Authors Richard Idro (corresponding author) and Judith Aloyo, Department of Paediatrics and Child Health, Mulago Hospital/Makerere University Medical School, Kampala, Uganda. Tel.: +256 41 531875; Fax: +256 41 532591; E-mail: email@example.com, firstname.lastname@example.org Lam Mayende, Masafu Health Centre IV, Busia, Uganda. Edward Bitarakwate, Department of Paediatrics, Kabale Regional Referral Hospital, Kabale, Uganda. Tel.: +256 486 22006; Fax: +256 486 22727; E-mail: email@example.com Chandy C John, Department of Pediatrics, Division of Pediatric Infectious Diseases, University of Minnesota Medical School, Minneapolis, USA. Tel.: +1 612 624 1966; Fax:+1 612 624 8927; E-mail: firstname.lastname@example.org George W Kivumbi, Child Health and Development Centre, Makerere University, Kampala, Uganda. Tel.: +256 41 541684; Fax: +256 41 531677; E-mail: email@example.com
Objectives Age and transmission intensity are known to influence the manifestations of severe falciparum malaria in African children. However, it is unclear how specific clinical features such as seizures, impairment of consciousness, or respiratory distress vary with the parasite load and transmission intensity. We examined how the peripheral parasite load varies with transmission intensity and how this influences the symptoms and manifestations of severe malaria in children under 5 years in three areas with different malaria transmission intensity across Uganda.
Methods We consecutively recruited 617 children with severe malaria presenting to three hospitals in areas with very low (51), moderate (367) and very high (199) transmission intensities and compared the age, admission parasite density and proportions of patients with different manifestations of severe disease.
Results The median age (months) was inversely proportional to transmission intensity and declined with rising transmission (26.4 in very low, 18.0 in moderate and 9.0 under very high transmission). The highest proportion of patients reporting previous malaria admissions came from the area with moderate transmission. The geometric mean parasite density (18 357, 32 508 and 95 433/μl) and the proportion of patients with seizures (13.7%, 36.8% and 45.7%, P < 0.001) from very low, moderate and very high transmission respectively, increased with rising transmission. A linear increase with transmission was also observed in the proportion of those with repeated seizures (9.8%, 13.4% and 30.2%, P < 0.001) or impaired consciousness (7.8%, 12.8% and 18.1%, P = 0.029) but not respiratory distress. The proportion of patients with severe anaemia (19.6%, 24.8% and 37.7%, P = 0.002) mirrored that of patients with seizures.
Conclusions These findings suggest that heavy Plasmodium falciparum parasitaemia may be important in development of seizures, severe malarial anaemia and impaired consciousness in children under 5 years of age but may not be important in the development of respiratory distress.
Age and transmission intensity are thought to influence the manifestations of severe malaria in sub Saharan Africa (Imbert et al. 1997; Snow et al. 1997; Modiano et al. 1998). In areas with similar transmission, age and immunity appear to be the most important risk factors influencing the type of manifestation (Imbert et al. 1997). In areas with different transmission, the risk of severe disease appears to be highest among populations exposed to low to moderate transmission (Snow et al. 1997). The apparently lower risk of severe malaria in very high transmission areas appears to be due to the development of clinical immunity early in life when other protective mechanisms may operate. However, how the peripheral parasite density varies with transmission or influences the different types of manifestations of severe disease is poorly described (Ellman et al. 1998). Differences in the occurrence of specific clinical features such as seizures, impairment of consciousness or respiratory distress with transmission or parasite load are also unclear.
Although malaria disease patterns related to transmission are best studied using incidence data (Snow et al. 1999), we used the prevalence of severe disease in hospital as a proxy measure. We examined how the peripheral parasite load varies with malaria transmission intensity and how this impacts on the symptoms and pattern of severe malaria in children under 5 years in three areas with different malaria transmission intensity across Uganda.
This was a descriptive study of children under the age of 5 years with severe malaria as defined by World Health Organization (WHO 2000) from three Ugandan hospitals in areas with different malaria transmission intensity.
Malaria transmission in Uganda can be divided into four main levels based on endemicity data: epidemic prone very low, low, medium to high (moderate) and very high transmission (Figure 1). The lowland plains of the northern and eastern parts of the country have very high transmission, whereas transmission is moderate in the central areas around the capital, Kampala. The highland area of Southwestern Uganda is characterized by epidemic prone very low transmission. Transmission rates vary from less than one infectious bite per year in the highlands to hundreds of infectious bites per year in some of the northern and eastern parts of the country (Omumbo et al. 1998; Lindblade et al. 1999; Ministry of Health 2000).
Although transmission may vary even within a single area, for the purpose of this study, we assumed homogenous transmission intensity in each transmission area. Three hospitals in areas with very low, moderate and very high transmission intensity were studied: Kabale Regional Referral Hospital, Mulago National Referral Hospital and Masafu mini-hospital, respectively. Masafu is in Busia district about 250 km east of Kampala, along the eastern border of Uganda with Kenya. Kabale is 400 km west of Kampala at the southwestern tip of the country and is bordered by Rwanda to the south and The Democratic Republic of Congo to the West. Although Kabale admits over 2000 children annually, malaria is only a problem during epidemic periods. In non-epidemic periods, Kabale has extremely low malaria vector density, with estimated entomological inoculation rates less than one infectious bite per person per year (Lindblade et al. 1999). Mulago, which is located in Kampala in central Uganda, is not only the national referral centre but also serves as a district hospital for Kampala. About 20 000 children annually are admitted through its paediatric emergency unit of whom about 30% have malaria (Idro & Aloyo 2004).
Chloroquine plus Sulphadoxine-Pyrimethamine given together has been the first line anti malaria drug combination in the country since 2000. However, drug resistance to this combination is widespread and the risk of clinical failure adjusted by genotyping over 28 days across the country ranges from 34% to 67% (Bakyaita et al. 2005).
Enrolment. To estimate the required sample size, we used the proportion of children with severe malaria presenting with severe anaemia in each of the three areas. In the low and high transmission areas, this proportion was estimated to be 10% and 30% respectively (Ndyomugyenyi & Magnussen 2004). Using a ratio of 1:4 cases in low and high transmission areas, we targeted a minimum sample size of 47 and 188 children respectively in the two areas. In the moderate transmission area, the proportion of patients with severe malaria anaemia was estimated to be 18% giving a minimum sample size of 214.
All children <5 years with suspected malaria presenting to any of the three hospitals had Giemsa stained blood smears obtained to confirm falciparum malaria infection. Patients were assessed for signs of severe malaria using the WHO 2000 criteria (WHO 2000) by clinicians who had a skills improvement training before the start of the study. Only children who fulfilled one or more of the criteria for severe malaria were recruited for the study (Table 1).
Table 1. Manifestations of severe malaria in areas with low, moderate and high transmission in Uganda
Manifestations of severe malaria
Kabale (n = 51, %)
Kampala (n = 367, %)
Busia (n = 199, %)
* Two or more generalized seizures in 24 h.
† In children 1 year and older, the child is unable to sit unsupported or stand although was able to before the illness and in children <1 year, child is unable to drink or breastfeed.
‡ Blantyre Coma Score <5 in children 9–59 months or <4 in children 3–8 months.
§ Age related tachypnoea with sustained nasal flaring, deep breathing or subcostal retractions.
** Haemoglobin <5 g/dl.
‡ Parasitaemia >20%.
‡‡ Blood glucose <2.2 mmol/l.
Being a referral hospital that receives patients from a wide area, in Mulago, we only recruited patients from within the urban areas of Kampala to reflect moderate malaria transmission. Enrolment was achieved within 3 months in Kampala and within 8 months in Busia; in Kabale, it stretched over a period of 17 months due to the extremely low numbers of patients with severe malaria. The Uganda National Council for Science and Technology and the Review Boards of Kabale Hospital and Makerere University Faculty of Medicine, approved the studies. Parents of all children gave written consent.
Haemoglobin (Hb) levels were determined with a colorimeter using Drabkin's method and blood glucose levels by glucometer readings (One Touch® SureStep®, Life scan, Johnson Johnson Company, USA Inc., 1996–2003). Drabkin's method is used for Hb estimation in most hospitals in Uganda. In Masafu mini-hospital, blood glucose testing was not performed consistently and so, results are not presented. Parasites were counted against 200 white blood cells (WBC) and parasite density calculated against a WBC count of 8000/μl. Cerebrospinal fluid (CSF) was obtained and examined in study subjects with suspected cerebral malaria. CSF was considered normal if CSF protein was <40 mg/dl and WBCs <10/mm2 (Berkley et al. 1999). To improve on the reliability of results, a skills’ improvement training in the above laboratory procedures was held for the laboratory technicians before onset of the study.
Ugandan national treatment guidelines for management of severe malaria, adapted from the World Health Organization guidelines, were used for treatment (Ministry of Health 2000; WHO 2000). Hypoglycaemic patients received 2 ml/kg of 25% dextrose as a bolus infusion. In Masafu, dextrose was administered to all children with impaired consciousness that did not have blood glucose measurements taken. Severely anaemic patients (Hb < 5 g/dl) were transfused with whole blood. All patients were given quinine 10 mg/kg eight-hourly for a total of 7 days.
Data management and statistical analysis
Data was entered using Epi Data 2.1a (Odense, Denmark) and analysed with SPSS 11.5 (Chicago 2003). Frequencies of the different manifestations were computed. Patients were then categorized into 12-month age bands (0.0–11.9, 12–23.9, 24–35.9, etc.) and the effect of transmission intensity and age on the manifestations of severe malaria in the three areas determined. The geometric mean parasite density for each age band, the proportion of patients with seizures, impaired consciousness, severe anaemia, respiratory distress or other forms of severe disease in the three regions were compared.
A total of 617 children were enrolled from the three sites. Fifty-one patients were recruited from the very low (Kabale), 367 from moderate (Kampala) and 199 from very high (Busia) transmission areas. The mean age was 19.2 (SD 13.4) months and 339 (55%) were male. The majority of patients were <3 years of age, especially in Busia compared to Kampala or Kabale. Figure 2 shows the age distribution.
Most patients were brought to hospital after 2–3 days of fever, the commonest presenting symptom. Other common symptoms included vomiting, cough and diarrhoea. The median duration of symptoms before admission, and the proportions of patients with these symptoms were similar across all the three transmission zones. However, cough was disproportionately more common in Kabale (Table 2).
Table 2. Admission characteristics of children with severe malaria in low, moderate and high transmission areas in Uganda
Demographic, clinical and laboratory characteristics
Kabale (n = 51)
Kampala (n = 367)
Busia (n = 199)
* Includes both a reported history of seizures and seizures observed at admission.
Sex, males (%)
Median age (IQR) in months
Median duration of fever (IQR) in days
Anti malaria drug before admission (%)
Previous admissions due to malaria (%)
Mean admission axillary temp (SD) in °C
Palpable spleen (%)
Mean haemoglobin (SD) in g/dl
Geometric mean parasite density/μl
Treatment before admission
Over 60% of patients received anti malaria treatment before admission. In Busia more patients received prior anti malaria treatment than in either Kampala or Kabale (Table 2). Chloroquine was the most common drug used. Other drugs included Sulphadoxine-Pyrimethamine (alone or in combination with Chloroquine), Quinine, and Amodiaquine. However, only 29% of patients from Busia and 58% from Kampala received anti malaria drugs within 24 h of fever onset.
Previous admissions to hospital
We used a history of admissions due to malaria as a proxy measure for the risk of severe disease in each area. Over 25% of patients in Kampala, the area with moderate transmission, and 18% in Busia but none in Kabale reported previous admissions with severe malaria (P < 0.001). However, the proportion of patients with a history of admission for conditions other than malaria were similar in all three sites, suggesting that the finding in Kampala was not just due to easier access to healthcare or a higher index for admissions in this referral unit. Unfortunately, we didn't establish the age at which patients were previously admitted.
Age, parasite density and seizures
The median age at admission was inversely proportional to transmission intensity declining with rising transmission. Children from Kabale and Kampala were three and two times older than those from Busia respectively (Table 2). On the contrary, the geometric mean parasite density and the percentage of patients with seizures increased with rising transmission intensity. The geometric mean parasite density in Busia was six times that of Kabale and three times that of Kampala. Seizures were 3.5 and 2.5 times more common in Busia and Kampala compared to Kabale respectively, P < 0.001. The proportion of patients with repeated generalized seizures was also highest in Busia, P < 0.001 (Table 1).
In Busia, seizures were most prevalent in children 1–2 years with the proportion of those with seizures declining rapidly among 3–4 year olds. In the areas with either very low or moderate transmission, seizures were most common in children 2–3 years of age and the decline in the proportion of children with seizures was less precipitous than in Busia. The proportion of seizures in each age group mirrored the geometric mean parasite density in the three regions, suggesting that high parasite density and young age may predispose children to seizures, and that children in very high transmission areas may be most at risk. Figure 3 shows the association between transmission intensity, parasite density, age and seizures.
We performed a logistic regression analysis to determine which among transmission intensity, age, duration of illness, and delay in anti malaria drugs (no antimalarial within 24 h of fever) was independently associated with seizures. Of the above risk factors, only malaria transmission intensity (adjusted OR 1.9 95% CI 1.3–2.7, P < 0.001) was independently associated with a child presenting with seizures.
Manifestations of severe malaria
Of the three main indicators of life threatening illness in children with malaria, (Marsh et al. 1995) significant differences with transmission intensity were observed in the proportion of patients with severe anaemia and impaired consciousness but not respiratory distress (Table 1). Overall, 309/617 (50.1%) patients had at least one indicator of life threatening illness. These were distributed as 21/51 (41.2%) under very low, 171/367 (46.6%) under moderate and 117/199 (58.8%) under very high transmission (P < 0.01).
Malaria with impaired consciousness
A linear increase in the proportion of patients with impairment of consciousness (P = 0.029) with transmission intensity was observed and this proportion was highest in the very high transmission area (Table 1). The median age (months) of children with impaired consciousness followed the overall age distribution in the three areas and was inversely proportional to transmission intensity: 33.0 in very low, 27.0 in moderate and 9.0 in very high transmission areas. We performed logistic regression analysis to determine risk factors independently associated with impairment of consciousness. Results show that age, transmission intensity, history of convulsions and presentation with respiratory distress but not severe anaemia were independently associated (Table 3). Hypoglycaemia was not included in the model due to incomplete data.
Table 3. Factors independently associated with impaired consciousness*
Adjusted OR (95% CI)
* Logistic regression analysis using stepwise enter method. Presentation with impaired consciousness was the dependent variable. Variables entered included duration of illness, age, seizures, transmission area, respiratory distress and severe anaemia on admission.
Malaria with severe anaemia
The mean admission Hb was inversely proportional to the transmission intensity. It was lowest in Busia, which not only had the highest transmission intensity but also the youngest children. The proportion of patients with severe anaemia (P = 0.002) and palpable spleens was similarly highest in Busia (P < 0.001, Table 2). Jaundice was however much more common in Kabale or Kampala than in Busia, probably as a result of the older ages of these patients (Table 1).
Malaria with respiratory distress
No significant differences were observed in the proportion of patients with respiratory distress in the three areas. In each area, the median ages (months) of children with severe malarial anaemia (28.5 in low, 18.0 in moderate and 8.0 in high transmission) was lower than that of those with impaired consciousness (33.0 in low, 27.0 in moderate and 9.0 in high transmission) while across transmission zones, the median ages of children with impaired consciousness or severe anaemia under low transmission were higher than those of children from high transmission areas.
The overall mortality was 27/617 [4.4%, (95% CI 2.9–6.4)]. Mortality was highest in Busia [17/199, 8.5% (95% CI 5.0–13.7)], followed by Kampala [9/367, 2.5% (95%CI 1.2–4.8)] and Kabale [1/51, 2% (95% CI 0.01–10.9)], P < 0.001. Deaths were associated with either impaired consciousness or respiratory distress: 14/87 (16.1%) of all children with impaired consciousness died compared to 13/530 (2.5%) of those without (P < 0.001) and 16/159 (10.1%) of children with any degree of respiratory distress died compared to 11/458 (2.4%) of those without (P < 0.001; Table 4). Although the higher mortality in Busia could have been due to the high proportion of patients with life threatening manifestations (less severe cases were admitted in Kampala and Kabale), some deaths were associated with delays in offering blood transfusion to severely anaemic children due to irregular blood supply to this rural unit.
Table 4. Mortality of patients with indicators of life threatening illness
Indicators of Life threatening illness
* Age related tachypnoea with sustained nasal flaring only (mild distress) or with deep breathing or subcostal retractions (severe distress).
Any degree of distress*
Deep acidotic breathing
This study set out to determine the relationship between transmission intensity, parasite density and manifestations of severe malaria in children <5 years in three areas with different transmission intensities. We found that children in areas with very high malaria transmission have the highest admission parasite densities and proportion of patients with life threatening features. However, patients in areas with moderate transmission may experience more frequent bouts of severe malaria.
As in earlier reports, severe malaria under high transmission affected much younger children compared to areas with moderate or low transmission (Snow et al. 1994; Modiano et al. 1999). The high, early and repeated exposure to malaria results in low Hb levels and a high proportion with severe anaemia (Snow et al. 1994; Greenwood 1997; Snow et al. 1999). We observed a paradoxical relationship between malaria transmission intensity and admissions with severe malaria: a higher proportion of patients under moderate transmission, over 25%, reported previous admissions with malaria compared to only 18% under very high transmission. Similar relationships have been described in other parts of East and West Africa (Rogier et al. 1999; Snow et al. 1999). The apparently lower risk of severe malaria in very high transmission areas appears to be due to the development of clinical immunity early in life when other protective mechanisms may operate (Snow et al. 1997). In areas of high transmission, children also get continuous exposure to infection and quickly develop protective immunity against severe disease so that during subsequent episodes, they suffer less severe forms or uncomplicated malaria (Barragan et al. 1998; Gupta et al. 1999; Rogier et al. 1999). The sharp decline in the proportion of patients in the older age groups and the exceptionally low number of patients older than 3 years in Busia compared to Kampala or Kabale lends credence to this observation (Figure 2). Such immunity to particularly non-cerebral severe malaria is thought to develop within 1–2 seasons (Gupta et al. 1999), and is demonstrable after the age of 7 months under conditions of high malaria transmission (Slutsker et al. 1996; Kitua et al. 1997). But it is possible that longer child years of exposure to malaria in Kampala (higher mean age) may account for the higher proportion of children with a history of previous admissions. Unfortunately, we did not establish the age at which these admissions occurred. It is also possible that in areas with high transmission, due to the high number of cases, a child may either have to be more ill (e.g. with convulsions) or younger (of more concern) to be admitted. A tendency to admit young children and treat older children as outpatients would therefore confound the association between risk of disease (re-admission) and transmission intensity.
The proportion of patients with seizures increased with rising transmission intensity and parasite load. Earlier reports suggested that falciparum malaria parasites might particularly be epileptogenic (Asindi et al. 1993). Akpede and colleagues observed that children with high parasite densities had a higher risk of developing seizures. They concluded that seizures in children with malaria are more likely to be a manifestation of cerebral dysfunction than being simply febrile in nature (Akpede et al. 1993). Comparisons between different types of malaria parasites provided further evidence to suggest that Plasmodium falciparum parasites may be epileptogenic: seizures are more common in falciparum than in vivax malaria despite a similar febrile response (Wattanagoon et al. 1994). We too, observed a striking relationship between the geometric mean parasite density and the proportion of patients with seizures, and between intensity of transmission and presentation with seizures. Under conditions of high transmission children are exposed to a high number of infectious bites and develop correspondingly high parasite densities (Slutsker et al. 1996). The sequestration of such high parasite loads in cerebral vessels may be responsible for initiating pathological mechanisms that may result in seizure disorders (Gimenez et al. 2003).
Previous studies have suggested that severe malarial anaemia is more common in areas of high transmission, while cerebral malaria or impaired consciousness is more common in areas of low or unstable transmission (Snow et al. 1997,1999; Gupta et al. 1999). In the present study, both severe malarial anaemia and impaired consciousness were most common in the area with very high transmission. The study demonstrates that impaired consciousness may be an important complication even in areas of high transmission. Although this could have been a result of limiting the age of study children to 5 years (impaired consciousness is more frequent in older children), the majority of patients in Busia were under 3 years of age.
Interestingly, the frequency of respiratory distress was similar in all three areas, suggesting that this complication may not be related to parasite load or frequency of parasite exposure. It can, however, be difficult to differentiate malaria with respiratory distress from pneumonia; yet X-rays were not performed to exclude pneumonias. It is therefore possible that the absence of an association between respiratory distress and transmission intensity could have been due to misclassification of the two conditions.
Mortality was highest in Busia. Although each hospital was the referral centre for the local population for individuals with severe illness, there were differences in health unit grades, resources and personnel. Due to the lower grade of Masafu mini-hospital, and the irregular supply of life saving blood for transfusion, the higher mortality in Busia cannot necessarily be attributed to the very high transmission in the area. Despite this shortcoming, the high proportion of children with multiple indicators of life threatening illness in Busia may still be a reason for the higher mortality (Marsh et al. 1995). The majority of children in Busia were also under 3 years, the age group associated with the greatest severe malaria mortality (Schellenberg et al. 1999). It is unlikely that differences in the underlying health of children in the three communities are the reason for the variation in mortality.
The Roll Back Malaria initiative recommends that children with febrile illnesses in sub Saharan Africa should receive effective anti malaria drugs within 24 h of fever onset (Alnwick 2000). African heads of State and Governments adopted this resolution in the Abuja declaration (Abuja 2000). The recommendation stems from the observation that childhood severe malaria develops within 48 h of fever onset (Greenwood 1997). Only 29% of patients in Busia, a rural area, reported antimalarial medication within 24 h of fever, a far cry from the Abuja targets. The malaria control and advocacy programs in Uganda therefore still have a long walk especially in rural areas to achieve the set targets. A higher proportion of patients in urban Kampala received antimalarial drugs within 24 h. Unfortunately, despite over 60% of patients reporting use of antimalarial drugs before admission, patients still developed severe malaria. One will question the effectiveness of the first line antimalarial drugs in not only resolving symptoms of uncomplicated malaria but also preventing the development of severe disease especially in view of the rising problem of antimalarial drug resistance (Kamya et al. 2002; Bakyaita et al. 2005).
In conclusion, the frequency of severe malarial anaemia, seizures both simple and repeated, and impaired consciousness as manifestations of severe malaria in children under 5 years of age increase with rising transmission intensity and parasite load, while respiratory distress occurs with similar frequency in areas with very low, moderate and very high transmission. The median age of presentation with severe malaria decreases with increasing transmission intensity. Our findings suggest that heavy P. falciparum malaria parasitaemia may be important in the development of seizures, severe anaemia and impaired consciousness, but may not be necessary in development of respiratory distress.
We thank all the clinicians who cared for the children, the laboratory staff, Joshua Kayiwa the data manager and all our study participants. This study received financial support from Makerere University Faculty of Medicine Small Research Grants, the Rainbow Babies and Children's Hospital/Children's Research Foundation Faculty Fund Grants and DANIDA/Torch project in Child Health development Centre, Makerere University. We have no conflict of interest to declare.