A survey was conducted among older adults aged 60 years and above from 5 June 2022 to 20 June 2022, 2 months after the mass COVID-19 outbreak in Shanghai, China. As an economic center in China, Shanghai serves as an important transportation hub to connect to other cities and regions in the country. As of April 2022, Shanghai had adopted strict control strategies to reduce transmission and to provide early diagnosis. Schools and businesses were closed, and public transport was shut down. Most residents were not allowed to leave their community but could only go outside for an essential reason (e.g., medical emergencies). Inclusion criteria to recruit participants were (1) people aged 60 years and above living in China; (2) people without severe hearing impairment; (3) individuals with a certain cognitive ability to correctly understand the questionnaire and who could independently complete the survey; and (4) those who voluntarily agreed to participate in the present study. The study protocol was approved by the Ethics Committee of Xinhua Hospital Affiliated with Shanghai Jiao Tong University School of Medicine (Approval number XHEC-D-2022-242).

A self-designed questionnaire containing 30 items was used in this study (Appendix A). The initial draft questionnaire was developed by reviewing relevant literature (24, 25). It was then modified and finalized after consultation with experts and pre-testing the instrument. The process of the pre-testing is described later in the manuscript. The final questionnaire collected information on COVID-19 vaccination status, demographic factors, internal risk perception of the pandemic, cognition and attitude toward the COVID-19 vaccine, individuals' health status, experience with COVID-19 infections, and experience with other vaccinations. Due to the differences in questionnaire item types and scoring methods for risk perception, vaccine knowledge, and vaccine attitude, scores were calculated separately for each domain before conducting a reliability analysis. Cronbach's alpha for the combined domains was 0.49.

On COVID-19 vaccination status, participants were asked whether they were vaccinated against COVID-19 or not. Participants who had received the COVID-19 vaccine were categorized into the vaccinated group, and those who had not received the vaccine were classified as the unvaccinated group. Those who were vaccinated were further asked about the number of doses they had received, the manufacturer of the vaccines, the reasons why they wanted to be vaccinated, and whether they had any adverse reactions after vaccination. Those who had not been vaccinated were asked why they did not want to be vaccinated.

On demographic characteristics, the following information was collected: gender, age, marital status (married or not married [including unmarried, divorced, and widowed]), education level (from primary school or below to university or above), pre-retirement occupation, monthly income, place of residence (Shanghai vs. other places), region (downtown vs. countryside), and living situation (solitude vs. live with others).

To assess the perception of the risk of COVID-19, which measured individuals' perceived risk of being infected with SARS-CoV-2, considering the infectiousness of the pathogen, protection measures, and demographic status, we developed three questions: (1) “Are you concerned about the COVID-19 pandemic?” (2) “Are you confident about the prevention and control of COVID-19 epidemic?” and (3) “Do you think that older adults are more likely to be infected with SARS-CoV-2 than young ones?” We assigned a score to the answer to each of the questions, with a higher score indicating a greater perception of risk (see Appendix B).

The cognition toward COVID-19 vaccines, which measured individuals' knowledge of COVID-19, was assessed using three questions concerning vaccination procedure, age limitation for vaccination, and types of COVID-19 vaccines available in China, respectively. For each item, if participants picked the correct answer, they received a score of 1 for that question, and they received a score of 0 if they selected a wrong answer or the response of “have no idea”. The sum of the score for all three items was calculated as the total cognition score toward COVID-19 vaccines, with the total score ranging from 0 to 3. The higher score indicated that the participants had better knowledge of COVID-19 vaccines. The attitude toward COVID-19 vaccines, which measured the extent to which individuals' trust in the efficacy and safety of COVID-19 vaccines, was assessed using two questions about the efficacy and safety of COVID-19 vaccines: (1) “Do you think vaccination is an important means to prevent and control SARS-CoV-2 infection?” and (2) “Do you think COVID-19 vaccines are safe?” Each answer was scored, with a higher score indicating a more positive attitude toward the efficacy and safety of COVID-19 vaccines (see Appendix B).

The questionnaire also covered information about the duration of lockdown (from <1 week to more than 2 months), history of coronavirus infection (oneself or if relatives got infected vs. never infected), history of other vaccinations (yes vs. no), and chronic disease situation (from no chronic disease to three or more chronic diseases) of participants. How each variable was coded is displayed in Appendix B.

The questionnaire was piloted in a group comprising 10 family members of healthcare workers who were aged between 60 and 90. During the pilot, we obtained feedback from participants to identify any problematic items. After deliberation with the study investigators and participants involved in the pilot, we revised and finalized the survey items. The instrument was administered in Chinese. On average, it took participants about 15 min to complete the questionnaire.

Considering that most older adults stayed at home due to the epidemic, 90% of the questionnaires were distributed online. The link and the QR code of the online survey were sent out through a WeChat “Friends circle,” a function that can be used to share personal photos or public website links. This questionnaire link and QR code could then be forwarded or shared by participants with friends in their WeChat contact list, whom they considered appropriate for this survey; their friends were also encouraged to send the link to their friends' networks. Such a snowball sampling process continued until a sufficient sample size was reached.

It is likely that many older adults, especially those who were very old, were not good at using the Internet. To complement the data collected through online questionnaires, we also collected data through a face-to-face survey in a geriatric ward of Xinhua Hospital in Shanghai. Approximately 10% of the sample included in this study was obtained through this approach from older adults above 75 years old, with the help of a trained investigator. This, to some extent, would help mitigate the bias if all the results were from the online survey.

This is an exploratory study, and we used a cross-section design to estimate the sample size. As of 5 May 2022, 86.23% elderly population over 60 years old had been vaccinated (26). Assuming the allowable estimation error is 3%, we estimated the minimum sample size to be 507, according to the formula =(Zα2)2π(1-π)E2, where n is the sample size, Zα2 is determined by the confidence interval, π is the overall proportion, and E is the allowable estimation error.

Wen Juan Xing, an online platform similar to Amazon Mechanical Turk, Qualtrics, SurveyMonkey, or CloudResearch, was used to administer the survey results. Each participant's responses were automatically entered into the Wen Juan Xing database through a mobile phone. A dedicated link and a QR code attached to the online survey were created for the purposes of this study, and participation in the survey was voluntary. Informed consent was deemed to have been given if participants clicked on the link or scanned the QR code. There was no monetary or non-monetary incentive for completing the survey. Due to limitations of the survey platform, participants were not able to review or make changes after they submitted their responses. To prevent multiple entries from being submitted from the same individual, cookies were used to assign a unique identifier to each entry.

Only completed questionnaires were analyzed in this study. Participants who terminated the survey without completing the form were deemed to have withdrawn their consent. Such data were not captured or analyzed.

A descriptive analysis was conducted for all the variables included in the analysis. The continuous variables (e.g., internal risk perception) were presented with mean and standard deviation, while categorical variables (e.g., gender) were presented with frequency. For the bivariate analysis, we compared the internal risk perception, cognition, and attitude between those who received COVID-19 vaccines and those who did not, using a t-test or analysis of variance (ANOVA). We also performed a path analysis to examine the association between vaccination and internal risk perception, cognition, and attitude toward COVID-19 vaccines using the AMOS software. Univariate and multivariate logistic regression models were used to examine the factors associated with COVID-19 vaccination after converting categorical variables into dummy variables. In the logistic regression models, whether the individual received COVID-19 vaccines served as the dependent variable (1 is assigned to those who were vaccinated and 0 is assigned to those who were unvaccinated). The independent variables included demographic information, internal risk perception, cognition and attitude toward COVID-19 vaccines, and individuals' health status and prior experience. Odds ratios (OR) with 95% confidence intervals (CI) were used to estimate the associations. For all the analyses, a P < 0.05 was considered to be statistically significant. SPSS 21.0 software was used for conducting statistical analyses.

The summary of participants' demographic characteristics, health status, and experience is provided in Table 1. Out of 725 participants, there were 319 male patients (44.0%) and 406 female patients (56.0%), where 465 patients (64.1%) were aged 60–70 years, and 63 patients (8.7%) were aged 80 years and above. The respondents mainly lived in Shanghai (82.1%), downtown (85.7%), had not left the community for more than 2 months (71.0%), with a high school education level or above (53.4%), were married (86.5%), lived with others (89.9%), and had an average monthly income of CNY?3,000–9,999 (77.8%). The most common occupations before retirement were workers/peasants (28.8%) and administrative staff (28.4%). A total of 199 (27.5%) of them had a history of other vaccinations. A total of 181 (25.0%) of the respondents had reported that they or their relatives had been infected with SARS-CoV-2. A total of 534 (73.7%) of the participants had chronic diseases, and 95 (13.1%) of them had three or more chronic diseases. The most common chronic diseases were hypertension, diabetes, and coronary heart disease.

Out of 725 individuals, 568 were vaccinated against COVID-19, with a vaccination rate of 78.3%. A total of 385 (53.1%) received booster injections, 163 (22.5%) received two doses, and 20 (2.8%) received only one dose (see Figure 1). The main vaccines were the Sinovac COVID-19 vaccine (65.5%) and the Sinopharm COVID-19 vaccine (BIBP) (23.2%). The top three self-reported reasons to get vaccinated (multiple selections) were “fear of COVID-19 infection (62.3%)”, “compliance with government recommendations (52.6%)”, and “contributing to herd immunity (41.2%)”. Self-reported reasons for unwillingness to get vaccinated (multiple selections) were “concerns about acute exacerbation of chronic diseases after vaccination (57.3%)”, “concerns about the vaccine's side effects (41.4%)”, “ineligibility for vaccination due to existing medical conditions (11.5%)”, and “perception of vaccine inefficacy against COVID-19 variants (11.5%)” (see Figure 2).

The average score of internal risk perception was 2.22 ± 0.31 in the vaccinated group and 2.14 ± 0.36 in the unvaccinated group. There was a statistically significant difference between the two groups (t = 2.64, P < 0.05). The cognition score of COVID-19 vaccines in the vaccinated group was 1.55 ± 0.67 compared to 1.13 ± 0.83 in the unvaccinated group. The difference was statistically significant between the two groups (t = 5.84, P < 0.05). The average score of attitudes toward COVID-19 vaccines was 1.81 ± 0.29 in the vaccinated group as compared to 1.56 ± 0.38 in the unvaccinated group. The difference was statistically significant between the two groups (t = 7.92, P < 0.05) (see Table 2).

Path analysis showed that the model fitted well (X²/df = 3.55, NFI = 0.86, IFI = 0.90, CFI = 0.89, and RMSEA = 0.06). From the total effects of various influencing factors on the vaccination behavior of older adults, we found that (1) the cognitive effect on vaccination behavior was relatively large, followed by the internal risk perception, and then the attitude toward COVID-19 vaccines; (2) cognition had a positive effect on vaccination behavior (with a total effect of 0.44, direct effect of 0.34, and indirect effect of 0.11). The higher the cognition of COVID-19 vaccines was, the more likely older adults are to receive the COVID-19 vaccines. Moreover, cognition successively affects vaccination behavior through internal risk perception with a direct effect of 0.49 and attitude with an indirect effect of 0.51; and (3) intrinsic risk perception had a positive effect on vaccination behavior with an indirect effect of 0.23. The higher the intrinsic risk perception of COVID-19, the more likely older adults received the vaccination, and this effect was generated through attitude with a direct effect of 1.05. (4) The attitude toward COVID-19 vaccines had a positive effect on vaccination behavior with a direct effect of 0.21. The more positive the attitude toward COVID-19 vaccines was, the more likely older adults were vaccinated (see Figure 3).

The bivariate logistic analysis shows that (Table 3, unadjusted odds ratio) age (OR = 0.48, P < 0.001), gender (OR = 0.68, P < 0.05), marital status (OR = 2.56, P < 0.001), monthly income (OR = 0.72, P < 0.01), place of residence (OR = 0.42, P < 0.01), living situation (OR = 0.52, P < 0.05), history of other vaccines (OR = 2.47, P < 0.001), number of chronic diseases (OR = 0.44, P < 0.001), internal risk perception (OR = 2.18, P < 0.01), cognition toward COVID-19 (OR = 2.18, P < 0.001), and attitude toward COVID-19 (OR = 9.01, P < 0.001) were associated with vaccination status in older adults (for more details, please see Appendix C).

The multivariable logistic analysis shows that age, measured as the continuous variable, was identified as a significant impact factor, and older respondents were less likely to accept COVID-19 vaccines (OR = 0.53, 95% CI 0.43–0.66, P < 0.001). A high level of cognition and a more positive attitude toward COVID-19 vaccines were important factors associated with the COVID-19 vaccination. Older adults with a high level of cognition toward COVID-19 vaccines were more willing to receive the vaccine than those who had a lower level of cognition (OR = 1.60, 95% CI 1.17–2.19, P < 0.01). Older adults with a more positive attitude toward COVID-19 vaccines were more willing to receive the vaccine than those with a positive attitude (OR = 9.22, 95% CI 4.69–18.09, P < 0.001). The respondents who lived in Shanghai were less likely to be vaccinated than those who lived in other cities (OR = 0.40, 95% CI 0.17–0.92, P < 0.05). Older adults with a lockdown period between 1 week and 1 month were less likely to be vaccinated than those who had a shorter lockdown period of <1 week (OR = 0.33, 95% CI 0.13–0.83, P < 0.05). Older adults who had a history of other vaccines were more likely to be vaccinated than those who had never obtained other vaccines (OR = 2.58, 95% CI 1.45–4.60, P < 0.01). Older adults with more numbers of chronic diseases were less likely to be vaccinated than those with fewer numbers of chronic diseases (OR = 0.49, 95% CI 0.38–0.62, P < 0.001) (see Table 3).

Several limitations need to be acknowledged. First, our study has a small sample size, and the data were primarily collected through an online survey platform. Therefore, the sample may not represent the geographic, cultural, and socioeconomic variations among Chinese older adults. Second, this is a cross-sectional study, which means that we cannot determine the causality of COVID-19 vaccinations and associated factors. Third, the online questionnaire is self-reported and may have a certain degree of information deviation. However, the online anonymous questionnaire is more helpful in obtaining information on sensitive topics than the face-to-face survey. Fourth, the reliability of the questionnaire remains low. One of the possible reasons for this low reliability is that we introduced different types of requestions and different numbers of options for questions in the questionnaire in the hope that this would allow the questionnaire to be easier for old adults to understand and facilitate the completion of the questionnaire. In future questionnaire design, we will seek to strike a balance between the cognitive abilities of older adults and the appropriate question types and number of options. Despite these limitations, this study provides important information on factors associated with COVID vaccination among older adults, whose protection is key to reducing COVID-19-related deaths.