Malaria epidemiology in the Korhogo area, Northern Côte d’Ivoire: baseline assessment prior to a randomized controlled trial

Abstract

Background

Widespread Insecticide resistance among malaria vector populations poses a significant threat to the efficacy of current vector control strategies. This highlights the urgent need for innovative approaches that integrate existing tools with those under development. The primary objective of this study was to characterize the malaria epidemiological profile in the Korhogo region prior to the implementation of complementary vector control tools alongside long-lasting insecticidal nets (LLINs). Both passive and active case detection methods were employed.

Methods

Passive case detection involved data collection from heath care facilities between July 2016 and July 2017, covering a total of 398,424 person-weeks from 26 villages in the Korhogo health district, northern Côte d’Ivoire. Active case detection was carried out through three cross-sectional surveys (CSS) conducted in September 2016, December 2016 and March 2017. During these surveys, individuals under 21 years of age were invited to attend clinical consultations. Those presenting with fever (> 37,5 °C) were tested for malaria with rapid diagnostic test (RDT), and individuals with a positive result were treated with artemisinin-based combination therapy (ACT). Active case detection during the surveys enabled the estimation of malaria case prevalence, defined as the proportion of population with symptoms and a positive RDT at the time of the survey. Additionally, in a randomly selected sub-sample of six villages, thick blood smears were analysed—regardless of symptoms—to assess the prevalence of malaria infection in the general population during CSS.

Results

The mean weekly malaria incidence throughout the year in the overall population was 0.25 cases per 100 person-weeks, corresponding to 130 cases per 1000 person-years. During CSS, malaria cases prevalence ranged from 13.4% during the dry season to 43.4% in the late rainy season. The highest malaria cases’ prevalence was observed in children under 5 years of age, with significant decline in older age groups. Thick blood smears from six villages revealed parasite prevalence between 85.4% and 100%. Plasmodium falciparum accounted for 97% of single infection, although mixed infections with Plasmodium malariae and Plasmodium ovale were also detected.

Conclusion

Active case detection conducted through three transversal surveys indicates that the study area exhibits characteristics of malaria transmission hotspot. The region’s dense hydrographic network provides a favourable ecological and epidemiological context for assessing additional vector control strategies to supplement universal LLINs coverage in the fight against malaria.

Background

Despite substantial public health progress achieved between 2000 and 2015 [1], malaria remains the leading cause of morbidity and mortality among children in sub-Saharan Africa. The observed reduction in malaria burden has largely depended on intensified efforts to control malaria vectors, improve diagnostic testing and ensure the availability, accessibility and adherence to antimalarial treatments. In 2017, at the onset of the present study, an estimated 219 million malaria cases and 435,000 malaria-related were reported globally. Africa accounted for 92% of these cases and 93% of deaths [2]. Notably, approximately 80% of global malaria deaths occurred in just only 18 countries, including Côte d'Ivoire [2].

In Côte d’Ivoire, malaria is the leading cause of outpatient consultations in health facilities. In 2017, it accounted for 33% of all medical consultations, with 3,557,891 cases (14 cases per 1000 inhabitants) and 3222 deaths [3]. To combat malaria, the National Malaria Control Programme (NMCP) of Côte d'Ivoire—alongside programmes in other endemic sub-Saharan African countries—has adopted the long-lasting insecticide treated nets (LLINs) as the primary vector control strategy. LLINs primarily target anthropophilic and endophagic mosquitoes, which preferentially bite humans indoors. Since 2012, NMCP has implemented large-scale LLIN distribution campaigns, contributing to a 48% reduction in Plasmodium prevalence [1].

Unfortunately, over the last two decades, insecticide resistance mechanisms have rapidly spread across Côte d’Ivoire, mirroring trends observed throughout sub-Saharan Africa [4, 5]. Resistance in malaria vectors is driven by behavioural or physiological mechanisms that reduce or prevent contact between the insecticide and its target site, thereby conferring a selective advantage to mosquitoes carrying such traits. Behavioural resistance enables vectors to avoid insecticide exposure altogether, while physiological resistance involves mechanisms that impair insecticide penetration, enhance its metabolism and excretion, or alter its molecular target. The selective advantages conferred by such mechanisms have led to the rapid spread of pyrethroid resistance among vector populations. Selective pressure is primarly attributed to the widespread use of pyrethroids in agriculture and their extensive deployment in public health interventions such as LLINs and indoor residual spraying (IRS). At present, pyrethroid resistance mechanisms are widespread in nearly all monitored mosquito populations in malaria endemic countries across Sub-Saharan Africa [6], including Côte d'Ivoire [4, 5, 7, 8], posing a serious threat to the effectiveness of current vector control strategies.

Despite a substantial decline in malaria-related morbidity and mortality since 2000, the World Health Organization (WHO) observed a deceleration of this downward trend between 2015 and 2019 [9]. Since 2019, malaria incidence has stagnated or even increased [1], underscoring the urgent need for integrated control strategies that curb the spread of resistance and preserve the efficacy vector control interventions.

To address this issue of insecticide resistance, the WHO recommends the development of complementary vector control strategies targeting malaria vectors that evade LLINs [10]. In this context, a randomized controlled trial (RCT) was conducted in northern Côte d’Ivoire to assess the additional benefit of supplementing universal LLIN coverage with larvicide (LAL), indoor residual spraying (IRS), and an enhanced information, education and communication (IEC) strategy in reducing malaria burden. Prior to village randomization, a one-year entomological and epidemiological monitoring campaign was carried out between 2016 and 2017 to inform the power analysis. Entomological findings were published in Zogo et al. [11]. The present study provides an updated overview of malaria epidemiology in the Korhogo health district, Northern Côte d'Ivoire, prior to the implementation of these complementary interventions.

Methods

Study area

The study was conducted in Korhogo health district, located 635 km north of Abidjan. Previous research carried out during the 1990 s and 2000 s reported malaria transmission rates ranging from 139 to 158 infective bites per person per year, with Anopheles gambiae sensu lato (s.l.) (84%) and Anopheles funestus group (5%) identified as the primary vectors [12, 13]. Between 2003 and 2005, approximately 80% of individuals screened during cross-sectional surveys were found to be infected, with average parasite densities ranging from 143 to 199 asexual stages of Plasmodium falciparum parasites per microlitre of blood [14].

The region experiences a tropical Sudanian-Guinean climate, characterized by two seasons: a rainy season from May to October, with two rainfall peaks typically occurring in June and September, and a dry season from November to April. The dry season is further divided into a cooler period from December to February and a hotter period from March to April. Average annual rainfall ranges between 1200 mm and 1400 mm, while average annual temperatures vary from 21 °C to 35 °C [11, 15].

The Bandama watershed drains the region, along with its tributaries, the Naramou and Solomougou rivers, which typically dry up during the dry season. Despite this, the hydrographic network remains relatively dense due to the high numbers of small water reservoirs constructed to support year-round agricultural activities [15]. The economy of Korhogo district is predominantly based on agriculture, with key crops including cotton, cashew nuts, mangoes, rice, maize, millet, groundnuts and vegetables. Rice is primarily cultivated during the rainy season in flooded lowlands, although it is also grown in irrigated areas near dams during the dry season.

A total of 26 villages were randomly selected from six administrative localities within the Korhogo health district—Napié, Kiémou, Dikodougou, Karakoro, Ballekaha and Siolokaha—based on inclusion criteria related to population size (between 100 and 600 inhabitants), accessibility during the rainy season, and a minimum distance of 2 Kms between villages (Fig. 1). Each locality equipped with at least one primary health care centre.

Fig. 1.

Map of the study area. Red dots indicate all villages included in the study, while green dots represent the six villages randomly selected for detailed analysis

Data collection

All data were collected using Android tablets equipped with the Open Data Kit (ODK) platform. A complete census of the 26 selected villages was conducted prior to the implementation of one-year baseline epidemiological surveillance system.

All individuals from 26 villages who sought care at the primary health centres were enrolled in the passive surveillance. Data collected included the patient’s village of residence, age, sex, reason of consultation (symptoms), temperature, clinical diagnosis, result of RDT (positive or negative or indetermined) and treatment prescribed. Malaria diagnosis was based on the presence of fever (measured or reported, or any fever-related symptoms within the last 48 h) in combination with a positive RDT result, as applied consistently across all primary health centres.

Three cross-sectional surveys (CSS) were conducted in the 26 selected villages during the year preceding the implementation of vector control interventions in September 2017, which aimed to complement universal coverage with LLINs. The first survey took place from 19 September to 14 October 2016, during the minor rainy season; the second from 23 November to 15 December 2016, at the onset of the cool dry season; and the third from 21 February to 15 March 2017, during the hot dry season.

Parasitological and clinical examinations were performed in the 26 villages. During each survey, villages were visited from one to two days, depending on their size and the availability of the population. Prior to each visit, community health worker (CHW) informed the residents of the upcoming survey. Individuals aged between 6-month and 21-year-old were invited to participate, accompanied by a legal guardian in the case of minors. During each CSS, every individual under 21 years of age who presented to the medical team, along with their guardian, received medical care as outlined below. All these participants—or their legal guardians in the case of minors—were informed about the study procedures and objectives, and were invited to provide written informed consent prior to participation.

Three teams—each composed of a nurse and a laboratory technician trained in standard operating procedures (including data collection, obtaining informed consent, performing rapid diagnostic testing, collected blood samples for thick and thin smears, administrating treatment, and more)—visited the villages under the supervision of a physician.

Each participant underwent a clinical examination, which included measurement of axillary temperature, weight, height, age, sex, mid-upper arm circumference, as well as assessment for symptoms as sweating, chills, headache, nausea or vomiting, diarrhoea.

Fever was defined as an axillary temperature ≥ 37.5 °C. Participants presenting with fever or reporting a history of fever within the previous 48 h were tested for malaria on-site using a rapid diagnostic test (RDT). The RDT used was the CareStart™ Malaria Pf (HRP2) Ag test, the same as that provided by the Ministry of Health through the NMCP to public health centres. Individuals with a positive RDT result were immediately treated with artemisinin-based combination therapy (ACT), in accordance with NMCP guidelines. Febrile illnesses not attributable to malaria were referred to the nearest health facility for appropriate care.

Finger-prick blood samples were collected from all participants for parasitological examination, including preparation of thick and thin blood smears. All slides were analysed by trained microscopists at Institute Pierre Richet (IPR) in Bouaké, Côte d’Ivoire. In the same six randomly selected villages as the entomological baseline study [11], parasite densities were estimated by counting the number of asexual Plasmodium parasites observed in 200 high-power microscopic fields on thick smears.

As part of quality assurance, a randomly selected subset representing 10% of all thick smears were re-examined by a senior technician at IPR.

Statistical analysis

Data were analysed using R statistical software [16].

The incidence rate was modeled using weekly case counts from passive case detection, stratified by age groups (0–5, 5–18, and > 18 years) and sex. A negative-binomial generalized linear model (GLM) was fitted, with the group size included as an offset. To account for temporal and spatial autocorrelation, a first-order autoregressive (AR1) covariance structure at the health center level and random intercepts for villages were included.

Models for case prevalence and for infection prevalence were fitted using individual-level data from CSS, applying binomial generalized linear regressions. Explanatory variables included age group, sex, LLIN possession, LLIN use, village and survey round. Case prevalence—defined as the proportion of participants presenting with fever or a history of fever within the past 48 h and a positive rapid diagnostic test (RDT) among the total survey participants—was assessed across all 26 study villages, whereas infection prevalence— defined as the proportion of participants with positive blood smears among the total number of individuals enrolled in the cross-sectional surveys—analysis was restricted to the six villages in which blood-smears were examined. Age categories for these models were derived from age quantiles (0–4, 5–8, 9–12, and 13–21 years), following verification of the linearity assumption of the logit transformation of age as a discrete quantitative variable.

Results

Socio-demographic characteristics of the study participants

The passive surveillance recorded data from residents of the 26 censused villages (total population 7662) who attended health centres for malaria episode (Table 1). The total follows up amounted to 398,424 persons-weeks. Between August 31, 2016 and August 31, 2017, a total of 1262 people was tested positive for malaria in the health centres.

Table 1.

Demographic characteristics of the village populations and participants in the passive surveillance survey

Features	Observation
	Census population	Consult HC*
	(N = 7662)	(N = 2346)
Age groups (years)
0–5	1478	929
5–18	2622	379
> 18	3562	799
Sex
Female	4127	1204
Male	3535	1141
Villages
BAP	183	35
FEL	220	53
FLA	233	196
GUE	171	54
KAG	290	130
KAR	443	18
KAT	300	83
KOG	223	20
KON	391	146
LAG	281	144
LAT	319	100
LLO	393	253
LOK	382	122
MOR	248	32
NAK	99	4
NAL	588	178
NOT	298	143
NOU	414	20
NOW	268	85
PEN	257	75
PES	231	122
TAG	320	92
TAH	147	11
TAK	470	182
YAY	189	23
YEN	304	25

All individuals from 26 villages who sought care at the primary health centres were enrolled in the passive suvey. NB: Missing data (error by omission): 239 for age and 01 for sex

^*Consultations at health centres

Across the three CSS, 9199 participants were enrolled in the 26 villages (Table 2). The sex ratio was balanced, with males representing 50.1% of the participants, and no significant difference in sex distribution was observed between the surveys (p = 0.982). Significantly more children aged 0–4 and 5–8 years attended CSS—each age group representing 28.6% of the participants—compared to those aged 9–12 and 13–21 years, which each represented 21.4% of the participants (p < 0.001).

Table 2.

Demographic characteristics of participants to the cross-sectional surveys

	Surveys			Total	p-value
	Sept-Oct. 2016 (Survey 1)	Nov-Dec. 2016 (Survey 2)	Feb-Mar. 2017 (Survey 3)
Sample size
N	3052	3234	2913	9199
Age groups: n (%)
[0–4]	901 (29,5)	885 (27,4)	842 (28,9)	2628 (28,6)	< 0.01
[5–8]	909 (29,8)	870 (26,9)	855 (29,4)	2634 (28,6)
[9–12]	622 (20,4)	728 (22,5)	617 (21,2)	1967 (21,4)
[13–21 +]	620 (20,3)	751 (23,2)	599 (20,6)	1970 (21,4)
Sex: n (%)
Female	1522 (49,9)	1622 (50,2)	1446 (49,6)	4590 (49,9)	0.982
Male	1530 (50,1)	1612 (49,8)	1467 (50,4)	4609 (50,1)
LLINs possession: n (%)
Yes	1291 (42,3)	1773 (54,8)	1278 (43,9)	4342 (47,2)	< 0.001
Sleep under LLIN the night before the survey: n (%)
Yes	599 (46,4)	1227 (69,2)	566 (44,3)	2392 (55,1)	< 0.001

The proportion of participants reporting ownership of LLINs varied significantly across the three surveys (p < 0.001), with the highest ownership rate of 55% observed during the second survey conducted in November–December 2016. Among LLINs owners, the proportion reporting having slept under a net the night before the survey ranged from 44.3% to 69.2%. LLIN use—defined as the proportion of net owners who reported sleeping under a LLIN the previous night—was significantly higher at the end of the rainy season (69.2% during November–December 2016) compared to other survey periods (p < 0.01).

Malaria incidence (passive detection)

The mean weekly malaria incidence over the year for the entire population was 0.30 case per 100 person-weeks, equivalent to 156 cases per 1000 person-years. The incidence exhibited a clear seasonal pattern with two peaks: one occurring in early rainy season (July–August) both in 2016 and 2017, and a second peak during the late rainy season (November 2016). Among children under five years of age, malaria incidence exceeded 1 case per 100 person-weeks during these peaks (Fig. 2). In contrast, incidence rates remained below 0.5 cases per 100 person-weeks in the 5–18 and > 18-year-old age groups.

Fig. 2.

Malaria incidence. The curves illustrate the incidence rate dynamics across age groups during the pre-intervention period (August 31, 2016–August 31, 2017). The red line corresponds to children under 5 years, the orange line to individuals aged 5–18 years, and the blue line to those over 18 years

The ratio of minimum to maximum village-specific incidence rates ranged from 0.06 to 3.12 relative to the mean incidence (0.30 case per 100 person-weeks), highlighting substantial local scale variability (Fig. 3). Analysis of health centre attendance relative to the mean incidence (0.30 case per 100 person-weeks) revealed that Dikodougou health centre reported a higher number of malaria cases compared to others, whereas Napie health centre reported fewer cases (Fig. 4).

Fig. 3.

Illustration of random village effects. Ratios range from 0.06 to 3.12, shown in blue or red depending on whether they fall below or above the average incidence (0.30 cases per 100 person-weeks), respectively

Fig. 4.

Variation in attendance at health centres. Points are coloured blue or red to indicate incidence rates above or below the average incidence (0.30 cases per 100 person-weeks), respectively

Malaria burden (active detection)

The highest malaria cases rates were recorded at the end of the rainy season (November–December 2016) (p < 0.001), while no significant differences were observed between the beginning of the rainy season (September–October 2016) and the dry season (February–March 2017) (p = 0.5).

Malaria cases rates were highest in the youngest age group (0–4 years old) and declined with increasing age, reaching the lowest rate in the oldest group (13–21 years) (p < 0.001). No significant difference was found between males and females (p = 0.9). Interestingly, individuals reporting ownership of LLIN had significantly fewer malaria cases (p = 0.034). Conversely, malaria case rates were significantly higher among those who reported having slept under a LLIN the night before the survey (p = 0.014).

Significant variations of malaria case prevalence were observed across villages. Combined results of the three cross-sectional surveys showed mean prevalence rates ranging from 10.9% to 33% (p < 0.001) (Table 3).

Table 3.

Prevalence of malaria cases detected by RDTs stratified by socio-demographic characteristics

	Observation		Cases (%)	Cases (95%CI)	OR	OR (95%CI)	p-value
	N	Malaria Cases (n)
Surveys							< 0.001
Survey 1	3052	423	13,9	[1, 7–12, 12–15]	1.00	–
Survey 2	3234	1404	43,4	[1, 7–36]	5.62	[4.92, 6.41]	< 0.001
Survey 3	2913	390	13,4	[2–7, 7–12, 12–14]	0.95	[0.81, 1.10]	0.5
Age groups							< 0.001
[13–21 +]	1970	212	10,8	[2, 5–9, 9–12]	1.00	–
[9–12]	1967	387	19,7	[18,0–21,5]	2.22	1.83–2.70	< 0.001
[5–8]	2634	730	27,7	[26,0–29,5]	4.15	3.47–4.96	< 0.001
[0–4]	2628	888	33,8	[32,0–35,6]	5.73	4.80–6.84	< 0.001
Sex							> 0.9
Female	4590	1104	24,1	[3, 8–22, 22–25]	1.00	–
Male	4609	1113	24,1	[4, 9–22, 22–25]	1.00	[0.90, 1.12]	> 0.9
LLINs possession							0.034
NO	4857	1128	23,2	[1–4, 4–22, 22–24]	1.00	–
YES	4342	1089	25,1	[4, 8–23, 23–26]	0.85	[0.74, 0.99]	0.034
Sleep under LLIN the night before the survey							0.014
No	1950	379	19,4	[3, 7–17, 17–21]	1.00	–
Yes	2392	710	29,7	[5, 9–27, 27–30]	1.22	[1.04, 1.44]	0.014
Villages							< 0.001
BAP	316	43	13,6	[3–8, 8–10, 10–17]	1.00	–
FEL	316	83	26,3	[4, 7–21, 21–30]	1.78	[1.14–2.79]	0.011
FLA	301	71	23,6	[1–7, 7–19, 19–28]	1.70	[1.08–2.68]	0.021
GUE	224	43	19,2	[6–9, 9–14, 14–24]	1.19	[0.72–1.97]	0.5
KAG	270	85	31,5	[2, 2–26, 26–36]	2.66	[1.70–4.16]	< 0.001
KAR	534	176	33,0	[1, 1–29, 29–36]	3.11	[2.08–4.63]	< 0.001
KAT	378	101	26,7	[4–22, 22–30]	1.95	[1.27–2.99]	0.002
KOG	281	60	21,4	[17,0–26,5]	1.45	[0.91–2.30]	0.12
KON	498	145	29,1	[3, 3–25, 25–32]	2.34	[1.56–3.51]	< 0.001
LAG	354	113	31,9	[3–9, 9–27, 27–35]	2.57	[1.68–3.94]	< 0.001
LAT	365	96	26,3	[22,0–31,0]	2.02	[1.31–3.11]	0.001
LLO	375	74	19,7	[16,0–24,1]	1.28	[0.82–1.99]	0.3
LOK	325	84	25,8	[4–9, 9–21, 21–29]	1.89	[1.22–2.94]	0.005
MOR	340	37	10,9	[8,0–14,6]	0.69	[0.42–1.14]	0.14
NAK	135	31	23,0	[7, 7–16, 16–29]	1.77	[1.02–3.10]	0.044
NAL	678	125	18,4	[5, 7–15, 15–21]	1.06	[0.70–1.58]	0.8
NOT	309	60	19,4	[2, 4–15, 15–24]	1.13	[0.71–1.80]	0.6
NOU	523	156	29,8	[1–9, 9–26, 26–32]	2.53	[1.69–3.79]	< 0.001
NOW	358	77	21,5	[1, 6–17, 17–26]	1.79	[1.15–2.79]	0.010
PEN	289	77	26,6	[21,9–32,0]	1.77	[1.13–2.78]	0.013
PES	284	54	19,0	[14,9–24,0]	1.16	[0.72–1.86]	0.5
TAG	482	95	19,7	[4, 5, 5–16, 16–23]	1.28	[0.84–1.96]	0.3
TAH	214	47	22,0	[16,9–28,0]	1.91	[1.16–3.12]	0.010
TAK	426	97	22,8	[19,0–27,0]	1.55	[1.01–2.37]	0.045
YAY	271	78	28,8	[4, 7–23, 23–33]	2.17	[1.38–3.42]	< 0.001
YEN	353	109	30,9	[3–9, 9–26, 26–34]	2.44	[1.59–3.74]	< 0.001
	N = 9199	Cases: n = 2217	% = 24,1	95%CI	OR	95%CI	p-value

Infection prevalence and parasite density in the 6 villages

The analysis focused on six randomly selected villages where malaria infection was assessed through blood smears examination (Table 4). The overall prevalence of Plasmodium infection was 93.1%.

Table 4.

Comparison of malaria case prevalence versus infection prevalence

		Infection						Cases
	N	n	%	95%CI	OR	95%CI	p-value	n	%	95%CI	OR	95%CI	p-value
Surveys							< 0.001						< 0.001
Survey 1	732	732	100,0	[99–100]	1.00	–		119	16.3	[1, 8–13, 13–19]	1.00	–
Survey 2	860	814	94.7	[92,9–96,0]	0.00	0.00, Inf	> 0.9	385	44.8	[1, 5–36]	4.97	3.83, 6.46	< 0.001
Survey 3	820	700	85.4	[6, 8–36]	0.00	0.00, Inf	> 0.9	140	17.1	[7, 8, 8–14, 14–19]	1.14	0.86, 1.51	0.4
Sex							0.5						0.4
Female	1266	1182	93.4	[6, 9–36]	1.00	–		329	26.0	[5–23, 23–28]	1.00	–
Male	1146	1064	92.8	[2, 2–36]	0.89	0.64, 1.24	0.5	315	27.5	[25,0–30,1]	1.09	0.89, 1.33	0.4
LLINs Possession							0.007						0.4
NO	1896	1767	93.2	[92,0–94,2]	1.00	–		457	24.1	[1–22, 22–26]	1.00	–
YES	516	479	92.8	[3–8, 8–36]	0.50	0.30, 0.83	0.007	187	36.2	[2–5, 5–31, 31–36]	1.12	0.86, 1.44	0.4
Sleep under LLIN the night before the survey							> 0.9						< 0.001
NO	481	444	92.3	[4, 6–36]	1.00	–		169	35.1	[31,0–39,5]	1.00	–
YES	35	35	100,0	[1–36]	0,00	0.00, Inf	> 0.9	18	51.4	[35,6–67,0]	4.74	2.30, 9.78	< 0.001
Villages							< 0.001						< 0.001
FEL	310	276	89.0	[85,1–93,0]	1.00	–		75	24.2	[3, 8–19, 19–29]	1.00	–
KAR	521	509	97.7	[96,0–98,7]	5.26	2.62, 10.6	< 0.001	176	33.8	[9, 9–29, 29–36]	2.00	1.40, 2.85	< 0.001
KON	488	463	94.9	[5, 5–36]	2.37	1.35, 4.17	0.003	137	28.1	[2–24, 24–31]	1.39	0.97, 1.99	0.076
LLO	370	339	91.6	[88,4–94,0]	1.28	0.74, 2.20	0.4	68	18.4	[1, 4–14, 14–22]	0.69	0.46, 1.04	0.078
NOU	519	475	91.5	[6, 8–36]	1.25	0.75, 2.09	0.4	141	27.2	[2, 5–23, 23–30]	1.46	1.01, 2.10	0.044
TAH	204	184	90.2	[3–6, 6–36]	1.40	0.75, 2.63	0.3	47	23.0	[3, 8–17, 17–29]	1.15	0.73, 1.83	0.5
Age groups							0.4						< 0.001
[13–21 +]	462	439	95.0	[6, 7, 7–36]	1.00	–		65	14.1	[2–5, 5–11, 11–17]	1.00	–
[9–12]	641	595	92.8	[6, 6–36]	0.81	0.47, 1.40	0.5	146	22.8	[2, 7–19, 19–26]	1.98	1.40, 2.79	< 0.001
[5–8]	650	597	91.8	[5–7, 7–36]	0.68	0.40, 1.15	0.15	179	27.5	[1–24, 24–30]	2.82	2.01, 3.95	< 0.001
[0–4]	659	615	93.3	[91,2–95,0]	0.91	0.53, 1.57	0.7	254	38.5	[3, 9–33, 33–36]	5.28	3.79, 7.36	< 0.001

Infection prevalence was defined as the proportion of participants with positive blood smears among the total number of individuals enrolled in the cross-sectional surveys

Case prevalence was defined as the proportion of participants presenting with fever or a history of fever within the past 48 h and a positive rapid diagnostic test (RDT) among the total survey participants

Fever and malaria prevalence according to RDTs

Symptomatic infections accounted for 29% (644/2246) of all malaria-positive individuals. This proportion varied significantly across surveys, age groups, LLIN usage and villages. The infection rates ranged from 85.6% to 100% and showed no significant variation across the three surveys (p > 0.9). However, the highest prevalence of clinical malaria cases was recorded during the second survey (November–December 2016; p < 0.001).

A similar pattern was observed across age groups. Although infection prevalence remained uniformly high (between 91,8% and 95,0%) without significant differences, the prevalence of symptomatic malaria cases decreased significantly with age (p < 0.001)(see Table 4).

With respect to spatial variability, symptomatic malaria case prevalence was the highest in Karafine (KAR) (p < 0.001), which also exhibited the highest infection prevalence (p < 0.001).

Plasmodium falciparum, Plasmodium malariae and Plasmodium ovale were detected either as mono-infections or mixed infections on blood smears (Table 5). The highest infection probability was associated with P. falciparum alone (p < 0.001).

Table 5.

Distribution of Plasmodium species identified in positive blood smears

Plasmodium species	Number of species
	N = 2246	%
P.f	2182	97.15
P.m	0	0.00
P.o	0	0.00
P.f + P.m	28	1.24
P.f + P.o	33	1.50
P.f + P.m + P.o	3	0.13

N: Number of occurrences; %: Percentage of occurrences; P.f.: Plasmodium falciparum; P.m.: P. malariae; P.o.: P. ovale; P.f. + P.m.: P. falciparum + P. malariae; P.f. + P.o.: P. falciparum + P. malariae; P.f. + P.m. + P.o.: P. falciparum + P. ovale + P. falciparum + P. malariae + P. ovale

Discussion

This study provides valuable insights into malaria epidemiology—both infection and morbidity—in the Korhogo health district prior to the implementation of complementary vector control strategies alongside universal LLIN coverage.

Malaria incidence

The malaria incidence observed in this study (156 cases per 1000 person-years) was markedly lower than the national incidence estimates reported for Côte d’Ivoire in 2016 (243.95 [154.08–366.00] cases per 1000 person-years), 2017 (254.20 [160.21–380.87]), and even 2023 (251.43 [170.09–360.79]) [17].

Prevalence of infection and clinical cases

Despite the relatively low incidence, both malaria infections and symptomatic cases prevalence recorded during the CSS were notably higher than those reported in a concurrent study conducted in Bouaké, central Côte d’Ivoire [18]. In Bouaké, symptomatic cases prevalence recorded in August 2016 was 2.42% (66/2725), infection prevalence was 84.88% (2313/2725). In contrast, in Korhogo at the same period, symptomatic cases prevalence ranged from 13.4% to 43.4%, and infection prevalence ranged from 85.4% to 100%, depending of the season.

As expected in such Sudanese Guinean climate, malaria case prevalence showed a clear seasonal variation [11]. Moreover, in line with well-established patterns in endemic regions, symptomatic cases prevalence decreased with age, consistent with the development of acquired immunity [19, 20]. Interestingly, infection prevalence did not vary significantly by seasons or age classes. This lack of variation may be due to extremely high infection prevalence throughout the study population, which could mask any subtle age-related difference.

Importance of a micro spatial approach

The high prevalence of both infections and symptomatic cases, supported by the entomological data from Zogo et al. [11, 21] suggests that the Korhogo district represents a hyper-endemic area according to the Yaoundé classification (1962) [22, 23], and may even constitute a transmission hotspot [24]. Understanding the drivers of such hotspots is essential for optimizing malaria surveillance and evaluating novel vector control tools [24, 25]. To achieve this, fine-scale spatial and temporal analysis are needed to model both epidemiological and entomological dynamics. While environmental and socio-economic studies have been conducted in various settings [26, 27], predictive tools remain under development [28]. Nevertheless, the present study confirms that the Korhogo area is highly endemic, underline the urgent need for additional interventions beyond LLIN distribution.

The study also documented the presence of both permanent and temporary mosquito breeding sites—such as inland valleys, rice paddies, swamps, and freshwater ponds— across the dense hydrographic network [15, 29], laying the groundwork for evaluating larval source management strategies as a complement to LLIN use.

Local socio-economic factors

However, spatial heterogeneity in malaria case prevalence is not solely driven by ecological factors [29]. Socio-economic determinants may also contribute significantly. This may include household income, woman’s economic autonomy, the presence and practices of traditional healers, and variability in the availability and activities of CHWs. Although detailed socio-economic data were not collected, future investigations should address these aspects to better understand the local epidemiology and inform the evaluation of RCTs.

LLIN ownership and use

Mass LLIN distribution remains the cornerstone of vector control in Côte d'Ivoire and other sub-Saharan countries [2]. LLINs ownership and use the night before the survey were recorded during the cross-sectional assessments. Overall, ownership was 47,2% and use was 55,1%, with both coverage (54,8%) and utilization (69,2%) being significantly higher during the dry-cold season. Although these rates were lower than those reported in a 2019 national study (ownership: 65.5%; use: 71.8%) [30], they were comparable to WHO-reported averages for sub-Saharan Africa in 2017 (ownership: 56%; use: 50%) [2].

Some studies have suggested that in absence of mosquito nuisance or during hot weather, discomfort (e.g. heat or unpleasant smell of insecticide) can deter LLIN use [31]. Surprisingly, malaria case prevalence was higher among individuals who reported sleeping under a LLIN the previous night. This contradicts numerous studies that demonstrate the protective effect of LLIN use [32], although some studies have found no association [33, 34]. This discrepancy could reflect limitations of self-reported data, which are subject to recall or desirability bias, especially when focused on a single night. Additional indicators and more objective measures of LLIN use should be considered in future studies.

Passive vs active case detection: contrasting even inconsistent results

The incidence captured through passive surveillance (156 cases per 1,000 person-years) appeared inconsistent with the high malaria case prevalence identified through cross-sectional surveys, which was higher than in comparable region [18]. This discrepancy underscores key methodological and contextual differences between passive and active case detection approaches. Recent studies have compared the two methods directly [35, 36]. For instance, Tiono et al. found that active detection identified significantly more cases, with a rate ratio of 1.32 [95% CI 1.13–1.54]. They concluded that passive surveillance would require a 30%–40% larger sample size in vaccine trials compared to achieve equivalent statistical power. Their study design ensured community awareness and involvement, potentially influencing health-seeking behaviour.

In contrast, passive case detection relied on routine health centre attendance without any awareness campaign in the control villages that could bias the population behaviour, while active case detection was conducted independently during CSS. Nevertheless, methodological differences in the data—incidence (count) vs prevalence (proportion)—make direct comparisons challenging. Addressing this gap is essential to accurately classify transmission intensity and to determine wether an area should be considered hyperendemic or holoendemic depending on the case detection strategy.

Conclusion

Active case detection through three cross-sectional surveys revealed that Korhogo health district constitutes a malaria transmission hotspot. With its dense hydrographic network and high infection prevalence, the area presents an optimal setting for the implementation and evaluation of implement the complementary vector control strategies to enhance the impact of universal LLIN coverage and reduce malaria transmission.

Author contributions

CP, LPAA, RKD, NM and FF conceptualized the study and coordinated its implementation. NBT supervised and conducted the field and laboratory activities alongside LPAA and SBA, performed the statistical analysis with support of NM and CP, and drafted the manuscript with CP and FF. MBZ, AK, IC, DDS and AS contributed to the field and laboratory work. AD managed the study database. PT oversaw data management, created the study map, and contributed to manuscript revision. LB-M and RKD reviewed and revised the manuscript. All authors read and approved the final version of the manuscript.

Funding

This study was conducted as part of the REACT project, supported by the French Initiative 5% (Grant No. 15SANIN213).

Data availability

All data generated or analysed during this study are included within this published article.

Declarations

Ethics approval and consent to participate

Ethical clearance for the study was obtained from the national ethical committee (N/Réf: 063/MSHP/CNER-kp, dated July 21, 2016). All symptomatic participants who tested positive by the rapid diagnostic test (RDT) during the study, received free and appropriate treatment. Participants were also informed of their right to withdraw from the study at any time without any consequences.

Competing interests

The authors declare no competing interests.