In reference to the study by Atanasova et al [22], this study combined 2 models to explore the key factors and mechanisms influencing the maintenance of public health behaviors after the COVID-19 pandemic. Based on the previous context, these 2 models are the MTM and the PADM, both widely used to explain and predict behaviors, and in our case, we target public maintenance of health behaviors. The MTM posits that 3 factors—emotional transformation, practical changes, and social environmental changes—influence long-term behavioral maintenance [18]. Meanwhile, the PADM emphasizes that environmental cues, social cues, warning messages, and information sources trigger core perceptions, including risk perception and efficacy perception, driving protective behavioral responses [20]. In the context of health behavior maintenance after the pandemic, the MTM helps predict the continuation of these behaviors, while the PADM aids in understanding how people adapt their decisions in response to evolving risks and available resources. Combining these 2 models offers a comprehensive analysis of the factors influencing behavior maintenance from cognitive, motivational, and environmental perspectives as pandemic risks decline (Figure 1).

A cross-sectional survey was conducted. From October to November 2023, we recruited participants from the general population in China using convenience sampling. The survey was conducted anonymously via the Sojump platform, a widely used online survey tool in China with a national user base. The eligibility criteria for participants were (1) voluntary completion of the questionnaire and informed consent to participate and (2) proficiency in using electronic devices.

The sample size was calculated using a formula guided by the Cochran technique [23], which is effective for large-population studies [24]. The calculation variables were n (sample size), z=1.96 (for a 95% confidence level), p=50% (the probability of selecting a choice, suitable for highly dispersed population), q=(1–p), and d=3.272% (allowed error percentage). According to the calculation, a sample size of 898 was determined.

We distributed the survey links and recruitment advertisements on social media via our research team members, using the Sojump to provide survey links and widely invite eligible participants who met the inclusion criteria. To ensure data quality, all questionnaire items were set as mandatory, and each IP address, device, and account was restricted to one submission only. Also, a time limit was imposed to exclude respondents finishing the survey in less than 2 minutes. To encourage participation, those who successfully completed the survey were rewarded with 3‐5 RMB (US$ 0.42-0.70). Ultimately, 1216 valid data were obtained.

The data analysis was performed using R Studio (R 4.2.3; R Core Team). The CFA was used to test the validity of the measurements, with standardized factor loadings and average variance extracted as indicators of convergent validity. The reliability of the measurements was examined using Cronbach’s α and composite reliability. Results of validity and reliability of measurements are shown in the Multimedia Appendix 1. Categorical variables were presented as frequency and proportion, and continuous variables were expressed as mean and SD. In the univariate analysis, normally distributed data were analyzed using t tests or ANOVA, and non-normally distributed data were analyzed using Wilcoxon rank-sum tests or rank correlation tests. Spearman correlation analysis was conducted to explore relationships between variables. Multivariate regression analysis was applied to examine impacts of demographic characteristics and key factors on the maintenance level of health behaviors. Structural equation modeling (SEM) was used to analyze pathways of factors influencing the public maintenance of health behaviors. A 2-sided P value below .05 was considered statistically significant.

A total of 726 females and 490 males participated in this survey, with an average age of 27.38 (SD 8.52) years. Among the participants, 12.83% (156/1216) reported having a chronic disease, and most of them had been infected with COVID-19 at least once (1054/1216, 86.68%). Over half rated their health as good (746/1216, 61.35%) and 73.27% (891/1216) indicated an increased concern for their health after the COVID-19 pandemic. Details are presented in Table 1.

For all participants, the overall maintenance score of health behaviors was 2.88 (SD 0.45) (Table 2). Factors such as age, education, career, average monthly income, marital status, living condition, whether one has a chronic illness, self-assessed health status, and the level of attention to personal health after the COVID-19 pandemic showed statistically significant associations with the public maintenance of health behaviors. Specifically, the factors of settlement type and time of COVID-19 infection showed no statistical differences with the public maintenance of health behaviors (Table 1). The results of the correlation analysis between variables reveal that all correlation coefficients were statistically significant, and the specific values of the correlation coefficients are shown in Table 3.

The demographic factors found to be statistically significant in the univariate analysis, along with core study variables, were sequentially incorporated into the multivariate regression analysis using a stepwise forward method. Specifically, the inclusion order prioritized (1) statistically significant demographic factors identified in univariate analysis, (2) variables from the MTM, and (3) variables from the PADM. By the way, 2-tailed t test for a regression coefficient was conducted and the degrees of freedom was 1175. The results indicated that average monthly income, marital status, increased health concern after the COVID-19, emotional transformation, practical changes, information sources, warning messages, and efficacy perception influenced the public maintenance of health behaviors, with an adjusted R² of 0.455 (Table 4).

In the SEM, the path coefficients could reflect the strength and direction of causality between variables. Compared with regression coefficients, path coefficients capture the net effect of one variable on another after controlling variables [32]. Standardized path coefficients convert these into a unitless value for direct comparison [33]. All standardized path coefficients in this study were within the normal range, as shown in Table 5. The modified model shows good overall fit to the actual sample data, with χ²/df of 4.92 (χ²₃₅₃=1737.201) and the goodness-of-fit index, comparative fit index, standardized root mean square residual, and root mean square error of approximation of 0.894, 0.923, 0.055, and 0.057, respectively. All indicators meet the requiremexcept for thet for goodness-of-fit index, which is slightly below 0.90.

The analysis shows that social cues have a positive influence on efficacy perception (β=.782, P<.001), which in turn has the most significant effect on the maintenance of public health behaviors (β=.386, P<.001). Emotional transformation and practical changes from the MTM also positively affect public health behavior maintenance (both β=.213, P<.001). Risk perception is found to have a negative impact on public health behaviors (β=−.099, P=.013). Social cues are the most influential factor for risk perception with a positive effect (β=.789, P<.001), while warning messages and information sources are negatively associated with risk perception (both β=−.064, P=.004).

This study found that the public maintenance of health behaviors was at a fairly low level (mean 2.88, SD 0.45) after the COVID-19 pandemic. Similar to the study of Wang and Gago [34], children exhibited health behavior maintenance issues after the pandemic, such as poor sleep, increased screen time, and changed daily activities. When examining the dimensions of health behavior maintenance, the public surveyed in this study maintained relatively stable social and psychological balance, daily life behaviors, and organized physical exercise. However, the maintenance of dietary behaviors showed significant fluctuation. This echoes the study by van Reekum et al [35], which highlighted the difficulty in sustaining dietary patterns formed during the pandemic once normal life resumed. Moreover, as the pandemic progressed, pandemic fatigue set in, leading to a decline in health behaviors due to the sustained high level of vigilance required [36]. Even a short-term decrease in health behavior maintenance may adversely affect health. Therefore, it is crucial to focus on public health behavior maintenance and its influencing factors after the pandemic and to implement effective and timely interventions.

In this study, factors such as average monthly income, marital status, and personal health concern after the COVID-19 pandemic were found to be associated with the public maintenance of health behaviors. Wambiya et al [37] similarly found that low-income groups, burdened by economic hardships and work demands, may neglect the role of health behaviors in promoting physical health. Meanwhile, spousal encouragement, support, and supervision can aid in maintaining health behaviors [38]. In addition, individuals with higher health concerns tend to have better health cognition, enabling them to practice health behaviors more effectively [39]. Notably, this study found no correlation between settlement type or whether individuals have been infected with COVID-19 and public health behavior maintenance. This is inconsistent with the study of Wollast et al [40]. The discrepancy may arise because most participants surveyed in this study had been infected with COVID-19 (86.68%). With frequent social interactions and population movements between urban and rural areas, resulting in widespread COVID-19 infections [41], which led to no significant differences in population distribution.

This study confirms that emotional transformation and practical changes in the MTM are positive predictors of health behavior maintenance, aligning with prior study [42]. Emotional transformation emphasizes redirecting negative emotions associated with health behaviors, such as reluctance and resistance, toward the goal of maintaining health behaviors [43]. Practical changes focus on identifying and correcting barriers and incorrect practices in implementing health behaviors through recording or reflection [44]. Throughout these 2 processes, individuals typically combine intrinsic self-motivation with extrinsic emotional support, seeking solutions that use both internal and external resources to overcome psychological barriers in maintaining health behaviors.

This study reveals that warning messages and information sources in the PADM negatively impact risk perception, which in turn undermines the public maintenance of health behaviors. This contradicts Cao and Liu’s [45] findings in their study on internet users. Generally, access to public health information is positively correlated with health behaviors [46]. However, in this study, the public might have encountered an abundance of conflicting and dubious health information, resulting in confusion, information overload, and distrust [47]. This aligns with Pedro et al’s [48] study, which could diminish sensitivity to health risks. Some studies point out that risk perception is a positive predictor of health behaviors [49,50], but excessive risk perception may trigger impatience or rebelliousness, decreasing individuals’ willingness and frequency of action [51]. Therefore, the bidirectional effects of information reception and risk perception on behavior warrant further exploration.

This study indicates a positive correlation between efficacy perception and the public maintenance of health behaviors, consistent with existing studies [52,53]. Individuals with stronger beliefs in their ability to maintain health behaviors exhibit higher self-control and willingness, highlighting the beneficial role of confidence and motivation in adhering to health practices [54]. Moreover, a greater perception of health behavior effects strengthens their belief in maintaining these behaviors [55]. However, this study found that information sources and warning messages do not influence health behavior maintenance through efficacy perception, conflicting with the hypothesis in the PADM [20]. This might be due to asymmetric information transmitted through these channels, leading to varying efficacy perceptions. When individuals perceive insufficient benefits, they tend to adopt a conservative behavior pattern and are less likely to persist in health behaviors [56].

This study found that social cues positively influence risk perception and efficacy perception, in turn affecting public health behavior maintenance, consistent with the PADM [20]. Individuals with greater exposure to social cues show heightened awareness of potential behavioral risks [57] and acquire more health-related information. This increased awareness and information boost their confidence in maintaining health behaviors and motivate them to sustain these behaviors. However, unlike previous studies [58,59], no statistical link was found between environmental cues and health behavior maintenance through risk and efficacy perception, nor any impact of social environmental changes on health behavior maintenance. This might be because most participants had COVID-19–related experiences during the study period. Due to frequent social interactions and population movements, people’s sensitivity to environmental cues and social environmental changes decreased compared to the pandemic period. Similar to a “gray rhino” event, individuals tend to overlook health risk warnings when they are accustomed to a situation [60].

The MTM and the PADM are both useful for studying behavior change mechanisms. The MTM focuses on individual psychology and emotional shifts, while the PADM emphasizes external environmental information. Despite their different focal points, they have not been integrated into previous studies. This study combines their strengths to analyze factors influencing public health behavior maintenance. It takes into account the public real-time health risk-benefit perception and behavioral responses after the pandemic, offering a new perspective on health behavior study and providing references for promoting public health behavior maintenance.

This study offers initial insights into the health behavior maintenance and its influencing factors among the Chinese public, but it has limitations that need further exploration. The results of this study rely on self-reporting by the participants, which may be prone to recall bias. Future research could use smart wearable devices to collect objective data on health behavior maintenance, such as daily step counts and sleep duration. In addition, the online survey with convenience sampling leans toward younger participants who are more tech-savvy, while they can provide unique perspectives on this study. To enhance data representativeness, future studies can combine online and offline recruitment channels, with regional stratified sampling to recruit more diverse participants, and embrace a longitudinal design to better establish causal relationships between factors.

Drawing on this study, we propose the following suggestions to enhance public health behavior maintenance. First, government, health departments, and medical institutions should be actively engaged as authoritative sources to promote public health behaviors, leveraging their authority and credibility. Second, integrate traditional, online, and social media to broaden information-sharing channels, innovate dissemination methods, and embed health concepts into public daily life. While propagandizing the benefits of health behaviors, it is also necessary to educate the public about the dangers of unhealthy behaviors and corresponding health warning messages. It is crucial to avoid false information and exaggerated information. Finally, enhance family and social support to help individuals manage negative emotions related to health behavior maintenance and encouraging self-motivation for effective health behavior management.

This study, integrating the MTM and the PADM, explores the status and influencing mechanisms of public health behavior maintenance after the COVID-19 pandemic. Our findings indicate that enhancing emotional transformation, practical changes, social cues, and warning messages can boost health behavior maintenance. However, attention should be given to pandemic-related fatigue, and the risk information should be communicated properly. Also, external support should be integrated to boost the public’s sense of efficacy in maintaining health behaviors. This study attempts to integrate the MTM and the PADM within the Chinese cultural context and offers perspectives on health behavior studies. It provides some strategies and recommendations for promoting the public maintenance of health behaviors.