This study was a secondary analysis of two rounds of random telephone surveys investigating COVID-19 vaccination uptake among community-dwelling Chinese-speaking individuals aged 65 years or above in Hong Kong, China [23]. STROBE checklist for cross-sectional study was shown in Multimedia Appendix 1. The first round was conducted during the fifth wave of the COVID-19 outbreak between February 14 and April 13, 2022. During the first round, strict COVID-19 control measures were implemented [24], including (1) closure of all playgroups, kindergartens, and primary schools (between January 14 and April 19, 2022); (2) closure of fitness centers, swimming pools, and sports premises (between January 5 and April 21, 2022); and (3) prohibition of social gatherings involving more than two persons (between February 8 and April 21, 2022). The number of daily confirmed COVID-19 cases reached its peak on March 2, 2022 (n=56,827) and dropped to 1043 on April 13, 2022. The second round of telephone surveys was conducted between May 11 and July 11, 2022. The number of daily confirmed COVID-19 cases increased slowly from 294 on May 11, 2022, to 2769 on July 11, 2022. A summary of the COVID-19 situation and its control measures in Hong Kong during the study period is presented in Figure 1.

Inclusion criteria of the participants were: (1) community-dwelling Chinese-speaking individuals aged 65 years or above and (2) having a Hong Kong ID card. The exclusion criterion was not able to communicate effectively with the study interviewers. We used the same data collection methods in both rounds of surveys and reported these details previously [23]. First, we input all household telephone numbers listed in the most updated telephone directories (approximately 350,000) into an Excel file. We then randomly selected 4000 numbers by using the function “select random cells.” Experienced interviewers carried out telephone interviews between 6 PM and 10 PM on weekdays and between 2 PM and 9 PM on Saturdays to avoid undersampling of working individuals. We considered a household to be nonvalid (one without an eligible participant) if no one answered five calls made at different time slots. If there was more than one individual in the household who was 65 years or above, the interviewers invited the person whose last birthday was the closest to the survey date to join the study. This practice was adopted to avoid clustering effects (ie, older adults living in the same household sharing a similar PA level and determinants of PA). Interviewers screened prospective participants for eligibility; briefed them about the study; and made guarantees of anonymity, their right to quit at any time, and that refusal to participate would have no consequences. Verbal informed consent was obtained. The interviewers also signed a form pledging that the participants had been fully informed about the study. The telephone interview took approximately 20 minutes to complete. No incentives were given to the participants.

Ethics approval was obtained from the Survey and Behavioral Research Ethics Committee of the Chinese University of Hong Kong (SBRE-19-187).

The target sample size for each round of the survey was 400. The sample size planning for the original study was explained in a previous publication [23]. In brief, assuming the proportion of participants having a low PA level of 30%-70% in the first round, this sample size was determined to be sufficient to detect a minimum between-round difference of 8.6% (power=0.80, α=.05; calculated with PASS 11.0 software).

There were no missing values in either round of survey. The differences in background characteristics between participants in the first and second rounds were assessed using χ² tests. After controlling for background variables with significant differences between rounds, the differences in the level of PA, MET-minutes/week, minutes of MVPA per week, and barriers to performing PA were compared using ordinal, logistic, or linear regression models. The level of PA, MET-minutes/week, and minutes of MVPA per week between rounds were further compared in subgroups of participants with and without a specific barrier to performing PA (ie, lack of physical capacity to perform PA or perception that PA would cause pain and discomfort). The subsequent analysis was performed among participants in the same round of the survey. Using a low level of PA as the dependent variable and background characteristics as independent variables, crude odds ratios were obtained using logistic regression models. The associations between barriers to performing PA and the dependent variable were then obtained by fitting a single logistic regression model involving one of the independent variables and all significant background characteristics. Adjusted odds ratios (AORs) and respective 95% CIs were obtained. SPSS version 26.0 (IBM Corp, Armonk, NY, USA) was used for data analysis, with P<.05 considered as statistically significant.

We called 3900 and 3840 households in the first and second rounds, 640 and 625 households had an eligible older adult, 245 and 255 refused to participate in the study, and 395 and 370 completed the telephone survey, respectively. The response rate was 62% and 59% in the first and second round, respectively. The characteristics of the participants are presented in Table 1. Approximately half of the participants were aged 70 years or above (first round, 50.1%; second round, 50.8%) and female. Most of the participants were married or cohabited with a partner, did not receive tertiary education (first round, 89.9%; second round, 89.2%), without full-time/part-time employment, and with a monthly household income lower than HK $20,000 (US $2580). Approximately one-fifth of the participants had at least one chronic condition (first round, 19.2%; second round, 18.9%). There was no difference in these characteristics between participants in the first and second rounds (P values ranged from .68 to .98). As compared to participants in the first round, more participants in the second round reported a history of COVID-19, completed the primary COVID-19 vaccination series (92.1% vs 78.7%, P<.001), and had received COVID-19 vaccine booster doses (58.9% vs 31.6%, P<.001).

The changes in PA are presented in Table 2. As compared to participants in the first round, fewer participants had a low level of PA in the second round. Participants in the second round had higher MET-minutes/week and minutes of MVPA per week than those in the first round. Subgroup analysis showed that older adults with some specific barriers to performing PA (ie, lack of physical capacity or perceived that PA would cause pain and discomfort) had a significantly lower PA level, MET-minutes/week, and minutes of MVPA per week compared to those of participants without such barriers (see Multimedia Appendix 3).

The changes in barriers to performing PA are also presented in Table 2. As compared to participants in the first round, fewer participants in the second round reported lack of space and a facility to do PA at home or being concerned about COVID-19 infection when doing PA. Fewer participants had experienced the closure of facilities and refusal by peers to do PA with them due to COVID-19 compared to those in the first round.

In univariate analysis, participants who completed the primary COVID-19 vaccination series and/or booster dose were less likely to have a low PA level in both rounds (Table 3). After adjustment for these significant characteristics, participants who perceived a lack of physical capacity to do PA and believed that PA would cause pain and discomfort were more likely to have low level of PA in both rounds. Do not have time and concern about COVID-19 infection during PA were associated with a low level of PA among participants in the first round, but not in the second round. Perceived lack of space and facility to do PA at home and cannot find people to do PA together were associated with a low PA level in the second round, but not in the first round (Table 4).

This was one of the first studies to track changes in the levels and barriers to performing PA among older adults before and after entering the “postpandemic era.” One of the strengths of this study is that it was based on a random and population-based sample. In response to the fifth wave of the COVID-19 outbreak, the Hong Kong government closed all sports centers and venues, and advised older adults to stay at home and limit their outdoor activities due to their vulnerability to COVID-19 [24]. Therefore, it was not surprising to observe that a sizeable proportion of older adults had a low PA level (over 40%), which was much higher than that reported at the time before COVID-19 (about 20%) [6,7]. Despite an increasing trend in daily confirmed cases, we observed a significant increase in PA level among older adults in the second round of the survey. However, the proportion of older adults having a low PA level (27.8%) was still higher than that reported at the time before COVID-19 (20%) [6,7]. Therefore, efforts are needed to facilitate older adults to resume and increase PA in the “postpandemic era.”

As compared to those in the first round, participants in the second round reported a much higher completion rate of the primary COVID-19 vaccine series (94% vs 78.7%) and the booster dose (58.9% vs 31.6%). Completing the primary COVID-19 series and/or the booster dose was associated with a higher PA level in both rounds. An increase in COVID-19 vaccination coverage might contribute to the increasing PA level in the second round. Vaccinated older adults might feel they are protected against COVID-19 and hence have fewer concerns to resume outdoor activities (eg, PA). Promoting COVID-19 vaccination and the booster dose might be useful strategies to improve PA among older adults in the future.

Our findings suggested some different barriers to performing PA applied to older adults in Hong Kong before and after the fifth wave of the outbreak. First, a higher proportion of older adults agreed that they did not have time for PA in the first round compared to those in the second round. Those who lacked time for PA were more likely to report a low PA level in the first round, but not in the second round. In Hong Kong, grandparents are the main caregivers of children [28,29]. A qualitative study suggested that the increased caring responsibility due to school closure and lack of childcare during the pandemic was a barrier to performing PA among older adults in the United Kingdom [21]. During the fifth wave of the COVID-19 outbreak, the Hong Kong government closed all playgroups, kindergartens, and primary schools between January 14 and April 21, 2022. During the same period, the Hong Kong government directed civil servants to work from home. However, most private organizations did not follow the work-from-home policy. Since civil servants only account for less than 2% of the entire working population in Hong Kong, most parents had to work as usual and could not take care of their children during school closure. As a result, the childcare responsibility suddenly increased among many grandparents, which reduced their availability to perform PA. Second, there was a significant decrease in concerns about COVID-19 infection over time. Such concern was associated with a lower PA level during the pandemic, but not in the “postpandemic era.” A decrease in daily confirmed cases and an increase in COVID-19 vaccination coverage might have reduced their concerns about COVID-19 infection. Third, lack of space and facilities for home fitness was a significant barrier to performing PA in the “postpandemic era,” but not in the first round of the survey. One possible explanation was that lack of time to perform PA caused by a sudden increase in childcare responsibility during school closure was an overwhelming barrier to performing PA during the outbreak. Therefore, although more participants indicated lacking space and a facility for home fitness in the first round of the survey, they might not consider it as a significant barrier. Health promotion in the “postpandemic era” should encourage older adults to perform home-based PA without extensive requirement of space or facilities, such as stationary aerobic exercise or weight/strength training. Although the Hong Kong government used a series of exercise videos on television to promote PA at home for older adults living in public housing estates during the pandemic [30], the efficiency of such programs in promoting PA for older adults should be further explored. In addition, facilitating older adults to form peer support groups is a useful strategy to promote PA in the “postpandemic era.” Forming such peer support groups was less feasible during the pandemic when group gatherings were prohibited [24]. Since the government already lifted the restriction on group social gatherings, health promotion in the “postpandemic era” should consider using such a strategy [31].

Some similar barriers have hindered older adults to perform PA both before and after the fifth wave of the COVID-19 outbreak. In this study, concern about PA capacity and that PA would cause pain and discomfort were associated with a lower PA level. These concerns were also barriers to performing PA in this group before the time of COVID-19 [7,20]. Sport scientists should introduce feasible options of PA suitable for older adults. Testimonials of older adults on how they overcome these physical and environmental barriers and stay active might also be useful. Studies suggested that communication among Chinese older adults is very effective due to the high level of rapport among people of similar age [32]. Chinese older adults prefer to seek information and opinion from their peers, which are considered to be more credible than other information sources [32].

In contrast to our hypothesis, the closure of facilities due to COVID-19 and its control measures were not significantly associated with the PA level in the first or the second round. A previous study suggested that older adults in Hong Kong changed their mode of PA to cope with the COVID-19 control measures. For example, older adults reduced swimming or going to the gym for exercise, but increased stretching, brisk walking, or other activities that rely less on access to sports facilities [10]. Such adaptations might have mitigated the impact caused by the closure of sports facilities.

This study had some limitations. First, it was a major limitation that we did not assess the functional limitations of the participants. In Hong Kong, 16.3% of community-dwelling older adults had a functional limitation [33]. Older adults with functional limitation had lower PA levels. Second, we did not gather more information on family members of the participants due to the limited length of the questionnaire. Family members would have an influence on participants’ PA level (eg, whether family members can share the childcare responsibility during school closure). Third, PA was self-reported. We did not use accelerometers or wearable devices to assess PA due to feasibility and resource constraints. This raised concerns about reliability and recall bias. Fourth, due to the limited length of the questionnaire, some determinants of PA among older adults were not covered by the study (eg, perceived benefits and self-efficacy of PA). Fifth, we did not measure participants’ physical capacity to perform PA. Sixth, compared to census data, people who were 75 years or above were undersampled in this study [34]. However, the distributions of gender and age were similar to those of recent random telephone surveys among community-dwelling older adults [35]. Furthermore, nonresponses would cause selection bias. We were not able to obtain characteristics of those who refused to participate, which might be different from those of the participants. Our response rate was comparable to those of previous random telephone surveys targeting older adults in Hong Kong [32,35,36]. Last but not least, this was a cross-sectional study and could not establish causal relationships.

The level of PA increased significantly after the drop of daily confirmed COVID-19 cases and the relaxation of COVID-19 control measures among older adults in Hong Kong. Different factors influenced older adults’ PA level during and after the fifth wave of the COVID-19 outbreak. Barriers to performing PA, such as perceived lack of space and a facility to perform PA at home, concerns about COVID-19 infection during PA, closure of facilities, and refusals made by peers to perform PA reduced significantly over time. Regular monitoring of the PA level and its associated factors should be conducted to guide health promotion and policy-making. Sport scientists should introduce suitable options for older adults with inadequate physical capacity or having functional limitations. In the “postpandemic era,” reactivating peer support groups and promoting home-based PA may increase the PA level among older adults. Health authorities should be aware of the negative impact of school closure on PA among older adults. If school closure has to be implemented in future waves of a COVID-19 outbreak, introducing PA options suitable for older adults to perform with their grandchildren may be helpful to alleviate its negative influence.