This cross-sectional study was conducted using an anonymous web-based questionnaire between February 19 and March 19, 2020. A convenience sampling method was used to recruit participants. Chinese residents aged ≥60 years without intellectual or cognitive impairment who agreed to participate in our survey were considered eligible. Those who were illiterate or unable to use electronic devices could ask others to help them fill out the questionnaire; however, it was emphasized that the answers must always reflect the older person’s own opinions. The link to the web-based questionnaire was sent through WeChat (which is similar to WhatsApp) and other social media platforms to the contacts of potential respondents. A brief introduction was presented at the beginning of the survey to inform the respondents of the purpose and content of this study and to instruct them on how to complete the questionnaire. Respondents who completed the questionnaire were entered in a draw for a monetary incentive of approximately ¥1-5 (US $0.15-0.76).

According to Kendall, multiple regression analysis demands a sample size of at least 5 to 10 times the number of independent variables. Thus, we required a maximum sample size of 670 cases, as 67 independent variables were included in this study. Returned surveys with unreasonably short answer times, incomplete information, or logic problems were deemed invalid. The effective sample size estimated for the study was 957 cases, with an invalid response rate of 30%. Finally, a total of 1501 questionnaires were returned in the study duration; after excluding 212 invalid questionnaires and 26 participants who reported having never heard about COVID-19, a total of 1263 surveys (84.1%) from all of the 31 provinces, municipalities, and autonomous regions in mainland China remained for statistical analysis.

Descriptive statistics were used to summarize the background factors and COVID-19 knowledge, perceived beliefs, and behaviors; the results are presented as frequencies (n) and percentages (%) or as means and SDs. Associations between background variables and behavior scores were examined by one-way analysis of variance (ANOVA) or independent sample t tests as appropriate. A multiple linear regression model using statistically significant background variables, knowledge scores, and factors of perceived beliefs as independent variables and scores of preventive behavior as outcome variables was applied to identify factors associated with preventive behaviors. The demographic variables of the respondents were entered in the regression model first to control for their effects. Then, the knowledge score was entered in the next block of the regression, and factors of perceived beliefs were simultaneously entered in the last block. Unstandardized regression coefficients (β) and 95% CIs are reported. All statistical analyses were performed in SPSS version 21.0 (IBM Corporation), and a P value <.05 was considered statistically significant.

The study protocol was approved by the Ethics Committee of School of Public Health, Zhejiang University (approval number: ZGL202002-2) before the formal survey. The questionnaire was designed to be anonymous and voluntary, and respondents were informed that submission of the questionnaire implied informed consent. The data were kept confidential, and the results did not identify the respondents personally.

Of the 1263 participants, 730 (57.8%) completed the questionnaire with the help of others, and 687 (54.4%) were registered permanent residents of rural areas. The mean age was 69.48 years (SD 6.72), over half of the respondents were female (697/1263, 55.2%), and approximately three-fourths were married or cohabitating (941/1263, 74.5%). Of the 1263 participants, 586 (46.4%) had an education level of primary school or below, and most respondents (844, 66.8%) had an average household income of between ¥600 and ¥6000 per month (US $91.61-$916.07). Moreover, 564/1263 participants (44.7%) self-reported their physical health status to be fair; 613 (48.5%) of the 1263 respondents lived in the eastern region, and nearly half of the participants (591/1263, 46.8%) lived in areas with medium risk of COVID-19. The majority of respondents (1050/1263, 83.1%) reported that entry and exit control was exercised in their community or village.

As shown in Table 3, region, marital status, education level, residence registration, and monthly household income per capita were significantly associated with preventive behaviors (all P<.05). Participants who lived in western regions had significantly lower level of behavior than participants from eastern regions (P=.01). Respondents who were married or cohabitating had significantly higher behavior scores (P<.001). Rural dwellers and respondents who attended primary school or below had lower behavior scores (both P<.001). The behavior scores of participants with an average household income of <¥600 per month (US $91.61) were significantly lower than those of participants with monthly incomes higher than ¥6000 (US $916.07; P=.02). Interestingly, a significant association between local community-level control measures and behavior score was observed (P<.001); respondents engaged in more preventive behaviors when epidemic control measures were exercised in their areas of residence.

The knowledge scores and perceived belief-related variables reported in Table 1, together with significant background variables, were included in the hierarchical linear regression models. In block 1, the demographic factors of region, marital status, education level, registered residence, monthly household income per capita, and local community-level control measures were entered first, accounting for 4.7% of the variance (R²=0.047, F_6,1256=5.575, P<.001). The score for knowledge about COVID-19 was entered into block 2 and contributed an additional 2.9% of the variance (R²=0.076, F_7,1255=8.603, P<.001). Finally, factors of perceived beliefs regarding COVID-19, including perceived severity, perceived susceptibility, perceived benefits, self-efficacy in preventing the disease, and confidence in the government, were entered in the third block; these factors were significant in predicting behaviors even after considering the effects of background and knowledge factors, and they contributed an additional 3.2% of the variance (R²=0.108, F_12,1250= 8.904, P<.001). The linear regression analysis of the preventive behaviors is presented in Table 4 and Table 5.

Education level and registered residence were generally nonsignificant, while some significant marital status and control measure differences were noted. Overall, respondents who were married or cohabitating (vs single, divorced, separated, or widowed, β=0.355, 95% CI 0.150-0.560) and who could not leave the house (vs free entry and exit, β=0.898, 95% CI 0.310-1.486) were more likely to have better preventive behavior scores (P<.01). After adjusting for background characteristics, we discovered that older people with higher COVID-19–related knowledge scores had significantly higher behavior scores (β=0.198, P<.001). Meanwhile, the results showed that 3 of the 5 components of perceived beliefs, namely, perceived susceptibility (β=0.263, 95% CI 0.022-0.504), perceived benefits (β =0.643, 95% CI 0.305-0.982), and self-efficacy in preventing the disease (β =0.468, 95% CI 0.223-0.713), were positively associated with preventive behaviors (all P<.05); however, perceived severity and confidence in the government were not significant predictors of preventive behaviors. Among the influencing factors, local community-level control measures showed the greatest impact on behaviors, followed by perceived benefits.

Older people, who are particularly vulnerable to acute diseases and their complications, are currently greatly threatened by the outbreak of COVID-19. Special effort should therefore be made to encourage older people to practice the behaviors suggested by the government to prevent the disease. An understanding of older people’s knowledge, perceived beliefs, and behaviors toward COVID-19, which affect the adoption of related health behaviors, can be a first step to prevent the spread of COVID-19 in this population.

In this study, it was found that Chinese older adults had good knowledge of COVID-19; the respondents had an overall correct rate of 87% on the knowledge questionnaire, which was slightly lower than a rate reported for the Chinese general population (90%) [40] but much higher than that reported for US residents (80%) [41]. The high correct rate was surprising, as previous studies suggested that knowledge scores are negatively related with age [42-44]. However, this finding may be due to the large amounts of publicity related to COVID-19 through various channels that are appropriate to the needs and characteristics of older people, such as vivid prints, marked banners, and broadcasts in dialect. The 2 items with the lowest correct rates are worth mentioning. Knowledge about modes of transmission, which has been proved to be a salient factor influencing the level of adoption of preventive measures [45], should be further improved among older people. Meanwhile, older people should also be informed that an effective vaccine for COVID-19 is not yet available and that taking precautions against the disease is still highly important.

The perceived beliefs of older Chinese residents toward COVID-19 were found to be optimistic overall. Although the majority of respondents (1141/1263, 90.3%) perceived that older adults would suffer from more severe symptoms if they were infected, relatively few respondents (190/1263, 15.0%) thought they were at high risk of acquiring the disease. Moreover, the majority expressed confidence in themselves (1048/1263, 83.0%) and the government (1160/1263, 91.8%) to stop the spread of COVID-19. This optimistic belief could be explained by the unprecedented epidemic control measures taken by the Chinese government after the outbreak of COVID-19 in the country and the concerted efforts of people all over China to prevent the spread of the disease. From January 20, 2020, when person-to-person transmission was confirmed and the Chinese public was notified of this finding, a series of nationwide public health emergency measures, including isolation and quarantine, close management of working and living spaces, and the Examine and Approve Policy on the resumption of work, were implemented by the Chinese government, health institutions, communities, companies, etc [46].

In addition to efforts from all sectors of society, high adherence to the lifestyle modifications suggested by the government and public health organizations greatly delayed the spread of the disease [47]. In our findings, 64.6%-86.6% of respondents reported practicing five major preventive behaviors more frequently in the preceding week. The reason for this finding may be that this study was conducted at a time when these key preventive measures were highly emphasized in China. Similarly high tendencies to adopt these precautionary behaviors during the pandemic have been found in other studies [35,48]. In particular, more than 80% of respondents reported increased handwashing and face mask wearing and decreased time spent in crowds, indicating that basic protective behaviors against COVID-19 had taken root among older people. The majority of participants avoided going out, which can potentially be attributed to the widely disseminated governmental propaganda and social media messages, which continuously instilled and reinforced people’s incentives to stay home (eg, personal safety, good citizenship, and contribution to the control of the national epidemic). Moreover, individuals may have felt pressure from the community to adhere to behaviors such as imposed lockdowns or quarantine of entire regions with suspected cases. Handwashing and wearing of face masks have long been regarded as significant preventive habits in the daily life of the public since they were proved to be efficacious in preventing influenza and SARS [49,50]. Lau et al [51] observed that hand-washing and face mask wearing were commonly practiced during the SARS outbreak, and these behaviors were sustained by a large proportion of the public even after the SARS epidemic subsided in Hong Kong. In contrast, the effectiveness of household ventilation and disinfection may have been neglected, with fewer participants reporting that they had practiced these two behaviors more frequently in the past week. Habit has a great impact on routine behavior—including hygiene behavior—and despite their best intentions, people may find it difficult to implement new measures during a pandemic if they were not previously in the habit of performing them [52,53]. Further education on practicing household ventilation and disinfection is needed, as these two measures are indispensable in preventing COVID-19.

In line with previous research [54-56], our study also found that individuals’ demographic characteristics had a significant influence on their preventive behaviors. Linear regression analysis of the behavior scores showed that participants who were married or cohabitating were more likely to report that they had complied with advocated protective behaviors more frequently; this finding was consistent with similar studies regarding SARS [57,58]. On the one hand, older adults who are married or cohabiting can receive support from their spouse or partner. Family support has been shown to influence older people’s health beliefs and self-care behaviors [59]. The more family support older people receive, the more attention they will pay to their own health, accompanied with increased willingness to acquire health knowledge and more positive preventive beliefs. On the other hand, according to the HBM, advice from family members can be regarded as an external cue to action, which is also a very important factor in increasing various preventive behaviors [60]. Consistent with research by Kong et al [48], age was not a significant factor of preventive behaviors in our study (P=.054); meanwhile, other studies have found that compared with younger participants, individuals aged 60 years or older implemented fewer preventive behavior changes [19,61]. Further larger-scale studies are warranted to assess the association between age and behaviors to prevent COVID-19.

Additionally, community-level control measures were found to be significantly associated with the practice of preventive measures. As COVID-19 rapidly spread from a single city to the entire country, the governmental “minimum contact strategy” was implemented nationwide; also, close management and screening of communities and villages to curb COVID-19 were subsequently implemented nationally in China [46]. Although the contribution of community control measures cannot be quantified, our study supports that community-level control measures against COVID-19 are related to significant increases in older adults’ practice of recommended preventive behaviors. We further confirmed that respondents who could not leave the house showed higher compliance with suggested preventive measures than those who could move freely as usual. According to the socioecological model, factors of structural, interpersonal, and personal levels determine health-related behaviors [62]. At the personal level, it has been found that older Chinese people feel they have an ethical duty (“filial piety”) to protect others, which facilitates their adherence to quarantine. In Wenzel’s study [63], “filial piety” suggested the ‘‘right and humane’’ way to act toward one’s family and others in the community and guided Chinese older adults in their use of strategies against SARS [63]. As social beings, people’s behaviors are subject to the influence of social relationships [64]. Chinese older people not only worry for themselves but also have concern for their significant others regarding COVID-19, which motivates their adoption of precautionary measures. Meanwhile, environmental manipulation and policies are crucial structural-level factors to encourage individuals to practice desired preventive behaviors. According to the WHO, social distancing/self-isolation and lockdown are two important nationwide social measures during a public health crisis [6]. The Chinese government adopted these strategies to mitigate the risk of COVID-19. Therefore, personal perceived ethical duty combined with community actions have facilitated China’s response to COVID-19. Particularly, community and home quarantine provided Chinese older adults with ample time to communicate with and influence their significant others. The abovementioned findings highlight the importance of control measures at the community level, which should be considered when planning pandemic control strategies in the future.

In addition to background factors, this study found that knowledge and perceived beliefs toward COVID-19 were significant predictors of preventive behaviors. In accordance with previous findings, this study confirmed that people with adequate knowledge about COVID-19 were more likely to take preventive measures in response to the pandemic than people who lacked such knowledge [56,65]; this suggests that health education aimed at improving people’s COVID-19 knowledge plays an important role in promoting preventive behaviors toward COVID-19. With regard to perceived belief–related factors, our study confirmed the positive correlation between self-efficacy in preventing COVID-19 and the adoption of preventive measures found in previous research [32,54]; however, bidirectional effects may exist between these two variables, as it has been suggested that perceived self-efficacy can encourage an individual to adopt certain behaviors to achieve desired outcomes, while success in performing certain behaviors can further augment perceived self-efficacy [66]. Contrary to previous findings [22,30,33], our study found that belief in the efficacy of local health authorities did not significantly predict the target COVID-19 preventive behaviors. The belief in government may not successfully translate into positive behavioral changes at the individual level; additionally, the positive relationship between trust in the government and the practice of preventive behaviors can be mediated by levels of perceived self-efficacy [20]. A relatively strong linear correlation between confidence in the government and self-efficacy (r=0.335, P<.001) was also found in this study.

Some of the items we used to assess the participants’ perceived beliefs were derived from the HBM, which includes perceived severity, perceived susceptibility, and perceived benefits. Although most of our respondents perceived that they might progress to serious symptoms if they were infected with COVID-19, perceived severity was not found to be significantly related to COVID-19–related preventive behaviors in this study. Perceived severity has been suggested to have relatively low relevance for preventive health behaviors; however, it may play an important role when individuals have already been diagnosed with certain diseases [25]. Therefore, the fact that the participants had not yet been diagnosed with COVID-19 as well as their low perceived susceptibility could account for the insignificant association between perceived severity and preventive behaviors. Among our participants, perceived susceptibility was found to be a significant correlate of COVID-19 preventive health behaviors. Perceived susceptibility has been consistently found to be a salient determinant of participation in preventive measures during an epidemic among both the general population and older people [23,33,67,68]. Other studies on health screening [69,70] and exercise behavior [71] also found that perceived susceptibility plays an important role in participation in protective behaviors. Despite the significant positive association between perceived susceptibility and engagement in protective behaviors, only 15.0% (190/1263) of the older people in this study perceived that they were highly vulnerable to COVID-19. This finding indicates that older people should be further warned about their higher vulnerability to the disease; moreover, to ensure the effectiveness of warnings, messages issued by public health communicators should be comprehensible, concise, and convincing [22]. Meanwhile, perceived benefits were found to be more prominent in affecting individuals’ adoption of preventive behaviors in the aforementioned HBM constructs [26]. When people deem certain preventive health behaviors to be effective in preventing a disease, they are motivated to engage in these behaviors [22,58,72,73]. Indeed, communications that increase individuals’ perceptions of the benefits of particular health-related behaviors have been proved to be successful in reducing health threats [74-76]. Health authorities should be reminded to exert effort to provide sufficient health communication messages to enhance older people’s perceived susceptibility and benefits in the context of the COVID-19 pandemic, which can greatly motivate these individuals to engage in preventive measures.

Our study has some limitations. First, no standardized tool for assessing knowledge, perceived beliefs, or practice of preventive behaviors on COVID-19 has been previously validated. We designed the questionnaire based on the latest official report from the WHO, the Chinese CDC, and the scientific literature; however, the depth of the survey may be limited, as we were obligated to conduct our investigation quickly and recruit as large a sample as possible. Secondly, selection bias may exist. The web-based survey only included people who have access to the internet, and older people who were infected or had close contacts with confirmed or suspected COVID-19 patients may not have wanted to participate in this survey; thus, the generalizability of our findings may be limited. However, web-based surveys may be the most appropriate method of data collection during an epidemic, as this method can prevent transmission; moreover, fielding the survey offline was not feasible, as strict epidemic control measures were being exercised in most parts of the country. Thirdly, this study was cross-sectional, and no casual effect statements concerning the relationship between targeting variables and the performance of preventive behaviors could be made. Fourthly, recall and social desirability bias may exist, as only retrospective self-reports of the participants were collected.

Despite the above limitations, this is one of few studies to assess the knowledge, perceived beliefs, and preventive behaviors toward COVID-19 among the older adult population in China. Meanwhile, a large number of respondents from all provinces in mainland China were recruited, enabling us to obtain a wide range of participants with various demographic backgrounds. Additionally, the investigation was conducted in the stable epidemic period (February 19 to March 19, 2020), when community-wide COVID-19 prevention activities were launched by local health authorities; thus, we were able to assess how community-level control measures influenced personal preventive behaviors. Therefore, our results are of practical significance for the design and implementation of health programs for the older population in the prevention of COVID-19 and other emerging epidemics.

Generally, the Chinese older adults in our survey demonstrated good knowledge, optimistic perceived beliefs, and appropriate behaviors toward COVID-19 during the pandemic, which are important factors to limit the spread of the disease. Higher behavior scores were found among older adults who were married and cohabitating and who were restricted by community-level control measures, suggesting that health education programs should pay more attention to people who are single and can move around freely. In addition, our findings suggest that good knowledge and appropriate perceived beliefs are associated with high levels of engagement in behaviors to prevent COVID-19. Therefore, to promote preventive behaviors, continued health intervention programs are advised among older people to improve their knowledge in certain aspects, including transmission modes and vaccines against COVID-19; additionally, older people’s perception of their own vulnerability to COVID-19 and the effectiveness of COVID-19–related preventive behaviors should be further emphasized.