Predictors of malaria-association with rubber plantations in Thailand

The Moo 2 village, a study site that belongs to the Chaiyarat Subdistrict (6 villages in total), is situated in a malaria transmission risk area (Figure 1), 25 km southwest of the Bang Saphan Noi Hospital and District Health Office; both agencies are responsible for malaria control. Due to the subtle geographic disparities and topographic altitudes 100–200 meters above sea level, the village consists of 7 hamlets, namely Ban Chong Samkaew, Ban Hin Tern, Ban Kok Ai Poek, Ban Mak Poo, Ban Pong Toei, Ban Subsomboon and Nuay Anurak. The village covers uplands, hills, hillside slope areas and valleys. More than 80% of the villagers exploited their land for rubber and oil palm plantations and to a lesser extent for fruit orchards. A census of 435 households including 1,896 local inhabitants was officially reported after 2009 and was used in both pre-surveys of existing households in 2010 and random sampling as mentioned below. The majority of the population has access to electricity and running water system. Other water sources available for domestic use and agriculture are streams, brooks, shallow wells and reservoirs. The average number of rain days per year exceeds 120. The average annual temperature is 27.6°C with a maximum 41°C and a minimum 14°C.

Supported by the GFM program since 2004 [9], Prachuap Khiri Khan Provincial Health Office has adopted the implementation of global malaria control strategies (RDT, ITNs/LLINs and IRS) at the provincial level (Figure 1) and has focused predominantly on reducing the mortality attributed to P. falciparum malaria in the 37 endemic villages (high-transmission A1/A2) (Figure 1). Subsequent to the initiation of operation coverage in the malaria transmission risk areas, the Moo 2 village effort commenced in mid-2008 because the number of malaria cases periodically reported for three consecutive years (2006–2008) had surged (Figure 2). The vector-borne disease control initiative centered not only on strengthening the capacity building of the communities, but also on promoting multisectoral partnerships under the supervision of the director of Bang Saphan Noi Hospital. The established alliance consisted of the Chaiyarat Subdistrict Administrative Organization, the Vector-borne Disease Control Unit (Bang Saphan) and two primary care units (formerly health centers), Chaiyarat Subdistrict Health Promoting Hospital and Ban Bang Charoen Subdistrict Health Promoting Hospital. The operational objective was to increase vector surveillance and control focused toward malaria prevention/control campaign activities that governed the household-level implementation of IRS and ITNs/LLINs together with public relations, community participation and health education, prior to and during the malaria transmission season.

According to WHO recommendations for active ingredient (ai) per square meter amounts [31], a 5% water-dispersible powder (WP) of deltamethrin at a concentration of 20 mg ai/m² was used for IRS, and only one 25% water-dispersible tablet (WT) of deltamethrin at a concentration of 50 mg ai/m² was used in treating mosquito nets or retreating ITNs. These measures can reduce the density of common indoor- and outdoor-biting vectors (Anopheles minimus and An. maculatus, respectively in South Thailand); however, nothing was known about the susceptibility of these vectors to effective doses of the insecticide used in the study area. Two trained village volunteers and malaria field workers distributed the ITNs/LLINs and performed focal IRS before (March-April) and during the peak (May-July) of the malaria transmission season (Figure 2).

The data on malaria-affected households retrieved from the electronic reporting system of notifiable diseases revealed 130 malaria cases reported during the period between 2006 and 2010 (Figure 2). This information was used in sampling the households as follows. Of the 435 previously surveyed households (Figure 3), a statistically required sample size of 339 assigned to a stratified two-stage random sampling was calculated at a 95% confidence level. Hypothetically, this was based on the proportion of 20% malaria-affected households reported during this period and the estimate with a 10% of margin of error. However, the 26 malaria-affected households that had reported cases had been reported before 2007 were not available for selection due to either their relocation or absence. Finally, a total of 313 randomly selected households had knowledgeable respondents aged ≥18 years of both genders. These respondents (72% coverage) from the 7 hamlets were from 70 malaria-affected households and 243 malaria-unaffected households. A malaria-affected household was defined as any household that had at least one member who experienced malaria for the first time in his/her lifetime from 2007 to 2010, regardless of the type of infections (single or mixed), infection episode (once or multiple), relapsing of P. vivax malaria and malaria foci (inside or outside the village). In this regard, the at-risk households with any member involved in work at either rubber plantations or natural rubber productions who occasionally became infected with malaria might have been psycho-sociologically affected by the household member’s ailment. This was because the disease resulted in a deviation from a normal lifestyle and caused loss of work days. It directly reduced family income, indirectly increasing patient costs. Moreover, the family members felt anxiety about whether the malaria-infected member would spread the disease to others. Of the 70 malaria-affected households that were followed up with 75% (98/130) of recorded malaria cases (Table 1), only 44% of the follow-up cases (21 male and 22 female) were recruited into the study as the respondents; 25% were not followed due to relocation, absence or death (Figure 3).

As noted earlier, the annual malaria prevention/control campaign activities (Figure 2), which are based on public relations, community participation and health education, had occurred in the village prior to the study. To evaluate the perception of malaria burden, all 313 respondents were, therefore, questioned about their knowledge of or hearing about malaria information through any information-conveying media and channels. In addition, they were questioned about whether they identified malaria as one of the top five public health problems affecting their family or the village community. Based on the survey responses, the 246 respondents who identified malaria as one of the top five public health problems affecting their family or the village community were labeled as malaria villagers (MVs), whereas the 67 remaining respondents who did not recognize malaria as a public health problem were labeled as non-MVs (Figure 3). These non-MVs were subsequently excluded from both the analysis of the proportions according to relevant health behavioral factors (knowledge, perceptions and practices) and the logistic model. For malaria mapping, the 70 malaria-affected households were asked to collect the coordinates (latitude and longitude) of their own houses and surrounding environments in the hamlets, using a global positioning system unit (eTrex Legend®, Garmin International, Inc., USA). This geographic information, by which their way-points were initially recorded in the field with a distance error ±10 meters, was transferred to a laptop running MapSource software version 6.15.7 and then manipulated using Google Earth maps. The spatial distributions of those premises within the studied village were displayed.

The closed-ended structured questionnaire comprised three domains, which consisted of knowledge, perceptions and practices. The Cronbach's alpha coefficient of knowledge and perception was 0.7. The multiple-choice questions were used to ask about the cause, mode of transmission, vector and breeding place, diagnosis, clinical manifestations (symptoms, severity and cause of death), prevention and control of malaria to discriminate between misconceptions and accurate conceptions of malaria among the 246 MVs. Based on the judgments (i.e., agree, disagree and uncertain) of the MVs, the perceptions of malaria were related to health behavior factors, which included perceived susceptibility, severity, benefits and barriers of the health belief model [32].

Perceived susceptibility was assessed based on the responses to the following statements: 1) malaria is not serious; everyone gets infected; 2) a normal, healthy person is insusceptible to malaria; 3) a rubber farmer/tapper is at a greater risk of malaria than others; 4) more people are infected during the rainy season than during the summer season; and 5) drinking alcohol or smoking while doing outdoor activities at night can prevent bites from malaria vector. Perceived severity was assessed based on the responses to the following statements: 1) malaria is a severe disease, and the infected person can be re-infected; 2) malaria cannot be cured; 3) the first infection is more severe than re-infection; 4) malaria causes income and work day loss; and 5) malaria causes death unless treatment is administered. Perceived benefit was assessed based on the responses to the following statements: 1) neighbors of a malaria-affected house should receive ITNs/LLINs; 2) you should sleep under ITNs/LLINs only when the inhabitants of a house is affected by malaria; 3) both IRS and ITNs/LLINs should be implemented for a number of vulnerable houses in which a malaria case occurs; 4) regular IRS can reduce the density of mosquitoes or human-vector contact; 5) every family member should sleep under ITNs/LLINs to prevent malaria; and 6) sleeping under mosquito-nets can prevent malaria whenever staying at smallholdings in hilly areas on rubber plantations. Perceived barriers to control were assessed based on the responses to the following statements: 1) you are reluctant to allow village volunteers to operate IRS at your house; 2) treating or retreating mosquito-nets with insecticides should be the responsibility of health personnel/VHV/malaria workers; 3) sleeping under ITNs/LLINs is uncomfortable; 4) you do not need to own or use ITNs/LLINs if you or your family owns a smallholding in an area on a rubber plantation; 5) it is unsafe to sleep under ITNs/LLINs; and 6) you or a family member are willing to treat or retreat ITNs once or twice a year.

Practicing preventive measures at the household level is focused primarily on frequency of use (i.e., regular, irregular and never) of a combination of preventive measures (i.e., physical, chemical, electrical or fumigated) to repel or kill adult mosquitoes and thereby reduce human-mosquito contact, in addition to the use of screened windows that may act as a physical barrier in the house of the respondents. Finally, ratings for knowledge, perceptions and practices were based on the following scale: <60% low, 60-79% moderate and ≥80% good.

The characteristics or individual explanatory factors for all 313 respondents (or 70 malaria-affected and 243 malaria-unaffected households) were described based on epidemiologic factors (i.e., socioeconomic, demographic and environmental) using descriptive statistics. In the initial univariate analysis, the chi-square test with Yates continuity correction (P < 0.05) on a 2x2 contingency table or Pearson’s chi-square test (P < 0.05) where appropriate, was used to analyze individual epidemiologic factor associated with malaria risk. Additionally, the household-level implementation coverage of the interventions, which included IRS (2007–2010) and ITNs/LLINs (2008–2010), was based on whether the at-risk households received either one of the intervention services once or twice a year over 4 consecutive years (2007–2010), regardless of the continuations of the interventions and the ITNs/LLINs owned per house or the number of persons who shared a net. Theoretical intervention effects might be related to complacency about malaria or to the people do not perceive the seriousness of the disease. Moreover, the utilization of mosquito nets was considered to be individually adapted behavior that the malaria cases may have practiced more routinely than the uninfected household members.

Similarly, all 246 MVs were used in the univariate analysis of the association between the health behavioral factors (knowledge, perceptions and practices) and malaria-affected MVs. The crude odds ratios (ORs) were computed to analyze the strength of their associations. In the multivariate analysis, the significant behavioral factors with ORs (P < 0.05 or P < 0.1) and those significant epidemiologic factors (P < 0.05) were included in a logistic model. The rates and adjusted odds ratios (aORs) for all of these variables and their 95% confidence intervals (CI) were calculated. The Wald’s test (P < 0.05) was used to test the statistical significance of each coefficient in the model to determine contributing predictors in a fitted model. Because a low number of cumulative malaria cases were reported in the study village, this multivariate analysis was performed collectively for the years 2007–2010 to estimate the effects of the behavioral factors on the implementation of malaria control strategies.

Regular IRS (or focal spraying) is aimed at reducing the density of Anopheles mosquitoes within at-risk households. This service also interrupts transmission within a number of houses when any malaria case is reported. Most study households covered by IRS services in the 4 past years were due to the unstable case morbidity in the study village. Similarly, a number of ITNs/LLINs were allocated freely to at-risk households to help vulnerable persons. In the study village, there must have been expansion of the combined intervention services to the target households, both the malaria-affected households and nearby malaria-unaffected households. As expected, all malaria-affected households that had access to IRS received ITNs/LLINs. Markedly, two-third of malaria-unaffected households covered by IRS received ITNs/LLINs. Some malaria-affected households, or even nearby malaria-unaffected households, particularly those uncovered by IRS and ITNs/LLINs are of interest. When the perceived barriers to implementation were examined, it was noted that the MVs felt reluctant to allow village volunteers or malaria field workers to operate IRS at their house; this may account for many households uncovered by IRS and ITNs/LLINs, as seen in Table 3. Moreover, both groups reduced the use of ITNs/LLINs because not all households that owned ITNs/LLINs used them, although almost the entire MV group believed in the potential benefits of ITNs/LLINs. The cultural factors that determine intra-allocation, ownership, retention and the use of ITNs/LLINs are considered to be significant [23, 49‐52]. We found that, as shown in Table 3, most malaria-affected households that owned ITNs/LLINs might have individually adapted the use of ITNs/LLINs because they used both nets/ITNs/LLINs intermittently and ITNs/LLINs only, whereas there were no reports of non-use. Similarly, most malaria-unaffected households that owned ITNs/LLINs neither used ITNs/LLINs nor slept under mosquito-nets, suggesting that cultural factors also determined sleeping-under-net behaviors. Therefore, in agreement with the observations of the perceived benefits of ITNs/LLINs, sleeping under mosquito nets, particularly ITNs/LLINs, was considered a positive-protecting behavior. However, there were some counterintuitive problems that most study households both unaffected and affected with malaria that owned ITNs/LLINs did not use them all year round whether their houses were treated with IRS before or during the peak of seasonal transmission.

Generally speaking, our findings were in agreement with previous findings [32, 49‐52] in that we found two main social factors for the non-use of ITNs/LLINs. The reasons were that the rectangular ITNs/LLINs owned were not large enough, i.e., neither appropriate for mother/husband who shared with children nor uncomfortable for adult persons who slept and that they were kept for the relatives or visitors who stayed at their houses. When questioned about the perceived barriers of the implementation, most MVs mentioned the individual or household role in treating/retreating the mosquito nets. Unlike complacency, the MVs felt that ITNs/LLINs were uncomfortable and unsafe for sleep. The MVs felt that they needed neither to own nor use ITNs/LLINs if they owned a smallholding in the area on a rubber plantation. This perception could explain why the MVs that received ITNs/LLINs did not use them or had intra-allocation of ITNs/LLINs with not everyone sleeping under ITNs/LLINs despite the perceived threat of malaria.

As expected, the household members who slept under the mosquito nets, particularly ITNs/LLINs, were more likely to be vulnerable in that they perceived of the risk only when any member developed malaria illness, and the persons that had experienced malaria in the past or recently practiced good behaviors more routinely than those who had not been infected. Thus, for example, the ITNs/LLINs owned by these study households were more likely to be used as directed by the village health volunteers and local health personnel than as practiced by their motivation or readiness [32] because of their concern about the benefits of ITNs/LLINs. Similar to the observations of the perceived benefits of IRS and ITNs/LLINs, the individually adapted behavior was considered a significant trade-off because the mosquito nets, including ITNs/LLINs, generally used in the study village were felt to be effective against malaria [49‐52]. This may be a reason why, in the model, the utilization of mosquito-nets (i.e., sleeping under nets/ITNs/LLINs intermittently and ITNs/LLINs only) had a significant association with malaria among the malaria-affected MVs. Nonetheless, it was not guaranteed that the greater increase in ITNs/LLINs coverage was related to the smaller decrease in a number of malaria cases in the transmission risk area on rubber plantations.

In the study village with malaria-associated rubber plantations, it was clear when the household members likely came into close contact with various bites of Anopheles mosquitoes based on their nighttime activities. Some vulnerability in how either a person or family acquired the infections depended on household members being involved with rubber tapping in rubber plantations at night and also with rubber-sheet processing in smallholdings both during the night and day, although a high coverage of IRS and ITNs/LLINs at the household level had been achieved. Generally when examined for the perceived susceptibility of malaria, the MVs regarded malaria acquisition as rubber farmer/tapper as an occupational risk. However, in the model, the MVs who were daily workers were at a higher risk than rubber farmers/tappers for malaria infection. Most daily workers settled in Ban Hin Tern, and generally, came from households of low to middle economic class. Notably, daily workers were hired to earn wages by job opportunity, and occasionally, they came into close contact with various Anopheles mosquitoes during different periods and at different places during rubber tapping and rubber-sheet processing in the hilly areas of rubber plantations, particularly when working in high endemic Moo 3 and Moo 4 villages (data not shown) (Figure 3). It is unlikely that most rubber farmers/tappers who worked at night in smaller rubber plantations at one place might have a lower chance of being subjected to infective bites of Anopheles mosquitoes. Perhaps the activities of these vulnerable persons such as revisiting more rubber plantations could have exposed them to multiple bites at multiple locations in the study village. Further investigation of this possibility is needed. Recently, the malaria incidence in the study village responsible for high transmission (A2) has resulted in the 2011 establishment of a malaria post in Ban Hin Tern.

The MVs, as outlined in this study, recognized malaria burden in the affected study village, implying that they should have increased malaria knowledge and awareness through the prevention/control campaign activities. With regards to the aspects of malaria knowledge, the MVs that had good knowledge scores were able to comprehend the mode of transmission (night biting of malaria vectors), diagnosis (blood examination) and clinical manifestations (symptoms). The MVs realized that the blood examination should be performed in a timely manner if an individual developed flu-like symptoms (fever, headache and myalgia) during the rainy season. These figures may imply a perceived health threat [30, 53]. In addition, the median days of illness prior to hospitalization and hospital adherence were 3 and 4 days respectively for both genders as seen in Table 1. In contrast to self-medication [53], the MVs sought medication as soon as they recognized malaria-like onset fever, suggesting that their household members with prodromal infections were less likely to delay treatment onset and that the case management was likely to be effective. However, our observations demonstrated that most MVs were more likely to have low-moderate knowledge scores than good knowledge scores. Similar to previous studies that investigated other malaria-endemic villagers [54], care takers of children under 5 years of age [26, 53] and pregnant women [30], the malaria-affected MVs had misconceptions about malaria regarding the cause, vector and prevention. These figures were in agreement with the observations of malaria perceptions. The malaria-affected MVs claimed that, in hilly areas on rubber plantations, mosquito density in the rainy season was higher than in the summer season when night-biting mosquitoes were thought to be a nuisance but were not considered dangerous [55] despite the perceived threat of malaria. This might substantially contribute to malaria risk in the model. A low-moderate level of malaria knowledge was significantly associated with the malaria-affected MVs whose misconceptions regarding mosquitoes and prevention might influence the promotion and use of mosquito nets and other preventive measures. However, it is unwise to suggest that these results indicate the pitfalls of a health education program that makes use of information, education and communication strategies or, rather, a perceived failure of malaria prevention/control campaign activities that preceded the study in the village.

Our study did not indicate a positive association between practices regarding malaria prevention and malaria, but, as seen in Table 4, our study indicate the MVs who had low-moderate practice scores might have practiced behaviors that increased exposure frequency of outdoor bites of Anopheles vectors in the study village. Clearly, the malaria-affected MVs had practiced preventive measures less frequently than the malaria-unaffected MVs, thus, the malaria-affected MVs who had low-moderate practices likely played a more significant role than those who had good practices. Additionally, the underlying factors for certain health practices that the MVs individually adapted were based on the severity of malaria illness symptoms they recognized, media campaigns they comprehended but not prolonged learning, advice from relevant people (family, village health volunteers and health staff) and the availability of preventive measures that they could afford. This figure may reflect the availability of individually adapted interventions applied in the study village. The MVs were more likely to practice both regularly and irregularly the combined preventive measures rather than just one; in this regard, the selection of preventive measures differed from one household to another. For examples, most MVs used an electric fan/mosquito coil or spray/mosquito repellent lotion or talcum powder, or fumigated natural plant products, as well as used electronic mosquito trap/electric mosquito repellents with a repellent mat or repellent liquid refill, before bedtime or during outdoor activities. In addition, they lit their own houses inside or outside all night. These findings are consistent with malaria knowledge in that the MVs commonly thought that the combined preventive measures were more effective against malaria than only sleeping under nets/ITNs/LLINs. Neither the protective effects of preventive health practices nor the internal and external factors and how they could influence different actions have been elucidated.