Quantifying the contribution of Plasmodium falciparum malaria to febrile illness amongst African children

Ursula Dalrymple, Ewan Cameron, Samir Bhatt, Daniel J. Weiss, Sunetra Gupta, Peter W. GethingeLife 2017 · DOI

TL;DR

This paper tackles the problem of overestimating malaria's burden in African children by developing a novel statistical model. It reveals that while many fevers are associated with malaria infections, only a fraction are caused by malaria, with most fevers in malaria-positive children stemming from other illnesses. This distinction is crucial for accurate disease burden assessment and effective fever management policies.

What Problem Does This Paper Address?

In sub-Saharan Africa, fever is a common symptom, and rapid diagnostic tests (RDTs) are widely used to detect Plasmodium falciparum malaria before prescribing antimalarials. While this improves resource allocation and reduces unnecessary drug use, it creates a diagnostic challenge: a positive RDT doesn't definitively mean malaria is the cause of the fever. This is because malaria infections are often asymptomatic, meaning individuals can carry the parasite without being sick from it. Consequently, current surveillance systems, which often equate RDT-positive fevers with clinical malaria, likely overestimate malaria's true burden. Conversely, non-malarial febrile illnesses (NMFIs) are likely underestimated, especially when they co-occur with asymptomatic malaria infections. This ambiguity hinders accurate understanding of disease burden and can lead to suboptimal case management.

Methods

The authors introduce a novel geospatial statistical approach, a hierarchical Bayesian multinomial model, to disentangle the contributions of malaria and NMFIs to febrile illness. This model leverages household survey data on fever history and RDT results.

Key methodological aspects include:

  • Multinomial Model: A Bayesian multinomial model is used to categorize fever outcomes into four groups: febrile with malaria, febrile without malaria, afebrile with malaria, and afebrile without malaria.
  • Covariate Integration: The model incorporates spatial predictions of P. falciparum prevalence (PfPR0-5) and environmental suitability for background fevers (derived from a boosted regression tree model).
  • Spatio-temporal Prediction: The model generates pixel-level (5x5 km) predictions of malaria-attributable fever (MAF) and non-malarial febrile illness (NMFI) across sub-Saharan Africa for the years 2006-2014.
  • Age Stratification: The model accounts for age-specific differences in malaria-attributable fever prevalence, particularly distinguishing between children under and over two years old.

Datasets

The study primarily utilizes:

  • Household Survey Data: Data from 38 cross-sectional, nationally-representative georeferenced surveys across 24 sub-Saharan African countries, collected between 2006 and 2014. These surveys included RDT and Giemsa-stained microscopy results for malaria, as well as two-week fever history and household wealth quintiles. Sources include the DHS Program, UNICEF Multiple Indicator Cluster Surveys, and national Ministries of Health.
  • Malaria Atlas Project (MAP) Data: Used for P. falciparum prevalence predictions (PfPR2-10).
  • Environmental and Socio-demographic Covariates: A suite of covariates (e.g., elevation, land cover, population density, temperature) were assembled and processed for use in the boosted regression tree model.

Key Insights and Ideas

  • The "Co-infection Conundrum": The most striking insight is that a significant proportion of fevers in malaria-positive children are not caused by malaria. This highlights the pervasive issue of co-infections and asymptomatic parasitemia in driving fever burden.
  • Overestimation of Malaria Burden: Current methods of reporting malaria cases based on RDT-positive fevers likely inflate the true burden of malaria by as much as two-thirds.
  • Increasing NMFI Burden: The proportion of fevers attributable to NMFIs has increased over the study period, underscoring the growing importance of these illnesses.
  • Age Matters: Younger children (under two years) are more susceptible to developing malaria-attributable fevers, even at similar malaria transmission intensities, compared to older children.
  • Policy Implications: The findings strongly suggest a need for improved diagnostic tools for NMFIs and a re-evaluation of fever case management guidelines, particularly in regions with high malaria endemicity but low malaria-attributable fever fractions.

Key Results

  • In 2014, 35.7% of all fevers in children under five were accompanied by a P. falciparum infection.
  • However, only 28.0% of these malaria-positive fevers (which equates to 10.0% of all fevers) were causally attributable to malaria.
  • The proportion of all fevers due to NMFIs increased from 82.5% in 2006 to 90.0% in 2014.
  • Countries like Niger, Gabon, and Nigeria showed particularly low fractions of malaria-attributable fevers among malaria-positive cases (less than 21%), indicating a high potential for overestimation of malaria burden in these regions.
  • The probability of a malaria-attributable fever increased with P. falciparum prevalence (PfPR0-5), but this gradient was steeper for children under two years old.

Limitations and Future Work

  • Recall Bias: The reliance on two-week fever recall from caregivers is subject to biases related to fever duration, frequency of episodes, socioeconomic status, and child's sex.
  • P. vivax Exclusion: Fevers caused by P. vivax malaria were categorized as NMFIs, as not all surveys tested for this parasite. This is considered a minor limitation due to the limited stable transmission of P. vivax in most of Africa.
  • Co-morbidity Complexity: The model does not explicitly account for the complex interactions of co-morbidities where multiple pathogens might contribute to fever severity.
  • Future Work: The authors emphasize the need for better diagnostic tests for NMFIs and suggest that current WHO guidelines for integrated management of childhood illness (IMCI) may be suboptimal for treating non-malarial fevers. They also highlight the importance of understanding the role of never-symptomatic infections in malaria transmission and elimination efforts.

Summary updated

Abstract

Suspected malaria cases in Africa increasingly receive a rapid diagnostic test (RDT) before antimalarials are prescribed. While this ensures efficient use of resources to clear parasites, the underlying cause of the individual's fever remains unknown due to potential coinfection with a non-malarial febrile illness. Widespread use of RDTs does not necessarily prevent over-estimation of clinical malaria cases or sub-optimal case management of febrile patients. We present a new approach that allows inference of the spatiotemporal prevalence of both <i>Plasmodium falciparum</i> malaria-attributable and non-malarial fever in sub-Saharan African children from 2006 to 2014. We estimate that 35.7% of all self-reported fevers were accompanied by a malaria infection in 2014, but that only 28.0% of those (10.0% of all fevers) were causally attributable to malaria. Most fevers among malaria-positive children are therefore caused by non-malaria illnesses. This refined understanding can help improve interpretation of the burden of febrile illness and shape policy on fever case management.
BibTeX
@inproceedings{dalrymple2017quantifying,
  author = {Ursula Dalrymple and Ewan Cameron and Samir Bhatt and Daniel J. Weiss and Sunetra Gupta and Peter W. Gething},
  title = {{Quantifying the contribution of Plasmodium falciparum malaria to febrile illness amongst African children}},
  booktitle = {eLife},
  year = {2017},
  doi = {10.7554/elife.29198},
}
Figure