Regional variation and epidemiological insights in malaria underestimation in Cameroon

Abstract

Background: Despite significant global effort to control and eradicate malaria, many cases and deaths are still reported yearly. These efforts are hindered by several factors, including the severe underestimation of cases and deaths, especially in Africa.

Methods: We used a mathematical model, incorporating the underestimation of cases and seasonality in mosquito biting rate, to study the malaria dynamics in Cameroon. Using a Bayesian inference framework, we calibrated our model to the monthly reported malaria cases in ten regions of Cameroon from 2019 to 2021 to quantify the underestimation of cases and estimate other important epidemiological parameters. We performed Hierarchical Clustering on Principal Components analysis to understand regional disparities, looking at underestimation rates, population sizes, healthcare personnel, and healthcare facilities per 1000 people.

Results: We found varying levels of case underestimation across regions, with the East region having the lowest (14 %) and the Northwest having the highest (70 %). The mosquito biting rate peaks once every year in most regions, except in the Northwest where it peaks every 6.02 months and in Littoral every 15 months. We estimated a median mosquito biting rate of over 5 bites/day for most regions with Littoral having the highest (9.86 bites/day). Two regions have rates below five: Adamawa (4.78 bites/day) and East (4.64 bites/day).

Conclusions: The low case estimation underscores the pressing requirement to bolster reporting and surveillance systems. Regions in Cameroon display a range of unique features contributing to the differing levels of underestimation. These distinctions should be considered when evaluating the efficacy of community-based interventions.

Keywords: Bayesian inference; Case under-estimation; Hierarchical clustering on principal components analysis; Malaria; Mathematical modeling; Vector-borne diseases.

Fig. 1

Schematic illustration of the compartmental model. An illustration of our model incorporating human and mosquito populations. The human population is divided into susceptible (S_h), exposed (E_h), infectious (I_h), and recovered (R_h) populations, while the mosquito population is divided into susceptible (S_v) and infected (I_v) mosquitoes. Solid black arrows show the transition of humans and mosquitoes through the different stages of malaria infection at the rate beside the arrows. Dashed black arrows show malaria transmission from humans to mosquitoes and vice versa.

Fig. 2

Monthly reported cases of malaria. The monthly reported cases of malaria in different geographical regions of Cameroon from January 2019 to December 2021, for different age groups, obtained from DHIS2 (Dhis2. https). Red bars represent the reported cases for <5 years age-group, and the green bars is for the ≥5 years age-group.

Fig. 3

Observed and estimated monthly malaria cases. The monthly reported malaria cases (black dots) and median predicted cases (solid lines) for each region of Cameroon. The darker bands are the 50 % CrI, while the lighter bands are the 90 % CrI.

Fig. 4

Percentage of malaria cases underestimation. The percentage of underestimation of malaria cases estimated for each region. Left panel: map visualization with the median underestimation percentage indicated with color intensity: high (reddish) and low (yellowish). Right panel: barplots showing the percentage of underestimation for each region with 90 % CrI in ascending order.

Fig. 5

Estimated region-specific malaria parameters: mean duration of natural immunity in months and malaria-induced per capita death rate per month. The bars are the median estimated values and error bars are for the 90 % CrI.

Fig. 6

Estimated region-specific malaria parameters: period of oscillation of mosquito biting rate in months, baseline mosquito biting rate per day and variation of baseline mosquito biting rate in percentage. The bars represent the median estimated values and error bars are for the 90 % CrI.

Fig. 7

Results from Hierarchical Clustering on Principal Components (HCPC). A: Dendrogram showing the clustering of the Cameroon regions. B: Significant variables characterizing the clusters, with p-values from ANOVA test.