Population Adherence to Infection Control Behaviors during Hong Kong’s First and Third COVID-19 Waves: A Serial Cross-Sectional Study

Background: Although COVID-19 has affected over 220 countries by October 2021, there is limited research examining the patterns and determinants of adherence to infection control measures over time. Aims: Our study examines the sociodemographic factors associated with changes in the frequency of adherence to personal hygiene and social distancing behaviors in Hong Kong. Methods: A serial cross-sectional telephone survey in the general population was conducted during the first (March 2020) (n = 765) and third wave (December 2020) (n = 651) of the local outbreak of the COVID-19 pandemic. Respondents were asked about their level of compliance with various personal hygiene and social distancing recommendations. Results: By the third wave, mask use increased to 100%, and throughout the study periods, >90% practiced frequent hand hygiene. However, adherence to social distancing measures significantly waned over time: avoidance of social gatherings (80.5% to 72.0%), avoidance of public places/public transport (53.3% to 26.0%), avoidance of international travel (85.8% to 76.6%) (p < 0.05). The practice of ordering food takeout/home delivery, however, increased, particularly among high-income respondents. Higher education, female gender and employment status were the most consistently associated factors with adherence to COVID-19 preventive practices in the multivariable models. Conclusions: In urban areas of this region, interventions to improve personal hygiene in a prolonged pandemic should target males and those with low education. In addition to these groups, the working population needs to be targeted in order to improve adherence to social distancing guidelines.


Introduction
The outbreak of novel coronavirus 2019 (COVID-19) has been a significant global public health threat since its emergence in December 2019. As of 3 October 2021, the SARS-CoV-2 virus has spread to over 220 countries with 234 million reported cases and 44 million deaths [1]. Although the novel coronavirus has shown high transmissibility and high case fatality rates in many regions of the world, national responses to COVID- 19 have varied greatly among the affected countries [2,3]. During the early phase of the pandemic, lockdown orders were made to restrict the movement of citizens in many countries across Europe and Asia [4][5][6]. By contrast, countries such as Sweden and Iceland Hong Kong did not undergo a lockdown despite 1755 local cases, but it was noted that the community widely adopted various preventive behaviors. Mask use rose and increased hand hygiene rose to approximately 95% of the population, while the avoidance of crowded places and home disinfection rose to over 80% of the population [16]. By contrast, a study conducted between 2005 and 2008 of H5N1 in which no human deaths occurred in Hong Kong, found that the levels of preventive behaviors that the public anticipated adopting in the event of human-to-human transmission declined (73.8% to 43.7% for mask use and 72.6% to 51.5% for avoidance of public places) [17]. There is, however, limited understanding of the pattern and how Health-Emergency Disaster Risk Management (Health-EDRM) personal preventive behaviors may evolve with the prolonged biological emergency.
Enactment of public health and social measures including physical and social distancing measures (e.g., lockdowns and restaurant/bar restrictions) and personal measures (e.g., hand hygiene/mask wearing) was shown to contribute to halt the spread of the disease [44]. In light of this, we aimed to examine the population adherence to these infection control behaviors in a region whose population is highly experienced in pandemic response due to a long history of zoonotic epidemics. It is theorized that due to Hong Kong's past experience with successive pandemics [45], the population would rapidly adopt high levels of personal hygiene (e.g., mask wearing) and social distancing guidelines in the first local wave. The study also will examine changes from the first wave of the pandemic when only 766 local cases and 5 deaths were confirmed in Hong Kong with the third wave of the pandemic when the total confirmed local cases had risen to 8725 with 179 deaths. It is theorized that the personal hygiene and social distancing behaviors would increase during the third wave due to higher perceived susceptibility (from much higher daily caseloads) and perceived severity of COVID-19 (from much higher numbers of deaths). In addition, the study aims to identify the sociodemographic risk groups for non-compliance with these preventive behaviors to allow for rapid interventions to these target groups. By shedding light on patterns of infection control compliance in a large metropolis with previous SARS and avian influenza pandemic experience, these findings may provide insights for post-COVID-19 pandemics in other regions of the world that have limited pandemic experience.

Study Design and Sampling
Serial cross-sectional telephone surveys were conducted using random digit dialing in Hong Kong at two time periods: (1) first wave of local epidemic (22 March to 1 April 2020) (n = 765) and (2) third wave of the local epidemic (n = 651) (15-29 December 2020). The sample sizes of 765 and 651 were accompanied with maximum margins of error of 3.5% and 3.8% at a 95% confidence level. The target population of the telephone survey were Cantonese-speaking Hong Kong residents aged 18 years or above, including individuals holding work or study visas. Stratified random sampling was used to ensure that the data collected were representative of the general population in Hong Kong in terms of district of residence. This data collection strategy has been used in similar local studies on infectious diseases [13,19]. Phone calls were made in the evening on weekdays and during the entire day on weekends to prevent overrepresentation of the unemployed. The household member with birthday closest to the interview date was selected for the baseline survey ("last birthday" method). After the purpose of the survey was explained and assurances of the confidentiality of response were made to potential respondents, oral consent was obtained from participants.

Data Collection and Data Management
Information about the respondent's sociodemographic and background characteristics was obtained in both surveys as the main exposure variables. Specifically, the surveys obtained information about the respondent's gender, age group, marital status (married versus all other), monthly household income, highest educational attainment, type of occupation and whether the individual had ever been diagnosed with a chronic health condition such as hypertension or diabetes (see Table 1). Since COVID-19 case distribution varied by geographic region, we also asked about the respondent's area of residence (18 districts in Hong Kong). Respondents were also asked about the channels by which they obtained their COVID-19 news: television, internet, smartphone applications (Yes/No).
For the outcome measures, respondents were asked about their compliance in adopting four personal hygiene measures (e.g., wearing a mask, handwashing) and four social distancing measures (e.g., avoidance of public transport) (see Table 2). For each protective measure, the level of adoption was assessed with four-point Likert scale responses (1 = always, 2 = usually, 3 = occasionally, 4 = never) during Wave 1 and Wave 3. Respondents were considered compliant with the preventive behavior if they responded that they "always" or "usually" engaged in the preventive behaviors, while those who responded that they only "occasionally" or "never" performed the behaviors were classified as non-compliant. These eight protective measures were used as outcome variables, while sociodemographic variables were used as predictors in the subsequent regression analyses.

Statistical Analysis
Proportion and frequency of responses were tabulated and shown in Tables 1 and 2. Cohen's w effect size was used to compare the demographics of respondents with the 2016 population by census in Hong Kong, while the chi-square test was used to examine population differences between Wave 1 and Wave 3 [46]. The four-point Likert scale responses for the preventive behaviors were dichotomized [47], whereby respondents were considered to be strong adherents if they responded 1: "Always" or 2: "Usually" to the item. Two-proportions z-test with continuity correction was used to test the difference of the personal protective measures between the two waves of studies. To examine the persistent sociodemographic factors independently associated with adherence with the personal hygiene and social distancing behaviors across time, we included all sociodemographic variables as candidate variables in the backward stepwise multiple logistic regression model, where the variable with the largest p-value was removed at each step. The multivariable models forced the time period (Wave 1 versus Wave 3) and age variable into the final model to adjust for the potential confounding effects. Adjusted odds ratio (AOR) in the logistic regression models with corresponding 95% confidence interval (95% CI) was used as a measure of independent association between an exposure and an outcome. Due to the 100% compliance with mask wearing in Wave 3, we did not further examine factors associated with compliance for this personal hygiene behavior. Separate analyses for the 1st and 3rd wave were conducted (see Appendix A) to examine factors associated with different protective measures adherence at two different study periods [48]. Statistical significance was set at α = 0.05. All analyses were performed using IBM SPSS 24 [49]. Ethical approval was received by the Survey and Behavioral Research Ethics Board of the university sponsoring the study.

Characteristics of the Study Sample
A total of 765 and 651 participants were recruited in the baseline survey and second survey, respectively (see Figure 1). Between the two serial cross-sectional samples (Table 1), the respondents in the second survey sample were slightly older, slightly less educated and had lower income levels. Descriptive statistics of the personal characteristics of the study samples are shown with the comparison to the characteristics of the most recent Hong Kong population census. Although the study samples were comparable to the census characteristics, the third wave sample were older and had lower education status and household income than the first wave sample. In this study, none of respondents reported that they had been infected by  and had lower income levels. Descriptive statistics of the personal characteristics of the study samples are shown with the comparison to the characteristics of the most recent Hong Kong population census. Although the study samples were comparable to the census characteristics, the third wave sample were older and had lower education status and household income than the first wave sample. In this study, none of respondents reported that they had been infected by COVID-19.     Table 2 showed the practice of preventive measures against COVID-19 in the first and third epidemic wave. Overall, the self-reported practice of most of the hygiene preventive measures listed were high except for the practice of bringing one's dining utensils when dining out. The cultural habit of bringing one's own dining utensils (e.g., bringing personal chopsticks) to restaurants, which is occasionally practiced by the elderly or very young children, was not commonly practiced during this epidemic, with only 1 in 12 respondents reporting to have done this in the early part of the pandemic, and these percentages dropped further in the later stages. Wearing face masks, handwashing with soap, and strict use of serving utensils for shared dishes were practiced widely but only mask wearing showed a statistically significant difference between the first and third waves, rising to 100% of the surveyed respondents.

Compliance Levels with Recommended COVID-19 Preventive Behaviors across Time
All four social distancing practices examined in this study showed statistically significant differences between the first and third waves of the study. Whereas respondents were significantly less likely to avoid international travel to high-risk regions, avoid social gatherings, avoid public places/public transport in the third wave, a much higher proportion of respondents adopted the use of food takeout/home delivery services to avoid eating in restaurants ( Table 2). * % of respondents who stated that they "Always" or "Usually" engaged in the behaviors. a Two-proportions z-test with continuity correction was used to test the difference between the two waves of studies. † This analysis only included people who will go outdoors for a meal during the epidemic. Table 3 shows the factors associated with the practice of various personal hygiene and social distancing measures after forcing the age variable and time period of the study (Wave 1 versus Wave 3). Due to the near ubiquitous use of face masks in Wave 1 and 100% compliance in Wave 3, we did not examine mask use further in these multivariable models.

Factors Associated with Compliance with Hygiene Measures against COVID-19
For personal hygiene behaviors, the results showed that while hand hygiene and the practice of bringing one's own dining utensils did not show any significant change between Wave 1 and Wave 3, the strict use of serving utensils had declined by Wave 3.

Factors Associated with Compliance with Social Distancing Measures against COVID-19
For the social distancing behaviors, the multivariable models revealed that the practice of all behaviors had significantly declined by Wave 3 except for the practice of ordering takeout/food delivery services, which had significantly increased during this time period.
During the study periods, females, those with higher education and non-employed respondents were significantly more likely to avoid social gatherings (AOR: 1.44-2.65). Similarly, avoidance of public places/public transport was more likely to be practiced by those with higher education and non-employed respondents as well as housewives and full-time students (AOR: 1.61-3.53). Respondents under 50 years of age, higher-income individuals and males were more likely to use restaurant takeout/home delivery services (AOR: 2.33-2.45). For avoidance of international travel, only housewives were more likely to avoid traveling abroad (AOR: 1.85, 95% CI: 1.09-3.16) as compared with white-collar workers. Having a chronic health condition was not associated with any of the preventive behaviors examined in this study.

Discussion
Overall, study findings reveal that residents in Hong Kong, a city with a long history of worldwide pandemics, were quick to adopt basic COVID-19 preventive practices. In contrast to many other countries in which citizens have resisted mask wearing as an infringement of civil liberties, there was a rapid uptake of mask use among Hong Kong residents during the first wave of the pandemic when the caseload was still relatively low. In addition to the near ubiquitous uptake of mask wearing in the first local wave of the pandemic, there was a high degree of compliance with hand hygiene measures by the general population. Our results were consistent with previous findings from an avian influenza epidemic showing high adherence to mask use and lower adherence to social distancing measures [13]. Although mask wearing and hand hygiene did not significantly decline between epidemic Wave 1 and Wave 3, the majority of social distancing practices waned over time. This may be attributed to the decreasing perceived disease severity across time [50]. Even though nearly all respondents routinely checked the daily COVID-19 situation in Hong Kong, their adherence to social distancing significantly decreased during the much larger third wave of the local pandemic. These findings suggest that combatting pandemic fatigue in adhering to social distancing guidelines will be the major focus of ongoing government infection control efforts.
Our study identified sociodemographic determinants of preventive-behavior uptake in Hong Kong during the COVID-19 pandemic. In addition to older age and lower education level, male respondents were more likely to have poorer hand hygiene, which was consistent with studies conducted abroad [27]. Since the elderly are at higher risk of COVID-19 mortality, hand hygiene should be strongly promoted in this age group. Higher education status but not higher income was commonly associated with the adoption of multiple preventive behaviors. It is possible that those with higher education may be better at evaluating public health messages and integrating the recommended actions into their daily life. It is also possible that those with lower education have lower perception about the transmissibility of the disease, incorrect ideas about the transmission routes and inadequate knowledge of the potentially severe morbidity from COVID-19. This suggests that increased knowledge rather than financial resources are the main drivers of adoption of these preventive behaviors. Past studies conducted in during SARS and H5N1 noted that misconceptions were prevalent in the population with regard to transmission routes [15,16]. Hence, government public health messages, which have mainly served to encourage mask wearing, should also include information about the potential serious sequelae of non-fatal disease and emphasize the possible airborne and non-airborne transmission routes so that less educated individuals can have a better understanding of the risks. Moreover, work-related considerations are found to be important influences in the practice of these preventive behaviors. Housewives and non-employed individuals who have less need to leave the home were more likely to engage in social distancing measures such as avoidance of public places, public transport and social gatherings. Despite allowances for working at home by many industries in Hong Kong, a large proportion of workers in retail, hospitality and educational sectors were still required to come to work. Hence, protective measures in the workplace and in public transport need to be emphasized and maintained, particularly in high-density urban areas.
As a likely result of intermittent prohibitions against evening dine-in services in Hong Kong restaurants since July 2020 [28][29][30], the use of food delivery/take out services greatly increased between the two study periods. Restaurant delivery services, which were rare in Hong Kong due to the close proximity of eating establishments to residential areas, quickly became available using mobile applications during this COVID-19 pandemic. It is unsurprising that younger people were more likely to use these services due to the digital nature of food ordering. Those from more affluent households were also more likely to use takeout/home delivery services since these services usually incur delivery costs or minimum spending requirements [31]. It is also likely that males used these food ordering services more frequently than females, who tend to prepare and consume homemade meals.
The Hong Kong government has recommended the strict use of serving utensils during communal meals since the SARS epidemic in 2003 since Chinese cuisine consists of communally shared dishes rather than individually served meals. Although the majority of the population adopted these recommendations during the COVID-19 pandemic, bluecollar workers, housewives, non-employed people and those with a low education were much less likely to do so. These subgroups may, therefore, be targeted for dining hygiene educational interventions in the future.
Compared with the personal hygiene measures, the uptake of social distancing measures was much lower and decreased significantly over time. Behavioral health models such as the Health Belief Model posit that perceived susceptibility and perceived severity of disease are directly associated with likelihood of adopting a health behavior. Although nearly all respondents sought information about COVID-19 from various news sources, the higher incidence of disease and death in the third wave did not result in the sustained adoption of social distancing measures. It is also likely that lower adherence to social distancing measures during the third wave reflects the population's fatigue with stringent measures such as the closure of bars, prohibition of restaurant dining in the evening hours and inability to travel outside of Hong Kong. However, we cannot rule out the possibility of lowered risk perception due to the population learning more about the disease [42]. Furthermore, when employees resumed working in offices, they were often urged through public reminders to observe personal hygiene measures, but public service messages did not discourage after-work gatherings. Difficulties in maintaining social distancing for workers and the lack of guidelines for workspaces may need to be addressed in a protracted pandemic. These findings indicate that in Hong Kong, despite strong adherence to mask use and hand hygiene practices, adherence to social distancing measures will be a challenge for long-term pandemic control. To improve Health-EDRM protection in urban contexts, periodic reassessment of population behaviors should be conducted by policy makers during prolonged biological emergencies.

Research Limitations
This study has several limitations. Firstly, although the land-based telephone list covered around 85% of land-based telephones in Hong Kong during the two study periods [51], the households that were not on the list of land-based telephone service (institutionalized individuals, dormitory residents and individuals who live in boats and non-conventional dwellings) were not included. Secondly, our study sample was slightly older than the general population of Hong Kong but was comparable with the latest population census in terms of gender and area of residence. The third-wave sample comprised slightly older and lower-SES participants than the first wave. Hence, there may be some confounding by these characteristics in reporting the prevalence of preventive behaviors. Thirdly, this study was based upon self-reported data that may be subject to reporting bias. However, since the participants of the survey were anonymous, any reporting biases should be moderate. Lastly, changes in preventive behaviors between the serial cross-sectional studies cannot be conclusively stated to be related to the time period of the study, due to possible residual confounding by sociodemographic and unmeasured lifestyle variables. A longitudinal cohort-based study would be more appropriate to examine long-term trends in behavior changes.

Conclusions
In examining pattern changes of information sources, attitude and practice of COVID-19 protective measures, our results showed that the general population rapidly adopted COVID-19 preventive behaviors and that personal hygiene preventive measures remained high during the first nine months of the pandemic in Hong Kong. Mask wearing and continued hand hygiene are not behaviors requiring concerted public health efforts in Hong Kong. Despite the rising caseloads and deaths, social distancing behaviors waned significantly over time. Pandemic fatigue may, therefore, present a larger challenge for governments facing protracted pandemics.
Hygiene and social distancing messages should be targeted not only at high-risk sociodemographic groups but also at those showing greater pandemic fatigue. Health education programs aiming at improving COVID-19 knowledge may be helpful to improve continued adherence by those targeted groups.

Informed Consent Statement: Not applicable.
Data Availability Statement: The datasets generated during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.