Estimating malaria risk behaviours and their determinants among at-risk populations in a pre-elimination setting, Sisaket Province, Thailand–Cambodia border

Abstract

Background

Human behavioural and occupational factors play a critical role in sustaining malaria transmission. This study aimed to identify factors associated with poor levels of knowledge, attitudes, and practices (KAP) and factors influencing risk behaviours among at-risk populations.

Methods

A survey of 402 at-risk individuals was conducted in Sisaket Province in late 2022, where substantial progress in malaria elimination has been achieved. Data was collected using a structured questionnaire. Descriptive statistics and logistic regression models were used to identify factors associated with KAP and malaria risk behaviours.

Results

Over half of the participants demonstrated good knowledge (51.5%), attitudes (64.7%), and practices (58.2%) regarding malaria transmission, treatment, and prevention. Poor attitudes were significantly more likely among individuals with lower education levels [primary or below: aOR = 2.2, 95% CI 1.2–4.0); senior primary: aOR = 2.5, 95% CI 1.4–4.8] and low socioeconomic status (aOR = 1.8, 95% CI 1.1–3.0). Poor practices were less common among individuals working in agriculture (aOR = 0.2, 95% CI 0.1–0.5). Risk behaviours were significantly more frequent among males (aOR: 2.8, 95% CI 1.2–6.8) and individuals with poor practices (aOR: 3.5, 95% CI 1.6–8.1).

Conclusions

Despite overall moderate KAP levels, gaps in attitudes and risk behaviours persist. Targeted strategies, such as social and behaviour change communication, should focus on improving attitudes among individuals with low education levels and reducing risk behaviours among males and those with poor practices by promoting adherence to effective preventive measures.

Background

Malaria remains a major vector-borne disease, contributing substantially to mortality and morbidity in many tropical countries [1]. Its transmission depends on the interaction of three key components: (i) Anopheles vectors, (ii) Plasmodium parasites, and (iii) human hosts [2]. These components vary considerably across different geographical regions [3, 4]. Thailand exemplifies this spatial heterogeneity, with high malaria burdens concentrated along its western, southern, and northeastern international borders, which collectively account for more than 90% all reported cases nationwide [5–7]. Transmission patterns in Thailand can be broadly classified into two clusters: (i) the western and southern regions, where malaria affects individuals across all age groups, and (ii) the northeastern region, where transmission predominantly affects the working-age population [5]. These patterns indicate that different geographical areas have unique factors influencing malaria transmission dynamics.

Regional differences in malaria transmission dynamics and subsequent malaria infection are driven by variations in both vector ecology and human behaviour. In some areas, malaria transmission persists within villages [8], whereas in others it primarily occurs outside residential areas [9]. These differences reflect the presence of diverse local malaria vectors that vary regionally [10]. On the human side, key contributing factors include residing in high-risk areas [11–13], engaging in behaviours that increase exposure to vectors (e.g., nighttime outdoor activities without protection) [14–16], and harbouring asymptomatic infections that act as reservoirs for transmission [17]. The critical role of human movement in malaria transmission was highlighted during the 1981 outbreak in Thailand, which recorded 473,210 cases [18]. This outbreak was largely driven by political conflict in Cambodia during the 1970s–1980s, which forced large number of people to cross into malaria-receptive areas of Thailand [7]. More recently, political conflict in Myanmar led to a large influx of migrants into malaria-prone areas of Thailand, coinciding with an increase in malaria cases from 3,266 in 2021 to 10,156 in 2022 and 16,656 in 2023 [5]. These population movements underscore the persistent vulnerability of Thailand’s border regions. Although malaria burden was historically high, Thailand has implemented aggressive malaria control and elimination strategies aimed at achieving elimination by 2024 [19]; this target has since been revised to 2026 (https://www.aplma.org/country/thailand) [7]. These efforts, together with local adaptations and [5–7]. the absence of major cross-border conflicts in recent years, have contributed to significant case reduction, moving some regions toward zero indigenous malaria transmission and positioning them for prevention of re-establishment (PoR) leading to malaria elimination [20].

Sisaket Province, located in northeastern Thailand along the Cambodia border, is in the elimination phase and is projected to meet the national elimination target by 2026. Historically, Sisaket experienced a high malaria burden, with positivity rates of 16.1% (1,510/9,403) in 2012, 11.5% (1,194/10,364) in 2013, 12.1% (913/7,542) in 2014, 11.3% (1,219/10,742) in 2015, and 7.6% (356/4,699) in 2016 [21]. Most malaria cases (98.7%, 5,418/5,490) occurred among Thai nationals, with only 1.3% (72/5,490) reported among migrant [21]. Substantial progress has since been achieved through implementation of tailored national malaria elimination strategies [7], resulting in a sharp decline in positivity rates to 0.1% (2/1,829) in 2023 and 0.4% (1/263) in 2024 [21]. Despite this progress, Sisaket’s location along an international border presents a continued risk for malaria reintroduction. Cambodia reported large numbers of malaria cases from 2015 to 2023 (68,109 in 2015, 43,380 in 2016, 76,804 in 2017, 62,582 in 2018, 32,196 in 2019, 9,962 in 2020, 4,382 in 2021, 4,047 in 2022, and 1,382 in 2023) [22]. Within Cambodia, the bordering provinces of Preah Vihear and Oddar Meanchey, adjacent to Sisaket, reported 2,026, 11,324, 4,527, 4,306, 8,016, 2,793, 487, 216, 274, and 51 from 2014 to 2023, respectively [7]. Entomological surveys conducted in Phu Sing and Khun Han Districts in 2018 identified Anopheles dirus sensu lato (s.l.) as the primary malaria vector, particularly in rubber-forest areas [23]. These findings highlight a persistent risk of re-establishment of malaria from the neighboring country and underscore the importance of vigilant surveillance and cross-border collaboration.

Despite the substantial decline in malaria cases and progress toward elimination in Sisaket Province, limited research has examined how human-related factors influence residual transmission. Knowledge, attitudes, and practices (KAP) surveys are widely used to assess community awareness and behaviours that may sustain or interrupt malaria transmission. Previous studies in Sisaket [24, 25] and along the Thai-Cambodian border [26–28] reported relatively high levels of KAP; however, gaps likely persist.

This study aimed to identify factors associated with poor KAP and risk behaviours, defined as prolonged exposure in high-risk areas without preventive measures. Understanding these behavioural determinants is essential for tailoring interventions to strengthen malaria elimination efforts and prevent re-establishment (PoR) in border areas.

Methods

Study locations

Three villages, Huai Chan (Moo 4), Kan Throm Tai (Moo 4), and Non Thong Lang (Moo 10) located in Khun Han District, Sisaket Province, were selected for this study (Fig. 1). These sites were chosen based on their history of high malaria positivity rates between 2012 and 2016 [21], which were 21.6% (366/1,691) in Huai Chan, 47.5% (77/162) in Kan Throm Tai, and 33.8% (23/68) in Non Thong Lang. In comparison, the overall positivity rate for Sisaket Province during the same period was reported at 12.1% (5,192/42,750). Between 2020 and 2024, positivity rates in the study areas declined markedly to 0.4% (6/1,468) in Huai Chan, 1.1% (1/87) in Kan Throm Tai, and 0% (0/247) in Non Thong Lang, consistent with the provincial average of 0.4% (70/16,136) [21]. These villages were selected based on two main criteria: (i) proximity to rubber-forest ecotypes that provide breeding habitats for An. dirus, the primary local malaria vector [23]; and (ii) residents’ predominant occupations involving work in these ecotypes [7].

Fig. 1.

Locations of three survey villages in Khun Han District, Sisaket Province: A Moo4, Huai Chan; B Moo4, Kan Throm Tai; and C Moo10, Non Thong Lang

The study area is characterized by mountainous terrain in the south, covered by forests extending to the Cambodian border, sloping northwards to rubber plantations, paddy fields, and residential areas. According to 2022 data from the Huai Chan and Kan Throm Health Promotion Hospitals (Ministry of Public Health), the population sizes were 858 inhabitants (265 households) in Huai Chan, 672 inhabitants (190 households) in Kan Throm Tai, and 535 inhabitants (131 households) in Non Thong Lang.

Study design and data collection

A cross-sectional survey was conducted using a structured questionnaire previously developed and validated by the research teams [29, 30], and adapted for this study. The questionnaire was initially drafted in English, translated into Thai, and back-translated to ensure linguistic and conceptual accuracy. It consisted of five sections with predefined response options. The first section assessed participants’ socioeconomic status, including age, gender, occupation, education level, housing type, and household facilities. The second section evaluated malaria-related knowledge, including causes of infection, symptoms, and prevention methods. Participants with a history of malaria were also asked about diagnosis and treatment. Each knowledge item offered multiple response options, and one point was awarded for each correct answer.

The third section explored participants'perceptions of malaria through statements such as whether malaria can lead to death, whether children experience more severe symptoms than adults, and whether patients should complete a full course of treatment. Responses were recorded on a three-point Likert scale (agree, uncertain, disagree) [31], and scored as three, two, and one point, respectively, with reverse scoring for negatively worded items. Perception scores were then categorized as good, neutral, or poor. The fourth section assessed behaviours that increase the risk of exposure to primary malaria vectors, while the fifth section evaluated practices for preventing mosquito bites. One point was awarded for each correct preventive measures, and responses of “don’t know” received zero points. Overall KAP scores were summed and classified as good or poor, and selected practice items were used to define risk behaviours.

Data were collected through face-to-face interviews conducted by trained village health volunteers (VHVs) recommended by local Health Promoting Hospitals. Before data collection, groups of three to four VHVs from each village received training from the research team on questionnaire administration and interview techniques. Role-playing exercises were conducted, with volunteers alternating between interviewer and respondent roles, to strengthen interviewing skills, improve familiarity with the questionnaire, and identify potential sources of misunderstanding or bias.

During data collection, the research team closely supervised the interviews to ensure that VHVs avoided leading questions and accurately recorded responses. Random spot checks, covering 25% of participants, were performed to verify accuracy and completeness. To minimize social desirability bias, all participants were assured of anonymity and confidentiality, which encouraged honest reporting. Trained VHVs recruited participants by providing verbal explanations of the study, and written informed consent was obtained from each participant before enrollment. Data collection was conducted over a 2-month period, from November to December 2022.

Sample size determination

The minimal sample size required for this study was calculated using the formula below:

$$n=\frac{Z^{2}P\left(1-P\right)}{d^2}$$

where:

n = required sample size.

Z = Z-score corresponding to the desired confidence level (e.g., for 95% Confidence level) = 1.96.

P = estimated prevalence or proportion of the characteristic of interest in the population as 58% of population in communities living nearby the forest in Southern Thailand knew the cause of malaria, in a study by Saita et al. 2023, so P = 0.58 [32].

d = margin of error, expressed as a proportion (set at 0.05)

$$n=\frac{1.{96}^2\times 0.58\left(1-0.58\right)}{0.{05}^2}=375$$

To accommodate potential incomplete questionnaires, a 10% contingency was added to the initially calculated sample size, resulting in a total required sample of 412 participants for the survey.

Participants and recruitment

Approximately 20% of residents from each village were selected using 2021 census lists. Eligibility criteria included: (i) residence in the study area for at least 1 year, (ii) age 18 years or older, and (iii) willingness to participate. Eligible individuals were stratified by gender to ensure balanced representation of males and females. Participants were then randomly selected through a lucky draw. VHVs approached the selected individuals, explained the study objectives and procedures, and obtained written informed consent before conducting the questionnaire interviews.

Data entry and analysis

Data from the questionnaires were entered into Microsoft Access (Microsoft 365, Version 2501). KAP scores were calculated separately, and participants were categorized as having “good” or “poor” levels based on the mean score for each domain. Those scoring above the mean were classified as good, while scores below the mean were classified as poor.

Socioeconomic status was assessed using an asset-based scoring approach commonly applied in social science research [33–35]. Ten items were included: (i) house wall material, (ii) house roof material, (iii) bicycle, (iv) motorcycle, (v) car, (vi) tractor, (vii) truck, (viii) television, (ix) refrigerator, and (x) mobile phone. House wall materials were scored as 1 (wood) or 2 (brick/concrete), and roof materials as 1 (thatched grass), 2 (galvanized iron), or 3 (tile). Each remaining asset was scored as 1 if owned and 0 if not owned. Scores from all ten items were summed to generate an overall asset score for each participant. Based on the mean distribution, participants were classified into “lower” or “higher” socioeconomic groups.

Risk behaviours were classified as “high risk” if participants met all of the following criteria: (i) engaged in nighttime activities in rubber plantations or forested areas [23, 27, 36, 37], (ii) spent more than 10 nights per month outdoors in these risk areas—similar to evidence from Vietnam showing that spending 12 nights in forests nearly tripled malaria risk [38], and (iii) did not use effective personal preventive measures, such as long-lasting insecticidal nets (LLINs) or long-lasting insecticidal hammock nets (LLIHNs) [30].

Descriptive statistics, including frequencies and percentages, were used to summarize population characteristics, Knowledge, Attitude and Practice items, and risk behaviours. Crude odds ratios (cOR) with 95% confidence intervals (CI) were estimated using univariate logistic regression to assess associations between explanatory variables and outcomes (KAP or risk behaviours). Variables from the univariate analyses (nine for KAP and twelve for risk behaviours) were included in multivariable logistic regression models to compute adjusted odds ratios (aOR) with 95% CIs, controlling for potential confounders. Associations were considered statistically significant if the 95% CI did not include 1. Statistical analyses were conducted using SPSS for Windows, Version 24.0. (IBM Corp., Armonk, NY, USA).

Results

Sociodemographic characteristics

402 residents participated in this study (Table 1): 136 from Huai Chan (15.9% of the village population), 160 from Kan Throm Tai (23.8%), and 106 from Non Thong Lang (19.8%). The sex distribution was balanced, with 50.2% male and 49.8% female participants. The majority were aged 50 years or older (36.6%), followed by 29.1% aged 41–50 years, and 34.4% aged 18–40 years. Most respondents had completed only primary school (63.7%), 32.3% had at least secondary education, and 4.0% had no formal schooling. Agriculture was the principal occupation for 86.0%, while 14.0% worked in other sectors. Mobility was low: 80.1% were native to their village, and 19.9% were migrants. Household sizes typically ranged from 3 to 5 members (48.0%), with smaller (1–3 members, 29.1%) and larger (> 5 members, 22.9%) families less common. Based on an asset-based index, 78.6% were classified in the “higher” socioeconomic stratum, and 21.4% in the “lower” stratum (Table 1).

Table 1.

Population characteristics of participants from survey sample data between November and December 2022 in Khun Han District, Sisaket Province, Thailand

Response options (n = 402)	n	Percent
Gender
Female	200	49.8
Male	202	50.2
Age distribution
18–30	61	15.2
31–40	77	19.2
41–50	117	29.1
> 50	147	36.6
Education level
High school or higher	130	32.4
Senior primary	99	24.6
Junior primary and below	173	43.0
Occupation
Non-agriculture	56	14.0
Agriculture	346	86.0
Locality
Born in local area	322	80.1
Moved in	80	19.9
Family members
1–3	117	29.1
3–5	193	48.0
> 5	92	22.9
Socioeconomic status
Higher	316	78.6
Lower	86	21.4

All percentages were calculated using the total number of participants (n = 402)

Knowledge of the population regarding malaria infection and treatment

Malaria knowledge among participants was presented in Table 2. Most correctly identified mosquito bites (89.6%) specifically Anopheles spp. (89.1%) as the route of transmission and recognized forest exposure as a risk factor (60.7%). Key symptoms such as fever (91.3%), shivering (92.5%), and headache (60.9%) were well known.

Table 2.

Knowledge of participants toward malaria infection, symptoms, treatment, and prevention among participants

Response options (n = 402)	Responses
	n	Percent
Do you know how people get malaria?
Hardworking	15	3.7
Live in forest area	244	60.7
Germ	153	38.1
Live with patients	24	6.0
Mosquito bite	360	89.6
Drinking water in forest	19	4.7
What is the malaria vector?
Aedes mosquitoes	48	11.9
Anopheles mosquitoes	358	89.1
Culex mosquitoes	0	0.0
What are the symptoms of malaria?
Fever	367	91.3
Shivering	372	92.5
Tiredness	112	27.9
Loss of appetite	68	16.9
Weight loss	19	4.7
Muscle ache	122	30.3
Squeamish	105	26.1
Headache	245	60.9
How can malaria be treated?
Herb	4	1.0
Self–healing	9	2.2
Take a rest	46	11.4
Take medication	381	94.8
What is the cause of malaria if it cannot be prevented?
I don't know who get malaria	187	46.5
I don't know how to prevent malaria	281	69.9
Can malaria be prevented?
Don’t drink water in a stream	18	4.5
Sleep inside the net	382	95.0
Prevent mosquitoes bite	301	74.9
Keep healthy	31	7.7
Regularly test malaria	87	21.6
Prophylaxis drugs	6	1.5
Can a person be reinfected with malaria after recovering from a previous infection
Possible	360	89.6
Impossible	2	0.5
Uncertain	40	10.0
Do you think that some individuals may not fully recover from malaria even after receiving treatment?
Possible	214	53.2
Impossible	142	35.3
Uncertain	46	11.4
Inefficient drugs	2	0.9
Taking a drug incomplete course	141	65.9
Severe germ	60	28.0
Intermittent treatment	62	29.0
Malaria drug resistance	144	67.3
Uncertain	3	1.4

Participants can respond to more than one answer for each question

All percentages were calculated based on the total number of participants (n = 402). Participants were allowed to select more than one response per question

Re-infection was considered possible by 89.6% of respondents, while 10.5% were either uncertain or believed re-infection to be impossible. Understanding of treatment failure included incomplete drug regimens (65.9%) and parasite resistance to antimalarial drugs (67.3%). Overall, the survey findings indicate a high level of knowledge regarding malaria infection, symptoms, treatment, and prevention among participants. An assessment of factors associated with poor knowledge revealed no significant influences within the studied population (Table 3).

Table 3.

Factors associated with poor knowledge, attitudes, and practices

Factor (n = 402)	N	Knowledge		Attitude		Practice
		Poor	aOR (95% CI)	Poor	aOR (95% CI)	Poor	aOR (95% CI)
		n (%)		n (%)		n (%)
Gender
Female (Ref)	200	100 (50.0)		94 (47.0)		115 (57.5)
Male	202	95 (47.0)	1.1 (0.7, 1.8)	96 (47.5)	0.9 (0.5, 1.4)	115 (56.9)	0.9 (0.5, 1.4)
Age (years)
18–30 (Ref)	61	29 (47.5)		29 (47.5)		39 (63.9)
31–40	77	43 (55.8)	1.5 (0.7, 3.1)	41 (53.2)	1.3 (0.6, 2.9)	46 (59.7)	1.3 (0.6, 2.9)
41–50	117	54 (46.2)	0.8 (0.4, 1.8)	48 (41.0)	0.5 (0.2, 1.1)	64 (54.7)	1.2 (0.5, 2.7)
> 50	147	69 (46.9)	0.8 (0.4, 1.9)	72 (49.0)	0.6 (0.3, 1.4)	81 (55.1)	1.2 (0.5, 2.7)
Education
High school and above (Ref)	130	62 (47.7)		51 (39.2)		83 (63.8)
Senior primary	99	48 (48.5)	1.2 (0.7, 2.2)	51 (51.5)	2.5 (1.4, 4.8)	53 (53.5)	0.8 (0.4, 1.4)
Junior primary and below	173	85 (49.1)	1.2 (0.7, 2.2)	88 (50.9)	2.2 (1.2, 4.0)	94 (54.3)	0.7 (0.4, 1.3)
Occupation
Non-agriculture (Ref)	56	27 (48.2)		25 (44.6)		47 (83.9)
Agriculture	346	168 (48.6)	1 (0.6, 1.9)	165 (47.7)	1.2 (0.6, 2.2)	183 (52.9)	0.2 (0.1, 0.5)
Marital status
Widow (Ref)	15	8 (53.3)		8 (53.3)		9 (60.0)
Marriage	316	154 (48.7)	0.8 (0.2, 2.3)	149 (47.2)	0.8 (0.3, 2.6)	174 (55.1)	1.1 (0.3, 3.7)
Single	71	33 (46.5)	0.8 (0.2, 2.7)	33 (46.5)	0.9 (0.3, 3.2)	47 (66.2)	1.6 (0.4, 5.7)
Total family members
1–3 (Ref)	117	59 (50.4)		55 (47.0)		63 (53.8)
3–5	193	93 (48.2)	0.9 (0.5, 1.4)	92 (47.7)	1 (0.6, 1.7)	105 (54.4)	1.1 (0.7, 1.7)
> 5	92	43 (46.7)	0.9 (0.5, 1.6)	43 (46.7)	1 (0.6, 1.8)	62 (67.4)	1.5 (0.8, 2.8)
Role of the respondent in family
Spouse (Ref)	133	72 (54.1)		62 (46.6)		73 (54.9)
Head of family	147	66 (44.9)	0.6 (0.4, 1.2)	73 (49.7)	1.2 (0.7, 2.2)	83 (56.5)	1.1 (0.6, 2)
Other members	122	57 (46.7)	0.7 (0.3, 1.3)	55 (45.1)	1 (0.5, 2.1)	74 (60.7)	0.9 (0.5, 1.9)
Residential status
Moved in (Ref)	80	39 (48.8)		46 (57.5)		51 (63.7)
Born here	322	156 (48.4)	1 (0.6, 1.6)	144 (44.7)	0.6 (0.3, 1)	179 (55.6)	0.7 (0.4, 1.2)
Socioeconomic status
Higher (Ref)	316	147 (46.5)		140 (44.3)		182 (57.6)
Lower	86	48 (55.8)	1.4 (0.9, 2.3)	50 (58.1)	1.8 (1.1, 3.0)	48 (55.8)	0.9 (0.6, 1.5)

All percentages were calculated by row, using the total number of participants with good and poor groups for each answer; aOR adjusted odds ratio; The aOR was calculated using a combination of factors in the analysis, including (i) gender, (ii) age, (iii) education level, (iv) occupation, (v) marital status, (vi) number of family members, (vii) role in family, (viii) residential status, (ix) socioeconomic status; CI: confidence interval; Ref: reference group; p-value <0.05 

Attitude of participants towards malaria

In addition to knowledge, the survey explored participants’ attitudes to gain deeper insights into their perceptions of malaria (Table 4). Overall, participants exhibited broadly positive attitudes across four dimensions: cause, epidemiology, treatment, and prevention.

Table 4.

Attitude of participants toward malaria

Attitude/response (n = 402)	Agree (%)	Disagree (%)	Not sure (%)
Causes
Suffering from malaria is a matter of fate	10 (2.5)	350 (87.1)	42 (10.4)
Everyone has a chance of getting malaria	384 (95.5)	6 (1.5)	12 (3.0)
The rich do not usually suffer from malaria*	38 (9.5)	271 (67.4)	93 (23.1)
Foreign workers suffer from malaria more often than Thai people	87 (21.6)	92 (22.9)	223 (55.5)
Epidemiology
Severe malaria can cause death	387 (96.3)	2 (0.5)	13 (3.2)
Local people no longer die from malaria*	169 (42.0)	42 (10.4)	191 (47.5)
Malaria is a problem in this village	144 (35.8)	132 (32.8)	126 (31.3)
Children experience more severe illness than adults	198 (49.3)	24 (6.0)	180 (44.8)
Treatment
Patients must adhere to medication to be cured	368 (91.5)	34 (8.5)	0 (0.0)
Taking malaria medication may cause side effects*	194 (48.3)	23 (5.7)	185 (46.0)
Free treatment prevents some patients from taking medication*	26 (6.5)	181 (45.0)	195 (48.5)
Prevention
Malaria can be controlled and prevented	321 (79.9)	10 (2.5)	71 (17.7)
Prevention of malaria is the responsibility of the authorities*	126 (31.3)	218 (54.2)	58 (14.4)
Insecticide spraying in the home can help prevent malaria	179 (44.5)	50 (12.4)	173 (43.0)
Malaria can be eliminated from our village	260 (64.7)	38 (9.5)	104 (25.9)

The percentages were calculated using the total number of participants (n = 402), while an asterisk (*) indicates negative statements

Regarding the cause of malaria, most respondents (87.1%) rejected the belief that the disease is determined by fate or supernatural forces. A majority (95.5%) acknowledged that everyone is susceptible to malaria, and 67.4% believed that individuals with higher economic status also have a chance of contracting the disease.

In terms of epidemiology, nearly all participants (96.3%) recognized that severe malaria can be fatal. Opinions about recent malaria-related deaths in their villages were divided: 42.0% believed no deaths had occurred, 10.4% thought deaths had occurred, and 47.5% were unsure. Views on whether malaria remains a major community problem were similarly mixed, with 35.8% agreeing, 32.8% disagreeing, and 31.3% unsure. Half of the respondents (49.3%) felt that children develop more severe symptoms than adults, while 44.8% were uncertain.

Regarding attitudes toward malaria treatment, nearly all participants (91.5%) recognized the importance of completing the full course of antimalarial medication. Despite this high level of awareness, 48.3% expressed concerns about medication side effects, and 46.0% were uncertain about this issue. Additionally, nearly half (48.5%) were unsure whether providing antimalarial treatment free of charge contributed to patient non-compliance.

In terms of malaria prevention, most respondents (79.9%) believed that malaria is both controllable and preventable. More than half (54.2%) agreed that malaria prevention is not solely the responsibility of healthcare workers, while 31.3% disagreed and 14.4% were uncertain. Regarding specific preventive measures, 44.5% believed that using insecticide sprays inside homes can prevent malaria, although half of the participants were unsure about its effectiveness. Nonetheless, optimism about malaria elimination was strong, with 64.7% believing that malaria can be eradicated from their villages.

Multivariate logistic regression analysis (Table 3) revealed that both formal education and household wealth significantly predicted “poor” attitudes toward malaria transmission, treatment, and prevention. Compared to respondents who had completed at least high school (Grade 9), those with only junior primary education or less were more than twice as likely to hold “poor” attitudes (aOR = 2.2, 95% CI 1.2–4.0). Similarly, senior primary graduates were also more likely to have “poor” attitudes compared to high school graduates (aOR = 2.5, 95% CI 1.4–4.8). These findings underscore the importance of tailoring behaviour-change interventions to communities with limited education and fewer economic resources. Additionally, individuals in the lower socioeconomic group were significantly more likely to exhibit “poor” attitudes than those in the higher group (aOR = 1.8, 95% CI 1.1–3.0).

Practices for malaria prevention and treatment

This survey revealed that participants used a variety of tools to prevent nighttime mosquito bites (Table 5). Electric fans were the most common method (72.4%), followed by mosquito coils (62.2%), untreated bed nets (51.5%), and insecticide-treated nets (ITNs; 44.8%). Less frequently used methods included topical repellents, smoke, and wearing long-sleeved clothing and pants.

Table 5.

Practice of participants to prevent mosquito bites and malaria treatment

Response options	Response
	n	Percent
During the night, how can you prevent mosquito bites? (n = 402)
Wire screen	14	3.5
Regular untreated bed net	207	51.5
ITNs	180	44.8
Topical repellents	132	32.8
Mosquito coil	250	62.2
Long shirt and pants	161	40.0
Electric fan	291	72.4
Insecticidal spray	10	2.5
Smoke	42	10.4
How is the medication taken (multiple answers)? (n = 111)
Take until the symptoms are gone, but not as prescribed by the doctor	5	4.5
Take all the medications prescribed by the doctor	106	95.5

Percentages were calculated based on the total number of participants (n = 402). Only participants who had experienced a malaria episode (n = 111) were asked questions regarding their malaria treatment with medication

Regarding regularity of use, electric fans again ranked highest, with 84.6% of users employing them consistently, likely due to their combined cooling effect and mosquito-repelling benefits. Regular use was reported by 60.7% for bed nets, 45.5% for mosquito coils, and 43.3% for ITNs. In contrast, only 14.4% of participants used topical repellents daily; 44.8% used them occasionally, while 40.8% never used them (Table 6).

Table 6.

Frequency of participants to use preventive measures

How often have you used the following preventive methods on average? (n = 402)	Every day (%)	Never used (%)	Sometimes (%)
Wire screen	40 (10.0)	345 (85.8)	17 (4.2)
Regular bed net	244 (60.7)	105 (26.1)	53 (13.2)
ITNs	174 (43.3)	153 (38.1)	75 (18.7)
Insecticidal spray	7 (1.7)	343 (85.3)	52 (12.9)
Fan	340 (84.6)	43 (10.7)	19 (4.7)
Mosquito coil	183 (45.5)	74 (18.4)	145 (36.1)
Topical repellents	58 (14.4)	164 (40.8)	180 (44.8)
Smoke	36 (9.0)	283 (70.4)	83 (20.6)

All percentages were calculated by row, using the total number of participants (n = 402) for each question

Sociodemographic related human activities

Participants were interviewed about their locations and activities during nighttime hours, from 6 PM to 6 AM. During the first time period (6–9 PM), 89% of individuals reported staying at home in the village. This proportion decreased slightly to 80% during the second period (9 PM–12 AM). A notable shift in activity occurred after midnight, as participants moved from their homes to rubber plantations. By the third (12–3 AM) and fourth (3–6 AM) time periods, 68% (n = 293) and 86% (n = 355) of participants, respectively, were reported to be at rubber plantations (Fig. 2).

Fig. 2.

Location of the surveyed population during the nighttime period (6 PM–6 AM)

Among rubber plantation workers, 81.8% reported tapping rubber on 11–20 nights per month. The most recent visits to rubber plantations were predominantly reported as “today” (72.2%), followed by “within this week” (23.4%). The duration of time spent at the plantations varied, with 57.6% staying for 1–6 h and 35.8% for 7–12 h.

Regarding the size of rubber plantations owned by participants, the most commonly reported ranges were 9,600–16,000 sqm (41.3%) and 17,600–32,000 sqm (38.0%). The proximity between participants’ households and their plantations was also assessed: 45.7% were located within 1–5 km, while 48.5% were more than 5 km away (Table 7).

Table 7.

Behavioural attributes of participants contributing to malaria risk while working in rubber plantations

Response options (n = 363)	Response
	n	Percentage
On average, how many days per month do you work in rubber plantations?
1–10 days	30	8.3
11–20 days	297	81.8
21–30 days	36	9.9
When was the last time you went to a rubber plantation?
Today	262	72.2
This week	85	23.4
Last week	11	3.0
> 2 weeks ago	5	1.4
How long do you work in rubber plantation per day?
1–6 h	209	57.6
7–12 h	130	35.8
> 12 h	24	6.6
What is the total area of your rubber plantation that you work?
1,600–8,000 sqm	46	12.7
9,600–16,000 sqm	150	41.3
17,600–32,000 sqm	138	38.0
> 32,000 sqm	29	8.0
What is the distance between your home and the rubber plantation?
Live in a garden/field	8	2.2
< 1 km	13	3.6
1–5 km	166	45.7
> 5 km	176	48.5

All percentages were calculated using the total number of participants working in rubber plantations (n = 363)

Sociodemographic predictors for risk behaviours in malaria prevention and treatment

A total of 140 participants reported spending more than 10 nights per month in rubber plantations and forested areas. Although all were exposed to the same high-risk environments, the use of effective preventive measures can significantly reduce malaria risk. Therefore, only a subset—28.6% (40 out of 140 individuals)—who reported never using LLINs or LLIHNs was classified as exhibiting"high-risk behaviours"for further analysis. Logistic‑regression analysis (Table 8) identified “poor” preventive practice as the strongest predictor of malaria risk behaviours, with significantly increased odds of risk behaviour (aOR = 3.5, 95% CI 1.6–8.1) compared to those with “good” practices. Additionally, males had significantly higher odds of engaging in risk behaviours than females (aOR = 2.8, 95% CI 1.2–6.8).

Table 8.

Sociodemographic predictors for risk behaviors in malaria prevention and treatment

Factor (n = 402)	Low risk	High risk	cOR (95% CI)	aOR (95% CI)
	n (%)	n (%)
Gender
Female (Ref)	188 (94.0)	12 (6.0)
Male	174 (86.1)	28 (13.9)	2.5 (1.2, 5.1)	2.8 (1.2, 6.8)
Age (years)
18–30 (Ref)	57 (93.4)	4 (6.6)
31–40	72 (93.5)	5 (6.5)	1 (0.3, 3.9)	0.9 (0.2, 4.2)
41–50	102 (87.2)	15 (12.8)	2.1 (0.7, 6.6)	2.6 (0.6, 10.9)
> 50	131 (89.1)	16 (10.9)	1.7 (0.6, 5.4)	3.5 (0.8, 15.8)
Education
High school and above (Ref)	116 (89.2)	14 (10.8)
Senior primary	88 (88.9)	11 (11.1)	1 (0.4, 2.4)	0.8 (0.3, 2.2)
Junior primary and below	158 (91.3)	15 (8.7)	0.8 (0.4, 1.7)	0.5 (0.2, 1.4)
Occupation
Non-agriculture (Ref)	50 (89.3)	6 (10.7)
Agriculture	312 (90.2)	34 (9.8)	0.9 (0.9, 2.3)	0.9 (0.3, 2.6)
Marital status
Widow (Ref)	14 (93.3%)	1 (6.7%)
Marriage	283 (89.6%)	33 (10.4%)	1.6 (0.2, 12.8)	1.1 (0.1, 10.7)
Single	65 (91.5%)	6 (8.5%)	1.3 (0.1, 11.6)	0.8 (0.1, 8.5)
Total family members
1–3 (Ref)	108 (92.3)	9 (7.7)
3–5	169 (87.6)	24 (12.4)	1.7 (0.8, 3.8)	1.9 (0.8, 4.6)
> 5	85 (92.4)	7 (7.6)	1 (0.4, 2.8)	0.9 (0.3, 2.7)
Role of the respondent in family
Spouse (Ref)	123 (92.5)	10 (7.5)
Head of family	129 (87.8)	18 (12.2)	1.7 (0.8, 3.9)	1 (0.3, 2.7)
Other Members	110 (90.2)	12 (9.8)	1.3 (0.6, 3.2)	1.3 (0.4, 4.4)
Residential status
Moved in (Ref)	74 (92.5)	6 (7.5)
Born here	288 (89.4)	34 (10.6)	1.5 (0.6, 3.6)	1.8 (0.7, 4.7)
Socioeconomic status
Higher (Ref)	284 (89.9)	32 (10.1)
Lower	78 (90.7)	8 (9.3)	0.9 (0.4, 2.1)	1.1 (0.4, 2.6)
Knowledge
Good (Ref)	186 (89.9)	21 (10.1)
Poor	176 (90.3)	19 (9.7)	1 (0.5, 1.8)	1 (0.5, 2)
Attitude
Good (Ref)	185 (87.3)	27 (12.7)
Poor	177 (93.2)	13 (6.8)	0.5 (0.3, 1)	0.5 (0.2, 1.1)
Practice
Good (Ref)	163 (94.8)	9 (5.2)
Poor	199 (86.5)	31 (13.5)	2.8 (1.3, 6.1)	3.5 (1.6, 8.1)

All percentages were calculated by row, using the total number of participants within each risk category (low or high) for each answer; cOR crude odds ratio, aOR adjusted odds ratio; The aOR was calculated using a combination of factors in the analysis, including (i) gender, (ii) age, (iii) education level, (iv) occupation, (v) marital status, (vi) number of family members, (vii) role in family, viii) residential status, (ix) socioeconomic status, (x) knowledge, (xi) attitude, and (xii) practice; CI confidence interval, Ref reference group; Risk behaviours were classified as “high risk” based on these three criteria: (i) participants'engagement in nighttime activities in both rubber plantations and forested areas, (ii) spending more than 10 nights per month in risk areas, (iii) failure to use effective preventive measures (LLINs and LLIHNs) to avoid mosquito bites while staying in risk areas. Participants who met all three of these criteria were classified as the “high risk” group; p=value <0.05

Discussion

This study examined human factors that continue to shape malaria transmission and potential risk of infection risk in Sisaket Province, which lies along Thailand’s international border with Cambodia. The aim was to identify factors associated with poor KAP and to determine factors contributing to risk behaviours among at-risk populations. The findings highlighted two key determinants that significantly influence malaria risk behaviours.

The first key factor was preventive practice, with higher risk observed among participants with poor practices compared to those with good practices. Malaria risk behaviours are closely linked to the use of effective preventive measures, such as LLINs and LLIHNs. Improved practices among at-risk populations in Sisaket Province may be partly attributed to malaria elimination initiatives. Between 2012 and 2023, large quantities of these preventive tools were distributed by Ministry of Public Health (MoPH) staff [7]. Notably, a major scale-up in 2018 was followed by a 56% reduction in malaria cases by 2019, accompanied by a 55% and 32% increase in coverage and access to LLIHNs and LLINs, respectively [7]. Additionally, the MoPH has implemented the 1-3-7 strategy since 2016 [6, 19, 39], and this approach was applied in the villages included in this study, consistent with national guidelines for both the control and elimination phases [7, 21]. Despite these efforts, uptake of preventive measures remains a challenge among a subset of the at-risk population. In this study, 28.6% of participants reported never using effective preventive measures, despite spending more than 10 nights per month in high-risk areas such as rubber plantations and forests. Such gaps likely reflect structural barriers rather than personal neglect. For instance, rural workers along the Thai–Myanmar border have reported low ownership and use of insecticide-treated nets (ITNs), citing the impracticality of deploying them in forest or plantation environments [29].

The second key factor was gender, with males exhibiting significantly higher risk behaviours than females. This finding is consistent with MoPH data, which show that from 2012 to 2024, 91.6% of confirmed malaria cases (7,035/7,682) occurred among males, compared with only 8.4% (647/7,682) among females [21]. A previous study in Sisaket Province similarly reported that males had a 2.7-fold higher risk of malaria infection (cOR = 2.7, 95% CI 1.4–5.6) [25]. Malaria transmission in the province has been associated with illegal logging and other forest-based work [7], activities predominantly undertaken by men. This occupational exposure likely explains the higher prevalence of risk behaviours and malaria incidence observed among male participants.

Although rubber tappers in Sisaket Province have previously been reported to account for a higher number of malaria cases [21] and to face increased risk of infection compared to other occupations [36] this study found no significant association between agricultural work and malaria risk behaviours. This finding may be explained by the adoption of better preventive practices among agricultural workers, as our analysis revealed a significant reduction in poor practices in this group compared to those in non-agricultural occupations.

This survey found generally good levels of KAP among at-risk populations regarding malaria transmission, treatment, and prevention. Most participants understood malaria transmission pathways, identified the responsible mosquito vector, recognized high-risk areas in their surroundings, and were aware of key symptoms, diagnostic methods, prevention strategies, and treatment options. These results are consistent with a previous survey [25] conducted in Sisaket Province during 2020–2021, which also found high levels of knowledge and positive attitudes toward malaria. Such awareness may have contributed to the marked decline in malaria cases in the province, particularly after 2019 [7]. Indeed, only one malaria case was reported in Sisaket Province in 2024 [21]. Despite strong knowledge, attitudes, and practices, particularly the use of ITNs, still require improvement. In this survey, only 40% of participants reported consistent ITN use, highlighting the need to strengthen efforts to promote and support appropriate prevention behaviours.

Although most participants demonstrated good overall KAP, gaps in attitudes persist, particularly among individuals with lower education levels. Addressing these gaps will require targeted social and behaviour change communication (SBCC) strategies aimed at improving attitudes of malaria risk and treatment compliance in this group. Furthermore, SBCC interventions should focus on enhancing adherence to ITN use among those with poor preventive practices and among males, who exhibited significantly higher risk behaviours than other groups.

Entomological monitoring in Sisaket has shown that An. dirus exhibits peak biting activity in the early evening, declining toward dawn [23]. In contrast, human mobility data reveal that most residents remain indoors in village areas during this early evening peak and typically move to rubber plantations after midnight. This temporal mismatch between peak vector activity and human presence likely contributes to the consistently low number of malaria cases, despite the presence of a highly competent vector. However, a subset of individuals stays overnight in rubber plantations and forested areas, placing them at significantly higher risk of malaria infection. These findings suggest that targeted interventions should prioritize those who spend nights in high-risk areas where An. dirus is active, rather than individuals who remain in villages. Efforts should emphasize the regular use of effective preventive measures, such as LLINs and LLIHNs. Ensuring access to these tools and promoting their correct and consistent use through SBCC strategies are essential.

Outdoor malaria transmission remains a major challenge in Sisaket Province and in many other regions globally [14–16, 29, 40, 41]. The use of personal protective measures, including topical repellents, is limited in rural communities due to several factors, such as poor accessibility, concerns about side effects (e.g., skin allergies), and low compliance [40, 42, 43].

There is clear evidence of a substantial reduction in malaria cases in Sisaket Province over recent decades, driven by multiple factors [7]. One key factor, highlighted by our survey, is the high level of KAP related to malaria among the local population. This strong community foundation supports the efforts of MoPH in controlling vectors and eliminating parasites, advancing progress toward malaria elimination. With only two malaria cases reported in 2023 and one case in 2024, Sisaket is well-positioned to transition into the prevention of re-establishment phase [7]. Achieve and sustaining zero indigenous cases will require ongoing community engagement, continuous education, and guaranteed access to effective preventive measures to protect at-risk populations from mosquito bites.

This study has several limitations. The cross-sectional design restricts the ability to infer causal relationships between risk behaviours and malaria transmission. Representativeness may be affected by the inclusion of multiple individuals from the same household, who are likely to share similar characteristics. Furthermore, since the study was conducted in only three villages within one district, the findings may not be generalizable to the entire province or other areas along the Thai-Cambodian border. Additionally, questionnaire-based data are subject to recall and social desirability biases, which could result in under- or overestimation of risk behaviours and self-reported malaria episodes. Despite these limitations, the findings provide valuable insights for tailoring targeted interventions to interrupt residual malaria transmission at the local level. To strengthen the evidence base, further studies should be conducted in other endemic areas to better characterize gaps in KAP and to identify factors influencing risk behaviours among at-risk populations. Such research will help inform tailored interventions and support Thailand’s goal of achieving malaria elimination and sustaining prevention of re-establishment.

Conclusions

Sisaket Province’s remarkable progress toward malaria elimination reflects multiple contributing factors, with human behaviour playing a central role. This study highlights that success is supported by strong community understanding of malaria transmission, diagnosis, treatment, and prevention. However, persistent gaps in attitudes and risk behaviours must be addressed through targeted SBCC strategies. Further research on the feasibility and acceptability of SBCC interventions to improve attitudes among at-risk populations and enhance adherence to LLIN and LLIHN use is warranted. Strengthening both attitudes and practices will not only advance malaria elimination in the province but also contribute to the control of other mosquito-borne diseases.

Abbreviations

aOR

Adjusted odds ratios

CI

Confidence interval

cOR

Crude odds ratios

GMS

Greater Mekong Subregion

ITNs

Insecticide-treated nets

KAP

Knowledge, attitude, and practice

MoPH

Ministry of Public Health

PoR

Prevention of re-establishment

Ref

Reference

Author contributions

S.P. conceived and designed the study. T.C. and S.P. contributed to funding acquisition. M.S. and S.P. organized and managed field and data collection. T.C., S.M., S.P. and N.J. supervised the research project. M.S., N.J. and P.L.A. analyzed and interpreted the data and drafted the manuscript. All authors have revised the manuscript and approved the final version for submission.

Funding

Open access funding provided by Mahidol University. This research received financial support from the Kasetsart University Research and Development Institute (KURDI) (Grant No. FF(KU) 51.68) to TC.; the Thailand Program Management Unit for Human Resources & Institutional Development, Research and Innovation (PMU-B), NXPO, Grant Number B17F640002 to NJ.; MS PhD student was supported by the High-Quality Research Graduate Development Cooperation Project between Kasetsart University (KU) and the National Science and Technology Development Agency (NSTDA), as well as the Franco-Thai scholarship program of the French Embassy in Bangkok and Campus France.

Data availability

All data supporting the conclusions of this article are included within the article and supporting materials.

Declarations

Ethics approval and consent to participate

This study was reviewed and approved by the Research Ethics Review Committee for Research Involving Human Research Participants, Kasetsart University (Certificate of Approval No. CAO63/035). All participants were informed with formal ethical clearance of the study protocol before commencing the study. Participants who did not meet the inclusion criteria were excluded from the study.

Consent for publication

All authors gave the consent for this publication.

Competing interests

The authors declare no competing interests.