Knowledge, attitudes and practices towards uncomplicated malaria management and LLIN use in Kumba and Muyuka, Cameroon: a cross-sectional study

Abstract

Introduction

Uncomplicated malaria (UM) remains a major public health concern in Cameroon. This study assessed knowledge, attitudes, prevention practices and care-seeking behaviours related to UM in Kumba and Muyuka, Southwest Region.

Methods

A cross-sectional survey was conducted between March and June 2025 among adults (≥18 years), residing in the area for at least 3 months. Multistage sampling yielded 638 valid responses in Kumba and 606 in Muyuka. Data were collected using structured interviewer-administered questionnaires. Descriptive statistics, Pearson correlation (r), Pearson χ² tests, analysis of variance and multivariate logistic regression were used to evaluate the Knowledge-Attitude-Practice (KAP) pathway and identify predictors of good and appropriate preventive and care-seeking practices.

Results

Comprehensive knowledge was poor: only 13.0% in Kumba and 14.9% in Muyuka demonstrated good knowledge although fever was widely recognised (86.1%). Long-lasting insecticidal net ownership was significantly higher in Kumba (81.7%) than in Muyuka (57.1%). Correlation analysis revealed that while knowledge and attitudes were independently associated with practices, knowledge did not significantly correlate with attitudes (r=0.045, p=0.114). In multivariate analysis, younger age, male gender, good knowledge, positive attitudes and good access to malaria information emerged as significant independent predictors of good preventive practices, with specific drivers varying between municipalities.

Conclusion

Despite high recognition of fever as a symptom, comprehensive malaria knowledge remains low. The weak association between knowledge and attitudes suggests that information alone does not drive behaviour change. Malaria control requires integrated, culturally sensitive strategies that address sociodemographic variations and contextual barriers rather than relying solely on information campaigns.

Clinical trial registration

Not applicable.

WHAT IS ALREADY KNOWN ON THIS TOPIC?

WHAT THIS STUDY ADDS

• Only 13.9% of the pooled participants demonstrated good comprehensive knowledge of malaria, although basic symptom recognition (fever) was high.

• The classical KAP sequential pathway was only partially supported; while knowledge and attitudes independently influenced practices, increased factual knowledge did not translate into more positive attitudes.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Malaria control efforts must move beyond information-only campaigns to address sociocultural beliefs, trust in health systems and contextual barriers.

• Public health interventions should be demographically targeted; particularly engaging women and older adults and specifically tailored to local municipality contexts.

Introduction

Malaria remains a leading cause of illness and death in sub-Saharan Africa (SSA), despite decades of control efforts.1,4 The WHO reported over 263 million malaria cases globally in 2023, with the African region accounting for 94% of cases and 95% of malaria-related deaths.^{4 5} In Cameroon, malaria is endemic and continues to pose a significant public health challenge, contributing to approximately 30% of outpatient consultations, 24% of all morbidity and 18.7% of reported deaths.^{6 7}

Malaria is caused by Plasmodium parasites, with Plasmodium falciparum being the most virulent species in SSA. Transmission occurs through infected female Anopheles mosquitoes and is facilitated by tropical and subtropical environments. Uncomplicated malaria (UM) is the most prevalent form of the disease, presenting with non-specific symptoms such as fever, chills, headache, nausea and muscle aches, without clinical or laboratory signs of severe organ dysfunction. It can be treated effectively with timely diagnosis and appropriate use of artemisinin-based combination therapies (ACTs).8,11 However, poor adherence to treatment guidelines and widespread self-medication continue to undermine malaria control efforts in many African countries, including Cameroon.^{1 3}

Since 2002, the Cameroonian government has implemented several strategies under the National Malaria Control Programme (NMCP) to reduce transmission. These include mass distribution of long-lasting insecticidal nets (LLINs) and community health education.¹² While indoor residual spraying (IRS) was historically part of vector control, its large-scale implementation in Cameroon has declined due to high operational costs and emerging insecticide resistance, making LLINs the primary vector control tool. Although LLIN ownership rates are generally high,13,15 usage remains inconsistent due to discomfort, heat and misconceptions surrounding their effectiveness.16,19 Inappropriate practices such as incomplete drug regimens and reliance on unregulated street medications are also common in areas like Kumba and Muyuka, complicating local control efforts.^{16 20 21}

Kumba and Muyuka, located in the Southwest Region of Cameroon, are both malaria-endemic municipalities.^{22 23} These urban and semiurban areas face environmental and infrastructural challenges such as poor drainage, open gutters, unmanaged waste and frequent flooding, which creates environments that are conducive to mosquito breeding in specific neighbourhoods including Kosala, Hausa Quarters, Three Corners and Kumba Town in Kumba, and Balong, Makata and Yoke in Muyuka.^{22 23} Kumba also hosts a large population of internally displaced persons (IDPs) due to the ongoing Anglophone crisis, which strains health services and increases vulnerability to disease outbreaks.

Studies on knowledge, attitude and practice (KAP) have shown that malaria knowledge does not always translate into appropriate health-seeking or prevention behaviours.^{15 16 24} Public servants and students typically exhibit better malaria-related KAP than less educated or economically disadvantaged individuals.^{16 25} Gender differences have also been observed in LLIN use and treatment-seeking patterns,^{26 27} while poor practices are associated with higher parasitaemia and lower haemoglobin levels, especially among children.^{15 28 29}

Although several studies have assessed KAP in other regions of Cameroon,30,32 there is limited data specific to Kumba and Muyuka. Understanding how local populations perceive and manage malaria is essential to tailoring effective context-specific interventions for these high-burden communities.

In this study, ‘UM management’ is defined at the community level and refers to behaviours related to the recognition of malaria symptoms, care-seeking behaviour and preventive practices, including the use of LLINs.³³ It does not include clinical case management processes such as treatment prescription, diagnostic procedures or referral systems. Within this community context, the use of ‘herbal remedies’ or ‘traditional healers’ refers to the reliance on non-standardised, plant-based or home concoctions rather than WHO-recommended ACTs.³⁴ While these practices diverge from formal clinical guidelines, they remain a reality of health-seeking behaviour in these communities.

The main objective of this study is to assess the KAP of community members regarding the management of UM in Kumba and Muyuka, Cameroon, to inform targeted, context-specific interventions.

Materials and methods

Ethics statement

This study followed the Declaration of Helsinki for research involving human participants World Medical Association.³⁵ Participants received study information and provided written informed consent (online supplemental S1 appendix). When literacy was limited, documented verbal informed consent was obtained. In such cases, the interviewer read the consent information aloud; on the participant’s agreement, the interviewer signed and dated a designated section of the consent form to verify and record their willingness to participate. All data were anonymised and stored securely.

Study design

A community-based cross-sectional study was conducted from 1 March to 31 March 2025 in Kumba and from 1 June to 30 June 2025 in Muyuka. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (see online supplemental S2 appendix).

Study area and setting

Kumba and Muyuka are located in the Southwest Region of Cameroon, a hyperendemic zone characterised by perennial, high malaria transmission that peaks during the rainy season. Plasmodium falciparum accounts for the vast majority of infections, transmitted primarily by the Anopheles gambiae complex.¹² In addition to the high malariometric burden, the area faces limited healthcare access, poor environmental drainage, high population density and a large influx of IDPs due to the ongoing Anglophone crisis.²² Kumba had a population of 144 413 and Muyuka 86 268 in 2024–2025.^{36 37}

Kumba serves as an economic hub in the region and comprises three administrative municipalities, Kumba I, Kumba II and Kumba III. The area is characterised by densely populated urban centres surrounded by periurban agricultural settlements. Muyuka is a single semiurban municipality with a strong agrarian base. Both areas experience similar challenges, including limited healthcare access, poor environmental drainage and an influx of IDPs due to the ongoing Anglophone crisis.²²

Study participants and target sample size

Adults aged ≥18 years who had resided in Kumba or Muyuka for at least 3 months prior to the study.

The minimum sample size per municipality was calculated using the Cochran’s formula for cross-sectional studies ($n_0=\frac{Z^2.p.(1-p)}{d^2}$) with an estimated proportion of the population with relevant knowledge or practices (p=0.546^{16 31}), a finite population correction and a design effect of 1.5. The final target sample sizes were 671 for Kumba (336 each for Kumba I and Kumba II) and 669 participants for Muyuka (see online supplemental S3 appendix; supplementry table 1).

Eligibility criteria

Eligible participants were those who had resided in either Kumba or Muyuka for at least 3 months, were aged ≥18 years and were able to communicate in either English or Pidgin.

Individuals with formal healthcare training (nurses, doctors, laboratory technicians) were excluded to prevent overestimation of community knowledge levels. Additionally, individuals with severe cognitive impairments or profound hearing/speech disabilities without available sign language interpreters were excluded due to an inability to independently provide informed consent or complete the interview. Participation was entirely voluntary, and individuals could withdraw at any time without consequence.

Sampling method

A multistage sampling strategy was used:

First stage: Kumba comprises three administrative municipalities (Kumba I, II and III), from which two were randomly selected to ensure representativeness while maintaining feasibility under prevailing security constraints. In contrast, Muyuka is a single municipality; therefore, it was included in its entirety.

Second stage: the number of quarters selected in each municipality reflected differences in population size, geographical spread and field accessibility. Kumba, being larger and more densely populated, required more quarters to achieve adequate coverage, whereas fewer quarters were sufficient in Muyuka (figure 1).

Figure 1. Multistage sampling procedure for participant recruitment.

Third stage: within selected quarters, streets were selected by balloting. Earmarked streets were listed, safer ones identified and names were folded and randomly drawn from a bowl.

Final stage: following street selection, households within each selected street were identified with the assistance of quarter heads and community health workers. Due to the absence of structured street layouts and household listings in some areas, purposive sampling was used. Households were selected based on accessibility, safety and willingness to participate. Within each selected household, one eligible adult (≥18 years) who had resided in the area for at least 3 months was recruited for interview. When more than one eligible adult was present, one participant was selected using simple random selection (eg, balloting). The unit of analysis for this study was the individual participant.

Data collection tool and method

Structured questionnaires were designed to assess participants’ KAP regarding UM management.

Interviewers were trained in ethical research practices, questionnaire design and administration before field work. The tool was developed in English and administered through face-to-face interviews in either English or Pidgin English,^{16 20} tailored to the Kumba and Muyuka context. To validate self-reported malaria experiences in the practices section, particularly for participants who demonstrated poor symptom knowledge earlier in the survey, interviewers probed using locally recognisable symptom descriptions in Pidgin (eg, ‘hot body’ for fever or ‘cold’ for chills) and cross-checked if these episodes prompted the participant to seek care or take antimalarials. The questionnaire was divided into five sections: demographic details, KAP and additional information on malaria prevention and control.

The knowledge section included both general and UM-specific knowledge-specific management, using a mix of multiple-choice and single-response items. Attitudes were assessed using Likert-scale and fixed-response questions, on perceived severity, personal concern and health-seeking behaviour. Practices were evaluated using behaviour-based items that explored the use of LLINs, treatment adherence and preventive actions. The final section addressed sources of information, challenges in malaria control and access to healthcare services. Each interview lasted approximately 15–20 min, providing adequate time for participants to respond thoroughly to all sections.

The questionnaire was pilot tested with 75 participants in Confidence Street and Owe Road. Based on feedback, open-ended questions were refined, and new items on LLIN ownership and use as well as the causative agent of malaria were included.

Data were recorded manually and responses were coded for quantitative analysis using a binary scoring system (0 for incorrect or negative responses; 1 for correct or positive responses). Aggregate domain scores were categorised into ordinal levels (eg, poor, moderate/fair, good) based on percentage thresholds (see online supplemental S3 appendix; supplemental table 2).

Variables and definitions

In this study, UM management was operationalised as a composite of community-level KAP relating to symptom recognition, care-seeking behaviour and prevention (including LLIN ownership and use), as measured through the structured questionnaire.

This study examined variables across three core domains: KAP towards UM management. The primary outcome variables were composite scores representing levels of KAP regarding UM management, while sociodemographic and behavioural characteristics served as independent variables.

Dependent variables

Each KAP domain was quantified using composite scores based on responses from relevant sections of the questionnaire (online supplemental S1 appendix). The scoring and classification followed established methods in the KAP literature.38,41

Knowledge score (K-score)

Calculated from 12 objective items related to both general malaria (Q10-Q15) and UM (Q16-Q21). Each correct answer received a score of 1, and incorrect or ‘I don’t know’ answers scored 0. Total scores were converted to percentages and categorised into: poor (<50%), moderate (50–74%) and good (≥75%).^{40 41}

Attitude score (A-score)

Derived from six questions (Q22-Q27), including two Likert-scale questions and four categorical ones. Responses reflecting a positive perception or health-seeking behaviour were coded as 1, while negative or uncertain responses were coded 0. Overall scores were expressed as percentages and grouped into: negative (<50%), neutral (50%–74%) and positive (≥75%).⁴¹

Practice score (P-score)

Assessed based on two ownership and utilisation of LLINs questions and nine behaviour-based questions (Q28–Q36). Responses indicating desirable prevention or treatment-seeking behaviours scored 1; undesirable behaviours scored 0. Percentage scores were then classified as: poor (<50%), fair (50%–74%) and good (≥75%).

Independent variables

The following sociodemographic and contextual variables assessed as potential predictors of KAP outcomes included age, gender, educational level, occupation, marital status, area and duration of residence, ownership and frequency of LLINs usage, and access to malaria-related information.

Data analysis methods

Data were entered, cleaned and analysed using the SPSS V.26.0.⁴² Descriptive statistics were used to summarise participants’ characteristics and KAP outcomes.

Scoring and categorisation

The scoring framework for the KAP and access to information domains is summarised in online supplemental table S2. Responses were scored using a binary system (1=correct/positive/desirable, 0=incorrect/negative/undesirable). Composite scores were then calculated as percentages and categorised into three levels: poor, moderate and good, following Bloom’s cut-off points as detailed in the table.

Reliability testing

Internal consistency was assessed using Cronbach’s alpha (α), a method suitable for the sample size following psychometric guidelines.⁴³ Internal consistency scores for each composite domain were as follows: knowledge (12 items; α=0.618), attitudes (6 items; α=0.367) and practices (11 items; α=0.358). The knowledge scale demonstrated moderate internal consistency, while the attitude and practice scales showed low internal consistency. These values are comparable to those reported in similar KAP studies, where heterogeneous behavioural and perception items often produce lower Cronbach’s alpha values.43,45 Lower Cronbach’s alpha values for attitude and practice domains are expected in KAP studies because these domains measure diverse behaviours, perceptions and decision-making processes rather than a single latent construct.^{46 47}

Inferential statistics

To examine associations between independent variables (age, gender, education) and KAP levels, Pearson χ² and one-way analysis of variance were initially performed. A p value <0.05 was considered statistically significant.

Variables that demonstrated significant bivariate associations (p<0.05) were included in multivariable logistic regression models to identify independent predictors of good preventive practices. To provide a comprehensive understanding of the data, analyses were conducted in two stages. First, a pooled multivariable model was run on the entire combined dataset (adjusting for municipality) to identify overall predictors. Second, municipality-stratified models were conducted separately for Kumba and Muyuka to identify context-specific predictors and observe how relationships differed by municipality. Multicollinearity was assessed using the variance inflation factor, with a threshold of <5 considered acceptable.^{48 49} Finally, to formally assess the Knowledge–Attitude–Practice (K → A → P) theoretical pathway, Pearson correlation analyses were conducted using continuous composite scores for KAP. These relationships were assessed for the pooled sample as well as stratified by municipality to determine whether knowledge predicts attitudes and whether either domain predicts preventive practices.

Data quality control

Data quality was assured through double entry verification and logical checks to minimise errors during data entry. Cronbach’s alpha was also used to assess the internal consistency of each KAP domain, with a threshold of ≥0.70 deemed acceptable for internal reliability. All incomplete or inconsistent questionnaires were excluded from the analysis. Responses related to LLIN ownership, usage the previous night and routine use, as well as malaria causative agent knowledge, were cross-checked for concordance. Any discrepancies led to the exclusion of 14 respondents.

Patient and public involvement

Patients and members of the public were not involved in the design, conduct, reporting or dissemination plans of this study.

Results

A total of 1500 questionnaires were administered equally between Kumba and Muyuka by trained interviewers. Following the exclusion of 126 questionnaires due to incompleteness and 14 due to discordance in LLIN ownership and usage patterns, 1244 questionnaires with valid responses 638 from Kumba (283 from Kumba I, 355 from Kumba II) and 606 from Muyuka were included in the final analysis.

Demographic characteristics of study participants

Among the 1244 participants, most were young adults (67.4%), with a higher proportion in Muyuka than Kumba (76.4% vs 58.9%, p<0.001). Females predominated (64.9%) and educational attainment was similar across municipalities. Students formed the largest occupational group (60.1%), more so in Muyuka (63.4%) than Kumba (57.1%, p=0.001).

Marital status differed, with more singles in Muyuka (57.9%) and more married in Kumba (54.4%, p<0.001). Mean duration of residence varied significantly, longer in Kumba (6.42±1.32 years) than Muyuka (1.86±0.85, p<0.001) (table 1).

Table 1. Sociodemographic characteristics of study participants by municipality.

Variable	Category	Total (n=1244)	Kumba (n=638)	Muyuka (n=606)	χ2/H/t	P value
Age group (years)	Young adults (18–29)	839 (67.4)	376 (58.9)	463 (76.4)	χ2= 47.94	<0.001
	Middle-aged (30–44)	360 (28.9)	226 (35.4)	134 (22.1)
	Older adults (45–60)	45 (3.6)	36 (5.6)	9 (1.5)
	Median (Q1–Q3)	25 (20–31)	27 (22–35)	23 (20–29)	H=57.34	<0.001
Gender	Male	437 (35.1)	211 (33.1)	226 (37.3)	χ2 = 2.43	0.119
	Female	807 (64.9)	427 (66.9)	380 (62.7)
Education level	No formal education	557 (44.8)	283 (44.4)	274 (45.2)	χ2 = 3.75	0.153
	Primary	407 (32.7)	198 (31.0)	209 (34.5)
	Secondary	280 (22.5)	157 (24.6)	123 (20.3)
Occupation	Student	748 (60.1)	364 (57.1)	384 (63.4)	χ2=15.50	0.001
	Self-employed	226 (18.2)	110 (17.2)	116 (19.1)
	Employed	203 (16.3)	117 (18.3)	86 (14.2)
	Unemployed	67 (5.4)	47 (7.4)	20 (3.3)
Marital status	Single	642 (51.6)	291 (45.6)	351 (57.9)	χ2=18.86	<0.001
	Married	602 (48.4)	347 (54.4)	255 (42.1)
Duration of residence	9–12 months	403 (32.4)	215 (33.7)	188 (31.0)	χ2 = 4.44	0.218
	1–5 years	374 (30.1)	188 (29.5)	186 (30.7)
	6–10 years	380 (30.5)	199 (31.2)	181 (29.9)
	>10 years	87 (7.0)	36 (5.6)	51 (8.4)
	Mean duration of stay±SD	4.20±2.54	6.42±1.32	1.86±0.85	t=−72.87	<0.001

Values are n (%) unless stated. Median (Q1–Q3) reported for age; mean±SD for duration of residence. Pearson’s χ2 used for categorical variables, Kruskal-Wallis for skewed data, and t-test for normally distributed variables. Statistically significant p values (<0.05) are shown in bold.

Knowledge of UM

Table 2 presents participants’ knowledge of UM. Most recognised malaria as mosquito-transmitted (97.8%) and parasite-caused (84.0%), with no significant differences between Kumba and Muyuka.

Table 2. Knowledge of uncomplicated malaria among participants in Kumba and Muyuka.

Knowledge variable	Total (n=1244)	Kumba (n=638)	Muyuka (n=606)	χ2	P value
Understanding of malaria
Caused by parasite	1045 (84.0)	527 (82.6)	518 (85.5)	1.914	0.167
Transmitted by mosquitoes	1217 (97.8)	626 (98.1)	591 (97.5)	0.517	0.472
Bacterial disease	119 (9.6)	67 (10.5)	52 (8.6)	1.325	0.250
Symptoms recognised
Fever	1071 (86.1)	545 (85.4)	526 (86.8)	0.491	0.483
Chills	446 (35.9)	236 (37.0)	210 (34.7)	0.738	0.390
Headache	578 (46.5)	287 (45.0)	291 (48.0)	1.151	0.283
Nausea and vomiting	467 (37.5)	237 (37.1)	230 (38.0)	0.086	0.769
Muscle aches	321 (25.8)	162 (25.4)	159 (26.2)	0.116	0.733
Prevention methods known
Use of LLINs	971 (78.1)	510 (79.9)	461 (76.1)	2.710	0.100
Indoor residual spraying	417 (33.5)	203 (31.8)	214 (35.3)	1.704	0.192
Protective clothing	382 (30.7)	195 (30.6)	187 (30.9)	0.013	0.911
Taking prophylaxis	485 (39.0)	240 (37.6)	245 (40.4)	1.033	0.310
Avoiding mosquito areas	246 (19.8)	124 (19.4)	122 (20.1)	0.095	0.758
Management options recognised
Use of antimalarial drugs	877 (70.5)	439 (68.8)	438 (72.3)	1.798	0.180
Hospital diagnosis	485 (39.0)	261 (40.9)	224 (37.0)	2.034	0.154
Herbal remedies	319 (25.6)	160 (25.1)	159 (26.2)	0.219	0.640
Knowledge of uncomplicated malaria
Yes	931 (74.8)	478 (74.9)	453 (74.8)	0.005	0.945
No	313 (25.2)	160 (25.1)	153 (25.2)
Diagnosis of uncomplicated malaria
Physical examination	497 (40.0)	252 (39.5)	245 (40.4)	0.171	0.918
Symptoms only	136 (10.9)	69 (10.8)	67 (11.1)
Laboratory test	611 (49.1)	317 (49.7)	294 (48.5)
Overall knowledge level
Poor	566 (45.5)	290 (45.5)	276 (45.5)	1.045	0.593
Moderate	505 (40.6)	265 (41.5)	240 (39.6)
Good	173 (13.9)	83 (13.0)	90 (14.9)
Mean knowledge score±SD	53.28±16.6	52.97±16.6	53.62±16.70	0.477	0.490

Values are presented as number (percentage). Participants could select more than one response for symptom and prevention variables. Comparisons between municipalities were performed using Pearson’s χ² test. Knowledge scores were categorised as poor (<50%), moderate (50–74%) and good (≥75%).

LLINs, long-lasting insecticidal nets.

Fever was the most commonly identified symptom (86.1%), followed by headache (46.5%) and chills (35.9%). LLIN use was the leading preventive method known (78.1%), while hospital diagnosis (39.0%) and herbal remedies (25.6%) were less cited. Overall, 74.8% reported knowing UM, but only 13.9% achieved good knowledge scores. Mean knowledge was similar across municipalities (53.3±16.6, p=0.490), indicating comparable awareness levels (table 2).

Attitudes towards UM management

Table 3 shows generally positive attitudes towards UM management in both municipalities. Most participants reported they would seek medical care immediately (92.1%) and trusted healthcare providers (83.5%). Only a small proportion reported self-medication or herbal remedy use (3.7%) or financial barriers to care (3.4%). Most preferred formal health facilities for treatment (94.1%) and expressed confidence in conventional malaria treatment (74.4%). Overall, positive attitudes were reported by 69.9% of participants, with no significant differences between Kumba and Muyuka across attitude variables (table 3).

Table 3. Attitudes towards uncomplicated malaria management in Kumba and Muyuka.

Attitude variable	Category	Total (n=1244)	Kumba (n=638)	Muyuka (n=606)	χ² (p value)
Would seek medical care immediately	Yes	1146 (92.1)	588 (92.2)	558 (92.1)	0.003 (0.956)
	No	98 (7.9)	50 (7.8)	48 (7.9)
Trust in healthcare providers	Yes	1039 (83.5)	533 (83.5)	506 (83.5)	<0.01 (0.983)
	No	205 (16.5)	105 (16.5)	100 (16.5)
Urgency and severity of malaria	Yes	497 (40.0)	250 (39.2)	247 (40.8)	0.321 (0.571)
	No	747 (60.0)	388 (60.8)	359 (59.2)
Self-medication/herbal remedies	Yes	46 (3.7)	24 (3.8)	22 (3.6)	0.015 (0.902)
	No	1198 (96.3)	614 (96.2)	584 (96.4)
Financial constraints	Yes	42 (3.4)	23 (3.6)	19 (3.1)	0.210 (0.647)
	No	1202 (96.6)	615 (96.4)	587 (96.9)
Health literacy/knowledge	Yes	479 (38.5)	245 (38.4)	234 (38.6)	0.006 (0.939)
	No	765 (61.5)	393 (61.6)	372 (61.4)
Preferred place for care	Clinic/hospital/health centre	1171 (94.1)	605 (94.8)	566 (93.4)	2.963 (0.397)
	Other (none, from god, pharmacy, TH)	73 (5.9)	33 (5.2)	40 (6.6)
Feelings toward conventional malaria treatment	Avoid whenever possible	67 (5.4)	32 (5.0)	35 (5.8)	3.252 (0.354)
	Prefer traditional remedies	120 (9.6)	64 (10.0)	56 (9.2)
	Hesitant but willing	131 (10.5)	76 (11.9)	55 (9.1)
	Confident and trusting	926 (74.4)	466 (73.0)	460 (75.9)
Overall attitude	Negative	23 (1.8)	10 (1.6)	13 (2.1)	0.964 (0.618)
	Neutral	352 (28.3)	186 (29.2)	166 (27.4)
	Positive	869 (69.9)	442 (69.3)	427 (70.5)

Values are presented as number (percentage). Comparisons between municipalities were performed using Pearson’s χ2 test. Percentages may not total 100 due to rounding. P<0.05 was considered statistically significant.

TH, traditional healer.

Practices towards UM management

Table 4 presents malaria prevention practices and care-seeking behaviours among participants in Kumba and Muyuka. LLIN ownership was significantly higher in Kumba (81.7%) than Muyuka (57.1%, p<0.001). Similarly, sleeping under an LLIN the previous night was more common in Kumba (69.7%) compared with Muyuka (42.9%, p<0.001). Frequency of LLIN use per week also differed significantly between municipalities (p<0.001), with more participants in Muyuka reporting never using LLINs. IRS, bush clearing and use of protective clothing or repellents were common preventive practices, with no significant differences between municipalities. Most participants sought care within 2 to 3 days after symptom onset, with a median of 3 days in both municipalities. Malaria treatment completion was reported by 82.1% of participants. Overall prevention practice levels were mostly fair (61.9%), while 26.5% demonstrated good practices and 11.6% poor practices, with no significant difference between municipalities (table 4).

Table 4. Malaria prevention practices and care-seeking behaviours in Kumba and Muyuka.

Practice variable	Category	Total (n=1244)	Kumba (n=638)	Muyuka (n=606)	χ2/F	P value
LLIN ownership	Yes	867 (69.7)	521 (81.7)	346 (57.1)	χ2=88.80	<0.001
	No	377 (30.3)	117 (18.3)	260 (42.9)
Slept under LLIN previous night	Yes	705 (56.7)	445 (69.7)	260 (42.9)	χ2 =91.21	<0.001
	No	539 (43.3)	193 (30.3)	346 (57.1)
Frequency of LLIN use per week	4–6 nights	299 (24.0)	183 (28.7)	116 (19.1)	χ2 =47.63	<0.001
	1–3 nights	288 (23.2)	119 (18.7)	169 (27.9)
	Never	206 (16.6)	76 (11.9)	130 (21.5)
Indoor residual spraying	Yes	824 (66.2)	410 (64.3)	414 (68.3)	χ2 =2.28	0.131
	No	420 (33.8)	228 (35.7)	192 (31.7)
Bush clearing around home	Yes	1212 (97.4)	624 (97.8)	588 (97.0)	χ2 = 0.75	0.388
	No	32 (2.6)	14 (2.2)	18 (3.0)
Use of protective clothing/repellent	Yes	724 (58.2)	371 (58.2)	353 (58.3)	χ2 = 0.001	0.971
	No	520 (41.8)	267 (41.8)	253 (41.7)
Time to seek care after symptoms (days)	1	100 (8.1)	53 (8.4)	47 (7.9)	χ2 = 1.07	0.957
	2–3	680 (55.4)	343 (54.4)	337 (56.4)
	4–7	356 (29.0)	186 (29.5)	170 (28.4)
	9–14	43 (3.5)	24 (3.8)	19 (3.2)
	15–21	21 (1.7)	11 (1.7)	10 (1.7)
	>21	28 (2.3)	13 (2.1)	15 (2.5)
	Median (Q1–Q3)	3 (2–5)	3 (2–5)	3 (2–5)	H ≈ 0.00	0.957
Completed malaria treatment	Yes	1021 (82.1)	514 (80.6)	507 (83.7)	χ2 = 2.03	0.154
	No	223 (17.9)	124 (19.4)	99 (16.3)
Overall prevention practice level	Poor	144 (11.6)	63 (9.9)	81 (13.4)	χ2 = 4.226	0.121
	Fair	770 (61.9)	397 (62.2)	373 (61.6)
	Good	330 (26.5)	178 (27.9)	152 (25.1)
	Median practice (Q1–Q3)	66.7 (55.6–77.8)	66.7 (55.6–77.8)	66.7 (55.6–77.8)	H=1.393	0.238

Values are presented as number (percentage) unless otherwise stated. Comparisons between municipalities were performed using Pearson’s χ² test for categorical variables and one-way analysis of variance (ANOVA) for continuous variables. Practice scores were categorised as poor (<50%), fair (50–74%) and good (≥75%). Bold numbers are significant p values.

LLINs, long-lasting insecticidal nets.

Factors associated with good prevention practices

Multivariate logistic regression analysis was conducted to identify factors associated with good UM prevention practices across both municipalities and within each municipality separately (online supplemental S3 appendix; supplementry table S3).

In the pooled analysis for Kumba and Muyuka combined, gender, age group, knowledge level, attitude level and access to malaria information were significantly associated with prevention practices. Females were less likely to demonstrate good prevention practices compared with males (AOR=0.73, 95% CI 0.55 to 0.97, p=0.030). Age was also significantly associated, with middle-aged adults (30–44 years) and older adults (45–60 years) less likely to have good practices compared with young adults (AOR=0.34, 95% CI 0.18 to 0.67, p=0.002; AOR=0.42, 95% CI 0.22 to 0.81, p=0.009, respectively).

Knowledge level showed a significant association with prevention practices. Using participants with ‘Good’ knowledge as the reference group, those with poor knowledge were significantly less likely to demonstrate good prevention practices (AOR=0.54, 95% CI 0.36 to 0.83, p=0.004), as were those with moderate knowledge (AOR=0.64, 95% CI 0.43 to 0.94, p=0.022). Attitude level was also associated with prevention practices; compared with those with positive attitudes, participants with neutral attitudes were less likely to have good practices (AOR=0.73, 95% CI 0.53 to 0.99, p=0.045). Access to malaria information was an important predictor: compared with participants with good access to information, those with poor access were significantly less likely to demonstrate good prevention practices (AOR=0.52, 95% CI 0.35 to 0.76, p=0.001).

Municipality, occupation and education level were not significantly associated with good prevention practices in the pooled model.

Municipality-specific analyses showed some variation in associated factors. In Kumba, older age, poorer knowledge and neutral attitudes were significantly associated with reduced odds of good prevention practices. In contrast, in Muyuka, age and access to information were the primary factors; specifically, participants with poor access to information were significantly less likely to demonstrate good practices compared with those with good access (AOR=0.44, 95% CI 0.25 to 0.76, p=0.004).

Overall, the regression analysis showed that demographic factors, knowledge levels, attitudes and access to malaria information were important predictors of good UM prevention practices, although the strength and significance of these associations varied between municipalities.

Assessment of the K → A → P pathway

To assess the Knowledge–Attitude–Practice (K → A → P) pathway, Pearson correlation analysis was conducted using continuous composite scores (see online supplemental S3 appendix; supplementry table S4a). Knowledge was not significantly associated with attitudes (r=0.045, p=0.114). However, both knowledge and attitudes showed weak positive correlations with practices (K → P: r=0.148, p<0.001; A → P: r=0.108, p<0.001).

Stratified analysis showed variation by municipality. Knowledge was weakly associated with attitudes in Kumba (r=0.083, p=0.037) but not in Muyuka (r=0.005, p=0.898). In both settings, knowledge and attitudes remained weakly correlated with practices.

Overall, the classical K → A → P pathway was not fully supported, with knowledge influencing practices more directly and varying by context.

Discussions

Summary of key findings

This study explored KAP related to UM in Kumba and Muyuka. Comprehensive knowledge was generally poor, with only 13.9% of the pooled sample achieving a good knowledge score, despite high recognition of basic symptoms like fever (86.1%). The ownership of LLINs was significantly higher in Kumba (81.7%) than Muyuka (57.1%) (p<0.001), though usage among owners was similar.

Multivariate logistic regression identified several key predictors of good preventive practices. In the pooled analysis, younger age, male gender, good overall knowledge, positive attitudes and good access to malaria information were all significantly associated with increased odds of good prevention practices. Importantly, the drivers of prevention varied by context: in Kumba, poor knowledge and neutral attitudes were the strongest predictors of poor practices, whereas in Muyuka, poor access to information was the primary barrier. Furthermore, correlation analysis indicated that the classical KAP sequential pathway was only partially supported; knowledge and attitudes independently influenced practices, but knowledge did not significantly predict attitudes.

Interpretation of findings

Knowledge of UM

Knowledge gaps remain despite general awareness of malaria, but limited comprehensive knowledge. Cameroonian studies show that over 90% understand mosquito transmission but fewer comprehend less common symptoms or causative agents.^{30 47 48} Our multivariate analysis confirmed that higher overall knowledge scores significantly predicted better prevention practices; however, only a small fraction of the population achieved ‘good’ knowledge.

Recognition that malaria is mosquito-borne was associated with significantly better prevention in Kumba, aligns with findings in Douala where transmission knowledge underpins protective behaviour.^{7 21 50} Access to malaria information increased odds of good prevention: two times in Kumba and fourfold in Muyuka.

Furthermore, multivariate analysis highlighted significant sociodemographic disparities. Middle-aged and older adults demonstrated poorer preventive practices compared with younger adults, and females were less likely to report good preventive practices than males. These findings underscore the need for demographic-specific educational outreach and targeted malaria prevention interventions. Similar demographic disparities in malaria prevention have been documented in other sub-Saharan African contexts, where younger adults often exhibit better health-seeking behaviours due to recent educational exposure, while gender dynamics can influence access to household resources like LLINs.^{15 27} Good access to malaria information significantly increased the odds of good prevention, proving to be the most critical driver in Muyuka.

Attitudes towards UM management

Confidence in conventional treatment (74.4%) and care-seeking willingness (92.1%) did not guarantee better prevention. In Kumba perceiving malaria as severe reduced preventive behaviour echoing urban findings where severity perception did not translate into action.^{21 51} In Muyuka trust in healthcare providers strongly predicted care-seeking and those practising prevention had more positive attitudes. This is similar to findings reported in Ethiopia by Addis & Gebeyehu Wondmeneh.³⁸

Practices towards UM management

The ownership of LLINs varied by town but usage remained low, consistent with patterns in the Tiko, Santa and Bamenda health districts, where ownership may outpace use due to discomfort or poor maintenance.^{14 18 52 53} Multimodal prevention behaviours, protective clothing, herbal remedies and awareness of anaemia suggest that integrated strategies, rather than nets alone, may yield stronger outcomes. Higher knowledge scores among those practising prevention further confirm that deeper understanding supports action.^{30 52 54}

The findings of this study align with other research conducted in SSA, particularly with regard to the challenge of translating knowledge and attitudes into effective malaria control practices. Studies in Nigeria, Tanzania and Uganda have similarly highlighted that while most people recognise the importance of LLINs and the threat posed by malaria, their usage is often inconsistent or influenced by competing priorities, such as financial constraints or a lack of perceived risk.^{19 55 56} Additionally, like our study, these studies revealed that despite high levels of malaria knowledge, self-medication remains a common practice, particularly in rural areas where healthcare access is limited.⁵⁷

The KAP framework assumes a sequential pathway from Knowledge to Attitude to Practice. In this study, this pathway was only partially supported. While both knowledge and attitudes were independently associated with preventive practices, knowledge was not significantly associated with attitudes in the pooled analysis. This weak and inconsistent K → A relationship suggests that increased factual knowledge about malaria does not necessarily translate into more positive health attitudes.

This finding aligns with evidence that knowledge alone is often insufficient to drive behavioural change, particularly in settings where cultural beliefs, prior experiences and trust in health systems strongly shape attitudes.^{58 59} In our stratified analysis, the relationship between knowledge and attitudes was significant in Kumba but absent in Muyuka, indicating that the KAP pathway is context-dependent. Notably, in both settings, knowledge and attitudes were weakly but consistently associated with practices, suggesting more direct or parallel influences on behaviour.

These results imply that malaria control strategies should move beyond information-based interventions and incorporate approaches that address community beliefs, trust in healthcare and sociocultural norms. Behaviour change frameworks that integrate both cognitive and affective components are more likely to produce sustained improvements in preventive practices.⁶⁰

Recommendations

Based on the findings of this study, several key recommendations are made for improving malaria prevention and management practices specifically in the municipalities of Kumba and Muyuka:

• Develop targeted education campaigns that emphasise less familiar symptoms, causative organisms and complications, with content adapted to local education levels and cultural context.

• Promote consistent nightly LLIN use by addressing practical barriers like heat and perceived suffocation through behaviour change communication and improved net design.

• Engage trusted healthcare workers and community leaders to deliver prevention messaging and encourage prompt care-seeking.

• Acknowledge the cultural reality of traditional and herbal remedies in community education programmes. Rather than alienating community members by strictly condemning these practices, health workers should encourage prompt diagnostic testing and the use of formal antimalarials (ACTs) as the primary treatment, while addressing traditional medicine as a potentially complementary, rather than alternative, approach.

Conclusions

General awareness of malaria transmission and symptoms is high in Kumba and Muyuka, but comprehensive knowledge and prevention behaviour remain suboptimal. High confidence in treatment and LLIN ownership alone are inadequate. Effective malaria control must combine deeper knowledge, access to credible information, cultural sensitivity and trusted local interventions tailored to community contexts.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.