Experience and perception of risk associated with knowledge, attitudes and practices regarding dengue in Riohacha, Colombia

Recognition of the determinants of knowledge, attitudes and practices (KAP), which could be conditioned by the experiences and perceptions of the population at risk, is essential for the control of dengue. The scope of this article is to estimate the relationship between the risk perception and dengue diagnosis experiences with KAPs on dengue in an endemic Colombian population. A cross-sectional study with multi-stage random sampling was conducted. Adjusted prevalence ratios (aPR) were estimated using regression models as measures of association. Of the 206 families interviewed, 7% know dengue is caused by a virus and less than 40% recognize other symptoms besides fever. As control strategies, 31% eliminate hatchery sites and 58% use fumigation, though 73% perceive the risk of dengue. The association was identified between the perception of the risk of dengue and knowledge about the vector (aPR = 3.32 CI95% 1.06–10.36), and the experience of diagnosis of dengue with the attitude towards dengue control (aPR = 1.61 CI95% 1.09–2.37). Risk perception and experience with dengue could become determinants of KAPs in relation to this disease.


Introduction
Dengue is a viral disease transmitted through the Aedes mosquito bite. These mosquitoes also participate in the transmission of other diseases, such as Yellow Fever, Chikungunya, and Zika 1 . No specific antiviral therapy is currently available. Therefore, action taken against the reproduction of the vector, and to avoid the mosquito bite is a crucial strategic approach 2 . Nonetheless, its management and implementation require the involvement of numerous public sectors 3 , where the community plays a leading role 4 .
The occurrence of dengue cases globally has shown an upward trend in the last decades. By 2008, a little over 1,2 million cases were registered in the Americas, South East Asia, and the Western Pacific. By 2013, these cases doubled. Moreover, by 2015, 2,35 million cases had been notified only in the Americas Region 5 . This increase has been accompanied by an elevated mortality rate in Latin American and Caribbean countries, tripling every ten years 6 .
Dengue is a primary public health issue in Colombia. It is characterized by an endemoepidemic and hyperendemic transmission pattern 3 . Notably, the department of La Guajira has reported seasonal outbreaks every four years between 1999 and 2010; and 34,7% of the cases were situated in its capital city, Riohacha 3 . The municipality of Riohacha's circumstances have favored the permanence of dengue in this population, such as urban overcrowded areas, poor sewerage system and waste disposal, limited control measures against the vector, and inadequate public health infrastructure 7 .
The PAHO and the WHO emphasize that the community must commit itself to the control measures. Thus, programs ought to be structured considering the perspectives within the community, identifying the barriers that prevent it from bringing about a change in behavior, and guiding the organization of such programs in consistency with local necessities and realities 8 . For this purpose, an important step is the acquisition of information on Knowledge, Attitudes, and Practices (KAP) concerning the disease 9 , which will help the community to be strategically involved through participatory research 10 .
Countless studies have been carried out on dengue KAP in endemic Latin American areas, for instance, Brazil 11,12 , Cuba 13 , Venezuela 14 , Puerto Rico 15 , and Peru 16 , as well as in other countries like Malaysia [17][18][19] , India 20 , Australia 21 , and Pakistan 22 . Nevertheless, these studies have not as-sessed how risk perception and the experience of dengue could determine KAP.
Evaluating this type of association could be a valuable contribution to the planning of prevention and control activities. Those mentioned above can provide a way to learn about the community that will be submitted to intervention, as well as beliefs that could facilitate or hinder a change in behavior.
Concerning the above, this research is the first in Colombia which aimed to estimate the association of risk perception and experience with dengue, with factors such as knowledge about the disease and its transmission, the application of practices, and a willing attitude aimed at the control of such arbovirus.

materials and methods
A cross-sectional analytical study nested in a cohort study was conducted in the department of La Guajira and other departments of the Colombian Caribbean region 23 . Concerning the selection of participants, a cluster and multistage sampling were conducted in which, initially, the city blocks were randomly selected, and for each block, the participant households were then selected randomly. The target population included only adults and permanent residents in the household selected to take part in the study. The participants who were planning to change their household in the following six months were excluded.
In the same visit, two questionnaires were conducted in order to collect data. The first one aimed to evaluate the KAP, validated by Caceres-Manrique et al. 24 , which contained 21 questions about knowledge and practices, 12 questions about attitudes, and nine questions concerning demographic data. Details on this questionnaire can be found in a previous article, which focused on the association between the educational level and the KAP 25 .
The second questionnaire aimed to assess the determinants of perceived demand in dengue vaccines 26 . This last questionnaire was built from a data collection instrument implemented in a multi-country study, carried out in Vietnam, Thailand, and Colombia 27 . It contains eight questions about the experience with the disease and four questions about dengue risk perception (Table 1). The latter are related to the Health Belief Model by Hochbaum and Rosenstock 28,29 . Thus, elements, such as the perceived susceptibility to becoming ill and the perceived severity in case of falling ill, were combined.
Regarding the collected data storage, an application in Access was employed, where data was double entered. The information was validated with the package Data Compare by Epi-Info. Concerning the description of the population, the absolute and relative frequencies were calculated in the case of the qualitative variables. On the other hand, the median and the interquartile range were calculated for the quantitative variables without a normal distribution.
As dependent variables, the KAP about dengue were considered, referred by the participants of the study. In the case of the open-ended questions, the responses were classified before the analysis. As independent variables, the following were considered: the perceived risk of falling ill; and the experience with dengue cases (diagnosed, hospitalized, and death from dengue) in the personal, family, and neighborhood levels. Moreover, as potentially confounding variables, the demographic variables including age, sex, educational level 25 , inhabitants per dwelling, number of working people, and the presence of under 18-years-olds in the place of household were taken into consideration.
Fisher's exact test and the Mann-Whitney U test were applied in the bivariate analysis depending on the nature of the variable. The associations identified, which p-value was lower than 0.20, were evaluated in a multiple model employing log-binominal regression. In the models, statistically significant variables (p < 0.05) were maintained, including those who modified the estimate of the independent variable of interest

Results
Two hundred six households from the municipality of Riohacha were invited to participate, from October 31 st to December 19 th , 2015, to apply the two questionnaires. In total, 99,5% (n = 205) of the households participated in the first questionnaire, and 99% (n = 204) in the second one.
Most of the participants were women (80%), and an educational level between primary and secondary education prevailed. Nearly half of the respondents (47%) were aged between 30 and 49 years. The median of the number of inhabitants per dwelling was 4, and minors were found in 81% of the households. Also, 98% were working people, most households had an income of up to 1 Colombian legal minimum wage (SMLV, in Spanish), equivalent to 206.67 USD by 2015 (Table 1).
Concerning dengue risk perception, we observed that 99% of the respondents consider dengue a severe disease, both for adults and children. Furthermore, most of the respondents admitted the probability of contracting dengue themselves (80%) or their children (73%) in the next five years. As for the experience with dengue, more than 20% of the respondents refer to earlier cases in their family and neighbors, including the experience of having had been diagnosed (23%) and hospitalized (21%) ( Table 1).
Concerning knowledge about dengue, most (75%) defined it as a disease and 7% related it to a viral etiology. Fever was the most referred symptom (95%). Meanwhile, the other symptoms that were questioned where recognized by less than 50% of the respondents. In what concerns knowledge of the vector, less than 40% of the respondents were aware that the name of the mosquito is Aedes aegypti and that it is identified by the white stripes in its legs. Most indicated their understanding that dengue is transmitted by a mosquito bite, that it reproduces in stagnant water, either clean or due to rainfall; and knew about the mosquito larvae ( Table 2).
Regarding practices, 97% indicated seeking medical care if a family member had dengue. Besides, in order to avoid dengue, the most prevalent actions were eliminating breeding areas and fumigating, each with a frequency of 31%. Regarding the measures taken to avoid the mosquito bite, 58% mentioned fumigating, and less than 15% referenced other practices ( Table 2).
The dengue control attitude manifested through participation in community actions showed a frequency of less than 40% for most of the respondents (Table 2). Also, 56% of them considered that the lack of information hindered measures against dengue. Concerning the preventive measures for dengue, 62% estimated that the local government is the one responsible for carrying these out, while less than 50% noted that each member of the community shares the responsibility (Table 2).
When evaluating the variables associated with KAP in multiple models, it was gathered that general knowledge about dengue (how it is transmitted, name, and vector's appearance) was positively associated with risk perception (PR 3.32 95%CI 1.06-10.36) and the educational level ( Table 3). As for the practices, measures against the adult stage of the vector were significantly less frequent among respondents who were not aware of whether their neighbors had been diagnosed with dengue (PR 0.57 95%CI 0.35-0.94).
Regarding the attitude towards dengue control, the attendance to meetings arranged by the community was directly associated with the experience of having been diagnosed with dengue (1.61 95%CI 1.09-2.37). This attitude was less frequent among those living with minors (0.56 95%CI 0.37-0.83). A positive association was observed between the experience with the hospitalization of a neighbor due to dengue and the acknowledgment of the responsibility of every person in the control of dengue (PR 1.52 95%CI 1.05-2.19). This attitude was less frequent among male respondents (Table 3).

Discussion
Risk perception of experiencing dengue was a factor associated with the general knowledge   1 Knowledge about how dengue is transmitted, its appearance and its name. 2 Considers that dengue is a serious dengue for children and adults, and considers possible contracting dengue in the next five years. 3 Technical and university. 4 Stagnant, clean, or rainfall water. 5 Fumigate, use a mosquito net, fan, or repellent. 6 The respondent, a family member, or a neighbor, was diagnosed with dengue by a doctor.
of the vector and the knowledge of its breeding areas. This could suggest that recognition of the epidemiological situation can be a determinant in the acquisition of knowledge about dengue. Educational level and age are essential variables of adjustment as they had already been recognized as factors associated with knowledge about dengue and its transmission in the study population 25 , as it was observed in Malaysia 17 and Brazil 11 . This suggests that the university community could be a facilitator in the knowledge transference. On a separate note, this also shows that the population with low educational levels would have significant barriers to acquiring knowledge relevant to the prevention of the disease 25 .
A study in Puerto Rico evidenced that people who had had a previous diagnosis of dengue were more knowledgeable 15 . Nevertheless, in this study, experience with dengue was not associated with dengue knowledge. This could be explained partially due to the small number of respondents with a personal history of the disease. Consequently, a study including a larger sample with a history of the disease was required in order to accurately assess this association.
While the study population is endemic to dengue and displayed some knowledge of the topic, this could be insufficient, as described in other cities in Colombia and South America 12,16 , and Asian countries such as India and Malaysia 20,18 . In this respect, the community itself expressed that the lack of information limits action taken against dengue. This observation is significant since, as stated by the COMBI experience 4 , the community itself should identify its weaknesses and manage what it needs to improve. Consequently, the comprehensive integration of all the sectors is imperative and relevant to solve this type of issue.
Concerning the practice of preventive measures, the respondents that did not know (or did not answer) whether any neighbor had been diagnosed with dengue showed a lower frequency of measures taken against the adult stage of the vector. This association suggests that unawareness of dengue cases around them could be associated with indifference to the control of this arbovirus. The mentioned above highlights the importance of communication among neighbors as a way of raising public consciousness and preparation in case of an outbreak 21 .
On another note, the study did not find any association between risk perception and practices against dengue, similar to a study carried out in Cuba 13 . Nonetheless, a study in Karachi, Pakistan 22 , endemic regions for dengue, determined that risk perception are a significant predictor of adequate preventive measures. These results highlight the importance of the regional differences, which must be considered when planning dengue control strategies.
Regarding water storage, it is essential to mention the low proportion of adequate practices observed in the population studied (Table  2), unlike findings in Malaysia 19 where more than 50% of the respondents had good practices in this regard. The unawareness of the vector's characteristics in its immature form may explain the findings in the population of La Guajira, as proposed in a study conducted in Venezuela 14 . Therefore, it is recommended that indications given to the community be complemented with information on this aspect.
Attitudes concerning control of dengue manifested in the attendance to meetings arranged by the community were associated positively with a history of diagnosis of dengue in the respondent, family member, or neighbor. Moreover, the experience of their neighbors being hospitalized due to dengue was associated positively with the reasoning that everyone in a community is responsible for taking preventive measures against dengue. Consequently, communication amongst community members is proposed as a cross-cutting element of the health education programs and actions 31 . These associations are relevant when planning strategies to accomplish empowerment, understood as "people's ability to work together, organize themselves, and mobilize resources to solve problems of common concern" 32 .
Although the study did not find any association between risk perception and dengue control attitude, there is evidence that carrying out activities to enhance risk perception promotes decision-making towards adopting an expected behavior 33 . It is also worth mentioning that risk perception is a highly polysemous construct, product of a cognitive process based on information that every person possesses from different contexts 28 . Also, the acceptability level, benefits associated with the risks, and the level of control intervene in the estimation of the risk. With this in mind, people tend to tolerate risk as long as it is associated with a benefit, or they consider such risk to be under control, which then leads to perceiving the risk to a lesser extent 29 .
Several studies that have analyzed dengue risk perception have suggested that it was not associated with a preventive behavior 34 . Other papers suggest that people who perceive the risk of catching the disease are more likely to adopt preventive practices 22 . Furthermore, interactions between experiences, perceptions, and decision-making were evaluated in Madeira Island, Portugal; a learning that lived experience due to an outbreak in the community improved risk perception and practices to some extent. Still, other perceptions and beliefs arose, leading to less preventive practices 35 . Meanwhile, in Sri Lanka, most of the population had a low-risk perception, and control measures were scarce, although it is a dengue endemoepidemic region 36 .
Due to the above, the complex connection between experience, perception, and KAP justifies carrying out studies in different regions in order to identify barriers and enablers in the acquisition of KAP. This knowledge will allow generating strategies towards social movements, dengue control, as well as other arboviruses.
Among the potential study limitations is the size of the sample, which might be insufficient to explore the associations between some of the variables, such as experience with dengue. Besides that, the cross-sectional design of the study hinders establishing a time connection between experience and dengue risk perception with the KAP. Nevertheless, regardless of those mentioned above, it is likely that experience and perception precede current knowledge, along with the attitudes and practices mentioned by the respondents. Consequently, we consider this study describes relevant associations to recognize communities that require a particular emphasis when carrying out dengue control campaigns.
While KAP descriptions are of a regional nature, the associations found could be useful to develop campaigns that improve KAP about dengue in other endemic countries. This way, it may be plausible to affirm that community cam-paigns could be mediated and enhanced by integrating the people who have been exposed and have a higher risk perception.
Although the study of KAP is a strategy that has been used for several decades and allows describing characteristics in the study population, it is worth saying that the present is the first study in Colombia that evaluates the associations in risk perception and experience with the KAP about dengue. Besides this, probability sampling, which reduces the risk of selection bias; the use of validated questionnaires and data auditing, which reduces the risk of reporting bias; and the evaluation of other potential confounding variables, are among the strengths of this research.
In conclusion, we observed that risk perception of suffering from dengue was associated with knowledge regarding this arbovirus, and that experience with dengue cases was related to a positive attitude concerning its control. These results show the importance of the individual and collective context in the planning of preventive strategies based on communication and social change.
Collaborations L Benítez-Díaz took part in the study design, analysis, and interpretation of the data, drafting the paper, review, and approval of the final version. FA Díaz-Quijano took part in the project design, coordination of data collection, participated in the data analysis planning, critical review for relevant intellectual content, and final approval of the manuscript. RA Martínez-Vega contributed to the project design, participated in data collection planning, data interpretation, critical review for relevant intellectual content, and the final approval of the manuscript.