502.

Bvass. I’.. Camubell. H.. O’Demnsev. T. I. D. & Greenwood. B. -M. (1991). Coincidence of malaria par&taemia and abnormal chest X-ray findings in young Gambian children. Journal of Tropical Medicine and Hygiene, 94,22-23.

Campbell, H., Byass, I’., Lamont, A. C., Forgie, I. M., O’Neill, K. I’., Lloyd, E. N. & Greenwood, B. M. (1989). Assessment of clinical criteria for identification of severe acute lower res- piratory tract infections in children. Lancet, i, 297-9.

Corbett, C. E. I’., Duarte, M. I. S., Lancellotti, C. L. I’., Silva, M. A. L. & Andrade, H. F. (1989). Cytoadherence in human falciparum malaria as a cause of respiratory distress. Journal of Tropical Medicine and Hygiene, 92, 112-120.

English, M., Waruiru, C. & Marsh, K. (in press). Transfusion for life-threatening respiratory distress in severe childhood malaria. AmericanJournal of Tropical Medicine and Hygiene.

Greenwood, B. M., Greenwood, A. M., Bradley, A. K:, Tulloch, S., Hayes, R. & Oldfield, F. S. J. (1987). Deaths tn infancy and early childhood in a well vaccinated rural, West African uonulation. Annals of Tronical Paediatrics. 7.91-99.

L&k&,, E. M., Campbell; C. C., Ruebush,‘T. K., Hightower, A. W., Wakube, W., Steketee, R. W. & Were, J. B. (1992). Ef- fect of blood transfusion on survival among children in a

- Kenvan hosoital. Lancet. 340.524528. MacPherson, G. C., Warrell, M. J., White, N. J., Looareesuwan,

S. & Warrell, D. A. (1985). Human cerebral malaria: a quan- titative ultra-structural analysis of parasitized erythrocyte se- questration. AmericanJournal of Pathology, 119,385-401.

Marsh, K., Forster, D., Waruiru, C., Mwangi, I., Winstanley, M., Marsh, V., Newton, C., Winstanley, I’., Warn, I’., Peshu,

N., Pasvol, G. & Snow, R. (1995). Life threatening malaria in African children: clinical spectrum and simplified prognostic criteria. New EnglandJournal of Medicine, 332,1399-1404.

Molyneux, M. E., Taylor, T. E., Wirima, J. T. & Borgstein, A. (1989). Clinical features and prognostic indicators in paediat- ric cerebral malaria: a study of 131 comatose Malawian chil- dren. Quarterly Journal of Medicine, 71,441-459.

O’Dempsey, T. J. D., McArdle, T. F., Laurence, B. E., Lamont, A. C., Todd, J. E. & Greenwood, B. M. (1993a). Overlap in the clinical features of pneumonia and malaria in African children. Transactions of the Royal Society of Tropical Medicine and Hvgiene, 87,662~665.

O’Dem&y, T. J.-D., Laurence, B. E., McArdle, T. F., Todd, J. E., Lamont, A. C. & Greenwood, B. M. (1993b). The effect of temperature reduction on respiratory rate in febrile illnesses. Archives of Disease in Childhood, 68,492-495.

Olumese, I?. E., Sodeinde, O., Gbadegesin, R. A., Natiu, O., Oguche, S. & Walker, 0. (1995). Respiratory distress ad- versely affects the outcome of childhood cerebral malaria. Transactions of the Royal Society of Tropical Medicine and Hy- giene, 89,634.

Onyango, F. E., Steinhoff, M. C., Wafula, E. M., Wariua, S., Musia, J. & Kitonyi, J. (1993). Hypoxaemia in young Kenyan children with acute lower respiratory tract infection. British MedicalJournal, 306,612-615.

Redd, S. C., Boland, I’. B., Kazembe, I’. N., Patrick, E., Tem- benu, R. & Campbell, C. C. (1992). Usefulness of clinical case-definitions in guiding therapy for African children with malaria or pneumonia. Lancet, 340, 1140-l 143.

Shann, F., Barker, J. & Poore, I’. (1989). Clinical signs that pre- dict death in children with severe pneumonia. Pediatric Infec- tious DiseaseJournal, 8,852-855.

Snow, R., Bastos De Azevedo, I., Lowe, B., Kabiru, E., Nevill, C., Mwankusye, S., Kassiga, G., Marsh, K. & Teuscher, T. (1994). Severe childhood malaria in two areas of markedly different falciparum transmission in East Africa. Actu Tropica, 57,289-300.

Spooner, V., Barker, J., Tulloch, S., Lehmann, D., Marshall, T. F., Kajoi, M. & Alpers, M. P. (1989). Clinical signs and risk factors associated with pneumonia in children admitted to Goroka Hospital, Papua New Guinea. Journal of Tropical Pe- diatrics, 35,295-300.

Taylor, T. E: & Molyneux, M. E. (1995). Something new out of Africa. New EnglandJournal of Medicine, 332, 1441-1442.

Taylor, T. E., Borgstein, A. & Molyneux! M. E. (1993). Acid base status in paediatric Plasmodium falczpamm malaria. Quar- terly Journal of Medicine, 86,99-109.

Walker, O., Salako, L. A., Sowumni, A., Thomas, J. O., Sode- ine, 0. & Bondi, F. S. (1992). Prognostic risk factors and post mortem findings in cerebral malaria in children. Transactions of the Royal Society of Tropical Medicine and Hygiene, 86, 491-493.

Waller, D., Krishna, S., Crawley, J., Miller, K., Nosten, F., Chapman, D. ter Kuile, F., Craddock, C., Berry, C., Hol- loway, I’. A. H., Brewster, D., Greenwood, B. M. & White, N. J. (1995). Clinical features and outcome of severe malaria in Gambian children. Clinical Infectious Disease, 21, 577-587.

WHO (1990) Acute respiratory infection in children; case manage- ment in small hospitals in developing countries. Geneva: World Health Organization, mimeographed document no. WHO/AR1 90.5.

Received 25 April 1996; revised 14 June 1996; accepted for publication 9July I996