The co-epidemic trend of COVID-19 and dengue in Southeast Asia needs serious attention. These two diseases have similar clinical symptoms. ^{20 , 21} The COVID-19 pandemic situation is a challenge in controlling dengue in Indonesia. The existence of a social restriction policy makes it difficult for volunteers to provide education manually door to door. The impact is increased dengue cases in several areas. ²²

Public awareness regarding dengue control efforts still needs to be improved. Knowledge related to dengue can influence attitudes toward dengue control. ^{23 , 24} A positive attitude can encourage action to control dengue. ^{25 , 26} A large number of water containers is a breeding ground for mosquitoes. This is a potential mosquito breeding site that needs to be controlled. Several studies also state that interventions are needed to increase knowledge, attitudes, and actions in controlling dengue. ^{27 – 29} Perceptions of the threat to the disease and the benefits of the interventions also influence attitudes towards dengue control. This is in accordance with the concept of the health belief model. ³⁰

Attitudes towards dengue control using digital technology need to be studied for the driving and inhibiting factors. Intervention with digital technology requires supporting facilities and infrastructure. ^{18 , 17} Digital educational media that can increase public knowledge is needed during the COVID-19 pandemic. Through socialization with digital educational media, prevention activities can be carried out widely despite a social distancing policy.

The development of digital technology has helped health services to provide fast and integrated services. Particularly in dengue control, digital interventions can help conduct surveillance of areas with mosquito density, larval density, and several dengue cases so that they can make a priority scale. ^{31 , 32} Utilization of digital technology in conducting digital data collection and education can provide early detection and response to dengue cases in an area.

The theoretical model adopts a health belief model between perceptions and dengue control behaviour. ^{33 – 35} The health belief theory is then modified by adding environmental variables and the need for digital technology. The hypotheses were compared with six latent constructs related to dengue control attitudes, influenced by perceptions of the threat of dengue, program benefits, environmental factors, program constraints, and technology needs ( Figure 1). ³⁶ The direction of the path shows the (+) and (-) effects of the relationship, and this study assessed the accuracy of the model and hypothesis with PLS-SEM.

Regarding potential bias in this study, online data collection means that respondents can answer questions repeatedly. Thus, to reduce bias, data validation was carried out based on names and addresses. Incomplete answers, this is done with a re-checking system and requires answering. Respondents also only represented the Denpasar City area, not representing Indonesia.

Various factors influence attitudes in efforts to control dengue. During the COVID-19 pandemic, social restrictions were carried out. ³⁷ Social restrictions impact dengue control programs. Based on the theory of behavior change that a person is motivated to make prevention efforts when they feel seriously threatened and feel the benefits of intervention. The use of digital technology in dengue surveillance is also encouraged by the benefits of these digital applications and their ease of use. The health belief model approach is used because it is appropriate to form a model for changing attitudes towards dengue control related to perceptions of dengue threat, benefits of dengue control programs, and perceptions of constraints. ^{38 , 39}

Environmental factors affect the transmission of dengue infection. Dengue is transmitted by the Aedes aegypti mosquito, which breeds in water containers. The large number of water containers in an environment can affect the density of mosquitoes. Climatic factors (temperature, humidity and rainfall), as well as population density, also increase dengue transmission. ^{40 , 41} Environmental factors can increase the potential for dengue transmission in an area. The perception of environmental factors is essential to analyze.

Perceived ease of use and perceived benefits influence the perceived need for technology. This is in accordance with the Technology Acceptance Model (TAM). ⁴² Several studies on TAM are related to public acceptance of an application. ^{43 – 46} The use of dengue surveillance applications is needed to collect data and education. Digital applications can record quickly and integrate. This study uses a combination of several theories to create a suitable model according to stakeholder needs. The determinants of these variables are by the need to develop a dengue control attitude model related to technological needs and perceptions of environmental factors.

Based on the literature discussed, the hypotheses that emerged to provide the scope of this study are shown in Table 1.

This cross-sectional study is conducted using an online survey with 6 variables. ⁴⁷ These include perceptions of the need for digital information systems, dangers of DHF, benefits of DHF control programs, program constraints, and environmental factors related to attitudes toward controlling DHF. Respondents answered with a Likert scale of 1-5, where 1, 2 3, 4, and 5 represent strongly disagree, disagree, neutral, agree, and strongly agree. The questionnaire was made by discussing with experts and testing about 30 respondents to measure the validity and reliability. Respondents were selected based on inclusion criteria, aged more than 17 years, having an address in Denpasar City for more than one year, and willing to answer questions. The results of the validity and reliability tests found that 46 of the indicators were declared valid. Invalid indicators are excluded and not used. The final questionnaire can be found as Extended data. ⁴⁷

It was then distributed online using a google form, and data collection was carried out in the Denpasar City area, which is endemic to DHF. Table 2 shows a description of the data from the composites and indicators, as well as the definitions of attitudes towards dengue control efforts with the other five composites.

Jumantik is a volunteer recruited from each village area to inspect, monitor, and control dengue vectors. They were given the task of conducting daily inspections to visit homes. The results of their activities are reported as vector entomological surveillance. This is part of community empowerment to carry out dengue control in their area actively.

The inclusion criteria were respondents who were over 17 years old and had resided in Denpasar City for more than six months. They are willing to fill out a research approval form and receive mobile phone credit from the internet provider for two. Even though 596 respondents filled in the data, only 515 fulfilled the requirements and were complete. Sampling was carried out with non-random sampling conducted online in a limited population with the consideration that respondents could not be visited directly due to the COVID-19 pandemic in the Denpasar City area, which had previously been permitted by the Licensing Service, Health Service, Head of Public Health Center, and Village Head.

The online survey was chosen because it was appropriate during the Covid-19 pandemic. Online surveys are more accessible and cheaper than manual surveys using door to door.

Measurement of sample size using the following formulation. S = Z 2 x P x 1 − P / M 2 S = 1.96 2 x 0.5 x 1 − 0.5 / 0.05 2

S = 384

Notes:

S = Sample size for infinite population

P = Population proportion (Assumed as 50% or 0.5)

M = Margin of error = 5%

Z = The Z-score wil be 1.96 if the confidence level is 95%

This means that: N = 384, z = 1.96, M = 0.05 and p = 0.5

This study consists of six variables with 46 indicators using a Likert scale of 1-5, where 1, 2, 3, 4, and 5 representing strongly disagree, disagree, neutral, agree, and strongly agree. Attitudes toward prevention strategies are a dependent variable that tends to act to regulate dengue in the surrounding environment through the use of vector control activities at breeding sites for mosquitoes. Therefore, nine indicators are measured, namely willingness to carry out a weekly movement to eradicate mosquito breeding areas, close water reservoirs, clean the environment regularly, filling in data on larval density weekly independently, providing assistance to dengue control programs, supporting students’ weekly larvae care activities, willing to be sanctioned when larvae are discovered, willing to pay a fine, and making efforts to eradicate mosquito breeding areas following the officer’s advice.

Perceptions of the benefits are related to the assessment of dengue control programs beneficial to the community. These consist of 7 indicators: jumantik volunteers always visit every month, the officers always provide information, the volunteers provide larvacide, the program is useful for preventing dengue cases, and students play a role for larvae, the dengue control program was supported, and the officer’s advice was followed.

The perception of being threatened with dengue is a condition that causes feelings of fear and vulnerability to outbreaks which consist of 6 variables. These include the risk of being infected with dengue fever and several families at risk of being infected. Dengue fever is a deadly, easily contagious, and dangerous disease that people are afraid of being infected.

Perceptions of program constraints are obstacles in carrying out activities related to facilities and pandemic conditions in dengue control. These consist of 8 variables of limited program funding, jumantik personnel, home visit activities due to social distancing and COVID-19, visiting time, which is during working hours from 8 to 10, the information provided, the larvicides, and smartphone facilities.

The need for digital technology is a public perception of the support for implementing the systems in dengue control. These consist of 8 variables, namely being willing to use mobile phones for dengue control programs, filling in data on websites, watching digital educational videos, sharing information with family, supporting digital information system programs, having Android phones that support the program, having social media applications such as WhatsApp, Facebook, Instagram, and others, but accustomed to using WhatsApp to communicate.

Perception of environmental factors is the surrounding conditions that affect the density of larvae and dengue cases, both natural and artificial. These consist of 8 variables, namely the rainy season affects the incidence of dengue, the number of water reservoirs affects the density of mosquitoes, the Aedes mosquitoes lay eggs in clean water, the bucket filled with water in bathroom containers, used bottles, tires, and vacant places have the potential to become a breeding place.

This study was analyzed using PLS-SEM with SmartPLS 3.0 software. It analyzed five variables related to attitudes towards dengue control. The PLS-SEM analysis uses two stages, and the first describes the measurement model connecting the constructs and indicators to the theory. In the second stage, the structural model determines the determinants of the relationship between the construction and the hypothetical model.

This study is part of a research carried out for the development of an integrated dengue control system. This study has been approved by the ethics committee of the Faculty of Public Health, University of Indonesia (Ket-416/UN2.F10. D11/PPM.00.02/2021). Before data collection, informants had received information about their goals, risks, and rights. In addition, a written consent form was given before the interview, and all information from participants is confidential and for this study only.

Composite mode A

The composite measurement model in mode A (attitude) was assessed in individual item reliability, construct reliability, convergent validity, and discriminant validity. First, the reliability of each item is analyzed through a loading factor, as seen in Figure 5. ³⁶

Table 4 shows the value of the measurement of validity and reliability. Cronbach’s Alpha value and composite were used to evaluate construct reliability. The values show that the construct exceeds the recommended cut-off of 0.7. Convergent validity was also proved because the construct’s extracted mean-variance (AVE) was higher than 0.500. Table 4 shows that the measurement model meets the criteria.

Table 5 presents discriminant validity results through the heterotrait-monotrait (HTMT) correlation ratio. All constructs reach discriminant validity because the confidence interval does not contain a zero value. This situation means that each variable is different from one another. The data examined in the measurement model show that the attitude construct measure is reliable and valid.

Composite mode B

The composite measurement model in mode B was assessed in collinearity between the outer weights’ indicators, significance, and relevance. First, removing the indicator is carried out when the value exceeds the variance impact factor (VIF = 3). As a result of this process, only the indicators shown in Table 2 are without collinearity. Second, the relevance of the weights is analyzed, and Figure 6 ³⁶ shows the indicators in construction for latent variables. Finally, it is possible to start a bootstrap with 10,000 sub-samples to assess significance. Indicators with insignificant weights but significant loadings of 0.50 or higher were considered relevant ( Table 6).

Structural model

The structural model is evaluated after verifying the correctness of the construction measurements. The path coefficients and their 10,000 resampling bootstrap significance levels are reported in Table 7 and Figure 6. ³⁶ Additionally, Table 7 shows that the VIF constructs range from 1,000 to 1,700, indicating no collinearity between variables. This study also assesses the quality by examining the overall predictive relevance of the model with a Q2 value above zero which indicates a fit in the prediction model. The magnitude of Q2 has a value of 0 < Q2 < 1, where the closer to 1, the better the model. The coefficient of determination (R2) also exceeds 0.1 for endogenous latent variables since the construct has an acceptable predictive power quality.

From Table 7, there is a direct influence of Perception of Environmental Factors on the Threatened Perception of Dengue, Program Constraints, Program Benefits, and Digital Technology Needs. Perception of Program Benefits and Threatened Perception of Dengue directly influences Digital Technology Needs. In general, Perception of Digital Technology Needs and Program Benefits directly influence Attitude Towards Dengue Control. Variables Perception of Digital Technology Needs and Perception of Program Benefits positively correlate to Attitude Towards Dengue Control.

VAF values above 80% indicate that the variable serves as a full mediator. The variable can be categorized as a partial mediator when the VAF value ranges from 20% to 80%. However, when the value is less than 20%, it can be concluded that there is almost no mediating effect. The value of VAF indicates that the proportion of Perception Of Digital Technology Needs from the pathway has no mediating effect (VAF<0.2 or 20%). Perception of Digital Technology Needs, Threatened Perception of Dengue, and Program Benefits can be categorized as partial mediators between Environmental Factors and Attitudes Towards Dengue Control (see the indirect effect in Table 7).

The strength is the development of a model that combines measurement of attitudes towards dengue control with environmental factors on the threatened perception of dengue, program constraints, program benefits, and digital technology needs. The commonly used model is the health belief, but a different approach combines the perceived need for digital technology, environmental factors, and health beliefs.

This study uses PLS-SEM analysis which selected because it is variance-based and estimates composite components and factors. ⁷⁰ The PLS analysis is a multivariate statistical technique that compares several responses and explanatory variables. ^{22 , 71 , 72} Through this approach, it is possible to make appropriate structural equations toward dengue control related to the perception of environmental factors on the threatened perception of dengue, program constraints, program benefits, and digital technology needs.

The use of online surveys is limited to certain areas and does not represent the whole of Indonesia, only Denpasar City. Generally, the respondents used were those with mobile phones and internet networks, and they were not randomly assigned.

This study adds to the literature and provides a comprehensive understanding related to attitudes in dengue control, perceptions of program benefits, perceptions of dengue threats, perceptions of constraints, perceptions of the need for digital technology, and perceptions of environmental factors. ^{39 , 73} This study also contributes to supporting the health belief model. ^{30 , 74} This study has added to the theoretical literature by developing a structural model related to dengue control attitudes, especially in Indonesia.

The results are helpful for policymakers to promote the use of digital technology in data collection of disease cases, surveillance, monitoring, and evaluation of health programs supported by socialization through social media that can influence perceptions of the benefits of the program. The community’s attitude toward controlling the disease is also related to the source of information that affects public perception. Policies to support digital facilities such as the availability of internet networks, computer facilities, mobile phones, and data packages affect the disease reporting system and its control. In the future, it is necessary to develop an integrated digital system for reporting disease cases and collecting data on the ecological environment, specifically larval density. This system should perform spatial mapping and predict the potential for a dengue outbreak to occur. Therefore, technology can be helpful in case surveillance for quick control measures.