Fever episodes in a holoendemic malaria area of Tanzania ...

1"

479TRANSACTIONS OF THE ROyAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (1992) 86, 479-482

Fever episodes in a holoendemic malaria area of Tanzania: parasitological and

clinical findings and diagnostic aspects related to malaria

Ingegerd Rooth1,2 and Anders Bjorkrnan2 1 Nyamisati Malaria Research Unit, Rufiji District, Tanzania; 2Department of

Infectious Diseases, Roslagstull Hospital, Stockholm, Sweden

insecticide spraying in the area, and only very limited useof bednets. No malaria chemoprophylaxis was dis-tributed in the village during the time of the study.

From 1986 to 1988 there was a research project in thevillage studying different aspects of malaria. The re-search team consisted of a physician, a nurse and 2 lo-cally trained assistants. A newly opened health care clinicoperated in conjunction with the research team. Theclinic was primarily run by a rural medical aid (RMA)with 2 years training. The clinic was open all workingdays from 08:00 to 16:00. After working hours the re-search team was directly available for emergency cases.Medical care was free. Drugs were available only throughthe clinic, thus all consultations and treatments were

properly recorded.

Slud.v groupsThe study included all children aged up to 9 years at-tending the clinic from 1986 to 1988. All clinical epi-sodes, whatever their aetiology , were recorded. Further ,during May-November 1987 the mothers of the childrenattending the clinic with fever as the dominant symptomwere asked what they thought was the child's disease.The RMA and the research physician then made theirpreliminary clinical diagnosis. These tentative diagnoseswere then correlated to the final one based on the resultsof microscopy of blood smears .

Four general surveys were made between 1986 and1989 including all inhabitants of the village. The parasi-taemias of the 'healthy' (asymptomatic) children, in therelevant age group, in these surveys were used as refer-ence comparison data.

Clinical evaluationsAnamnestic data were recorded for children with epi-

sodes of any disease and physical and appropriate labora-tory examinations were done by the RMA or the researchphysician. Symptoms of fever, vomiting and diarrhoeawere specifically asked about. The duration and type offever were recorded for children with fever episodes. Allthose with fever as the main symptom were examined bythe research team. TemperatUres were measured with aPhillips electronic thermometer. An axillary temperature>37.5°C was considered as elevated CSCHMITf , 1984;TRAPE et al., 1985). Fever episodes lacking symptomssuggesting diseases of other origin, and with microscopi-cally confIrmed parasitaemia, were considered as clinical

malaria episodes .

IntroductionAbout 110 million cases of clinical malaria occurworldwide evc:ry year, with 80% in sub-Saharan Africa.The mortality is high, up to 2 millions (WHO, 1989).Most are small children and nearly all fatal cases occur inrural areas before the children reach hospital. This isprobably mainly due to difficulties in identification andtreatment of the clinical episodes of malaria in peripheral

health clinics.Drug resistance is an increasing problem (BJbRKMAN

& PHILLIPS-HoWARD, 1990). Chloroquine is no longerreliable as presumptive treatment and restricted use ofantimalarial drugs to reduce development of resistance isincreasingly important. The treatment of malaria is fur-ther complicated by the higher costs and limited availa-bility of alternative drugs to chloroquine.

Thus early and correct diagnosis is a key issue in thecontrol of malaria mortality and morbidity , in which pe-ripheral health clinics and village health workers playamajor role. However, the diagnostic perfortnance ofhealth units 'without microscopical examination, a com-mon situation in rural health centres, has been little stu-died. Furthero1ore, even with microscopy, the diagnosisof clinical malaria is complex in endemic areas with ahigh prevalence of asymptomatic parasitaemias. Theability of mothers to identify clinical malaria episodes atan early stage is also important and needs to be evalu-ated. Two studies, in Nigeria (MOLINEAUX & GRAMIC-CIA, 1981) and Thailand (HONGIVATANA et at., 1985),have reported low recognition and understanding of ma-laria disease whereas, in another study from Kenya,women sho~'ed a high recognition rate of malaria symp-tomatology (SPENCER et at., 1987).

We report the results of a 3 years study embracing allepisodes of acute illness in children up .to 9 years of ageattending a c:linic in a holoendemic village situated on the

Tanzanian cc>ast.

Material an,d MethodsStudyareaNyamisati is situated in the Rufiji delta in the man-~ove swamips surrounding the outflow of Rufiji riverInto the Inclian Ocean. Malaria is holoendemic (spleenrate in children 2-9 years > 75%) and perennial, withpeak transmission towards the end of the rainy season(March to May) (Unpublished observations). There is no

00 TT~~-'Author for correspondence: lngegerd Rooth, Roslagstull Hospi-

tal, Box 5651, S-114 Stockholm, Sweden.

480

because of general weakness. Two had pica syndrome.All had had symptoms for more than 4 d, and most re-ported fever some time earlier .

The types of fever in relation to malaria are sum-marized in Tables 1 and 3. In children under 5 years oldwho reported fever, elevated temperature was recordedin 61% of the malaria episodes and 97% of other diseases(P<0.001). In 5-9 years old children the temperatur;:was elevated in 60% of malaria cases and 71% of th.'others, Of the children with clinical malaria and fever formore than one day, 98% reported intermittent fever .

Severe manifestations of malaria were uncommon. Nocerebral symptoms were observed in malaria patientsduring the study period although 10 mothers reportedconvulsions in connection with high fever in their child-ren, Eleven small children had haemoglobin valuesunder 70 g/litre, 3 having less than 50 g/litre. All hadcome to the clinic more than 4 d after the onset of symp-toms. The mean haemoglobin level for the other childrenwith malaria was 95 g/litre in the 0-4 years age grouJ'and 98 gllitre in the 5-9 years group. The occurrence 01less severe symptoms such as vomiting and diarrhoea isshown in Table 3. The difference in frequency of vomit-ing in children with clinical malaria compared with otherdiagnoses was significant in the under 5 years age group(P=0.05) but not in older children (P=0.1 ).

Plasmodium falciparum accounted for 97% of the mala-ria episodes and P, malariae for 3%, The geometric meanparasite densities in clinical malaria cases, children withfebrile illness of other aetiology and asymptomatic child-ren are shown in Table 4, The densities in malaria episodes were more than 10-fold higher than those in otherfebrile illnesses. Even the 'healthy' asymptomatic child-ren had higher parasite densities than children with Qtherfebrile illness (P<0'05 for the age group 0-2 years).

Nine children with malaria had parasitaemias in excessof 200 000 parasites /1!1. All came to the clinic within 3 dafter the onset of fever, Twenty-three children with> 100 000 parasites /1!1 came to the clinic 1-7 after onsetof symptoms. The frequencies of the different levels ofparasitaemias in malaria and non-malaria patients areshown in Table 5. The malaria episodes with fewer tha:'400 parasites/1!1 all represented the first clinical episodesof 16 children below one year of age.

Laboratory examinationsThick and thin blood films from fingerprick blood

were made at the time of all clinical episodes and stainedwith Giemsa's stain. After identification of the species,parasite counts were estimated against leucocytes assum-ing 81)00 white blood cells per 1!1. A blood film was con-siden~d negative if no parasite was detected in 100 oil-im-mersjon microscope fields of the thick film at amagI:tification of x 1000. Microscopy was performed bythe S:lffie individual (the research physician) throughoutthe study.

Stati~;tical methodsx2 analysis was performed to test differences in dis-

tributions and Student's t test to evaluate differences be-tweel1 means. Mean parasite densities were calculated asgeometric densities, i.e. among those with detectablepara~;ites only.

Ethical considerationsEthical approval for all the different components of the

malaria research project was obtained from the ethicalcommittee of the National Institute of Medical Research,Tan:~nia and the Karolinska Institute, Sweden. Thestudies were explained several times directly to the vil-lagers and the village authorities during meetings, at-tended by both men and women but on different occa-sion', for cultural-religious reasons.

ResultsClinical episodes

In all, 249 children attended the clinic with 668 epi-sodf's of illness; 395 were diagnosed as clinical malaria( =malaria episodes) and 273 received other diagnoses,mainly common cold, influenza, pneumonia, measles,gastroenteritis, scabies, conjunctivitis and otitis. All thecases diagnosed as clinical malaria had positive bloodslidl:s. There was only one fatality, an infant who diedfror!1 pneumonia,

T'he duration of fever in children with clinical malariaor other diagnoses is summarized in Table I and the

Table I. Reported fever in children 0-9 years old withmalaria episodes and clinical episodes of other origin

Number ofMalaria Other clinicalepisodes episodes

Diagnosis of febrile illnessBetween May and November 1987 we studied the pre-

liminary diagnostic abilities of the mothers, the RMAand the research physician for all children brought to theclinic with fever as the only or main symptom. The finaldiagnosis was based on the results of microscopy of bloodsmears. Of 164 fever cases, 118 were classified as malariaepisodes, 10 as influenza, 2 as common cold, 3 as prob-ably tooth eruptions and the other 31 as fever episodeswith unspecified diagnosis. Nine additional childrencame to the clinic with breathlessness as primary symp-tom in addition to fever. They were not considered asprimarily 'fever patients' by the mothers and were all di-agnosed by the RMA and doctor as pneumonias.

Only 2 mothers mentioned malaria as a possible diag-nosis and the blood films from those children were nega-tive. The correlation between the preliminary diagnosesmade by the RMA and the research physician after an-amnestic recordings and physical examination and the

395a 273

385 (100%)83 (21%)

287 (75%)5 (1%)

10 (3%)0

200 (100%)94 (47%)

4 (2%)87 (43%)

O15 (8%)

Toral

Rel)Qrting continuingfever

.~ 1 d'0:Intermittent 2-3 d<:ontinuous 2-3 dIntermittent ~4 d<:Ontinuous ~4 d

"The 10 children without reported continuing feverattended the clinic with general weakness and had beenfebrile earlier .

nwnber of days with fever before reaching the clinic areshown in Table 2. Ten children with high parasitaemiasdiagnosed as malaria did not complain of fever but came

Table 2. Days of fever in children 0-9 years before attending the clinic with episodes of malaria

No. of

episodes

Mean

duration (d)

Duration of fever2-3 d

Age(years)

No. of

children 2!:4 d~l d

2704580

395

2.02.52.32.1

204 (76%)36 (80%)52 (65%)

292 (74%)

t

481

No. of

children

No. of

episodes Intermittent fever Vomiting Diarrhoea

Children 0-4 yearsMalaria episodesOther febrile illnesses

Children 5-9 yearsMalaria epi~odesOther febrile illnesses

44 (14%)

11 (7%)

27 (90/0)10 (7%)

9670

308152

241 (78%)4 (3%)

3 (4%)3 (6%)

3630

77

49

56 (73%)O

12 (16%)2 (4%)

Table 4. Meal~ parasite densities by age in children with malaria episodes and other febrile illnesses compared toasymptomatic children from four surveys"

9570 (8318, 11015)

324 (257,410)602 (582, 624)

10070 (8876,18113)Malaria episoclesOther febrile

illnessAsymptomatic

339 (215, 525)

513 (335, 785)

"Results expressed as numbers of parasites/l1l (including positive blood films only), with 95% confidence intervals in

parentheses.

Table 5. Differerlt levels of parasite densities in episodes of malariaand other febrile illnesses compared to those in asymptomatic children

Other

febrile

illness

Asympatomaticchildren

Malariaepisodes

100

201

219

654

Ko. of childrenKo. of examinations

Parasite densit).

>10000/1'1>1000/11!>400~11-399111

132

395

10 (2%)

121 (19"/0)

232 (35%)

502 (77%)

190 (48%)

340 (89"/0)

379 (96%)

395 (100%)

O (00/0)7 (3%)

17 (8%)86 (43%)

final diagnose's are shown in Table 6. Although thephysicians showed better specificity in malaria diagnosisthan the RM,\, 48% of the non-malaria episodes werestill wrongly diagnosed as malaria. All rnisclassified pa-tients were seen on the first day of illness.

DiscussionPrompt and early treatment of a malaria episode is es-

sential to prevent severe illness and death. In Nyamisati,despite easy access to medical care, only 21% of thechildren sick 'with malaria were brought to the clinic onthe first day of illness but 96% came within 3 d of theonset of symptoms. In a study in Togo, altogether only20% of all suspected malaria cases were brought to theclinic, and of those only 17% came on the first day of dis-ease. Home tJ'eatment with antimalarial drugs was more

Table 6. Accura~y of clinical diagnosis of febrile episodes by a rural

medical aid and by a physician j

Clinical diagnosis made byRural medjcal aid Medical doctor

Not NotMalaria malaria Malaria malariaTrue statUs

117

22

139

1

24

25

114

40

154

46

10

common in Togo (DEMING et al., 1989). Fever of differ-ent aetiology is common in young children, and is notconsidered by the villagers as dangerous in itself. It istherefore easily understandable that mothers wait acouple of days to see if the child gets well.

No fatality from malaria occurred in our study. Thissuggests that, when the majority of children are treatedwithin 2-3 d, the high mortality often due to malaria insuch areas is prevented. Early treatment also probablyprevented severe anaemia, which was seen only in child-ren who came to the clinic more than 4 d after onset ofillness. Whether early treatment also prevented cerebralcomplications is more uncertain as cerebral malaria wasnot seen in the village and seldom seen in the DistrictHospital (ROOTH, 1984). Perhaps cerebral malaria is un-common in highly endemic areas.

In more than 95% of the malaria episodes, fever wasreported as the main symptom. However, raised tem-perature was recorded in only 60% of cases. In 78%,fever was reported to be intermittent compared to only2% in cases of other illness. Of the patients with malariaepisodes of more than one day's duration, 98% reportedintermittent fever. Many children were thus probablybrought to the clinic in a non-febrile phase of an inter-mittent fever. When mothers report intermittent fever ,this does not necessarily mean a cyclic fever with regularintervals but, rather, irregular fever. They often talkabout 'night fever' and 'fever up and down'. The fever inP. falciparum episodes is frequently erratic. It can evenbe continuous at the beginning of the illness, becomingparoxysmal in the second or even the third recurrence(HARINASUTA & BUNNAG, 1988; KWIATKOWSKI &GREENWOOD,1989).

Many viral infections (e.g. , influenza and measles)show fever as main symptom on the fIrst day of illnessbefore more specific catarrhal symptoms develop. Con-versely, clinical malaria can present with a range ofsymptoms normally associated with other infections.Symptoms like vomiting and diarrhoea, seen almost asoften in malaria episodes as in other febrile diseases, thushave no diagnostic value. Similar findings were reportedin a study in Zimbabwe (BASSET et al., 1991).

We consider that a parasite density in excess of 400parasites per III is a valid cut-off level for diagnosingclinical malaria in 1-9 years old children with fever as themain symptom. Using this criterion, only 8% of febrilecases of other aetiology would be falsely diagnosed ashaving malaria and given unnecessary treatment. In in-fancy , the first clinical malaria episode often occurs at

97%13%74%

99"10

52%

84%

96%

Malaria 118Not malaria 46Total 164

Sensitivity (malaria diagnosis)Specificity (malaria diagnosis)Predictive value (malaria

diagnosis)Predictive value (not malaria

diagnosis)

60%

,

482

low parasite density; here the cut-off point is the level ofpatency. Previously, 5000 parasites/~l has been sug-gc:sted as a cut-off point for treatment (TRAPE et at.,1~)85). In Nyamisati, this would have resulted in misseddj,agnosis and treatment of one-third of the clinical mala-ria episodes.

The possibility of having both malaria and another fe-brile infection must be considered. Recently, however,we showed that at least 2 viral infections, measles and in-fluenza, are usually associated with lower parasite den-sities than those seen in asymptomatic children (ROOTH& Bj()RKMAN, in press).

The mothers brought their children to the clinic be-cause of fever. They were well aware of different types offc'ver, but could not recognize a malaria episode. How-e'/er, they apparently distinguished children with pneu-monia because they were not considered as primarily sickwith 'fever' but rather with breathlessness. The RMAcorrectly diagnosed and treated almost all the clinical ma-laria episodes but also considered many other febrile epi-sodes to be due to malaria. As the concern for malaria ishigh, the RMA rather overdiagnosed malaria in order notto miss any patient in the absence of microscopical exam-ination. However, even an experienced physician identi-tied only half of the febrile episodes not related to malariainfection without microscopical examination.

In rural areas in sub-Saharan Africa, identification andtJ.eatment of most malaria episodes is made at villageIc:vel without microscopical examination. Many febrileepisodes are then falsely diagnosed as malaria, resultingill unnecessary treatment with antimalarial drugs. Thisnlay be reduced if a specific fever history is obtainedfrom the mothers. However, microscopical examinationshould be promoted as much as possible in differenthealth care settings for optimal diagnosis of clinical epi-sodes of malaria.

AcknowledgementsWe thank Professor W. Kilama. Director General, National

Institute for Medical Research, Dar es Salaam, for his support,M. Willcox for kindly reviewing and commenting on the manu-s.:ript, and Bo Nilsson for assistance with statistical evaluation.This study received support from the Swedish PentecostalC:hurch and from the Forenade Liv Mutual Group Life Insur-ance Company, Stockholm, Sweden.

ReferencesBasset, M. T., Taylor, P., Bvirakare, J., Chiteka, F. & Govere.

E. (1991). Clinical diagnosis of malaria: can we improve?Journal of Tropical Medicine and Hygiene, 94,65-69.

Bjorkman, R. & Phillips-Howard, P. A. (1990). Drug-resistantmalaria: mechanisms of development and inferences for mala-ria control. Transactions of the Royal Society of Tropical Me-dicine and Hygiene, 84, 323-324.

Deming, M. S., Gayibor, A., Murphy, K., Jones, T. S. &Karsa, T. (1989). Home treatment of febrile children withantimalarial drugs in Togo. Bulletin of the World Heallh Or-ganization,67, 695-700.

Harinasuta, T. & Bunnag, D. (1988). The clinical featUres ofmalaria. In: Malaria: Principles and Practice of Malariology,Wernsdorfer, W. H. & McGregor, I. (editors). London &New York: Cht1rchill Livingstone, pp. 709-734.

Hongivatana, T. , Leerapan, P. & Chaiteeranuwatisiri, M.(1985). Knowledge, Perception and Behaviour of Malaria.Bangkok: Centre for Health Policy Studies, Mahidol Univer-sity, Monograph Series, p. 5.

Kwiatkowski, D. & Greenwood, B. M. (1989). Why is malariafever periodic? A hypothesis. Parasitology Today, 5, 164-166.

Molineaux, L. & Grarniccia, G. (1981). The Garki Project: Re-search on the Epidemiology and Conlrol of Malaria in the SudanSavanna of West Africa. Geneva: World Health Organization,

p.39.Rooth, I. (1984). Annual Report. Dar es Salaam, Tanzania:

Mchukwi Hospital (P .0. Box 20027).Rooth, I. & Bjorkman, A. (in press). Plasmodiumfalciparum in-

fections are suppressed during concomitant measles or in-fluenza but not during pertussis. American Journal of TropicalM edicine and H.vgiene .

Schmitt, B. D. (1984). Fever in childhood. Pediatric PatientEducation, pp. 929-936.

Spencer, H. C., Kaseje, D. S., Roberts, J. M. & Huong, A. V.(1987). Symptoms associated with common diseases in Sa-radidi, Kenya. Annals of Tropical Medicine and Parasitolog::,81,128-134.

Trape, J. F., Peelman, P. & Morault.Peelman, B. (1985). Crite-ria for diagnosing malaria among a semi-immune populationexposed to intense and perennial transmission. Transactions ofthe Royal Society of Tropical ,Wedicine and Hygiene, 79, 435-442.

WHO (1989). World malaria situation. Weekly EpidemiologicalRecord, 64,241-248.

Received 2 September 1991; revised 28 January 1992;accepted for publication 29 J anuary 1992

IAnnouncement

The Tropical Health and Education TrustFellows of the Society have always been actively involved in many tropical countries in establishing and

developing medical schools and other training institutions. But some of these schools, particularly in poorerAfri.can countries, face severe hardships. Students have no books, there is no foreign exchange for journals,equipment lacks spares, research cannot be supported and external aid is directed towards primary healthcare.

The Tropical Health Education Trust has starred to relieve,.with support from many individuals, Trustsand organizations, some of these disadvantages.

Basic books have been sent to all the rural hospitals in two African countries, sets of books have beengiven for students in a number of others. Links between medical schools overseas and home departmentshave been starred with fellowships for students in training and research methods also.

The Tropical Health and Education Trust aims to extend support like this to more countries, hospitals,medical schools and students and needs funds to do it: Fellows of the Society who would like to take thisopportunity to help our colleagues overcome some of their obstacles can do so through a single gift, afour-year or a deposited covenant, or even through a legacy.

Trustees include: R. M. Anderson, K. P. W. J. McAdam, E. H. 0. Parry (Chairman), D. A. Warrell.For more information about THET please write or telephone: 21 Edenhurst Avenue, Fulham, London,

SW6 3PD, UK. Tel: 071 927 2411, Fax: 071 637 4314.