Each year 500 million infections and up to 2.7 millions deaths are attributable to malaria [1], about 90% of these deaths occur in children in sub-Saharan Africa [2]. Eighty percent of the deaths occur during the first 24 hours following admission [3–5]. Despite a better understanding of pathophysiology and management of malaria, childhood mortality remains unacceptably high [6]. The acquisition of malaria immunity is closely linked to the level of transmission and severe Plasmodium falciparum infection is very rare after the age of 5 years in highly endemic areas. Presentations of severe malaria are different at different ages and in areas with different levels of transmission. In Gabon, a country of about 1.2 million people, malaria is the main cause of neurological, haematological and infectious emergencies at the Centre Hospitalier de Libreville (CHL), the country's tertiary referral centre [7]. Malaria transmission is hyperendemic and perennial with an entomological inoculation rate of 50 per person-year [8]. Previous studies in Africa have shown three frequent presentations of severe falciparum malaria: cerebral malaria, metabolic malaria (hyperlactaemia, acidosis or respiratory distress) and severe anaemia [3, 5, 9]. The case fatality rate of severe anaemia, however, is low and in some studies it is not an independent predictor of death [9, 10]. In Africa, malaria mortality remains high for a number of reasons including limited access to healthcare and increased drug resistance [6]. Better classification of severe malaria could aid clinicians caring for children with severe malaria to avoid diagnostic delays, identify the children most likely to die and thus improve management by targeting resources to the sickest children. This prospective observational study was designed to determine the clinical and laboratory features that identify those children most severely ill with malaria.

This study was carried out in the CHL between 1st August 2000 and 31st July 2002. The CHL is the largest public hospital in the country, situated in Libreville, the capital city (pop. 500,000). Services at the CHL include an Emergency Department (20 beds), a Paediatrics Unit (80 beds) and an Intensive Care Unit (12 beds). Ethical permission for the study was granted by the Gabonese Ministry of Health.

Febrile children were referred the study team and seen on admission by a Malaria Clinical Research Unit (MCRU) clinician, summary data were recorded on a pro-forma sheet. A blood sample (2 ml) was drawn (anti-coagulated with EDTA) for quantitative examination of a blood film for malarial parasites using the Lambaréné method [11], measurement of haemoglobin concentration, white cell count, platelet count (STKS, Coulter Corporation) and blood lactate and glucose concentrations. Within 15 minutes of blood sampling, blood lactate and glucose concentrations were measured using Accusport™ hand held analyser (Bohringer, Manheim, Inc., Germany) and One-touch Analyzer (LifeScan, Inc., USA) [11]. Blood films were defined as negative if there were no asexual forms of P. falciparum in 100 high power fields of a thick film. The schematic process of the inclusion is shown in Figure 1.

Febrile children (or those with history of fever in the last 48 hours) were considered for inclusion in the study if they were: aged 0 to 10 years of age (inclusive), had malaria (> 2 asexual forms of P. falciparum seen on blood film) and had one or more of the following features of severity: [9, 12, 13]: Blantyre coma score (BCS) ≤ 2 defining cerebral malaria, repeated observed seizures (3 or more observed in 24 hours), lactate concentration in whole blood or capillary blood ≥ 5 mmol/L, glucose concentration in whole blood or capillary blood ≤ 2.2 mmol/L, severe anaemia (haemoglobin concentration of < 5 g/dL) and/or haematocrit concentration < 15%). Children were excluded from the study if informed consent was not obtained from a relative or if an alternative diagnosis was made clinically or by investigation (such as cerebrospinal fluid examination, chest radiography or blood culture).

Respiratory distress was defined as the presence of one or more of these features [14]: abnormalities in respiratory rate (according to the age), rhythm (Kussmaul's or Cheyne-Stokes's breathing) and signs of distress such as nasal flaring, intercostal or subcostal recession.

Malaria remains a serious health problem in sub-Sahara Africa. It was the most common reason for neurological emergencies during 2001 at the CHL [7]. This study was designed to describe the epidemiology, clinical and laboratory presentations of severe falciparum malaria in childhood presenting to CHL, in order to improve the diagnosis, classification and appropriate management of malaria. It is not possible to exclude absolutely all children with alternative diagnoses on clinical examination and simple investigation alone, which is a problem shared by all other similar studies on severe malaria [3–5, 9, 10, 14, 16–18]. The small numbers with alternative diagnoses should not affect the conclusions of this or other studies. Most cases (92%) of severe malaria were in children less than 5 years old. Similar observations have been made in another group of children hospitalized for malaria in Gabon [19]. Elsewhere, severe malaria tends to occur in older children [20]. Differences in the age of presentation of severe malaria may be the result of lower background immunity or other undefined variables [21]. The study confirms the stable and perennial transmission of malaria in Gabon, which contrasts with reports from other countries in West Africa where malaria transmission is predominantly at the end of the long rainy season [22, 23].

Fever is a characteristic feature of P. falciparum infection, but a sizeable proportion of these children (25%) with severe malaria were afebrile on admission as observed elsewhere [23]. Self-medication with antipyretic or antimalarial agents was common (about 50% of the children) and may contribute to this finding. There are obvious implications for the diagnosis of malaria, which may be underestimated using clinical criteria alone.

Severe anaemia was the most frequent feature of severity in this study, but was associated with decreased mortality. A similar observation in a recent Ghanaian study showed a better outcome in children with severe anaemia [17]. These findings confirm that severe malaria anaemia has a lower case fatality rate than other complications of severe malaria, consistent with several other studies where severe anaemia was not an independent predictor of in-hospital mortality [9, 10, 18]. The case fatality rate of severe anaemia without other markers of severe malaria is 1 to 2%, where blood transfusion is available [3, 9, 10, 14, 18] raising questions about the value of severe anaemia as a defining feature of the syndrome of severe malaria.

Despite the increasing toll of HIV infection, and the continuing burden of diarrhoeal disease, malnutrition and respiratory tract infections, malaria remains a major cause of childhood death in endemic regions [1]. The overall case fatality rate of severe malaria in the study was 8.9% (52 deaths/583 cases), which is in keeping with studies from other geographic areas, where case fatality rates range between 8 and 40% [4, 5, 9, 14, 16, 20, 22, 24, 25]. Most of these deaths (90% in this study) occurred in the first twenty-four hours of hospital admission, a finding also in keeping with other studies [5].

The independent prognostic indicators in this study were cerebral malaria, respiratory distress, hypoglycaemia and hyperlactataemia. These observations are entirely consistent with a large number of studies where the independent predictors of a fatal outcome in malaria are impaired consciousness and metabolic dysfunction (as measured by hyperlactataemia, hypoglycaemia, acidosis or respiratory distress) [3, 5, 9, 10, 14, 16–18]. The metabolic complications of malaria are complex and a number of interrelated measures have been used in different studies. Severe malaria is associated with a metabolic acidosis [16] and hyperlactataemia [5]. Respiratory distress has been associated with acidosis and hyperlactataemia in some studies [26]. These features of metabolic malaria probably all result from increased anaerobic metabolism due to tissue hypoxia [27].

Estimates of the prevalence of hypoglycaemia have been reported in Africa, ranging from 8% to 34% [28, 29]. In severe childhood malaria hypoglycaemia results from impaired gluconeogenesis and increased tissue demand for glucose [27, 28] and quinine induced hyperinsulinaemia. Blood glucose concentrations should be monitored in all children hospitalised for malaria especially those who receive quinine.

The definition of hyperlactataemia used in this study was a blood lactate concentration higher than the conventional cut-off (≥ 5 mmol/L). This was necessary because of the limitations of the analyser used, but probably means that the frequency of true hyperlactataemia was underestimated. The Accusport™ analyser used has been shown to have poor agreement with "gold standard" machines [11, 15, 30].

Hyperlactataemia is a frequent and serious complication of severe malaria in childhood [5, 9, 10], which may be due microcirculatory sequestration of parasitized erythrocytes resulting in increased production of lactate by anaerobic glycolysis [31]. A recent study showed a correlation between hyperlactataemia and high plasma glutamine levels in severe malaria. This correlation may reflect impaired gluconeogenesis [31]. Lactate disposal is proportional to blood lactate concentration and can be increased by dichloroacetate [27, 32]. Lactic acidosis, as confirmed in this study, is an established strong predictor of a fatal outcome in falciparum malaria in African children [5, 9, 10] and may prove a target for further interventional studies to improve survival. Respiratory distress was present in 31% of these children. This is higher than the frequency reported in other studies of severe malaria: 4.9% in Burkina [20], 6.4% in Togo [22] and 13.7% in Kenya [14].

These differences may partly be explained by low inter-observer agreement for this variable, geographical variations in disease pattern as well as differing definitions of severe malaria. Results from many studies consistently show that respiratory distress is a life-threatening syndrome in childhood malaria [14, 33]. Respiratory distress was significantly associated with both hyperlactataemia and cerebral malaria in this study. The Blantyre coma score has long been established in children as a good indicator of cerebral dysfunction in malaria [3] and has enabled better standardization of studies on cerebral malaria in African children. The case-fatality rate associated with cerebral malaria (22%) is similar to that in Gambian children (27%) [4] but is higher than that observed for Kenyan (17%) [14] and Malawian children (15%) [3]. It has been postulated that with the higher the level of malaria transmission, immunity is acquired earlier, perhaps altering the presentation of severe malaria from predominantly a cerebral syndrome to that of severe anaemia [34, 35].

The clinical and laboratory presentations of severe malaria are described in a hospitalized population of children in Gabon. The severe cases are likely to be only the "tip of the iceberg", many children living far from health care units may die whilst travelling to the nearest hospital. Most deaths from malaria occurred in the first 24 hours of admission, which highlights the need for early recognition of the most severely ill children. Early diagnosis and classification of severe malaria would allow appropriate management, including basic adjunctive therapy such as to prevent hypoglycaemia, and better use of scarce healthcare resources. Together these improvements could contribute to a reduction in the intractably high mortality due to the disease.