New insights into the epidemiology of malaria relevant for disease control

Transcription

1 New insights into the epidemiology of malaria relevant for disease control R W Snow** and K Marsh** *Nuffield Department of Clinical Medicine, University of Oxford, Oxford, UK, f Kenya Medical Research Institute/Wellcome Trust Collaborative Programme, Nairobi, Kenya and *Kenya Medical Research Institute, Coastal Unit, Kilifi, Kenya Correspondence to DrRWSnow, Kenya Medical Research Institute/Wellcome Trust Collaborative Programme, PO Box 43640, Nairobi, Kenya Despite over 100 years of scientific investigation, malaria remains the leading cause of death among children living in sub-saharan Africa. Our understanding of the epidemiology of clinical malaria has, until recently, been hampered by a paucity of empirical data from endemic settings. A striking feature of Plasmodium falciparvm malaria is that, compared to infection and mild disease, severe complications and death are rare. Perhaps the single most important factor which ameliorates the risk of asymptomatic infection progressing to life-threatening pathology is the development of clinical immunity. Examination of recent epidemiological evidence suggests that the speed with which clinical immunity is acquired is dependent upon the frequency of parasite exposure from birth. Consequently, the age at which disease presentation peaks, the clinical spectrum of disease and the life-time risks of disease appear to be a function of the intensity of transmission within a given community. These observations are discussed in relation to control measures aimed at reducing P. falciparum exposure and the need to understand better the processes by which children naturally acquire clinical immunity before more rational statements can be made about their wide-spread use in Africa. Take any contemporary textbook on malaria and turn to the section on epidemiology. One will find details of the distributions of human malarias and their vectors across the globe and detailed treatments of the relationship between vector dynamics and the basic reproduction rate of infection in man. Far less attention will be paid to the clinical consequences of infection and almost nothing on the epidemiological patterns of disease that can lead to death. There is a historical basis for treating the epidemiology of malaria as a vector-borne parasitic infection rather than a potentially fatal disease. The 50 years following Ross's identification of the vectors of malaria saw a rapid expansion in our knowledge of the epidemiology of vector-parasite dynamics, the prevailing view being that at last malaria could be eradicated. Despite successes in several parts of the world, the assumption that malaria was a single British Medical Bulletin 1998;54 (No. 2): C The British Council 1998

2 Tropical medicine: achievements and prospects problem amenable to a single set of solutions (as typified by the statement shown in the Box below) was misguided and doomed to failure. In 1975, the World Health Organization (WHO) admitted defeat and set new goals focused on the control of the clinical consequences of infection rather than interrupting transmission. The research community was slow to respond to these new challenges with many epidemiologists still entrenched in the ideology of eradication. Indeed, some have maintained that the only thing eradicated by 1975 was the malariologist. By far the greatest challenge facing the malaria control community today is in sub-saharan Africa, where over 90% of the global morbidity and mortality burden is concentrated primarily among children and caused by Plasmodium falciparum. This review will focus on recent changes in our epidemiological understanding of P. falciparum malaria as it effects the patterns and distribution of clinical malaria among children living in Africa. 777/s conference recommends to governments responsible for the administration of African territories that malaria should be controlled by modern methods as soon as feasible, whatever the original degree of endemicity and without awaiting the outcome of further experiments. Conference on malaria eradication in Equatorial Africa held in Kampala, Clinical outcomes of P. falciparum infection The clinical consequences of P. falciparum infection are often divided into mild and severe, potentially life-threatening manifestations. This of course provides an arbitrary distinction as one will always precede the other and not all complications of mild disease are life-threatening. Fever and a variety of symptoms such as headache, body pains, cough and diarrhoea are the most common features of mild disease. For most individuals in endemic settings, these clinical events will be either effectively treated, treated with only partial success due to inappropriate drugs or dosages or the patient will spontaneously recover in the absence of any intervention. A minority of these events evolve further to pathological complications. Previous attempts to define the severe complications of falciparum malaria derived mainly from experiences of non-immune adults 1 and it is now clear from comprehensive studies of the clinical presentations of paediatric malaria in The Gambia, Kenya and Malawi that the spectrum of severe malaria is significantly different in African children 2 " British Medial Bulletin 1998;54 (No. 2)

3 Malaria morbidity in African children General features of malaria morbidity and mortality in relation to infection The risk of disease susceptibility Perhaps the most striking distinction between the study of infection and disease is that across all stable endemic settings every individual will experience infection, often repeatedly from birth; every individual will experience at least one mild clinical attack during his or her life-time; but only a relatively small proportion of these will develop severe complications carrying a risk of a fatal outcome. It has been suggested that approximately 0.25% of infections will result in death 5. Recently, two large case-control studies have examined a range of putative host, parasite and environmental protective mechanisms implicated in reducing the risks of severe life-threatening disease compared to mild disease and those asymptomatically infected in The Gambia and Kenya. A striking feature of studies of risk factors for malaria is the number of host genetic polymorphisms that have been shown to confer protection from severe disease: these include sickle-cell trait, P- and a-thalassaemia, G6PD deficiency, HLA class I and II alleles, Band 3, spectrin, Lewis and Kid Js(a) red cell types and TNFa promoter polymorphisms. The variety and number of these protective traits also suggests that others await description. Only a limited number of susceptibility genes have been identified 6 ' 7. Hill suggests that most of the protective alleles operate at reducing risks of disease progression 8. Together, these characteristics constitute the polygenic control over Africa's leading cause of parasite-related death and may profoundly influence the micro- and macro-epidemiology of severe disease, however, to date there have been few epidemiological descriptions of the summed genetic control. On the other side of the morbidity equation, malaria parasites also show marked phenotypic and genetic variation. Here, investigation has been limited by the difficulty in defining parasite molecules critical for pathogenesis. Most attention has been focused on a series of parasite products expressed on the surface of the infected red cell and mediating microvasculature sequestration. The picture is complex, but emerging evidence suggests that such polymorphism may be related to parasite virulence 9 and the recent cloning of the gene family coding for these proteins should open the way to a clearer understanding. There is much debate over the role of a host's nutritional status and disease susceptibility and the dogma, that undernutrition is protective, stems from relatively few clinical observations coupled with experimental laboratory-based evidence 10 ' 11. Recent evidence from Kenya suggests that poor nutritional status is a risk factor for severe British Medical Bulletin-\998;5HNo. 2) 295

4 Tropical medicine: achievements and prospects malaria (Marsh et al, unpublished data). The limitations of a casecontrol approach mean that these relationships need to be tested further through detailed prospective studies. Environmental, behavioural and socio-economic risks have so far suggested little influence over an individual child's risk of developing severe disease 12. These results probably reflect the homogeneity in these features within the populations studied, as it would be hard to argue that the total absence of first-line treatment, whether through physical isolation, poverty or behaviour would not influence the risks of developing severe disease. Perhaps the most important factor affecting the epidemiology of malaria as a disease is the acquisition of immunity. Presently there are no well defined immunological markers which relate to clinical protection. There have been several attempts to relate, through crosssectional and prospective studies, immune responses to the risks of clinical malaria but where these have shown significant protection the effects have been weak and always include a large proportion of 'protected' non-responders 13. It is not within the scope of this paper to discuss the mechanisms of acquired immunity but rather we hope to demonstrate certain epidemiological features of immunity through an examination of age and exposure patterns of disease. The age pattern of disease susceptibility The second most important distinction between infection and disease is that severe disease and death from malaria declines with increasing host age much faster than the host's ability to regulate fever, parasite densities and ultimately infection per se. These features suggest important distinctions in the definitions of immunity where, for epidemiological simplicity, one may classify them as anti-severe disease, anti-fever and anti-parasitic immunity. A comparison of parasite prevalence curves under varying levels of transmission pressure indicates that, while the basic shape of the curve remains the same, it may be shifted to the left or the right at higher or lower levels of transmission, respectively 14. Given the dearth of epidemiological descriptions of disease, whether the same is true for disease has remained the subject of much debate. Malaria mortality and its association with transmission intensity In an earlier attempt to examine the relationship between transmission intensity and the risks of fatal outcome, we abstracted all the data 296 British Medical Bulletin 1998,54 (No. 2)

5 Malaria morbidity in African children Fig. 1 Possible relationships derived from studies in Africa of malaria-specific mortality among children under 5 years and the entomological inoculation rate x i 4 o Entomological Inoculation rate 1000 related to malaria-specific childhood mortality recorded in communities where estimates of the intensity of P. faldparum exposure had been documented through entomological investigations (Fig. I) 15. These data have proved of limited value in defining an empirical relationship between intensity of transmission and death from malaria for a number of reasons, including the insensitivity of indirect techniques for ascertaining cause of death 1617 ; the problems in defining host-parasite exposure through the entomological inoculation rate; and confounding factors such as the presence or absence of effective clinical management for malaria between the sites and over time. The latter cannot be underestimated: at Keneba in The Gambia 18 and Mlomp in Senegal 19, childhood mortality was reduced to remarkably low levels through the provision of well funded, well staffed and comprehensive essential clinical services. Conversely, the emergence of resistance to Africa's leading anti-malarial, chloroquine, has led to significant rises in severe disease and mortality 20. Nevertheless, Figure 1 does highlight several important points: (i) there is an amazingly limited amount of data on one of the most fundamental relations in the epidemiology of malaria; and (ii) under the lowest transmission intensities mortality must rise sharply, where after the relationship remains unclear. Patterns of morbidity in relation to transmission intensity Field-based epidemiological studies of mild morbidity frequently use fever and accompanying high parasite densities as characterising a British Medical Bulletin 1998;54 (No. 2) 297

6 Fig. 2 Age-specific patterns of mild, clinical malaria at Ifakara, Tanzania (T. Smith unpublished data); Navrongo, Ghana 25 ; Dielmo, Senegal 26 ; Bo, Sierra Leone (G. Barnish and BM Greenwood unpublished data); Ndiop, Senegal 27 ; and Farafenni, The Gambia 2 ". Cross-sectional, point prevalence parasite rates among the childhood community are shown in parentheses. Afltfttl) B S Ifakara, Tanzania (85%) 3 4 Again) Bo, Sierra Leone (60%) 7 6 S Navrongo, Ghana (74%) Ndlop, Senegal (33%) e Dielmo, Senegal (77%) 3 4 Ao*(yn) Farafonnl, The Gambia (32%) o D n Q. to 3 I 01 Q. "D o

7 Malaria morbidity in African children clinical event. These events are detected either through active surveillance of well defined cohorts of children or passively detected at referral centres. The latter best reflects what the local community perceive as ill-health, although by definition will not detect all mild, transitory clinical events. The former relies heavily upon the attribution of a febrile event to the associated parasitaemia, which in many endemic settings will be present among the majority of asymptomatic hosts and statistical methods have been developed to attribute risk for specific parasite density cut-offs 21. Recently, the use of passive detection of disease events presenting to local hospitals with upgraded diagnostic and research facilities have been employed in the epidemiological study of severe, life-threatening malaria 22 " 24. Such approaches provide useful information on the patterns of severe potentially life-threatening disease and differ from studies of mild disease in that they describe events involving pathologies and complications associated with a higher risk of a fatal outcome. Such studies require carefully selected drainage populations to ensure easy access to hospital and estimates of the childhood populations exposed to risk to derive minimal estimates of the age-specific rates of severe disease within and between communities. Age-patterns of malaria morbidity according to intensity of P. falciparum exposure A comparison of six active-case detection studies of mild morbidity conducted mainly in seasonal transmission areas of West Africa has indicated a variety of case-definitions. Because the studies include weekly, monthly and single-cross sectional approaches, these 'rates' (Fig. 2) should be regarded as arbitrary units. Nevertheless, two points emerge: (i) among the populations experiencing the highest intensity of transmission (expressed here simply as the prevalence of P. falciparum infection in the childhood population) rates of mild morbidity decline throughout childhood, whilst the lower intensity transmission settings experience little change in the rate of morbidity; and (ii) under all transmission conditions (with the exception of Ifakara), the rate of malaria morbidity is lower in the first year of life compared to the second year of life. To describe the patterns of severe morbidity across a range of endemicities common to sub-saharan Africa prospective paediatric admission surveillance was maintained at hospitals in The Gambia and Kenya situated next to five distinct populations with markedly different intensities of P. falciparum transmission intensity 23. The age-specific malaria admission rates for these five communities are shown in Figure 3. The mean ages of paediatric malaria admissions below 10 years of age declined from 63 months at the lowest intensity transmission spectrum British Medical Bulletin 1998;54 (No. 2) 299

8 Fig. 3 Age-specific patterns of severe, clinical malaria at five sites in Kenya and The Gambia. Cross-sectional, point prevalence parasite rates among the childhood community are shown in parentheses i 60 i Siaya, Kenya (83%) Sukuta, The Gambia (37%) 100 Kilifi South, Kenya (74%) Bakau, The Gambia (2%) Kilifi North, Kenya (49%) o T3 3 CL < to Q. D O

9 Malaria morbidity in African children (Bakau) to 17 months at the highest end of the transmission spectrum (Kilifi south and Siaya). Consistent with observations of mild disease (Fig. 2) rates of severe disease begin to decline earlier under more intense transmission. However, these declines may occur earlier for severe disease than for mild disease and particularly evident for the communities which experience low-to-moderate transmission. The relation between rates of disease and transmission intensity The age effects described above would not have any direct implications for attempts to reduce transmission intensity from high to moderate or low levels if by doing so the cumulative risk were concomitantly reduced throughout childhood. This has been the subject of much recent debate 15>27)29. Whilst intervention studies allow the examination of the short-term impact of reducing parasite exposure on morbidity, descriptive epidemiological studies allow an examination of the effects of life-time parasite exposure upon disease risk. A study of parasite densities among children living in a high transmission area of Kenya 30 showed that increased frequency of exposure to infected mosquitoes was related to an increasing risk of high parasite densities, although the authors were unable to distinguish whether this simply reflected increasing risks of superinfection. Two studies have assessed the entomological inoculation rate in relation to the incidence of severe disease within defined communities 31 ' 32. These studies were conducted over short periods of observation, were unable to control for mobility between exposure settings and, in the case of the Kenya study, were unable to adjust for the marked over-dispersion of vector populations within the sampling units 31. Nevertheless, neither of these studies demonstrated a linear rise in disease risk with increasing intensity of sporozoite inoculation. The studies of severe morbidity (Fig. 3) 23 showed that, at the most intense levels of P. falciparwn transmission, the risks of developing severe disease were highest during the first 2 years of life, thereafter the risks declined rapidly. By contrast, communities which encountered lowto-moderate rates of P. falciparum challenge experienced a much more extended period of risk in childhood lasting from the first year of life through to the fifth birthday. At the lowest end of the transmission spectrum, disease risk appears to be spread evenly across the entire childhood period. The total disease risk for children below 10 years of age is shown in Figure 4a. These data do not support the notion that the higher the natural parasite exposure risk from birth the higher the disease risk throughout childhood; on the contrary, they suggest a decline in overall malaria rates in childhood under the highest levels of P. falciparum challenge. This paradoxical decline in disease risk was British Medical Bulletin 1998;54 (No. 2) 301

10 Tropical medicine: achievements and prospects o Q o Q. I 10 4 I 3 I * Fig. 4 (a) Total rates of severe, clinical malaria among children aged 0-9 years at Bakau (A), Sukuta (B), Kilifi North (Q, Kilifi South (D) and Siaya (E). Intensities of P. falciparum transmission for each community are shown in Figure 3 u. (b) Total rates of cerebral malaria among children aged 0-9 years at Bakau (A), Sukuta (B), Kilifi North (C), Kilifi South (D) and Siaya (E). Intensities of P. falciparum transmission for each community are shown in Figure 3". much more pronounced for cerebral malaria which was notable by its rarity under high transmission (Fig. 4b) an observation made in other studies 22 ' 24. However, in contrast to these other studies, the rates of severe malaria anaemia at Kilifi north, Kilifi south and Siaya, where haemoglobin levels were recorded on every admission, did not support the idea that this manifestation of severe disease (and, therefore, death) increases with transmission intensity. Only at Bakau, where the highest mean and median 302 British Medial Bulletin 1998;54 (No. 2)

11 Malaria morbidity in African children ages of disease presentation and the lowest levels of P. falciparum transmission were recorded, did the risk of disease in childhood remain low. Mechanisms of disease protection in relation to parasite exposure Host-parasite interactions are complex in semi-immune populations and yet the net result is in survival for the greater majority of exposed hosts but a numerically significant cost in terms of mortality. The latter comprising those with the wrong genes with the wrong parasite at the wrong time. Against this background, an individual's ability to acquire mechanisms which render him or her clinically immune will depend upon their individual life-time experiences with the parasite. The most plausible explanation for the patterns of mild and severe malaria described above is that a given amount of exposure is required to develop effective clinical immunity. When infection rates are high, these experiences occur early in life furnishing the child with an acquired resistance to the clinical consequences of infection. During this window of time, other mechanisms may also operate 33 " 34 to modulate disease severity and the young infant enjoys a period of clinical protection whilst undergoing active immunisation. Furthermore, the period of active immunisation may be effective much longer in relation to cerebral malaria as this appears to be inherently rare in younger children for reasons that are not clear. There is now good evidence that clinical protection is, at least in part, strainspecific 35 and transmission intensity will define the frequency and speed with which an individual encounters the entire repertoire of the local parasite population 36. The combined effects of an attenuated disease-risk period and reduced disease susceptibility early in life may be all that is required to explain the apparent paradox of reduced cumulative risk of severe disease under intense parasite exposure. However, there may be additional mechanisms of protection which operate under high transmission and involve the constant interaction between infection and the immune system. It is striking that, under the highest intensities of transmission, over 80% of the childhood population are asymptomatically infected at any one time throughout childhood. Often these infections will involve multiple parasite populations acquired continuously throughout the year. The constant stimulation of T-cell responses through recognition of conserved or previously encountered parasite antigens may be important in maintaining clinical immunity. In areas of very low intensity transmission (Bakau, in Fig. 4a) the lack of development of immunity is probably balanced by the low frequency British Medical Bulletin 1998;54 (No. 2) 303

12 Tropical medicine: achievements and prospects of infection, hence disease rates will be low and the risks may continue throughout life. Obviously, spreading the disease risks over all age classes would provide no resultant benefit for the population 27. However, assuming stable transmission, the risks of severe morbidity and death are unlikely to be high enough to ever reach the life-time (before age 80 years) risks shown for any of the other communities shown in Figure 3. Others have argued in favour of an age-related effect of immune acquisition where protection from the severe consequences of infection are more attenuated the older the host The evidence on which this is based derives from studies of non-immune migrants into endemic areas of Indonesia 37. Although the study was unable to demonstrate clearly complete lack of previous exposure to P. falciparunt among the migrant children and their parents, the concept deserves further attention. Evidence from control programmes aimed at reducing parasite exposure on the long-term consequences upon immunity Direct effects upon mortality The statement shown in the Box, related to the use of DDT to interrupt transmission in stable, hyperendemic parts of Africa and stimulated much heated debate among the research community A similar position prevails today with the advocation of the universal promotion of insecticide-treated bednets throughout Africa 40. The scientific debate continues, although with the advantage of a better epidemiological description of disease patterns under a variety of natural transmission settings. But what direct evidence do we have that vector control will achieve a sustained impact upon mortality in Africa?. Until recently, evidence of the effects of reducing parasite exposure in Africa on mortality in childhood came from studies conducted in Pare Taevta (Tanzania), Kisumu (Kenya) and Garki (Nigeria). At Pare Taveta, a 3 year programme of residual house-spraying with dieldrin resulted in an infant mortality reduction from 165 per 1000 live births to 78 per 1000 live births. On completion of the programme, concerns were raised about a 're-bound' in mortality and efforts were introduced to improve clinical management of febrile events despite which the next year saw infant mortality rise to 132 per 1000 live births and after 7-8 years mortality among children aged 1-4 years returned to pre-intervention 304 British Medical Bulletin 1998;S4 (No. 2)

13 Malaria morbidity in African children Indirect effects upon mortality levels 41. At Kisumu, residual spraying of fenitrothion successfully reduced the parasite exposure among infants by 96% accompanied by a 48% reduction in the infant mortality rate Proxopur spraying in combination with mass drug administration at Garki, Nigeria showed similar effects to the study in Kisumu where, over a 2 year period of intervention, significant effects were achieved in the reduction in both parasite exposure and death in infancy 14. Although both the Kisumu and Garki studies were more carefully controlled than the Pare Taevta study, neither allowed the estimation of the age-specific mortality effects over a sustained period of intervention nor after aggressive intervention was suspended. Efficacy trials of the effects of insecticide-treated bednets or curtains against childhood mortality have been conducted in settings with estimated annual entomological inoculation rates ranging from between infective bites in Burkina Faso 44, between in Ghana 45, between in Kenya 46 and less than 10 in The Gambia 47. The trials provided a range of protective efficacy estimates against all-cause childhood mortality over the 1-2 years following their introduction, ranging from a non-significant reduction of 14% in Burkina Faso to significant effects of 17% in Northern Ghana, 33% on the Kenyan Coast and 63% in The Gambia. Despite these dramatic differences in protective efficacies, according to the intensity of transmission these comparisons of protective efficacies should be treated with caution (especially when comparing potentially averted deaths) as they do not allow for the marked differences in other competing risks for mortality between these communities, such as the use and access to essential clinical services. Several authors have argued that the removal of malaria through reduced parasite exposure may reduce the risks of other causes of mortality. These views are supported by the observation of a greater than expected reduction in crude mortality rates in Sri Lanka following a malaria eradication campaign using DDT which began in Using historical trend analysis, the 'indirect' effects were between 2 4 times those of the 'direct' effects 48. However, recent analysis of these data suggests there were significant improvements in the general mortality patterns before the introduction of DDT and that these have not previously been used in the interpretation of the overall trends in mortality 49. Alonso and colleagues 47 suggest that insecticide treated bednets may confer indirect protection against other causes of death following the observation of a reduced verbal autopsy diagnosed acute respiratory infection (ARI) death rate among insecticide-treated bednet British Medical Bulletin 1998;54 (No. 2) 305

14 Tropical medicine: achievements and prospects users. Conversely, in the Gharki project, the seasonal peaks of mortality attributed to malaria were removed following intervention but mortality remained high overall 14. Some have argued that this latter observation supports a notion of replacement mortality with dominant risk factors of childhood mortality being non-disease specific so that the removal of one cause will not reduce the chances of that child's overall mortality risk when presented with another pathogen 50. During a recent trial of insecticide-treated bednets in Kenya, paediatric admissions due to malaria were significantly reduced by over 40% whilst there were no effects upon the admission rates with either ARI or diarrhoeal disease during the same period 46. Furthermore, studies on the epidemiology of paediatric malaria requiring in-patient care demonstrated almost identical rates of ARI admission between each of five communities despite a log order difference in the intensity of P. falciparum transmission between the sites 23. To date, therefore, the evidence for indirect or replacement mortality effects is weak; nevertheless, these possible effects deserve further investigation in particular how this may be mediated through nutritional status. Implications of new insights into malaria epidemiology for vector control in Africa It is to be regretted that none of the organisations promoting vector control either through residual spraying or more recently bednets set up long term model control zones in which the fundamental question of delayed acquisition of effective clinical immunity could be studied. Presently, we must rely upon epidemiological observations of immunity as judged by changing age patterns of disease risk according to intensity of parasite exposure. Clearly, from any starting point, an intervention to reduce transmission would be expected to lead to an initial reduction in severe disease but the most important question is whether this would be maintained in the cohort of new births exposed to the 'new' transmission conditions. The data presented above are consistent with the view that effective control of parasite exposure amongst populations traditionally exposed to intense P. falciparum challenge may lead to a rise in the age at which an effective anti-disease immune response is acquired, thereby increasing the susceptibility to the severest forms of falciparum malaria (cerebral malaria) and may serve to cause a paradoxical rise in disease risk throughout childhood. For sustained, significant gains to be achieved it is probably necessary to reduce parasite exposure to extremely low levels and, with current interventions, this is only likely to be achievable in areas of low-to-moderate transmission. At the very least, it seems 306 British Medical Bulletin 1998;54 (No. 2)

15 Malaria morbidity in African children Acknowledgements References inappropriate to treat malaria morbidity and mortality as a single problem across Africa and it seems reasonable that the variations in the patterns of disease should be used to guide control. Available mathematical models of malaria designed for control programmes consider only parasite and vector interactions and there is an urgent need for models which can predict the epidemiological patterns of morbid and fatal outcomes of infection based upon usable criteria of parasite exposure. The present tools used to define endemicity (the parasite ratio, spleen rates, infant conversion rates, entomological inoculation rates) should be re-considered in the context of disease outcome. It is all too easy to call for more research but the epidemiological descriptions of malaria as a fatal disease remain scarce and data that does exist highlight our poor understanding of how clinical immunity develops or could be affected by intervention. Moving from a position of phenomenology to mechanisms is difficult, however, there are signs of a new collaborative spirit between basic, field and theoretical scientists which, hopefully, will unravel the mysteries of malarial immunity and pave the way for a more rational approach to designing and implementing new control measures. The authors are grateful to Drs Brian Greenwood and Sunetra Gupta for their helpful comments and to the Wellcome Trust for their support as part of their senior fellowships programme (RWS and KM ). 1 Warrell DA, Molyneux ME, Beales PR Severe and complicated malaria. Trans R Soc Trop Med Hyg 1990, 84 (suppl 2): Marsh K, Forster D, Waruiru C et al. Indicators of life-threatening malaria in African children: clinical spectrum and simplified prognostic criteria. N Engl J Med 1995; 332: Molyneux ME, Taylor TE, Wiriama JJ, Borgstein A. Clinical features and prognostic indicators in paediatric cerebral malaria: a study of 131 comatose MaJawian children. Q]M 1989; 71: Brewster DR, Kwiatowski D, White NJ. Neurological sequaelae of cerebral malaria in children. Lancet 1990; 330: Greenwood BM, Marsh K, Snow RW. Why do some African children develop severe malaria?. Parasitol Today 1991; 7: McGuire W, Hill AVS, Allsop CEM. Variation in the TNFcc promoter region associated with susceptibility to cerebral malaria. Nature 1994; 371: Fernandez-Reyes D, Craig AG, Kyes SA et al. A high frequency African coding polymorphism in the N-terminal domain of ICAM-1 predisposing to cerebral malaria in Kenya. Human Mol Gen 1997: 6: Hill AVS. Malaria resistance genes: a natural selection. Trans R Soc Trop Med Hyg 1992; 86: British Medical Bulletin 1998;54 (No. 2) 307

16 Tropical medicine: achievements and prospects 9 Newbold CL, Warn P, Black G et al. Receptor specific adhesion and clinical disease in Plasmodium falciparum. Am J Trop Med Hyg 1997; 57: Hendrickse RG, Hasan AH, Olumide LO, Akinkunmi A. Malaria in early childhood. An investigation of five hundred seriously ill children in whom a clinical diagnosis of malaria was made on admission to the children's emergency room at University College Hospital, Ibadan. Ann Trop Med Parasitol 1971; 65: Edirisinghe JS, Fern EB, Targctt GAT. Dietary suppression of rodent malaria. Trans R Soc Trop Med Hyg 1981; 75: Koram KA, Bennet S, Adiamah JH, Greenwood BM. Socio-economic determinants are not major risk factors for severe malaria in Gambian children. Trans R Soc Trop Med Hyg 1995; 89: Molineaux L. Plasmodium falciparum malaria: some epidemiological implications of parasite and host diversity. Ann Trop Med Parasitol 1996; 90: Molineaux L, Gramiccia G. The Gbarki project: research on the epidemiology and control of malaria in the Sudan savanna of West Africa. Geneva: World Health Organization, Snow RW, Marsh K. Will reducing Plasmodium falciparum transmission alter malaria mortality among African children? Parasitol Today 1995; 11: Snow RW, Armstrong JRM, Forster D et al. Childhood deaths in Africa: uses and limitations of verbal autopsies. Lancet 1992; 350: Todd JE, De Francisco A, O'Dempsey TJD, Greenwood BM. The Limitations of verbal autopsy in a malaria endemic region. Ann Trop Paediatr 1994; 4: Lamb W, Foord FA, Lamb CM, Whitehead RG. Changes in maternal and child mortality rates in three isolated Gambian villages over ten years. Lancet 1984; 2: Pison G, Trape JF, Lefebvre M, Enel C. Rapid decline in child mortality in a rural area of Senegal. Int J Epidemiol 1993; 22: Greenberg AE, Ntumbanzondo M, Ntula N, Mawa L, Howell J, Davachi F. Hospital-based surveillance of malaria-related paediamc morbidity ad mortality in Kinshasa, Zaire. Bull World Health Organ 1989; 67: Smith T, Armstrong Schellenberg J, Hayes R. Attributable fraction estimates and case definitions for malaria in endemic areas. Stat Med 1994; 13: Snow RW, Bastos de Azevedo I, Forster D et al. Severe childhood malaria in two areas of markedly different Plasmodium falciparum malaria transmission in East Africa. Ada Trop 1994; 57: Snow RW, Omumbo JA, Lowe B et al. Relation between severe malaria morbidity in children and level of Plasmodium falciparum transmission in Africa. Lancet 1997; 349: Slutsker L, Taylor TE, Wirima J, Steketee RW. In-hospital morbidity and mortality due to malaria-associated severe anaemia in two areas of Malawi with different patterns of malaria infection. Trans R Soc Trop Med Hyg 1994; 88: Binka FN, Morris SS, Ross DA, Arthur P, Aryeetey ME. Patterns of malaria morbidity and mortality in children in northern Ghana. Trans R Soc Trop Med Hyg 1994; 88: Trape JF, Rogier C, Konate L et al. The Dielmo project: a longitudinal study of natural malaria infection and the mechanisms of protective immunity in a community living in a holoendemic area of Senegal. Am J Trop Med Hyg 1994: 51: Trape JF, Rogier C. Combating malaria morbidity and mortality by reducing transmission. Parasitol Today 1996; 12: Greenwood BM, Bradley AK, Greenwood AM et al. Mortality and morbidity from malaria among children in a rural area of The Gambia, West Africa. Trans R Soc Trop Med Hyg 1987; 81: Greenwood BM. Malaria transmission and vector control. Parasitol Today 1997; 13: McElroy PD, Beier JC, Oster CN et al. Predicting the outcome in malaria: correlation between rate of exposure to infected mosquitoes and level of Plasmodium falciparum parasitaemia. Am J Trop Med Hyg 1994; 51: Mbogo CNM, Snow RW, Khamala CPM et al. Relationships between Plasmodium falctparum transmission by vector populations and the incidence of severe disease at 9 sites on the Kenyan Coast. Am J Trop Med Hyg 1995; 52: British Medical Bulletin 1998;54 (No. 2)

17 Malaria morbidity in African children 32 Trape JF, Quinet MC, Nzingoula S et al. Malaria and urbanization in Central Africa: the example of Brazzaville. Part V: Pernicious attacks and mortality. Trans R Soc Trop Med Hyg 1987; 81 (suppl 2): McGregor IA. The passive transfer of human immunity. Am J Trop Med Hyg 1965; 13: Gilles HM. The development of malarial infection in breast-fed Gambian infants. Ann Trop Med Parasitol 1957; 51: Bull P, Lowe B, Korkok M, et al. Parasite antigens on the infected red cell surface are targets for naturally acquired immunity to malaria. Nature Med 1998; 4: Gupta S, Hill AVS, Kwiatkowski D et al. Parasite virulence and disease patterns in Plasmodium falciparum malaria. Proc Natl Acad Sci USA 1994; 91: Baird JK. Host age as a determinant of naturally acquired immunity to Plasmodium falctparum. Parasitol Today 1995; 11: Wilson DB, Garnham PCC, SweLlenberg NH. A review of hyperendemic malaria. Trop Dis Bull 1950; 47: World Health Organization. Report of the malaria conference in Equatorial Africa. World Health Organisation Technical Report Series 1951; 38: World Health Organization. Meeting on insecticide-impregnated materials. TDK News, WHO 1996; 50: 3 41 Pringle G. Experimental malaria control and demography in a rural East African community: a retrospect. Trans R Soc Trop Med Hyg 1969; 63 (Suppl): S2-S18 42 Payne D, Grab B, Fontaine RE, Hempel JHG. Impact of control measures on malaria transmission and general mortality. Bull World Health Organ 1976; 54: Fontaine RE, Pull JH, Payne D et al. Evaluation of fenitrothion for the control of malaria. Bull World Health Organ 1978; 40: Habluetzel A, Diallo DA, Esposito F et al. Do insecticide-treated curtains reduce all-cause child mortality in Burkina Faso? Trop Med Int Health 1997; 2: Binka FN, Kujabe A, Aduik M et al. Impact of permethrin impregnated bednets on child mortality in Kassena-Nankana district, Ghana: a randomised controlled trial. Trop Med Int Health 1996; 1: Nevill CG, Some ES, Mung'ala VO et al. Insecticide-treated bed nets reduce mortality and severe morbidity from malaria among children on the Kenyan coast. Trop Med Int Health 1996; 1: 139^6 47 Alonso PL, Lindsay SW, Armstrong JRM et al. The effect of insecticide-treated bed nets on mortality of Gambian children. Lancet 1991; 337: Molineaux L. The impact of parasitic diseases and their control with an emphasis on malaria and Africa. In: Vallin J, Lopez A. (eds) Social Policy and Mortality Prospects. liege, Belgium: Ordina Editions, 1985; Langford CM. Reasons for the decline in mortality in Sri Lanka immediately after the second World War: a re-examination of the evidence. Health Trans Rev 1996; 6: Bradley DJ. Malaria: old infection, changing epidemiology. Health Tran Rev 1992; 2: British Medical Bulletin 1998,54 (No. 2) 309