INTRODUCTION
Zika virus (ZIKV) infection, a re-emerging infection that has been declared as Public Health Emergency of International Concern in year 2016 by World Health Organization (WHO), affected travel and trading, has recaptured the world’s attention after almost a century (Musso et al. 2014; WHO 2016). Recent outbreaks associate microencephaly and Guillain-Barre syndrome with Zika infection (Hughes et al. 2016). The Flavivirus is transmitted primarily by vector mosquitoes Aedes sp., and secondary routes, including sexual transmission, maternal to child transmission, blood transmission and possibly through various body fluids (Anderson 2017; Musso et al. 2015), including vaginal and seminal fluid, with longest incubation of 6 months reported in seminal fluid. Also, the virus seems to evolve fast; from 2004 to 2015, the spectrum of disease advanced from mostly asymptomatic (80% of the time), to haematospermia, microencephaly in new-born, and Guillain-Barre syndrome as acute inflammatory ascending paralysis (Brasil et al. 2016; Karwowski et al. 2016; Roze et al. 2016; Turmel et al. 2016; Uncini et al. 2017; Saba Villarroel et al. 2018).
Despite hosting the Asia lineage of ZIKV (Marchette et al. 1969; Pond 1963) which exploded massive outbreaks of Zika related microencephaly and Guillain-Barre syndrome in Latin America (year 2007-2016) (Musso et al. 2014), Malaysia only reported one suspected local Zika transmission in 2019 since the discovery of the P6-740 strain in 1954 (Smithburn 1954; Lim et al. 2017; Malay Mail 2019). There was no reported neurological complications or birth defect associated with Zika in the surveillance pathways of Zika, but under-reporting is possible due to the challenge of screening and diagnosis (Lim et al. 2017). Forest fringes represent the zone of sylvatic-urban cycle transition for Zika (Wolfe et al. 2001). It is worth investigating the forest fringe population’s knowledge and perception towards the disease as ZIKV infected mosquito was discovered locally in 1954 (Smithburn 1954).
Knowledge and risk perception towards Zika infection would influence the prevention and control of the disease, not only in terms of vector breeding site destruction, but family planning and screening for blood donation (Villarroel 2018). Only scanty questionnaires were developed to assess the level of knowledge, attitude, and practice of the population towards the Zika infection for effective risk communication (Lim et al. 2017; Arief et al. 2017; Rosales et al. 2017); yet mostly focus on clinical populations rather than the public, resulting in systematic bias (Wong et al. 2017; Cheema et al. 2017; Whittemore et al. 2017). A validated questionnaire based on general question bank of World Health Organization (WHO 2016) has been developed for communities with ZIKV transmission or those at risk, to serve as the first proper screening tool for knowledge and risk perception towards Zika infection in Malaysia. This would aid in tailoring health education materials on the ZIKV that is suitable for local communities (Wong et al. 2020).
MATERIALS AND METHODS
A population based cross-sectional study recruited 433 adult respondents in forest fringe areas of Malaysia from September 2019 to March 2020. Respondents completed the assisted-administered, validated questionnaire consisting of knowledge and perception domains on Zika infection via Google Form with the single attempt and compulsory modes set. The methodology of sampling in this study is shown in Figure 1.
Instrument
The English language validated questionnaire had 68 questions made up of ‘knowledge’ and ‘perception’ domains; 14 questions (63 items) tested on ‘knowledge’ towards Zika infection, and five items / questions tested on risk perception of ZIKV (Wong et al. 2020). The copyright number for the Bahasa Malaysia version is UKM.IKB.800-4/1/3215. Available answer options were ‘Yes’, ‘No’ or ‘Don’t know’. Correct answers were given 1 mark, while ‘don’t know’ yielded 0 mark. Risk perception items were displayed as a 5-ratings Likert scale, consisting of ‘Strongly agree’, ‘Agree’, ‘Neutral’, ‘Disagree’ or ‘Strongly disagree’. Three items (P1, P4, P5,) tested on positive perception and two items (P2, P3) tested on negative risk perception. For items tested on positive perception, “Strongly agree" carries a maximum score of 5, while "Strongly disagree" carries minimum score of 1. For items tested on negative perception, “Strongly disagree" carries a maximum score of 5, while "Strongly agree" carries minimum score of 1. All negative perception items' scores were inverted for Rasch analysis.
Statistical Analysis
Knowledge and perception scoring were analysed using a unique dual-instruments approach with IBM Statistical Product and Service Solutions Statistics (SPSS) (Armonk, NY, USA) and Rasch Model analysis (Winsteps) (Winsteps, Oregon, USA) to provide bidirectional analysis of person’s ability and item’s ability (Rosnah et al. 2015). This approach ensures respondents’ ability in answering the question items were on well-targeted question item, removing bias of possible language barrier and culture comprehensibility (Azrilah 2010). The validation made has a proven model of fitness and suitability of 5-rating Likert scale in the measurement of risk perception by (Wong et al. 2020). Normality of data distribution was checked with kurtosis and skewness, confirmed by formal Kormogorov-Smirnov test; continuous variables were reported in mean + 2 SD for normally distributed data, and median (25th, 75th centile) for non-normally distributed data. Frequency and percentage were used for categorical variables. Both knowledge and perception levels were categorized into ‘good’ and ‘poor’ levels based on mean score in both SPSS and Rasch (Wright map). Scores above mean level (normality satisfied) (in SPSS and in "summary statistic of person" in Rasch), or above median (if normality not satisfied in SPSS) were considered as of "good" knowledge and "good" perception. For Rasch analysis, question items within Mean + 1 SD of the ‘person map of item’ (left side of Wright map) were well-targeted items, beyond which were difficult, while items below -1 SD of were too easy to test the ability of the respondents. Person measure correlation with r<0.32 was considered respondents with erratic performance, in which they could answer very difficult item correctly but getting easy item wrong. Bivariate analysis for discrete proportions were tested using Pearson Chi Square for dichotomous outcomes, reported as X2 (df), with confidence interval of 95%, p-value of <0.05 as significant, with the power of study set as 80%. Results were analysed in consideration of possible biases.
RESULT
After meticulous data cleaning, a total of 433 respondents completed the questionnaire. The sociodemographic characteristics of the respondents areb shown in Table 1.
Knowledge on Zika Infection
All respondents answered all the questions with the highest score achieved being 59/63 (0.5% of respondents) while 1.8% respondents scored 0 (lowest) for the knowledge domain. Median score for knowledge domain was 43/63 (25th centile = 35, 75th centile = 49) by SPSS, while mean + 2 SD score of 39.6 + 11.8/63 from Rasch complimented the result. Both analyses showed that less than half of the respondents had good knowledge on Zika infection, as showed in Table 2.
Rasch analysis was used to access the person’s ability on knowledge level towards Zika infection among forest fringe population in Malaysia. Person’s ability measured a Log 2.92 to Log -6.02. The score reflects a poor knowledge level on Zika infection rather than random answers, as they tended to get more correct answers accidently (Azrilah 2010; Rosnah et al. 2015). From the Wright map in Figure 2, 44.4% items were well-targeted, 12.7% of the items in knowledge domain were able to test the most able respondents. There were 6.5% respondents who were able to answer the hardest knowledge items (above 1 SD). The "knowledge" items were easy to endorse and well-fitted to test the respondents at Malaysia forest fringe.
For the hardest endorse knowledge item that was item K10: ‘Is there treatment to cure Zika?’, with the correct answer being ‘No’, only 15.5% answered correctly, while 60.3% respondents answered ‘Yes’. Respondents possibly assumed that symptomatic treatment as a curative treatment. Respondents may have answered this based on their previous perception towards Dengue fever, which presents similarly to a Zika infection, with the former being much more familiar to the local population. However, respondents did not understand the pathophysiology of the virus infection. There was a discrepancy of understanding between the knowledge level of researchers and the public. The easiest endorsed item was K11-4, answer "Yes" to ‘Seek medical advice’, for symptomatic treatments for Zika. Almost all (91.9%) of the respondents would seek medical advice if they contracted the Zika infection. This is a proxy to show that majority of the respondents had good health seeking behaviour and established trust towards local healthcare facilities/services.
Perception of Risk
The perception domain of questionnaire consisted of five items to test respondents’ risk perception towards the vulnerability, transmission, complication, prevention, and health seeking behaviour on Zika infection. Items in ‘perception’ domain were all presented in a five level Likert scale, where participants answered ‘Strongly agree’, ‘Agree’, ‘Not sure’, ‘Disagree’ or ‘Strongly disagree’. Perception item P1, P4, P5 tested on perception towards a positive health seeking behaviour, where ‘Strongly agree’ was given a score of 5 and ‘Strongly disagree’ was given a score of 1; item P2, P3 tested on perception towards negative health seeking behaviour with ‘Strongly disagree’ given a score of 5 and vice versa. The score of negative perception items were inverted for the Rasch analysis. The five level Likert scale has a logits of 2.72 (1.4<s<5.0), indicating good separation between the rating options.
Rasch analysis showed good model of fitness for both item and person’s measure for risk perception towards Zika infection, indicating suitability of perception items towards assessment of targeted population. Overall, 2/3 of the questioned items were well targeted. The bubble chart in the Rasch analysis showed perception for item P1 and P3 were well-fitted to test the respondents’ ability (Figure 3). Item P1 was ‘I am at a risk group of getting Zika infection’, with ‘Strongly agree’ given a score of 5. Item P3 was ‘I feel safe from Zika infection when I enter forest fringe even if I do not use mosquito net/repellent/using mosquito coil/wearing protective clothing’, with ‘Strongly disagree’ given a score of 5.
There were 3.5% respondents who achieved a perfect score, while 0.2% had the lowest score of 8/25. Mean score for risk perception was 18.04 + 3.1/25 by both SPSS and Rasch. Less than half of the respondents had good risk perception (41.8% by SPSS and 42% by Rasch), result that are summarised in Table 2. Overall, this set of questions were too easy to endorse (Wright map), with 9.3% respondents having risk perception above +1 SD and 8.3% below -1 SD.
Both knowledge and perception scores were subjected to bivariate analysis against sociodemographic/ socioeconomical status with Pearson’s Chi Square using SPSS, as summarised in Table 3.
Respondents from the state of Perak generally had a significantly higher level of knowledge and risk perception towards Zika infection. Sociodemographic/socioeconomical status which influenced knowledge level significantly included state of residence (X2 = 7.95 (1), p <0.005, OR = 0.579, 95% CI (0.395, 0.848)); ethnicity (X2 = 10.292 (1), p<0.001, OR = 1.945, 95% CI (1.292, 2.927); education level (X2 = 7.590 (1), p = 0.006, OR =1.845, 95% CI (1.19, 2.861); household income group (X2 = 15.568 (1), p<0.0001, OR = 0.352, 95% CI (0.207, 0.600) and being in the reproductive age group (15-49 years) (X2 = 4.403 (1), p = 0.036, OR = 1.590, 95% CI (1.029, 2.455). The forest fringe population in Perak had 42% lower odd of poor knowledge, with bumiputra (natives) having almost two times better knowledge levels than non-bumiputras. Those with higher education levels had 84.5% better odd performance in knowledge level. Respondents from a higher household income group has 30% lower odd of poor knowledge level, while those who were in their reproductive age had 59% higher odd of good knowledge level.
Sociodemographic/socioeconomic status which influenced risk perception significantly include state of residence (X2 = 33.057 (1), p<0.0001, OR = 3.207, 95% CI (2.142, 4.803); ethnicity (X2 = 24.776 (1), p<0.0001, OR = 0.330, 95% CI (0.211, 0.516); location of workplace (X2 = 8.617 (1), p= 0.003, OR = 0.330, 95% CI (0.211, 0.516); regular and recent jungle visit (X2 = 3.945 (1), p = 0.047, OR = 1.516, 95% CI (1.004, 2.289); (X2 = 7.273 (1), p = 0.007, OR = 1.786, 95% CI (1.169, 2.128); Reproductive Age Group (15-49 years) (X2 = 6.812 (1), p = 0.009, OR = 0.564, 95% CI (0.366, 0.869). Forest fringe population from Perak had 3.2 times higher odd of good knowledge level than those in Pahang; working in the forest fringe had 67% lower odd of poor knowledge level while respondents who enter jungle regularly and those who had recent jungle visit has 1.5 times and 1.78 times higher odd of risk perception level, respectively. Women in the reproductive age group had 44% lower odd of poor risk perception level. The bivariate analysis of all covariates against the knowledge level of respondents is summarised in Table 3.
Association of Risks Factors with Knowledge and Risk Perception Level of Respondents from Forest Fringe Population in Malaysia
All significant factors were subjected to multivariate analysis using Multiple Logistic Regression in SPSS identify confounders and effect modifiers. Table 4 summarises the factors significant for knowledge and perception levels with covariates adjusted.
Non-bumiputra, high education level, high income group, and recent jungle visit had better knowledge level than their counterparts. Bumiputra (Malay, Orang Asli (aborigine) and Orang Asal (natives of Sabah/Sarawak) had 60% lower odd of good knowledge (95% CI 0.377; 0.961); high education level has 1.8 times higher (95% CI 1.123; 2.943) knowledge level than their counterparts. High household income group (M40 & T20)(Kirat 2020) and recent jungle visit (<14 days) both had 2.1 times higher odd of good knowledge level than those below the poverty line (95% CI 1.181; 3.792) and without recent jungle visit (<14 days) (95%CI 1.288; 3.344).
Adult, menopaused women, from Perak had better risk perception level for Zika infection, as shown in Table 4. Adult (18-59 years) had 2.86 times higher odd of good risk perception, (95% CI 1.213; 6.716); women of reproductive age group had 53.7% lower odd of good risk perception (95% CI 0.334; 0.864) and respondents from Perak has 3 times higher odd of good risk perception than respondents from Pahang (95% CI 0.334; 0.864).
DISCUSSION
Almost 2/3 (68.25%) of the question items used to test the knowledge level of respondents in forest fringe population of Malaysia was well-targeted and easy to endorse. The hardest item to answer was item K10: ‘Is there treatment to cure Zika?’ (No). Respondents possibly assumed admission to hospital is considered as a form of treatment, and assuming a discharge from hospital is equivalent of being cured. This assumption is without understanding the pathophysiology of immune response towards the viral infection (Klimpel 1996). Most of the study population mistook asymptomatic spectrum of Zika infection or gave low emphasis of disease in the country as ‘cure’.
The result indicated geographical factor, social status (such as education, earning level), and prior experience with jungle related activities affects the knowledge level of respondents towards a subject, similar to a study conducted in Brunei (Chaw et al. 2018). Respondents from higher household income had better knowledge towards Zika infection. They could be potential advocates to provide health education to their at-risk neighbours to prevent a Zika infection. A similar approach was proposed by Desjardins et al. (2020). Health education materials should be more accessible to the lower odd groups to raise knowledge level and reduce inequity, hence narrowing the gap for universal health coverage (Lyratzopoulos et al. 2012).
However, risk perception towards the Zika disease is affected differently as compared to how knowledge was affected. Risk perception is largely affected by prior experiences, which forms a behaviour (Ohman 2017). Most respondents had good perception towards prevention of Zika (item P3, P5) and health-seeking (item P4), but majority did not perceive that they were vulnerable to Zika infection (item P1) and risk of complications (item P2). Adult, women beyond reproductive age group from Perak were having better risk perception than their counterparts, at 2-3 times higher odd. Risk perception denotes prior personal experiences or comprehension towards a subject (Carpenter 2010; Mohd Shahrol et al. 2021). In the case of Zika infection, respondents’ risk perception level was most likely influenced by their prior experience towards the more common hyperdemic Dengue infection in Malaysia (Ministry of Health Malaysia 2017). The risk perception level was not related to socioeconomy status or prior experience of forest-related activity. The poor risk perception towards vulnerability was most likely due to poor emphasis on the diseases because of underreporting of cases (National Public Health Laboratory MOH 2016; Woon et al. 2019). This could possibly be affected by the input and pathway of surveillance system for Zika in Malaysia, which focuses on international ports (MOH 2019). The poor risk perception level reported in the present study was similar to Southeast Asian (Heang et al. 2012; Leung et al. 2015; Moi et al. 2017) and the Middle Eastern countries (Cheema et al. 2017); however lesser than respondents from Latin American regions (Carabali et al. 2018), possibly due to a higher incident of Zika infection and related microencephaly cases that has sparked more widespread awareness (Korhonen et al. 2016; Krauer et al. 2017; Zammarchi et al. 2015; Zanluca et al. 2015).
With poor knowledge and risk perception levels towards Zika, Malaysia is in a precarious state to curb this re-emerging disease. Autochtonous transmission of Zika may have long been affecting Malaysia, but many cases could have been mistaken as dengue fever due to the similarity in presentation (Musso et al. 2015; Pond 1963; Duong & Dussart 2017). The low prevalence of Zika in Malaysia could be due to false assurance given by the current surveillance system and a low funding interest for Zika-related research in the Southeast Asian region (WHO 2017). There are likely more cases in this region than currently reported, such as in Thailand (Ruchusatsawat et al. 2019). The recent report of a local ZIKV transmission case in the Malaysian Navy proves that autochthonous transmission of Zika does exist (Malay Mail 2019). The Asian lineage of Zika which mutated to American lineage was responsible for the Public Health Emergency of International Concern in year 2016 (Hu et al. 2019). The missing link of the Asian lineage ZIKV mutation in Southeast Asia may provide answer to how other flavivirus family members, particularly Dengue virus, mutated (Musso et al. 2015). Recent evidence has shown that the dengue virus is sexually transmissible (ECDC 2019). The economic power of most developing countries is inadequate to cope with the health expenditure of Zika complications, particularly microencephaly and Guillain-Barre syndrome, as compared to Latin America and European continents (Jaydip et al. 2014; Kuadkitkan et al. 2020). Therefore, preventive public health effort should be strengthened, particularly in raising the knowledge and risk perception levels of the residence towards Zika infection in order to prevent uncontrolled outbreaks which may further exhaust the Gross National Income of the country (World Bank 2020).
CONCLUSION
In conclusion, majority of the forest fringe population in Malaysia have poor knowledge and poor risk perception towards Zika infection. The questionnaire was well-targeted and easy to endorse. Non-bumiputra of higher education level, higher income status and recent jungle visits were found to have better knowledge. Adult, menopaused women from the state of Perak had better risk perception. This tool would provide insight for the Ministry of Health Malaysia to develop Zika related health education material to the forest fringe populations in Southeast Asia.