Population coverage of artemisinin-based combination treatment in children younger than 5 years with fever and Plasmodium falciparum infection in Africa, 2003–2015: a modelling study using data from national surveys

Adam Bennett, Donal Bisanzio, Joshua Yukich, Bonnie Mappin, Cristin Fergus, Michael Lynch, Richard Cibulskis, Samir Bhatt, Daniel J. Weiss, Ewan Cameron, Peter W. Gething, Thomas P. EiseleThe Lancet Global Health 2017 · DOI

TL;DR

This study introduces a more accurate indicator for tracking malaria treatment coverage in young children across Africa by incorporating rapid diagnostic test (RDT) results alongside fever symptoms. Despite significant investments in malaria control, the study reveals that in 2015, only 19.7% of children with a confirmed malaria infection and fever received appropriate treatment, highlighting persistent gaps in access and service delivery.

What Problem Does This Paper Address?

Monitoring the effectiveness of malaria treatment programs is crucial for public health. Traditionally, coverage has been measured by the proportion of children with fever who received an antimalarial. However, this metric is flawed because it doesn't confirm if the fever was actually caused by malaria. This can lead to an overestimation of appropriate treatment, as children without malaria might receive antimalarials, while those who truly need them might not. This study addresses this by developing and applying a more precise indicator that combines reported fever with a positive Plasmodium falciparum diagnosis via RDT, aiming to provide a clearer picture of who is actually receiving the correct treatment for malaria.

Methods

The researchers employed a modeling approach using cross-sectional data from national household surveys (Malaria Indicator Surveys, Demographic and Health Surveys, and Multiple Indicator Cluster Surveys) conducted between 2003 and 2015 across sub-Saharan Africa.

Key methodological steps included:

  • Developing a predictive model for RDT status: For surveys where RDT data was missing, a logistic regression model was built using available RDT results. This model predicted the likelihood of a child having a positive RDT based on factors like age, wealth, household ITN ownership, urban/rural residence, season, and malaria transmission intensity.
  • Imputing ACT coverage: A generalized additive mixed model (GAMM) was used to estimate ACT coverage for country-years where survey data was unavailable. This model incorporated the relationship between ACT coverage and ACT availability (ACTcap), along with time, country, and UN Africa subregion.
  • Meta-analysis of individual-level factors: Random-effects meta-analyses were conducted to identify individual, household, and community-level factors associated with ACT coverage. This allowed for the examination of how variables like caregiver education, household wealth, ITN ownership, and residence influenced treatment decisions.

Datasets

The study utilized data from three major types of nationally representative household surveys:

  • Malaria Indicator Surveys (MIS)
  • Demographic and Health Surveys (DHS)
  • Multiple Indicator Cluster Surveys (MICS)

These surveys were collected across 33 countries in sub-Saharan Africa between 2003 and 2015. Data on ACT and other antimalarial distributions from the WHO World Malaria Report were also incorporated. Population estimates were derived from the UN Population Prospects database, and malaria transmission intensity data came from the Malaria Atlas Project (MAP).

Key Insights and Ideas

  • The limitations of traditional indicators: The study strongly emphasizes that simply looking at fever cases and antimalarial treatment is misleading. The inclusion of RDT status is a critical step towards accurately assessing appropriate treatment.
  • Persistent low coverage: Despite years of investment and policy shifts towards ACTs, the actual coverage for children who need treatment (fever + positive RDT) remains alarmingly low. This suggests that upstream issues in healthcare access and service delivery are major bottlenecks.
  • Disparities in access: The research highlights significant disparities. Children in rural areas, from poorer households, and those whose caregivers have no education or whose households lack ITNs are less likely to receive ACTs. This underscores that the most vulnerable populations are often the least likely to get appropriate care.
  • The role of the public sector: Children receiving care from public providers were significantly more likely to receive an ACT compared to those treated in the private sector, even when seeking care. This points to potential issues with ACT availability, provider knowledge, or adherence to treatment protocols in the private sector.
  • Improvements are happening, but slowly: While coverage has increased over the study period, the pace of improvement is not sufficient to meet the needs of the population. The study notes that the introduction of RDTs and increased ACT availability have contributed, but systemic issues remain.

Key Results

  • Overall Coverage: In 2015, only 19.7% (95% CI 15.6–24.8) of children younger than 5 years with a fever and a confirmed P. falciparum infection received an ACT.
  • Trends: ACT coverage across sub-Saharan Africa gradually increased between 2005 and 2015, with a slight acceleration between 2009 and 2011.
  • Geographic Variation: ACT coverage was significantly higher in eastern Africa compared to central and western Africa.
  • AMFm Scheme Impact: Coverage was higher in countries participating in the Affordable Medicines Facility malaria (AMFm) scheme.
  • Country-Level Variation: Coverage ranged from a low of 0.6% in Somalia to a high of 70.2% in Uganda in 2015.
  • Predictors of ACT Use: In meta-analyses, children were more likely to receive an ACT if they lived in an urban area, had higher household wealth, had a caregiver with any education, or lived in an area with higher malaria prevalence.
  • Public vs. Private Sector: In 2015, coverage for children treated at public providers was almost twice that of those using private providers.

Limitations and Future Work

The authors acknowledge several limitations:

  • Time lag between RDT and treatment: The time lag between RDT measurement and fever reporting means a positive RDT at the time of survey doesn't definitively confirm malaria infection during the recent fever episode.
  • RDT sensitivity and specificity: The relatively low sensitivity of some RDTs and the possibility of hrp2 gene deletion in parasites mean that negative RDT results don't entirely rule out infection.
  • Accuracy of predictions: While models showed good accuracy, there's inherent uncertainty in predicting RDT status and coverage.
  • Recall bias: The limited number of surveys that captured the exact number of days since fever started restricts the ability to fully account for recall bias in caregiver reports.
  • Sample size limitations: Many surveys are not powered to provide representative estimates for children with fever, potentially affecting subnational heterogeneity analysis.
  • Proxy measure: The indicator of ACT treatment for fever and a positive RDT is a proxy and doesn't capture crucial aspects like correct dosage, patient adherence, or exploration of alternative fever causes.

Future work should focus on:

  • Improving data collection: Robust routine collection of test results and treatment outcomes at the health system level is crucial for ongoing monitoring.
  • Health facility surveys: Periodic health facility surveys with exit interviews could complement household data.
  • Addressing systemic barriers: Continued efforts are needed to improve healthcare access, service delivery, and ACT availability, especially in rural and low-income areas.
  • Integrated community case management: Scaling up integrated community case management is essential for improving access.

Summary updated

Abstract

US President's Malaria Initiative and Medicines for Malaria Venture.
BibTeX
@inproceedings{bennett2017population,
  author = {Adam Bennett and Donal Bisanzio and Joshua Yukich and Bonnie Mappin and Cristin Fergus and Michael Lynch and Richard Cibulskis and Samir Bhatt and Daniel J. Weiss and Ewan Cameron and Peter W. Gething and Thomas P. Eisele},
  title = {{Population coverage of artemisinin-based combination treatment in children younger than 5 years with fever and Plasmodium falciparum infection in Africa, 2003–2015: a modelling study using data from national surveys}},
  booktitle = {The Lancet Global Health},
  year = {2017},
  doi = {10.1016/s2214-109x(17)30076-1},
}
Figure