*2.3. Enzyme Immunoassay Test*

Enzyme-linked immunosorbent assay (ELISA) was used to detect dengue IgG and IgM antibodies. The IgG antibody was measured through indirect ELISA (LOT: 01P20A006, Inverness Medical/Panbio, Windsor, Australia). The IgM antibody was tested via capture ELISA (LOT:01P30A002, Inverness Medical/Panbio, Windsor, Australia). Undefined results were confirmed through the colloidal gold method (LOT: DEN141001, Inverness Medical/Cortez, Calabasas, CA, USA). The details of the assays could be found in a previous publication [19].

### *2.4. Case Definition and Control Selection*

Among the 3136 serum samples, 305 and 103 were identified as IgG-antibody positive and IgM-antibody positive, respectively, and 34 were positive for both antibodies. Thus, 374 individuals with antibody-positive samples were defined as dengue-infected, and those willing to participate in the questionnaire survey were selected as the members of the case group.

In this study, 256 patients with dengue infections opted to fill in the questionnaires, and 19 questionnaires missing most of the important information were eliminated. Eventually, 237 cases were included in the case group.

The controls were selected through frequency matching from individuals who tested negative for IgG and IgM. Specifically, candidate controls were stratified in accordance with the age, gender, and community information of the case group and selected through convenience sampling, wherein participants volunteered to become part of the samples, from each layer. Additionally, age matching was carried out according to ≤15 years, 16–30 years, 31–50 years, 51–65 years and ≥66 years. Finally, 237 controls were selected.

### *2.5. Data Collection and Analysis*

The phone questionnaire investigation was conducted by trained investigators. Subjects who interrupted the telephone investigation and whose questionnaire information contained logic errors were interviewed face-to-face to verify the integrity and validity of their information. Just as shown in Supplementary Materials File 1, the main contents of the questionnaire included general demographic characteristics (age, gender, blood type, and average household income). It also included personal life activities, such as activities in the park, outdoor sports (such as hiking, mountain climbing, and camping), and outbound tourism experience. Moreover, the questionnaire presented questions related to environmental sanitation (domestic sewage and garbage managemen<sup>t</sup> and participation in community hygiene managemen<sup>t</sup> interventions); housing situation, such as the age and area of domiciles and living floor; and living conditions (average numbers of occupants per room, use of air conditioning, quality of indoor daylight, and presence of animals or aquatic plants on property). It also had questions on mosquito protection status (use of mosquito nets and pesticide) and residential surroundings (presence of junk yards, ponds, or construction sites within 200 m).

"Contact with patients with DF" was defined as previous living or working experience with patients with DF in the past years in his/her life. The definition of "outdoor activities in parks" was established as outdoor activity for at least two times per week and for more than half an hour each time. "Participation in outdoor sports" was defined as participation in activities at least twice a year for more than half an hour each time. The definition of "good indoor daylight quality" was established on the basis of the minimum requirement for sunshine in general residences of not less than 2 h on extremely cold days. The definition of "good ventilation" was given to an open-ventilation area of not less than 5% of the floor area of each domicile. The use of mosquito repellent referred to the use of mosquito coils and insecticide vaporizers. "Occasionally" was defined as less than once a week, and "often" was defined as at least once a week.

Epidata 3.1 software (Epidata Association, Odense, Denmark) was used to establish a database of individual risk factors for dengue infection among residents in Guangdong Province. All data were

analyzed by SPSS statistics 23.0 software (SPSS Inc., Chicago, IL, USA). The χ2 test was used to test for differences in demographic characteristics between cases and controls. A univariate unconditioned logistic regression analysis was applied for the preliminary screening of variables identified by using the questionnaire but not for the variables of general demographic characteristics. An unconditioned logistic regression analysis method for multivariate analysis was employed to analyze the relative importance of statistically significant variables in univariate analysis. Additionally, considering the rule of frequency matching design, the age, gender and community information variables were also introduced. Then, stepwise regression was used to establish a regression equation. *p* < 0.05 was set as the significance level of the χ2 test and multivariate analysis. However, to avoid missing important factors, *p* < 0.1 was set as the significance level in univariate analysis. In addition to odds ratios (ORs), 95% confidence intervals (CIs) were used to express associations.
