Secondary analysis of data from the 2019 and 2021 Behavioral Risk Factor Surveillance System (BRFSS) surveys, which are national annual surveys that investigate health-related behaviors and health conditions of noninstitutionalized US adults aged ≥18 years, was performed [20]. The surveyor used the random-digit–dialing technique to generate random phone numbers for telephone and mobile phone interviews across 50 states in the United States, the District of Columbia, Guam, Puerto Rico, and the US Virgin Islands. The latest BRFSS survey was conducted in 2021. The median survey response rate of the 2019 BRFSS survey and 2021 BRFSS survey was 49.4% and 44%, respectively [20,21].

Cases for which data regarding sociodemographic variables or chronic health conditions were missing and individuals who refused to answer questions were excluded from the study. We included 272,543/418,268 (65.2%) and 278,064/438,693 (63.4%) individuals from the 2019 and 2021 BRFSS 2021 surveys, respectively.

Data were analyzed using the R statistical software (R Foundation for Statistical Computing). Logistic regression analysis was used to investigate which independent variables were associated with practicing healthy lifestyles (dependent variable). These independent variables included the sociodemographic characteristics, health status, and health-related behaviors. To inform future research, the latest 2021 data set was analyzed. Because the percentage of the excluded subjects was as high as 36.6% and the missing rate of outcome variables were associated with various covariates (missing at random instead of missing at completely random), multiple imputations were conducted to estimate the missing variables among the excluded subjects. Five imputed data sets were created by the multiple imputation via chained equations (MICE) package in R.

MICE imputation starts by a simple imputation such as mean imputations on all variables with missing value. Then, these imputed missing values are updated by running a series of regressions, whereby each variable with missing value is regressed upon other variables repeatedly until convergence. For cases with imputations, logistic regression analyses were performed for each of the imputed data sets. The regression coefficients were combined using the pool() function within the MICE package. In addition, we used the detectseparation package in R to check if there was any multicollinearity in our logistic regression analysis. The results of the 2021 data set logistic regression analyses were presented both for cases without missing data and for those with imputations. A P value <.05 was considered statistically significant.

The surveys were conducted by US federal agencies where verbal consent was obtained from each participant prior to the data collection [20,21]. The BRFSS data set are allowed for secondary data analysis without permission required. No institutional review board approval or patient informed consent was required for secondary data analysis. The participants’ personal information had been deidentified in the data set prior to being publicly available [20,21].

We analyzed 550,607 respondents (272,543 and 278,064 from the 2019 and 2021 BRFSS surveys, respectively; Table 1). The majority of the respondents were female (n=284,876, 51.7%) and resided in urban areas (n=468,547, 85.1%); 35.2% (n=193,850) of the respondents were aged 65 years and older. Regarding health and related behaviors, 83% (n=456,846) of individuals were perceived to have a good or better health status. Hypertension (n=276,148, 50.2%) and high cholesterol (n=248,440, 45.1%) were the most commonly reported chronic diseases. The percentage of respondents who performed physical activity was greater in 2021 than in 2019 (214,943/278,064, 77.3% vs 204,733/272,543, 75.1%); in contrast, the percentage of smokers was lower in 2021 than in 2019 (36,983/278,064, 13.3% vs 38,662/272,543, 14.2%). The percentage of practicing a healthy lifestyle nonsignificantly decreased from 4% in 2019 to 3.6% in 2021.

Table 2 shows the results of the logistic regression analysis for cases without missing data in 2019 and 2021 who practiced healthy lifestyles; similar results were obtained for 2019 and 2021. In 2021, female individuals (odds ratio [OR] 1.95; P<.001) who had a high level of education (OR 1.99; P<.001), were single (OR 1.10; P=.02), unemployed (OR 1.10; P=.001), and had perceived good health (OR 1.62; P<.001) were found to have a higher likelihood of practicing healthy lifestyles. Compared with adults aged 65 years and above, young adults aged 18-64 years tended not to practice healthy lifestyles (OR 0.53-0.66; P<.001). Individuals with a low household income have a low likelihood of practicing healthy lifestyles (OR 0.72-0.79; P<.001).

Individuals with chronic health conditions had a low likelihood of practicing healthy lifestyles. The data in Table 2 show that individuals with NCDs such as hypertension and diabetes had a low likelihood of practicing healthy lifestyles (OR 0.48-0.83; P<.001).

For the cases with imputations, the weighted estimates of the logistic regression analysis results from 5 imputed data sets are presented in Table 3 (n=438,693). The rates of missing outcomes of healthy lifestyles in each variable are reported in Table S1 in Multimedia Appendix 1; briefly, the rate varied from 14.1% (61,858/438,693) to 27.3% (119,823/438,693). The weighted estimates were in line with those for cases without missing data (Table 2). Multicollinearity was not detected in all regression analyses.

This study assessed the determinants of practicing healthy lifestyles at the national level using 2019 and 2021 BRFSS survey data. The distribution of sociodemographic variables and health and related behaviors were similar for both the surveys. However, the tendency to practice healthy lifestyles slightly decreased in 2021. The logistic regression analysis results indicated that female individuals residing in urban areas with a higher level of education were more likely to practice healthy lifestyles, whereas younger individuals with a low household income were less likely to do so. Although chronic health conditions hindered individuals from practicing healthy lifestyles, those with perceived good or better health were more likely to practice healthy lifestyles.

However, the overall rate of practicing a healthy lifestyle was still very low. Fang et al [22] analyzed 2013 BRFSS data and reported that only 1.7% of the assessed individuals with hypertension practiced healthy lifestyles. Other studies have shown that 2.1% of healthy adults and 6% of health care providers practiced healthy lifestyles [5,6]. Moreover, although more individuals performed physical activity, the percentage of individuals with optimal body weights and following healthy diets decreased. High BMI and unhealthy diet were major risk factors for disability and premature death worldwide [23,24]. In this study, we found that the individuals’ residences considerably affected their decisions to practice healthy lifestyles. Rural residents reported that they received limited access to health-related information from health care providers [25]. Although the overall number of health care providers, such as physicians, nurse practitioners, and physician assistants, in the United States significantly increased between 2009 and 2017, the urban-rural disparities remained significant; the number and density of health care providers increased to a greater extent in urban areas than in rural areas [26]. Therefore, a healthy lifestyle should be strongly promoted, and urban-rural health disparities should be tackled urgently.

In concordance with previous studies [9,11,27], our results showed that individuals with a low household income, who were male, and had high school or lower levels of education were less likely to practice healthy lifestyles. Although married people were more likely to practice healthy lifestyles, results from another national survey revealed that such individuals were more likely to be overweight or obese [28]. In contrast, we found that young people (aged <65 years) were less likely to practice healthy lifestyles. Thus, early health promotion is required because the risk of developing NCDs exponentially increases after 50 years of age [29].

Practicing healthy lifestyles among individuals with hypertension were focused because hypertension has been the leading risk factor for global disease burden worldwide for a decade [23,30]. Although certain guidelines recommend healthy lifestyles and medical treatments to effectively control blood pressure [31-33], our analysis showed that individuals with hypertension were less likely to practice healthy lifestyles. A Finnish cohort study involving 41,225 participants followed up over 4 years reported that the initiation of antihypertensive treatment significantly decreased the tendency of individuals with hypertension to practice healthy lifestyles (OR 0.55-0.92) [34]. As blood pressure can indeed be controlled with medications, less attention has been paid to individuals with hypertension practicing healthy lifestyles. Approximately 82.3% of individuals with hypertension in the United States use antihypertensive medications [35]; they might not practice healthy lifestyles as their blood pressure is medically controlled.

Concern about around those with diabetes practicing healthy lifestyles have been raised, because these individuals showed the least likelihood of doing so. A healthy lifestyle has been indicated in various diabetes management guidelines [36,37]. However, our findings revealed that individuals with diabetes were the least likely to practice healthy lifestyle, showing the smallest OR among all variables. Individuals with diabetes have emphasized that education and support from health care providers can enhance their self-care tendencies [38]. Hence, new methods for diabetes care should be used.

Our study findings have strong implications for clinical practice, research, and policy making. The low likelihood of practicing healthy lifestyles highlights the need for strengthening current practices. Health education supported by printed materials, phone calls, and SMS text messaging may improve individuals’ adherence to healthy lifestyles and medical treatments [39-43]. Given the rapid increase in smartphone ownership and advancing technologies [44], digital interventions such as website and smartphone apps have been shown to slightly improve adherence to a healthy lifestyle [45-47]. Supportive methods and digital interventions can be integrated into current health promotion activities to facilitate healthy lifestyles and widen the possibility of receiving health-related information and support from health care providers, especially for rural residents. Moreover, digital interventions can help overcome the barriers to public health measures caused by COVID-19 [18]. However, we noted that several sociodemographic characteristics can hinder individuals from practicing healthy lifestyles. Additionally, there may be associations between depression, ethnic background, and socioeconomic status [48]. Future studies can explore the associations and focus on developing specific interventions that promote healthy lifestyles among such individuals.

Rural residents showed a low likelihood of practicing healthy lifestyles; this highlights the need of support from policy makers. A cohort study of 25,014 middle-age and older adults with a 20-year follow-up revealed that performing any type of physical activity could significantly reduce the likelihood of hospitalization to a mean of 0.42 days per year; this reduced likelihood of hospitalization could reduce health care expenditures by approximately 7% [49]. Moreover, some amount of the saved health care expenditure could be granted as tax deductions to individuals who participate in healthy lifestyle programs and provide evidence of practicing such lifestyles (ie, reduced body weight). Such tax deductions can attract more people toward healthy lifestyles, further decreasing the overall health care expenditure. The shortage of health care providers may be managed through the use of digital interventions for providing education and support to residents in rural areas. However, more interventions should be developed to attract health care providers to work and remain in rural areas. For example, increased training in rural areas, the payment of locum relief, and loan repayment may help do so [50].

This study has 4 key limitations. First, the variables in the secondary analysis were limited to the data set used. For example, only 1 question in the data set investigated the individuals’ physical activity habits. Moreover, the question was rather general and inquired whether the individuals performed any type of physical activity or exercise during the past 30 days. Two essential components—frequency and intensity—were not explored. Thus, our study findings might not adequately represent the level of physical activity as recommended by the WHO [51]. Second, all variables were reported by individuals via phone call. Therefore, we might have encountered reporting biases arising from the participant’s recall from memory, and the interviewers may have recorded the responses incorrectly. To address such issues, the CDC has created a program to identify errors or conditions of concern due to data entry by the interviewers [20,21]. Third, although the phone numbers used in each year were not repeated, it was unclear if some numbers would be used in the other survey years. Hence, it was possible that some people might have responded to both surveys in this study. Fourth, the overall rate of missing data was high at 36.6% in 2021, which could have impacted the generalizability of the results. However, we performed multiple imputations, and the pooled results of the imputed data were in line with those from the cases without missing outcomes. Hence, our imputed data provided an accurate estimation of the influence of sociodemographic variables and chronic health conditions on practicing healthy lifestyles and enhanced the generalizability of our findings.

Practicing a healthy lifestyle can have numerous health benefits across all age groups. We found that very few people practiced healthy lifestyles before and during COVID-19. Some sociodemographic variables and chronic health conditions were identified as significant factors affecting individuals’ decisions to practice healthy lifestyles. Several interventions were suggested to strengthen the promotion of healthy lifestyles at the community level and to attract more health care providers to work and remain in rural areas.