Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Clinical Features and Outcome in Children with Severe Plasmodium falciparum Malaria: A Meta-Analysis

  • Laurens Manning ,

    laurens.manning@uwa.edu.au

    Affiliation School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia, Australia

    ⨯
  • Moses Laman,

    Affiliation School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia, Australia

    ⨯
  • Wendy A. Davis,

    Affiliation School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia, Australia

    ⨯
  • Timothy M. E. Davis

    Affiliation School of Medicine and Pharmacology, University of Western Australia, Fremantle Hospital, Fremantle, Western Australia, Australia

    ⨯

Abstract

Background

Although global malaria mortality is declining, estimates may not reflect better inpatient management of severe malaria (SM) where reported case fatality rates (CFRs) vary from 1–25%.

Methods

A meta-analysis of prospective studies of SM was conducted to examine i) whether hypothesized differences between clinical features and outcome in Melanesian compared with African or Asian children really exist, and ii) to explore temporal changes in overall and complication-specific CFRs. The proportions of different SM complications and, overall and complication-specific CFRs were incorporated into the meta-analysis. Adjustments were made for study-level covariates including geographic region, SM definition, artemisinin treatment, median age of participants and time period.

Findings

Sixty-five studies were included. Substantial heterogeneity (I2>80%) was demonstrated for most outcomes. SM definition contributed to between-study heterogeneity in proportions of cerebral malaria (CM), metabolic acidosis (MA), severe anemia and overall CFR, whilst geographic region was a significant moderator in for CM and hypoglycemia (HG) rates. Compared with their African counterparts, Melanesian children had lower rates of HG (10% [CI95 7–13%] versus 1% [0–3%], P<0.05), lower overall CFR (2% [0–4%] versus 7% [6–9%], P<0.05) and lower CM-specific CFR (8% [0–17%] versus 19% [16–21%], P<0.05). There was no temporal trend for overall CFR and CM-specific CFR but declining HG- and MA- specific CFRs were observed.

Interpretation

These data highlight that recent estimates of declining global malaria mortality are not replicated by improved outcomes for children hospitalized with SM. Significant geographic differences in the complication rates and subsequent CFRs exist and provide the first robust confirmation of lower CFRs in Melanesian children, perhaps due to less frequent HG.

Citation: Manning L, Laman M, Davis WA, Davis TME (2014) Clinical Features and Outcome in Children with Severe Plasmodium falciparum Malaria: A Meta-Analysis. PLoS ONE 9(2): e86737. https://doi.org/10.1371/journal.pone.0086737

Editor: Quique Bassat, University of Barcelona, Spain

Received: September 24, 2013; Accepted: December 14, 2013; Published: February 6, 2014

Copyright: © 2014 Manning et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This study was funded by a National Health and Medical Research Council (NHMRC) grant (#513782). LM was supported by a Basser scholarship from the Royal Australasian College of Physicians and an NHMRC scholarship, ML by a Fogarty Foundation scholarship, and TMED by an NHMRC Practitioner Fellowship. No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Although most of the 500 million clinical cases of malaria due Plasmodium falciparum estimated to occur globally are uncomplicated [1], some (1–4%) are considered severe enough to require hospitalization and/or parenteral antimalarial treatment [2]. In children, the term ‘severe malaria’ is currently applied when asexual forms of P. falciparum are detected in the peripheral blood and there is impaired consciousness or coma, prostration, multiple seizures, hyperlactatemia or metabolic acidosis, severe anemia, dark urine, hypoglycemia, jaundice, respiratory distress, persistent vomiting, abnormal bleeding, shock, and/or renal failure [3]. These clinical criteria reflect the World Health Organization (WHO) 2000 definition of severe malarial illness [3] that is based on descriptive studies of African children and Asian adults with severe P. falciparum infections. It has evolved from previous iterations, published in 1986 [4] and 1990 [5], that were developed to identify patients most at risk of death and to provide standard criteria for research studies [3].

Both the proportions of children with different clinical features defining severity and their subsequent case fatality rates (CFRs) will depend on selection/recruitment strategies, the sensitivity of malaria diagnosis/speciation and the definitions of the features themselves, as well as inpatient management including identification of co-incident disease, and social, cultural and genetic factors that could influence presentation and clinical course. Apparent inconsistencies between published studies may reflect differences in one or more such factors. Indeed, reported CFRs in severe childhood malaria studies vary from 1–25% [6], [7], with the suggestion that Melanesian children have a better prognosis than African children [8], [9]. However, there has also been a temporal change in global malaria-associated mortality, with a peak in 2004 and subsequent fall to an estimated 1,238,000 deaths in 2010 [10]. These estimates have been derived from vital registration (VR) of cause of death and verbal autopsy (VA) studies, augmented by estimates of the contribution of malaria to non-specific causes of death such as anemia and fever based on local malaria epidemiology. Case fatality rates (CFRs) of children hospitalized with severe malaria are, however, not incorporated into overall malaria mortality estimates. Because of this, it cannot be inferred that improved inpatient management of such cases has contributed to the recent decline in malaria-attributable mortality [10].

In the light of this background, we performed a standardized review and meta-analysis to examine whether i) there are significant differences between clinical features and outcome in Melanesian children with severe malaria and those in cases from Africa or Asia, and ii) there have been changes in overall and complication-specific CFRs over time. We incorporated geographic region, the year in which the study was performed, the case definition used, patient age and antimalarial therapy as potential moderators of outcome.

Methods

The supporting PRISMA checklist is available as Checklist S1.

Search strategy and selection criteria

A search of all descriptive studies of severe childhood falciparum malaria published before the end of 2012 was conducted using Medline. The following MeSH terms were employed: “child*” OR “pediatric” OR “paediatric” AND “severe malaria” OR “cerebral malaria” OR “severe anemia” OR “severe anaemia” OR “severe malarial anemia” OR “severe malarial anaemia” OR “metabolic acidosis” OR “lactic acidosis”. The bibliographies of key papers were reviewed and additional studies obtained. Publications in French were included but those describing children with severe P. vivax or mixed P. vivax/P. falciparum malaria were not.

The definition of severe malaria was categorized as ‘all malaria requiring admission’, ‘clinician-determined severe malaria’, WHO 1990 [5] and WHO 2000 [3]. The latter two definitions were applied if explicitly stated or if the clinical features were consistent with the published criteria. Unless otherwise stated, included studies were prospective with observational, case-control or interventional designs and with clearly defined clinical features of severity and/or reported CFRs. In antimalarial treatment trials, the proportions of children with specific features of severity in each treatment arm were pooled but CFRs were calculated by allocated treatment with artemisinin derivatives or with quinine/aminoquinoline therapy. Studies were grouped according to whether they were conducted in Africa, Asia or Papua New Guinea (PNG; Melanesia). The median or mean age of children reported for each study was dichotomized as <4 or ≥4 years. Analysis of temporal trends was based on the year the study was conducted, or the median for those running for more than a year, rather than the year of publication.

Studies focusing on children with only severe malarial anemia, cerebral malaria, hypoglycemia and metabolic acidosis were included only in CFR analyses of the relevant specific clinical feature. The term ‘deep coma’, ‘unrousable coma’, ‘comatose not responding to painful stimuli’ were considered synonymous. This equated to a Blantyre Coma Score (BCS) [11] of ≤2 in most instances, but deep coma was defined as a BCS ≤3 in three studies [12]–[14] and, in another [15], the Adelaide coma scale was used. For publications reporting ‘deep coma’, subgroups of children with BCS scores of 0, 1 and 2 were also extracted when available.

Definitions of severe malarial anemia (hemoglobin <50 g/L) and hypoglycemia (blood/plasma glucose <2.2 mmol/L) were uniform across time, geographic region and WHO definition. Either hyperlactatemia (blood/plasma lactate ≥5.0 mmol/L), or serum bicarbonate ≤15 mmol/L or ≤12.2 mmol/L [16] or base excess <−8 mmol/L, were considered indicative of metabolic acidosis. A severe overlap syndrome of deep coma, severe anemia, and metabolic acidosis, hyperlactemia or respiratory distress, in a single child was reported in a number of studies and was, therefore, considered a separate presenting clinical feature.

Statistical analysis

The R statistical package ‘metafor’ was used for meta-analyses [17]. Briefly, all outcomes of interest were considered proportions for random effects models. Study level covariates or ‘moderators’ including geographic region, the definition of severe malaria used, treatment with an artemisinin derivative, patient age and the year of study were incorporated as appropriate into mixed effects models. If a variable of interest in an individual study had a value of 0 but was measured, a value of 0.01 was assumed to allow its inclusion in the meta-analysis.

Heterogeneity was determined using the I2 statistic, a measure of the percentage of total variability due to between-study heterogeneity. Values <50% suggest a lack of significant heterogeneity. The τ2 statistic is a measure of total heterogeneity before and after incorporation of moderator variables, allowing the contribution of moderator variables in mixed effects models to be quantified as a relative proportion of the overall between-study heterogeneity. The mixed effects model with the best fit for each outcome of interest was determined by comparison between derived models using backward stepwise testing, incorporating moderators with a statistically significant effect on heterogeneity, and by using Aikake's Information Criterion and analysis of variance testing at each step to identify the most parsimonious model.

The results of random effect models are depicted using forest plots. In each such plot, the study is denoted by author and year of publication with the exception of the Severe Malaria in African Children (SMAC) studies [18] and the important studies comparing artesunate with quinine therapy in Asia and in African children (denoted as AQUAMAT [19] and SEQUAMAT [20], respectively). Publication bias was assessed by visual inspection of funnel plots. For each outcome of interest, funnel plots with (the final mixed effects model) and without (random effects) moderator effects were generated. Meta-regression plots for outcomes of interest according to the year of the study were also generated displaying a meta-regression line and 95% confidence intervals (CI95).

Registration

This study is registered with PROSPERO (CRD42013005510).

Results

Studies included in the meta-analysis

The initial Medline search yielded 4,629 citations. A PRISMA [21] flow diagram outlining the determination of which studies were included in the meta-analysis is shown in Figure 1. Sixty-five studies were included (see Table 1) [2],[6],[7],[11]–[15],[19],[20],[22]–[75]. One was reported as a retrospective study, but the data were of good quality and collected prospectively [30]. Most (n = 55) were from Africa, with 6 and 4 studies from Asia and PNG, respectively. The SMAC network of studies contributed 10 data points to the meta-analysis. The data on proportions of different clinical features from the 6 SMAC sites are published in a single paper [73], while additional data from the same study sites are detailed elsewhere [36], [44]. One study included data in two publications from the same cohort [74], [75].

thumbnail
Download:
Figure 1. Flow diagram showing studies included in meta-analysis.

https://doi.org/10.1371/journal.pone.0086737.g001

thumbnail
Download:
Table 1. Description of potential moderator covariates and the number of studies with individual moderators included in meta-analysis.

https://doi.org/10.1371/journal.pone.0086737.t001

The meta-analysis included studies performed from 1987 to 2009 (median 1999) with average patient age ranging from 1 to 8 (median 3) years. Three highly cited articles, all of which reported studies of Melanesian patients, were captured by the search strategy but excluded from analysis. The first, a retrospective study of morbidity surveillance in a cohort of PNG children, did not have detailed data available for ascertainment of clinical manifestations and did not report overall or phenotype-specific CFRs [76]. In the second study, in which all malaria hospitalizations in Irian Jaya, Indonesia were prospectively identified [77], it was not possible to determine the proportions of each clinical manifestation because the group with multiple features was not defined. Furthermore, overall and phenotype-specific CFRs were not reported according to age and Plasmodium species, making it difficult to identify children with P. falciparum. The third study reported clinical data collected from children recruited to the SEQUAMAT intervention trial [78]. Because patients were stratified in 10-year age groups, it was not possible to extract important clinical data that supplemented those provided in the original report [20].

Of the studies included, only one systematically reported HIV status in children with severe malaria. Blood cultures were taken from at least some of the children with severe malaria in 7 studies [2], [12], [41], [47], [65], [79], but were routinely performed and reported in only one [2]. The proportions of deeply comatose children with BCS of 0, 1 and 2 were presented in 14, 13 and 13 studies, respectively, but the BCS-specific CFRs were only provided in 6, 3 and 3 studies, respectively.

Heterogeneity in descriptive studies of severe malaria

There was substantial heterogeneity (I2>80%) for all but three outcomes of interest. The exceptions were the proportions and CFR of children presenting with the overlap syndrome (I2 71.9 and 0%, respectively) and CFR for children presenting with a BCS of 0 (I2 0%). The overall heterogeneity (I2), total estimated heterogeneity (τ2) and the effect of moderators on τ2 for the proportions of clinical features in descriptive studies are shown in Table 2. The same metrics for overall and feature-specific CFRs, and for BCS scores, are shown in Tables 3 and 4, respectively. Funnel plots for each outcome of interest did not indicate publication bias after the effects of moderators were incorporated into final mixed effects models (data not shown).

thumbnail
Download:
Table 2. Heterogeneity and the effect of moderators on heterogeneity for the proportions of children presenting with different clinical features in descriptive studies of severe Plasmodium falciparum malaria.

https://doi.org/10.1371/journal.pone.0086737.t002

thumbnail
Download:
Table 3. Heterogeneity and the effect of moderators on heterogeneity for overall and clinical feature-specific case fatality rates in descriptive studies of severe Plasmodium falciparum malaria.

https://doi.org/10.1371/journal.pone.0086737.t003

thumbnail
Download:
Table 4. Heterogeneity and the effect of moderators for proportions and case fatality rates according to Blantyre Coma Score in deeply comatose children from descriptive studies of severe Plasmodium falciparum malaria.

https://doi.org/10.1371/journal.pone.0086737.t004

The definition of severe malaria used in each of the studies had a significant impact on heterogeneity for the proportions of deep coma, metabolic acidosis and severe anemia and hypoglycemia, accounting for 50%, 45% and 14% of total heterogeneity, respectively.

Geographic region was a significant moderator in the proportion of deep coma and hypoglycemia, accounting for 10% of heterogeneity in each. For the proportion with deep coma, region was not included in the most parsimonious mixed effects model that had definition of severe malaria and average patient age as the only significant moderators. When compared with African children, the proportion of children presenting with hypoglycemia was significantly lower in PNG children (10% [CI95 7–13%] versus 1% [0–3%], P<0.05; see Table 5).

thumbnail
Download:
Table 5. Geographic differences in the clinical features and outcome for severe malaria (data are proportions [%] with 95% confidence intervals [CI95] from random effects models).

https://doi.org/10.1371/journal.pone.0086737.t005

Forty-five studies reported overall CFRs. A forest plot showing random effects models for overall mortality according to region is shown in Figure 2. The definition of severe malaria and region were significant moderators of heterogeneity, accounting for 11% and 16%, respectively. The final mixed effects model contained both variables. After adjustment for the definition of severe malaria used, the overall CFR was lower in PNG children when compared with African children (2% [0–4%] versus 7% [6–9%], P<0.05; see Table 5). Geographic region was also a significant moderator of heterogeneity in CFRs due to deep coma and severe anemia, accounting for 27% and 26% of total heterogeneity, respectively. PNG children had a significantly lower mortality due to cerebral malaria than African children (8% [0–17%] versus 19% [16–21%], P<0.05), while Asian children also had a higher mortality due to severe anemia than African children.

thumbnail
Download:
Figure 2. Forest plot showing random effects model of overall mortality due to severe malaria according to region.

https://doi.org/10.1371/journal.pone.0086737.g002

The average patient age in studies included in the meta-analysis was a significant moderator and was incorporated into the final mixed effects model for proportion of deep coma in severe malaria (18% of total heterogeneity) and mortality due hypoglycemia (33% of total heterogeneity). Artemisinin use was not a variable in any final mixed effect model for any outcome of interest.

For children with deep coma, the proportions in BCS categories were highly heterogeneous, with at least some of the heterogeneity accounted for by the year the study was performed. There were no apparent regional differences in the proportions of children with BCS of 0, 1 or 2, although no Asian studies were represented. The CFRs for children with a BCS of 0 had no heterogeneity (Table 4; CFR 38% [27–49%]). As a consequence, there were no regional differences observed. However, although limited to an analysis of 3 studies only, PNG children with a BCS of 2 had a lower mortality than African children (Table 5; 0% [0–4%] versus 7% [1–13%], P<0.05).

Temporal trends in severe malaria mortality rates

There was no significant temporal trend in overall CFR (Figure 3A). The year in which the study was done was a significant moderator and contributed significantly to heterogeneity for hypoglycemia- and metabolic acidosis- associated CFRs, accounting for 35% and 45% of total heterogeneity, respectively (Figures 3C and 3D, respectively). The meta-regression line demonstrates declining CFRs due to hypoglycemia and metabolic acidosis. This is in contrast to the CFR due to deep coma that has remained constant over the past 30 years (Figure 3B).

thumbnail
Download:
Figure 3. Temporal trends in overall case fatality rates (CFR) due to severe malaria (figure 3A) deep coma (figure 3B), hypoglycemia (figure 3C) and metabolic acidosis (figure 3D) (dashed lines represent 95% confidence intervals of predicted mortality rates).

https://doi.org/10.1371/journal.pone.0086737.g003

Discussion

The present systematic review and meta-analysis of descriptive and interventional studies of severe childhood P. falciparum malaria demonstrates that there are significant regional differences in the proportions of clinical features and outcome. PNG children present less frequently with hypoglycemia than African children, and they have lower CFRs overall and due to cerebral malaria. Asian children have a higher CFR associated with severe anemia than African children. The present analyses also show that, in contrast to population studies of malaria-attributable mortality [10], there has been no reduction in overall CFR or cerebral malaria-associated CFR over time despite declining CFRs due to both hypoglycemia and metabolic acidosis.

There was substantial overall heterogeneity between studies for nearly all outcomes that was only partly accounted for by moderator covariates. The definition of severe malaria used in each study had a large effect on heterogeneity for overall CFR and the relative proportions of deep coma, severe anemia and metabolic acidosis. When compared to ‘all malaria admissions’, the WHO 1990 definition had increased rates of deep coma, while metabolic acidosis and severe anemia were more frequent under the WHO 1990 and WHO 2000 definitions. This finding reflects the fact that the WHO 2000 definition and ‘all admissions’ have least restrictive inclusion criteria. The inclusion of prostration, impaired consciousness and at least two seizures in the WHO 2000 definitions are relevant examples.

Regional differences accounted for at least some of the overall heterogeneity between studies. When compared with their African counterparts, PNG children (in the only representative studies from Melanesia or Oceania) had lower overall and cerebral malaria-specific CFRs. The proportions of children from PNG with deep coma who had a BCS of 0, 1 or 2 were comparable to those in African studies. The CFRs for a BCS of 0 were extremely high but without heterogeneity. For children with BCS of 2, the CFR was lower in PNG children. These findings reflect epidemiological studies, historical information [80] and other data collected from Vanuatu and elsewhere in Melanesia in the 1980s and 1990s [8], [9]. Even though a Melanesian survival advantage is not universally endorsed [81], our analyses are the most robust to date and may reflect factors such as widespread availability of effective antimalarial therapy, immunological cross-protection from exposure to P. vivax, and a high prevalence of protective genetic polymorphisms such as South Asian Ovalocytosis (SAO) [82]–[84].

Locally prevalent protective genetic polymorphisms are, however, unlikely to account for the observed differences in CFRs. First, SAO prevalence is highly heterogeneous and been identified in coastal PNG peoples with Austronesian language lineage, but not elsewhere in Melanesia [85]. Second, and more importantly, the selection pressure maintaining this polymorphism in the population is protection against incident uncomplicated infection and possibly severe disease due to P. vivax rather than a survival benefit mediated via protection against cerebral malaria due to P. falciparum [86]. Other undiscovered novel local genetic polymorphisms may underlie the low Melanesian CFRs but, despite advanced bio-informatic and high-throughput molecular technologies, may be hard to identify because of the low number of outcomes and the consequent need for very large sample sizes.

Cross-sectional and longitudinal surveys in Melanesia have shown negative correlations between the prevalence of different Plasmodium species [82], [87], and mixed-species infections tend to be substantially less common in symptomatic compared to asymptomatic patients [87]. Epidemiologic evidence from Thailand has also shown that the incidence of severe malarial illness, but not mortality, is lower in patients with P. falciparum when there is a mixed infection with P. vivax [88]. In addition, P. vivax co-infection appears to abrogate the nadir in hemoglobin after treatment for falciparum malaria [89]. These observations have been interpreted as evidence that P. vivax protects against P. falciparum clinical disease [9], [88]. However, recent clinical data showing more severe clinical features and a substantially higher CFR in children with evidence of mixed P. falciparum/P.vivax infections in PNG is inconsistent with this concept [2].

Two alternative explanations for the low CFRs in Melanesian children with severe malaria include the low rates of hypoglycemia in this setting, as demonstrated in the present meta-analysis, and a more recent observation that concomitant bacteremia, particularly non-typhoidal salmonellae (NTS), occurs much less frequently than in African children with severe malaria. A very low prevalence of hypoglycemia in Melanesian children with severe malaria (1%) might account for lower CFRs observed in comatose children as well as overall CFR. Our analyses suggest that an absolute difference in hypoglycemia between PNG and other children of 8% might account for only 2% of the differences in overall CFR. Differences in nutritional status and hepatic glycogen reserves, pre-referral treatment with quinine with attendant hyperinsulinemia and as yet undiscovered genetic polymorphisms in glucose metabolism are possible explanations for geographical variability in hypoglycemia rates.

Concomitant bacteremia in severe childhood malaria in African children is well described, occurring in 5–16% of children [90]–[96], and is an independent risk factor for mortality with associated CFRs of up to 33%. NTS are an important cause of concomitant bacteremia in Africa, accounting for 7–58% of bacterial isolates and reflecting malaria-associated down-regulation of humoral and cellular immunity and/or disruption of the gut mucosa by microvascular sequestration [97]. Variability in the proportions of concomitant bacteremia and causative organisms between studies could be due to differences in local bacterial epidemiology (such as S. aureus nasal carriage rates or circulating S. pneumoniae serotypes), HIV seroprevalence and sickle cell disease rates, pre-treatment with antibiotics, and nutrition status [97]. Due to logistic challenges and cost in resource-poor settings, few studies have performed blood cultures systematically in children with severe malaria. Indeed, of the 65 studies in the meta-analysis, only 7 reported that blood culture were performed, usually as deemed clinically necessary rather than as a routine investigation. Although limited to a single study, a striking absence of concomitant bacteremia in admission blood cultures in children with severe malaria was observed recently in PNG [2]. In addition, no NTS were isolated from blood cultures from other children presenting with severe non-malarial illness [2]. This suggests that, in contrast to Africa where culture-positive NTS can be found in up to 4% of healthy children [98], these pathogens are uncommon in PNG and perhaps other parts of Melanesia. HIV prevalence in malaria endemic areas could also confound the results of the present study. However, standardised testing for HIV was performed in only one study and further analysis was not possible as a result.

There were no significant temporal trends for overall CFR or for CFRs due to deep coma and severe anemia in the present study. In addition, artemisinin use was not a significant moderator of any outcome of interest. These observations appear surprising in the light of recent models demonstrating that estimates of global malaria-attributable mortality have declined since 2004 [10], and the publication of the AQUAMAT [19] and SEQUAMAT [20] studies that showed a mortality benefit for artesunate over quinine in Asian and African children with severe malaria. However, models showing declining global malaria mortality do not incorporate inpatient data despite the fact that a significant proportion of children dying in tropical countries present to a health facility during their final illness [99]. Mortality estimates based on VA and VR may have limited validity in malaria-endemic countries [100], [101], while declining malaria-attributable mortality in these models are likely to reflect the success of malaria control programs at a population level rather than better in-patient care.

The lack of change in CFR due to cerebral malaria over recent decades could reflect diagnostic uncertainty, with a true decline masked by misdiagnosis of other causes of encephalopathic illness as cerebral malaria [102]. Alternatively, a significant proportion of children may present late in their illness with complications that are unresponsive to optimal resuscitation, and antimalarial and supportive therapy. In the present study, the lack of heterogeneity for, and magnitude of, the CFR for children presenting with a combination of cerebral malaria, severe anemia and evidence of metabolic acidosis highlights this point. A lack of heterogeneity was also observed for CFR in children with a BCS of 0. Regardless of geographic region or other moderators, the CFR for these specific groups of moribund children were 32% and 38%, respectively. On a more positive note, our data demonstrate significant improving temporal trends for CFRs due to metabolic acidosis and hypoglycemia. This is likely to reflect incorporation of routine use of intravenous 10% w/v dextrose and judicious use of fluid resuscitation [103].

Artemisinin use was not a significant moderator of any outcome of interest in the present analyses. Only one third of the studies involved an artemisinin derivative, nearly all of which were comparative intervention trials. The artemisinin formulations used varied widely in terms of derivative (artemether, artesunate or artemether) and route of administration (intramuscular, intravenous or rectal). Due to the limited number of studies and the different treatment regimens, we pooled the data for all preparations and routes of administration. This may have masked a beneficial effect of intravenous artesunate that was clearly demonstrated in AQUAMAT [19] and SEQUAMAT [20]. Despite this possible limitation, it is important to note that, even in a research setting, the overall mortality for African children treated with parenteral artesunate was still 8.5% and the CFR for deep coma was 18% [19]. Although both lower than observed for quinine treatment, these CFRs are consistent with the temporally stable respective mean estimates shown in Figure 3.

Our study had other limitations. First, the four Melanesian studies eligible for inclusion were from the same regional hospital in PNG. Each study was independent and conducted at different times, but the applicability of the present findings to other Melanesian geographical areas is unknown. Second, because of methodological issues described in the Results section above, two studies of severe malaria in Melanesian children were excluded from the analysis [76], [77]. From the tabulated data provided by Tjitra et al. [77], severe anemia and coma occurred in 20% and 1.7%, respectively, of all children admitted with malaria. Although these percentages are inaccurate because children with more than one severe manifestation were not counted, the rate of severe anemia is consistent with other Melanesian studies. By contrast, the percentage with deep coma appears lower than our data but may have been up to ∼3% if those children with more than one clinical manifestation had been included. The overall CFR for patients across all age groups with severe falciparum malaria was 2.2%, lower than in studies of patients from an Asian ethnic background. After adjustment for other variables, Asian ethnicity did not remain an independent risk factor for mortality, but the mortality trends for a predominantly Melanesian population in the Tjitra et al. study [77] accord with the low mortality observed in our meta-analysis. In the study published by Genton et al. [76], the limited ascertainment of clinical features and absence of CFRs precluded its incorporation in the present analysis. It is important to note that, in both excluded studies, blood glucose concentrations were not recorded. Therefore, neither would have influenced the striking regional differences observed in rates of hypoglycemia. A final possible limitation is that there were no studies from South America and, although we included studies in French, we did not search the Portuguese or Spanish literature. This may limit the generalizability of our results beyond Africa, Asia and Oceania, but it is unlikely that the omission of non-English, non-French literature would have impacted on our main findings [104].

In conclusion, the present meta-analysis supports the contention from individual studies and observations that there is a lower CFR for PNG children when compared with African children presenting with severe malaria. Although this may be mediated in part by low rates of hypoglycemia and concomitant bacteremia, there is still substantial heterogeneity across all studies. Although there is good evidence that CFRs due to hypoglycemia and metabolic acidosis have been declining, the lack of temporal change in overall and cerebral malaria CFRs in an era of improving access to parenteral artesunate should be a catalyst for renewed efforts to refine the diagnosis of severe malaria, improve pre-referral practices, optimise inpatient management and continue the search for effective adjunctive therapies.

Supporting Information

Checklist S1.

The supporting PRISMA checklist.

https://doi.org/10.1371/journal.pone.0086737.s001

(DOC)

Author Contributions

Conceived and designed the experiments: LM TMED. Performed the experiments: LM ML. Analyzed the data: LM WD. Contributed reagents/materials/analysis tools: WD. Wrote the paper: LM TMED.

References

  1. 1. Snow RW, Guerra CA, Noor AM, Myint HY, Hay SI (2005) The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature 434: 214–217.
  2. 2. Manning L, Laman M, Law I, Bona C, Aipit S, et al. (2011) Features and prognosis of severe malaria caused by Plasmodium falciparum, Plasmodium vivax and mixed Plasmodium species in Papua New Guinean children. PLoS One 6: e29203.
  3. 3. World Health Organisation (WHO) (2000) Severe falciparum malaria. Trans R Soc Trop Med Hyg 94 Suppl 1: S1–90.
  4. 4. World Health Organisation (WHO) (1986) Severe and complicated malaria. World Health Organization Malaria Action Programme. Trans R Soc Trop Med Hyg 80 Suppl: 3–50.
  5. 5. World Health Organisation (WHO) (1990) Severe and complicated malaria. World Health Organization, Division of Control of Tropical Diseases. Trans R Soc Trop Med Hyg 84 Suppl 2: 1–65.
  6. 6. Karunajeewa HA, Reeder J, Lorry K, Dabod E, Hamzah J, et al. (2006) Artesunate suppositories versus intramuscular artemether for treatment of severe malaria in children in Papua New Guinea. Antimicrob Agents Chemother 50: 968–974.
  7. 7. Imbert P, Yen KK, Migliani R, Sokhan C, Martin YN, et al. (1994) [Malaria in children at the Sihanoukville Hospital (Cambodia)]. Bull Soc Pathol Exot 87: 253–260.
  8. 8. Bastien P (1999) More on Malaria Mortality in the Pacific. Parasitol Today 15.
  9. 9. Maitland K, Williams TN (1998) Malaria mortality: the pacific enigma. Parasitol Today 14: 258–259.
  10. 10. Murray CJ, Rosenfeld LC, Lim SS, Andrews KG, Foreman KJ, et al. (2012) Global malaria mortality between 1980 and 2010: a systematic analysis. Lancet 379: 413–431.
  11. 11. Molyneux ME, Taylor TE, Wirima JJ, Borgstein A (1989) Clinical features and prognostic indicators in paediatric cerebral malaria: a study of 131 comatose Malawian children. QJM 71: 441–459.
  12. 12. Imbert P, Gerardin P, Rogier C, Ka AS, Jouvencel P, et al. (2002) Severe falciparum malaria in children: a comparative study of 1990 and 2000 WHO criteria for clinical presentation, prognosis and intensive care in Dakar, Senegal. Trans R Soc Trop Med Hyg 96: 278–281.
  13. 13. Crawley J, Waruiru C, Mithwani S, Mwangi I, Watkins W, et al. (2000) Effect of phenobarbital on seizure frequency and mortality in childhood cerebral malaria: a randomised, controlled intervention study. Lancet 355: 701–706.
  14. 14. Di Perri G, Di Perri IG, Monteiro GB, Bonora S, Hennig C, et al. (1995) Pentoxifylline as a supportive agent in the treatment of cerebral malaria in children. J Infect Dis 171: 1317–1322.
  15. 15. Varandas L, Julien M, Van Lerberghe W, Goncalves L, Ferrinho P (2000) Independent indicators of outcome in severe paediatric malaria: maternal education, acidotic breathing and convulsions on admission. Ann Trop Paediatr 20: 265–271.
  16. 16. Manning L, Laman M, Townsend MA, Chubb SP, Siba PM, et al. (2011) Reference intervals for common laboratory tests in melanesian children. Am J Trop Med Hyg 85: 50–54.
  17. 17. Viechtbauer W (2010) Conducting meta-analyses in R with the metafor package.. Journal of Statistical Software 36: 1–48.
  18. 18. Helbok R, Kendjo E, Issifou S, Lackner P, Newton CR, et al. (2009) The Lambarene Organ Dysfunction Score (LODS) is a simple clinical predictor of fatal malaria in African children. J Infect Dis 200: 1834–1841.
  19. 19. Dondorp AM, Fanello CI, Hendriksen IC, Gomes E, Seni A, et al. (2010) Artesunate versus quinine in the treatment of severe falciparum malaria in African children (AQUAMAT): an open-label, randomised trial. Lancet 376: 1647–1657.
  20. 20. Dondorp A, Nosten F, Stepniewska K, Day N, White N (2005) Artesunate versus quinine for treatment of severe falciparum malaria: a randomised trial. Lancet 366: 717–725.
  21. 21. Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ (Clinical Research Ed) 339: b2535.
  22. 22. Aceng JR, Byarugaba JS, Tumwine JK (2005) Rectal artemether versus intravenous quinine for the treatment of cerebral malaria in children in Uganda: randomised clinical trial. BMJ 330: 334.
  23. 23. Achan J, Byarugaba J, Barennes H, Tumwine JK (2007) Rectal versus intravenous quinine for the treatment of childhood cerebral malaria in Kampala, Uganda: a randomized, double-blind clinical trial. Clin Infect Dis 45: 1446–1452.
  24. 24. Adam I, Idris HM, Mohamed-Ali AA, Aelbasit IA, Elbashir MI (2002) Comparison of intramuscular artemether and intravenous quinine in the treatment of Sudanese children with severe falciparum malaria. East Afr Med J 79: 621–625.
  25. 25. Al-Taiar A, Jaffar S, Assabri A, Al-Habori M, Azazy A, et al. (2006) Severe malaria in children in Yemen: two site observational study. BMJ 333: 827.
  26. 26. Allen SJ, O'Donnell A, Alexander ND, Clegg JB (1996) Severe malaria in children in Papua New Guinea. QJM 89: 779–788.
  27. 27. Anberber S, Tessema T, Yimer A (2003) Severe malaria in the under-fives–clinical featrues, management and outcome in a district hospital. Ethiop Med J 41: 301–310.
  28. 28. Assimadi JK, Gbadoe AD, Agbodjan-Djossou O, Ayewada K, Goeh-Akue E, et al. (2002) [Treatment of cerebral malaria in African children by intravenous quinine: comparison of a loading dose regimen to a regimen without a loading dose]. Arch Pediatr 9: 587–594.
  29. 29. Barennes H, Munjakazi J, Verdier F, Clavier F, Pussard E (1998) An open randomized clinical study of intrarectal versus infused Quinimax for the treatment of childhood cerebral malaria in Niger. Trans R Soc Trop Med Hyg 92: 437–440.
  30. 30. Bassat Q, Guinovart C, Sigauque B, Aide P, Sacarlal J, et al. (2008) Malaria in rural Mozambique. Part II: children admitted to hospital. Malar J 7: 37.
  31. 31. Bobossi Serengbe G, Ndoyo J, Gaudeuille A, Longo JD, Bezzo ME, et al. (2004) [An update on severe pediatric malaria in Central Africa hospital units]. Med Mal Infect 34: 86–91.
  32. 32. Bojang KA, Van Hensbroek MB, Palmer A, Banya WA, Jaffar S, et al. (1997) Predictors of mortality in Gambian children with severe malaria anaemia. Ann Trop Paediatr 17: 355–359.
  33. 33. Cao XT, Bethell DB, Pham TP, Ta TT, Tran TN, et al. (1997) Comparison of artemisinin suppositories, intramuscular artesunate and intravenous quinine for the treatment of severe childhood malaria. Trans R Soc Trop Med Hyg 91: 335–342.
  34. 34. Carme B, Bouquety JC, Plassart H (1993) Mortality and sequelae due to cerebral malaria in African children in Brazzaville, Congo. Am J Trop Med Hyg 48: 216–221.
  35. 35. Dzeing-Ella A, Nze Obiang PC, Tchoua R, Planche T, Mboza B, et al. (2005) Severe falciparum malaria in Gabonese children: clinical and laboratory features. Malar J 4: 1.
  36. 36. Evans JA, May J, Ansong D, Antwi S, Asafo-Adjei E, et al. (2006) Capillary refill time as an independent prognostic indicator in severe and complicated malaria. J Pediatr 149: 676–681.
  37. 37. Gay-Andrieu F, Adehossi E, Lacroix V, Gagara M, Ibrahim ML, et al. (2005) Epidemiological, clinical and biological features of malaria among children in Niamey, Niger. Malar J 4: 10.
  38. 38. Gbadoe AD, Lawson-Evi K, Badayodi A, Geraldo A, Guedenon J, et al. (2006) [Severe malaria: assessment of the treatment of anaemia and cerebral manifestations in a tropical intensive cares unit]. Arch Pediatr 13: 1554–1555.
  39. 39. Genton B, al-Yaman F, Alpers MP, Mokela D (1997) Indicators of fatal outcome in paediatric cerebral malaria: a study of 134 comatose Papua New Guinean children. Int J Epidemiol 26: 670–676.
  40. 40. Giha HA, Elghazali G, TM AE, IE AE, Eltahir EM, et al. (2005) Clinical pattern of severe Plasmodium falciparum malaria in Sudan in an area characterized by seasonal and unstable malaria transmission. Trans R Soc Trop Med Hyg 99: 243–251.
  41. 41. Huda SN, Shahab T, Ali SM, Afzal K, Khan HM (2003) A comparative clinical trial of artemether and quinine in children with severe malaria. Indian Pediatr 40: 939–945.
  42. 42. Idro R, Aloyo J, Mayende L, Bitarakwate E, John CC, et al. (2006) Severe malaria in children in areas with low, moderate and high transmission intensity in Uganda. Trop Med Int Health 11: 115–124.
  43. 43. Idro R, Aloyo J (2004) Manifestations, quality of emergency care and outcome of severe malaria in Mulago Hospital, Uganda. Afr Health Sci 4: 50–57.
  44. 44. Issifou S, Kendjo E, Missinou MA, Matsiegui PB, Dzeing-Ella A, et al. (2007) Differences in presentation of severe malaria in urban and rural Gabon. Am J Trop Med Hyg 77: 1015–1019.
  45. 45. Kremsner PG, Radloff P, Metzger W, Wildling E, Mordmuller B, et al. (1995) Quinine plus clindamycin improves chemotherapy of severe malaria in children. Antimicrob Agents Chemother 39: 1603–1605.
  46. 46. Luckner D, Lell B, Greve B, Lehman LG, Schmidt-Ott RJ, et al. (1998) No influence of socioeconomic factors on severe malarial anaemia, hyperparasitaemia or reinfection. Trans R Soc Trop Med Hyg 92: 478–481.
  47. 47. Marsh K, Forster D, Waruiru C, Mwangi I, Winstanley M, et al. (1995) Indicators of life-threatening malaria in African children. N Engl J Med 332: 1399–1404.
  48. 48. Mockenhaupt FP, Ehrhardt S, Burkhardt J, Bosomtwe SY, Laryea S, et al. (2004) Manifestation and outcome of severe malaria in children in northern Ghana. Am J Trop Med Hyg 71: 167–172.
  49. 49. Modiano D, Sirima BS, Sawadogo A, Sanou I, Pare J, et al. (1998) Severe malaria in Burkina Faso: influence of age and transmission level on clinical presentation. Am J Trop Med Hyg 59: 539–542.
  50. 50. Moyou-Somo R, Tietche F, Ondoa M, Kouemeni LE, Ekoe T, et al. (2001) Clinical trial of beta-arteether versus quinine for the treatment of cerebral malaria in children in Yaounde, Cameroon. Am J Trop Med Hyg 64: 229–232.
  51. 51. Neequaye J, Ofori-Adjei E, Ofori-Adjei D, Renner L (1991) Comparative trial of oral versus intramuscular chloroquine in children with cerebral malaria. Trans R Soc Trop Med Hyg 85: 718–722.
  52. 52. Namutangula B, Ndeezi G, Byarugaba JS, Tumwine JK (2007) Mannitol as adjunct therapy for childhood cerebral malaria in Uganda: a randomized clinical trial. Malar J 6: 138.
  53. 53. Oduro AR, Koram KA, Rogers W, Atuguba F, Ansah P, et al. (2007) Severe falciparum malaria in young children of the Kassena-Nankana district of northern Ghana. Malar J 6: 96.
  54. 54. Olumese PE, Bjorkman A, Gbadegesin RA, Adeyemo AA, Walker O (1999) Comparative efficacy of intramuscular artemether and intravenous quinine in Nigerian children with cerebral malaria. Acta Trop 73: 231–236.
  55. 55. Planche T, Agbenyega T, Bedu-Addo G, Ansong D, Owusu-Ofori A, et al. (2003) A prospective comparison of malaria with other severe diseases in African children: prognosis and optimization of management. Clin Infect Dis 37: 890–897.
  56. 56. Rahman ME, Samad R, Rahman MR, Alam AK, Ferdous A (2001) Prognostic factors relating to outcome of severe malaria among children in Bangladesh. Bangladesh Med Res Counc Bull 27: 1–8.
  57. 57. Ranque S, Poudiougou B, Traore A, Keita M, Oumar AA, et al. (2008) Life-threatening malaria in African children: a prospective study in a mesoendemic urban setting. Pediatr Infect Dis J 27: 130–135.
  58. 58. Schapira A, Solomon T, Julien M, Macome A, Parmar N, et al. (1993) Comparison of intramuscular and intravenous quinine for the treatment of severe and complicated malaria in children. Trans R Soc Trop Med Hyg 87: 299–302.
  59. 59. Schellenberg D, Menendez C, Kahigwa E, Font F, Galindo C, et al. (1999) African children with malaria in an area of intense Plasmodium falciparum transmission: features on admission to the hospital and risk factors for death. Am J Trop Med Hyg 61: 431–438.
  60. 60. Taylor TE, Wills BA, Kazembe P, Chisale M, Wirima JJ, et al. (1993) Rapid coma resolution with artemether in Malawian children with cerebral malaria. Lancet 341: 661–662.
  61. 61. Thuma PE, Bhat GJ, Mabeza GF, Osborne C, Biemba G, et al. (2000) A randomized controlled trial of artemotil (beta-arteether) in Zambian children with cerebral malaria. Am J Trop Med Hyg 62: 524–529.
  62. 62. Tripathy R, Parida S, Das L, Mishra DP, Tripathy D, et al. (2007) Clinical manifestations and predictors of severe malaria in Indian children. Pediatrics 120: e454–460.
  63. 63. van Hensbroek MB, Onyiorah E, Jaffar S, Schneider G, Palmer A, et al. (1996) A trial of artemether or quinine in children with cerebral malaria. N Engl J Med 335: 69–75.
  64. 64. Walker O, Salako LA, Omokhodion SI, Sowunmi A (1993) An open randomized comparative study of intramuscular artemether and intravenous quinine in cerebral malaria in children. Trans R Soc Trop Med Hyg 87: 564–566.
  65. 65. Waller D, Krishna S, Crawley J, Miller K, Nosten F, et al. (1995) Clinical features and outcome of severe malaria in Gambian children. Clin Infect Dis 21: 577–587.
  66. 66. White NJ, Waller D, Crawley J, Nosten F, Chapman D, et al. (1992) Comparison of artemether and chloroquine for severe malaria in Gambian children. Lancet 339: 317–321.
  67. 67. Winkler AS, Salmhofer G, Helbok R, Kalokola F, Matuja W, et al. (2008) In-hospital risk estimation in children with malaria–early predictors of morbidity and mortality. J Trop Pediatr 54: 184–191.
  68. 68. Krishna S, Waller DW, ter Kuile F, Kwiatkowski D, Crawley J, et al. (1994) Lactic acidosis and hypoglycaemia in children with severe malaria: pathophysiological and prognostic significance. Trans R Soc Trop Med Hyg 88: 67–73.
  69. 69. Pasvol G, Newton CR, Winstanley PA, Watkins WM, Peshu NM, et al. (1991) Quinine treatment of severe falciparum malaria in African children: a randomized comparison of three regimens. Am J Trop Med Hyg 45: 702–713.
  70. 70. Kremsner PG, Valim C, Missinou MA, Olola C, Krishna S, et al. (2009) Prognostic value of circulating pigmented cells in African children with malaria. J Infect Dis 199: 142–150.
  71. 71. Kochar DK, Tanwar GS, Khatri PC, Kochar SK, Sengar GS, et al. (2010) Clinical features of children hospitalized with malaria–a study from Bikaner, northwest India. Am J Trop Med Hyg 83: 981–989.
  72. 72. Rahman ME, Samad R, Rahman MR, Alam AK, Ferdous A (2001) Prognostic factors relating to outcome of severe malaria among children in Bangladesh. Bangladesh Medical Research Council Bulletin 27: 1–8.
  73. 73. Helbok R, Kendjo E, Issifou S, Lackner P, Newton CR, et al. (2009) The Lambarene Organ Dysfunction Score (LODS) is a simple clinical predictor of fatal malaria in African children. Journal of Infectious Diseases 200: 1834–1841.
  74. 74. Kremsner PG, Radloff P, Metzger W, Wildling E, Mordmuller B, et al. (1995) Quinine plus clindamycin improves chemotherapy of severe malaria in children. Antimicrobial Agents and Chemotherapy 39: 1603–1605.
  75. 75. Luckner D, Lell B, Greve B, Lehman LG, Schmidt-Ott RJ, et al. (1998) No influence of socioeconomic factors on severe malarial anaemia, hyperparasitaemia or reinfection. Transactions of the Royal Society of Tropical Medicine and Hygiene 92: 478–481.
  76. 76. Genton B, D'Acremont V, Rare L, Baea K, Reeder JC, et al. (2008) Plasmodium vivax and mixed infections are associated with severe malaria in children: a prospective cohort study from Papua New Guinea. PLoS Med 5: e127.
  77. 77. Tjitra E, Anstey NM, Sugiarto P, Warikar N, Kenangalem E, et al. (2008) Multidrug-resistant Plasmodium vivax associated with severe and fatal malaria: a prospective study in Papua, Indonesia. PLoS Med 5: e128.
  78. 78. Dondorp AM, Lee SJ, Faiz MA, Mishra S, Price R, et al. (2008) The relationship between age and the manifestations of and mortality associated with severe malaria. Clinical Infectious Diseases 47: 151–157.
  79. 79. Imbert P, Sartelet I, Rogier C, Ka S, Baujat G, et al. (1997) Severe malaria among children in a low seasonal transmission area, Dakar, Senegal: influence of age on clinical presentation. Trans R Soc Trop Med Hyg 91: 22–24.
  80. 80. Lambert S (1949) Malariology. In: Boyd M, editor. Malariology. Philadelphia and London: W.B. Sauders & Co. pp. 824–825.
    • 81. Saul A (1998) Malaria mortality: the pacific enigma-reply. Parasitol Today 14: 293–294.
    • 82. Maitland K, Williams TN, Bennett S, Newbold CI, Peto TE, et al. (1996) The interaction between Plasmodium falciparum and P. vivax in children on Espiritu Santo island, Vanuatu. Trans R Soc Trop Med Hyg 90: 614–620.
    • 83. Maitland K, Williams TN, Peto TE, Day KP, Clegg JB, et al. (1997) Absence of malaria-specific mortality in children in an area of hyperendemic malaria. Trans R Soc Trop Med Hyg 91: 562–566.
    • 84. Genton B, al-Yaman F, Mgone CS, Alexander N, Paniu MM, et al. (1995) Ovalocytosis and cerebral malaria. Nature 378: 564–565.
    • 85. Jarolim P, Palek J, Amato D, Hassan K, Sapak P, et al. (1991) Deletion in erythrocyte band 3 gene in malaria-resistant Southeast Asian ovalocytosis. Proc Natl Acad Sci U S A 88: 11022–11026.
    • 86. Rosanas-Urgell A, Lin E, Manning L, Rarau P, Laman M, et al. (2012) Reduced risk of Plasmodium vivax malaria in Papua New Guinean children with Southeast Asian ovalocytosis in two cohorts and a case-control study. PLoS Medicine 9: e1001305.
    • 87. Stanisic DI, Mueller I, Betuela I, Siba P, Schofield L (2010) Robert Koch redux: malaria immunology in Papua New Guinea. Parasite Immunol 32: 623–632.
    • 88. Luxemburger C, Ricci F, Nosten F, Raimond D, Bathet S, et al. (1997) The epidemiology of severe malaria in an area of low transmission in Thailand. Trans R Soc Trop Med Hyg 91: 256–262.
    • 89. Price RN, Simpson JA, Nosten F, Luxemburger C, Hkirjaroen L, et al. (2001) Factors contributing to anemia after uncomplicated falciparum malaria. Am J Trop Med Hyg 65: 614–622.
    • 90. Prada J, Alabi SA, Bienzle U (1993) Bacterial strains isolated from blood cultures of Nigerian children with cerebral malaria. Lancet 342: 1114.
    • 91. Akpede GO, Sykes RM (1993) Malaria with bacteraemia in acutely febrile preschool children without localizing signs: coincidence or association/complication? J Trop Med Hyg 96: 146–150.
    • 92. Enwere G, Van Hensbroek MB, Adegbola R, Palmer A, Onyiora E, et al. (1998) Bacteraemia in cerebral malaria. Ann Trop Paediatr 18: 275–278.
    • 93. Berkley J, Mwarumba S, Bramham K, Lowe B, Marsh K (1999) Bacteraemia complicating severe malaria in children. Trans R Soc Trop Med Hyg 93: 283–286.
    • 94. Bronzan RN, Taylor TE, Mwenechanya J, Tembo M, Kayira K, et al. (2007) Bacteremia in Malawian children with severe malaria: prevalence, etiology, HIV coinfection, and outcome. J Infect Dis 195: 895–904.
    • 95. Bassat Q, Guinovart C, Sigauque B, Mandomando I, Aide P, et al. (2009) Severe malaria and concomitant bacteraemia in children admitted to a rural Mozambican hospital. Trop Med Int Health 14: 1011–1019.
    • 96. Were T, Davenport GC, Hittner JB, Ouma C, Vulule JM, et al. (2011) Bacteremia in Kenyan children presenting with malaria. J Clin Microbiol 49: 671–676.
    • 97. Scott JA, Berkley JA, Mwangi I, Ochola L, Uyoga S, et al. (2011) Relation between falciparum malaria and bacteraemia in Kenyan children: a population-based, case-control study and a longitudinal study. Lancet 378: 1316–1323.
    • 98. Calis JC, Phiri KS, Faragher EB, Brabin BJ, Bates I, et al. (2008) Severe anemia in Malawian children. N Engl J Med 358: 888–899.
    • 99. Nolan T, Angos P, Cunha AJ, Muhe L, Qazi S, et al. (2001) Quality of hospital care for seriously ill children in less-developed countries. Lancet 357: 106–110.
    • 100. Mpimbaza A, Filler S, Katureebe A, Kinara SO, Nzabandora E, et al. (2011) Validity of verbal autopsy procedures for determining malaria deaths in different epidemiological settings in Uganda. PLoS One 6: e26892.
    • 101. Todd JE, De Francisco A, O'Dempsey TJ, Greenwood BM (1994) The limitations of verbal autopsy in a malaria-endemic region. Ann Trop Paediatr 14: 31–36.
    • 102. Taylor TE, Fu WJ, Carr RA, Whitten RO, Mueller JS, et al. (2004) Differentiating the pathologies of cerebral malaria by postmortem parasite counts. Nat Med 10: 143–145.
    • 103. Maitland K, Kiguli S, Opoka RO, Engoru C, Olupot-Olupot P, et al. (2011) Mortality after fluid bolus in African children with severe infection. N Engl J Med 364: 2483–2495.
    • 104. Juni P, Holenstein F, Sterne J, Bartlett C, Egger M (2002) Direction and impact of language bias in meta-analyses of controlled trials: empirical study. International Journal of Epidemiology 31: 115–123.