The underutilization of preventive cardiovascular measures in patients with cancer: an analysis of the Behavioural Risk Factor Surveillance System, 2011–22

Abstract

Aims

This study aimed to characterize the influence of a cancer diagnosis on the use of preventive cardiovascular measures in patients with and without cardiovascular disease (CVD).

Methods and results

Data from the Behavioural Risk Factor Surveillance System Survey (spanning 2011–22) were used. Multivariable logistic regression models adjusted for potential confounders were applied to calculate average marginal effects (AME), the average difference in the probability of using a given therapy between patients with and without cancer. Outcomes of interest included the use of pharmacological therapies, physical activity, smoking cessation, and post-CVD rehabilitation. Among 5 012 721 respondents, 579 114 reported a history of CVD (coronary disease or stroke), and 842 221 reported a diagnosis of cancer. The association between cancer and the use of pharmacological therapies varied between those with vs. without CVD (P-value for interaction: <0.001). Among patients with CVD, a cancer diagnosis was associated with a lower use of blood pressure-lowering medications {AME: −1.46% [95% confidence interval (CI): −2.19% to −0.73%]}, lipid-lowering medications [AME: −2.34% (95% CI: −4.03% to −0.66%)], and aspirin [AME: −6.05% (95% CI: −8.88% to −3.23%)]. Among patients without CVD, there were no statistically significant differences between patients with and without cancer regarding pharmacological therapies. Additionally, cancer was associated with a significantly lower likelihood of engaging in physical activity in the overall cohort and in using post-CVD rehabilitation regimens, particularly post-stroke rehabilitation.

Conclusion

Preventive pharmacological agents are underutilized in those with cancer and concomitant CVD, and physical activity is underutilized in patients with cancer in those with or without CVD.

Lay Summary

•This paper compared the use of preventive cardiovascular measures, both pharmaceutical and non-pharmaceutical, in patients with and without cancer.

•In patients with cardiovascular disease and cancer, there is a lower use of preventive cardiovascular medications compared with those with cardiovascular disease but without cancer. This includes a lower utilization of blood pressure-lowering medications, cholesterol-lowering medications, and aspirin.

•Patients with cancer reported lower levels of exercise but higher levels of smoking cessation compared with those without cancer

Graphical Abstract

Graphical Abstract.

Summary of study results regarding key differences in the utilization of preventive measures based on the presence or absence of cancer.

See the editorial comment for this article ‘Secondary prevention of cardiovascular disease: unrecognized opportunity to improve survival in cancer patients’, by A. Vallakati and B. Konda, https://doi.org/10.1093/eurjpc/zwad183.

Introduction

Cancer and cardiovascular disease (CVD) are the two leading causes of death globally.^1,2 Due to the shared risk factor profile between cancer and CVD,^3,4 the cardiovascular toxicity of anti-neoplastic medications,⁵ as well as the success of contemporary cancer therapies in increasing overall survival, CVD has emerged as a major cause of morbidity and mortality in patients with cancer.⁶ Accordingly, the appropriate use of preventive cardiovascular measures, both pharmacological and non-pharmacological, is of crucial importance to the well-being and longevity of this growing patient population. Indeed, the latest data from the Surveillance, Epidemiology, and End Results programme estimates a lifetime cancer risk of nearly 40%.⁷

There is conflicting data on whether the implementation of preventive cardiovascular measure is reduced in the presence of a cancer diagnosis. Compared with the general population, one study reported the underutilization of preventive agents in cancer survivors,⁸ one reported no significant differences,⁹ while another study reported higher utilization rates in patients with cancer.¹⁰ Additionally, there are little data as to whether the difference (if any) in the use of preventative strategies is most prominent in those with or without clinically diagnosed CVD.

Therefore, we aimed to evaluate the relationship between having a cancer diagnosis and the use of preventive pharmacological/non-pharmacological cardiovascular measures in patients with and without CVD using a nationwide US survey.

Methods

Data source

We analysed data from the Behavioural Risk Factor Surveillance System (BRFSS), an annual, nationally representative, telephone survey that collects data on socioeconomic characteristics, health-related behaviours, and medical conditions from US adults.¹¹ The BRFSS uses a stratified random sampling design. We analysed data from 2011 to 2022 as data prior to 2011 used a different weighting method and did not include cell phone-only respondents. The questions used by the BRFSS to obtain data on the variables used in our analysis are listed in Supplementary material online, Table S1.

Variables and outcomes

Our outcomes of interest were self-reported use of medical therapies, including blood pressure-lowering therapies (among people with hypertension), use of lipid-lowering therapies (among patients with dyslipidaemia), and use of aspirin; levels of physical activity, including engagement in leisure-time physical activity (LTPA), meeting guideline-recommended thresholds for aerobic activity and strength training; use of post-CVD rehabilitation services after myocardial infarction (MI) and/or stroke; and smoking habits, including smoking cessation attempts and current smoking status. Physical activity recommendations used by the BRFSS are based on the 2008 Physical Activity Guidelines for Americans published by the US Department of Health and Human Services.¹² The explanatory variable of interest was a self-reported diagnosis of cancer.

Statistical analysis

The association between cancer and these preventative cardiovascular measures was investigated using logistic regression models with these outcomes as the dependent variable and a diagnosis of cancer as the independent variable. We used multivariable models to adjust for potential confounders as several factors that are associated with the presence of cancer are also likely to be associated with our outcomes of interest.

For instance, the incidence of cancer increases with age, and older people are more likely to adhere to medical therapy than younger patients.^13,14 Similarly, cancer may be associated with greater contact with the healthcare system, resulting in an increase in the use of preventive cardiovascular measures. Additionally, cancer is associated with the presence of other cardiovascular morbidities,¹⁵ possibly due to shared risk factors,³ and the presence of such comorbidities may influence the decision of the patient and/or provider to initiate preventive agents.

Therefore, potential confounders adjusted for in our models included sociodemographic factors (age, sex, race, income, educational attainment, and employment status), access to healthcare (health insurance, length of time since last check-up, and the presence of a personal doctor), and the presence of other CVD risk factors (diabetes, hypertension, dyslipidaemia, and smoking).

Further, previous work has suggested that the presence of two major disease burdens (CVD and cancer) may be more detrimental to a patient’s ability to manage their disease than the presence of one major disease (cancer only). This is suggested by previous research showing that patients with both CVD and cancer have difficulty in affording medical expenses due to the high cost of treatment,¹⁶ that polypharmacy is negatively associated with medication adherence,¹⁷ and that greater disease burdens and complexities may adversely impact a patient’s adherence due to a greater workload and a decreased functional capacity.¹⁸

Accordingly, we investigated whether the concomitant presence of CVD (defined as a self-reported history of coronary artery disease or stroke) modified the association between cancer and our outcomes of interest. This was formally investigated through the inclusion of an interaction term between cancer and CVD. If a statistically significant interaction was detected (P-value for interaction term: <0.05), we reported effect sizes of cancer separately for those with and without CVD. If no statistically significant interaction was detected, the overall effect sizes were reported without stratification.

Additionally, because of the known association between outdoor sun exposure during activities and skin cancer,^19,20 our analysis of physical activity excluded patients with skin cancer to minimize this source of confounding.

Effect sizes are primarily reported as average marginal effects (AMEs). The AME represents absolute differences in using a given preventive measure per 100 individuals and is obtained using multivariable logistic regression models after adjusting for the aforementioned potential confounders (sociodemographic characteristics, healthcare access factors, and CVD risk factors). The AME has three important advantages over the odds ratio (OR). First, it has a more intuitive interpretation and is less prone to misinterpretation as compared with the OR.²¹ Second, it is more effective at conveying the magnitude of an association than the OR.²² Third, it is less sensitive to model misspecification compared with the OR.²²

Additionally, we conducted a regression analysis to assess the relationship between the average life expectancy of the type of cancer reported by each respondent and the likelihood of using blood pressure-lowering medications, lipid-lowering medications, and engaging in LTPA. This was done to investigate whether cancers with a longer life expectancy had a different impact on the utilization of preventive cardiovascular measures. The results of this analysis are presented as the AME per 1 year increase in life expectancy. The average life expectancies of each cancer were obtained from the latest report published by the American Cancer Society, which uses the Surveillance, Epidemiology, and End Results programme, the North American Association of Central Cancer Registries, and the National Center for Health Statistics to compile data on cancer outcomes.² For 2022, the report included data on the average life expectancies for cancers of the breast, colorectum, oesophagus, kidneys and renal pelvis, liver and intrahepatic bile ducts, lungs and bronchi, melanoma, non-Hodgkin lymphoma, oral cavity and pharynx, ovaries, pancreas, prostate, urinary bladder, uterine cervix, and uterine corpus.

All analyses were conducted using R, version 4.2.0.²³ Because of the survey design of the data set, all calculations were performed using the ‘survey’²⁴ package using weights provided by the BRFSS. The code used to produce the results will be made available upon contacting the first author. A P-value of <0.05 denoted statistical significance.

Results

Baseline characteristics

From 2011 to 2022, a total of 5 012 721 survey respondents were included in the BRFSS and a total of 4 945 822 had available data on cancer and CVD status. Of those, a total of 579 114 reported a history of CVD (defined as a self-reported history of coronary artery disease or stroke), 842 221 reported a history of cancer, and 170 815 reported a history of both CVD and cancer. The characteristics of patients with and without cancer are reported in Table 1. ٌRaw utilization rates of preventive measures, without adjustment for potential confounders, are presented in Supplementary material online, Table S2. Further stratification of the baseline characteristics according to the presence or absence of CVD is presented in Supplementary material online, Tables S3 and S4. Data on participants with missing data on the status of cancer or CVD (1.33%) are presented and compared with participants with available data on both variables in Supplementary material online, Table S5. On average, those with missing data were more likely to be disadvantaged in terms of healthcare and socioeconomic status, with lower incomes and educational attainment as well as a greater proportion of racial/ethnic minorities, less access to healthcare, and a greater prevalence of comorbidities.

Table 1.

Baseline characteristics of Behavioural Risk Factor Surveillance System respondents (2011–22) stratified by the presence or absence of cancer

Characteristic	Respondents without cancer	Respondents with cancer
Number of respondents (N)	4 148 404	842 221
Sociodemographic characteristics (%)
Age
18–24	14.3	1.2
25–29	9.1	1.3
30–34	10.3	2.1
35–39	8.7	2.3
40–44	9.2	3.7
45–49	7.9	4.6
50–54	9.4	8.2
55–59	7.9	9.7
60–64	7.6	12.9
65–69	5.5	13.2
70–74	4.1	13.6
75–79	2.9	12.0
80 or older	3.2	15.1
Sex
Male	49.3	43.5
Female	50.7	56.5
Race/ethnicity
Non-Hispanic White	60.6	84.4
Non-Hispanic Black	12.5	5.6
Other non-Hispanic	7.3	2.7
Non-Hispanic multiracial	1.4	1.2
Hispanic	18.2	6.1
Educational attainment
Never attended elementary	0.3	0.2
Attended elementary	4.7	3.7
Attended high school	9.2	7.9
Graduated high school	28.4	27.2
Some college	30.6	31.6
College graduate	26.9	29.4
Employment status
Employed	59.6	33.4
Homemaker	6.4	5.7
Retired	14.4	46.3
Student	6.3	0.6
Unemployed	13.3	14.1
Income
$10 000 or less	6.3	4.6
$10 000–$15 000	5.3	5.5
$15 000–$20 000	7.7	7.3
$20 000–$25 000	9.0	9.2
$25 000–$35 000	10.6	11.4
$35 000–$50 000	13.3	14.5
$50 000–$75 000	15.0	15.6
$75 000 or more	32.8	32.0
Factors related to healthcare access (%)
Health insurance
No	14.8	4.8
Yes	85.2	95.2
Last check-up
Within the past year	70.6	85.1
Within the past 2 years	13.8	8.2
Within the past 5 years	8.2	3.7
5 years or more ago	7.4	3.1
Personal doctor
No	23.9	7.2
Yes	76.1	92.8
Medical comorbidities (%)
Diabetes mellitus
No	90.2	81.8
Yes	9.8	18.2
Hypertension
No	70.3	47.2
Yes	29.7	52.8
Dyslipidaemia
No	67.1	49.1
Yes	32.9	50.9
No	57.0	73.5
Yes	43.0	26.5
Coronary artery disease
No	94.6	84.5
Yes	5.4	15.5
No	97.4	93.0
Yes	2.6	7.0
Cardiovascular disease a
No	92.8	80.4
Yes	7.2	19.6

Defined as a self-reported history of coronary artery disease or stroke.

The association of cancer with the use of preventive pharmacological agents

A reported history of cancer was associated with a lower utilization of pharmacological therapies among patients with CVD but not those without CVD (Table 2 and Figure 1). Among those with CVD, patients with a cancer diagnosis were less likely to report using blood pressure-lowering medications compared with those without a cancer diagnosis {AME: −1.46% [95% confidence interval (CI): −2.19% to −0.73%]}. In contrast, among those without CVD, there was no difference between patients with and without cancer [AME: 0.06% (95% CI: −0.65% to 0.77%)] (P-value for interaction between cancer and CVD: <0.001).

Table 2.

Differences between patients with and without cancer in terms of the use of preventive cardiovascular measures

Variable	Odds ratio [95% CI]	Average marginal effect [95% CI]a	P-value
Use of pharmacological agents
Use of blood pressure-lowering medications
Cardiovascular disease absent	1.01 [0.95–1.07]	0.06% [−0.65% to 0.77%]	0.86
Cardiovascular disease present	0.80 [0.71–0.89]	−1.46% [−2.19% to −0.73%]	<0.001
Use of lipid-lowering medications
Cardiovascular disease absent	1.06 [1.00–1.12]	0.98% [−0.02% to 1.98%]	0.055
Cardiovascular disease present	0.84 [0.75–0.95]	−2.34% [−4.03% to −0.66%]	0.006
Use of aspirin
Cardiovascular disease absent	0.99 [0.92–1.08]	−0.13% [−1.59% to 1.33%]	0.86
Cardiovascular disease present	0.71 [0.61–0.83]	−6.05% [−8.88% to −3.23%]	<0.001
Use of non-pharmacological measures
Engagement in leisure-time  exercise or physical activity	0.89 [0.87–0.92]	−2.19% [−2.80% to −1.58%]	<0.001
Meeting aerobic physical activity recommendations	0.95 [0.92–0.98]	−1.23% [−2.08% to −0.38%]	0.005
Meeting strength physical activity recommendations	0.97 [0.93–1.00]	−0.61% [−1.28% to 0.05%]	0.071
Outpatient rehabilitation following myocardial infarction or stroke	0.76 [0.60–0.98]	−4.29% [−7.74% to −0.84%]	0.015
Outpatient rehabilitation following myocardial infarction	0.92 [0.76–1.11]	−1.85% [−6.04% to 2.34%]	0.39
Outpatient rehabilitation following stroke	0.76 [0.60–0.98]	−5.49% [−10.39% to −0.58%]	0.028
Attempted to quit smoking	1.14 [1.09–1.19]	3.09% [2.00% to 4.17%]	<0.001
Current smoker	0.94 [0.92–0.97]	−1.03% [−1.54% to −0.52%]	<0.001

Average marginal effects represent absolute differences per 100 individuals and were obtained from multivariable logistic regression models adjusted for age, sex, race, education, employment, income, insurance status, frequency of medical check-ups, the presence of a personal doctor, smoking, the presence of diabetes, hypertension (for analyses other than that for blood pressure-lowering medications), dyslipidaemia (for analyses other than that for lipid-lowering medications), and year of survey. For outcomes where heterogeneity was present between those with and without cardiovascular disease (use of blood pressure-lowering therapies, lipid-lowering therapies, and aspirin), estimates are provided for each group separately. For outcomes with no such heterogeneity, overall effect sizes are presented.

Figure 1.

Differences in the use of preventive cardiovascular measures in people with vs. without cancer.

Similarly, in those with clinical CVD, patients with a cancer diagnosis were less likely to report using lipid-lowering medications compared with those without a cancer diagnosis [AME: −2.34% (95% CI: −4.03% to −0.66%)]. Conversely, among those without clinical CVD, there was no significant difference between patients with and without cancer [AME: 0.98% (95% CI: −0.02% to 1.98%)] (P-value for interaction between cancer and CVD: <0.001).

Additionally, among those with CVD, patients with a cancer diagnosis were less likely to report using aspirin compared with those without a cancer diagnosis [AME: −6.05% (95% CI: −8.88% to −3.23%)]. In contrast, among those without CVD, there was no significant difference between patients with and without cancer [AME: −0.13% (95% CI: −1.59% to 1.33%)] (P-value for interaction between cancer and CVD: <0.001).

The association of cancer with the use of preventive non-pharmacological measures

A history of cancer was independently associated with a lower likelihood of engaging in LTPA [AME: −2.19% (95% CI: −2.80% to −1.58%)]. As opposed to pharmacological therapies, there was no evidence of this association varying according to the absence or presence of CVD.

Additionally, a history of cancer was associated with a significant reduction in the probability of meeting recommendations for aerobic activity [AME: −1.23% (95% CI: −2.08% to −0.38%)], and there was no evidence that this association varied depending on the presence of CVD. The association between a history of cancer and meeting recommendations for strength activity was not statistically significant and did not vary according to the presence of CVD.

Furthermore, a cancer diagnosis was associated with a lower likelihood of participating in post-CVD rehabilitation [AME: −4.29% (95% CI: −7.74% to −0.84%)]. This was primarily driven by lower engagement in post-stroke rehabilitation [AME: −5.49% (95% CI: −10.39% to −0.58%)], as differences in post-MI rehabilitation [AME: −1.85% (95% CI: −6.04% to 2.34%)] were lower and not statistically significant.

Compared with patients without cancer, those with a history of cancer were more likely to attempt smoking cessation [AME: 3.09% (95% CI: 2.00% to 4.17%)], with no heterogeneity based on underlying CVD. Patients with cancer were also less likely to report current smoking [AME: −1.03% (95% CI: −1.54% to −0.52%)], with no evidence of interaction by CVD status. The corresponding relative effect sizes (presented as the OR) are shown in Table 2. Unadjusted raw utilization rates of the aforementioned preventive measures are shown in Supplementary material online, Table S2.

Association of cancer life expectancy with likelihood of using preventive measures

Among patients with cancer, there was no statistically significant association between the average life expectancy of the reported cancer diagnosis and the likelihood of taking blood pressure-lowering therapies [AME: 0.05% (95% CI: −0.02% to 0.12%)], lipid-lowering therapies [AME: 0.02% (95% CI: −0.08% to 0.12%)], or engaging in LTPA [AME: 0.05% (95% CI: −0.02% to 0.12%)].

Discussion

Our analysis of a nationally representative US sample showed that, in patients with CVD, a cancer diagnosis was associated with a lower likelihood of using blood pressure-lowering medications, lipid-lowering medications, aspirin, or CVD rehabilitation services. Additionally, a cancer diagnosis was associated with a lower likelihood of engaging in physical activity and meeting aerobic exercise recommendations regardless of CVD status. These findings can be explained by a combination of factors, namely the psychosocial implications of a cancer diagnosis such as fatigue, depression, and other physical limitations;^25,26 the challenges of polypharmacy; the compounded financial barriers imposed by a cancer diagnosis; and the lack of primary care involvement and follow-up in the context of a cancer diagnosis.

In the setting of cancer diagnosis and treatment, it is understandable that the time and effort of both patients and healthcare teams may shift to cancer as a more immediate concern than the prevention of CVD, as the latter is a more remote concern. This is made more likely by the extensive time and effort necessary to adequately manage a cancer diagnosis. Accordingly, the healthcare team may be less likely to prescribe the use of several preventive agents due to a perceived lack of importance relative to cancer treatment or concerns about possible side effects and drug interactions.

Polypharmacy is an important potential driver for the underutilization in this patient population, as the simultaneous prescription of several medications is associated with poor utilization.²⁷ Additionally, many preventive agents cause adverse events that may require discontinuation, particularly in a population that is vulnerable to drug-related toxicities.²⁸ Further, concerns about potential drug interactions may hinder the use of preventive agents while patients are undergoing treatment.²⁸ Moreover, the economic burden imposed by cancer treatment also compounds the financial barriers faced by patients within the US healthcare system. Previous analyses have highlighted the importance of financial barriers to the use of appropriate blood pressure-lowering and lipid-lowering medications, with higher fees being associated with significantly lower utilization rates.²⁹ In patients with cancer, for whom cancer treatment can come at an exorbitant price, these barriers are likely exacerbated.

The financial burden of a cancer diagnosis is further accentuated when managing a concomitant chronic disease such as CVD, since patients with concomitant CVD are at a greater risk of financial hardship compared with those with cancer alone.¹⁶ This may partially explain why our analysis showed that the underutilization of pharmacological agents was seen in those with both cancer and CVD as opposed to those with cancer alone.

Several steps may be taken to remedy this underutilization of preventive measures. Proper communication and regular follow-up play a major role in patient attitudes towards the management of their disease;³⁰ moreover, health literacy is positively correlated with adherence in numerous studies.³¹ Implementing a three-part intervention at the clinician level would greatly improve patient utilization of treatment. First, cancer care must consist of a well-connected and communicative multidisciplinary team including oncologists, cardiologists, and primary care physicians (PCPs).³² Second, ensure that clinicians, namely PCPs and cardiologists, communicate the importance of preventive measures to patients. This would involve highlighting both the risk that CVD poses to the patient and the benefit of utilizing preventive measures to minimize the risk of CVD and chemotherapeutic cardiotoxicity.³³ Third, consistent follow-up to identify non-adherence and tackle its causes is vital to ensure long-term success.

Because PCPs often have the greatest long-term contact with patients, they may be the best-positioned members of the care team to ensure utilization of appropriate preventive treatment. Although the role of the PCP and an interdisciplinary approach in cancer care has been outlined by several investigators,^34–36 studies continue to report inconsistent levels of involvement by PCPs and a lack of interdisciplinary communication.^35,37,38

At the policy level, programmes targeting patients with cancer aimed at promoting the uptake of appropriate pharmacological and non-pharmacological interventions need to be implemented. The effectiveness of implementation programmes in improving the uptake of guideline-directed therapies has been previously documented;^39,40 therefore, initiatives tailored to patients with cancer can help reduce the underutilization of appropriate preventive measures in this patient population.

Finally, our analysis identified two additional noteworthy findings. First, smoking cessation was more likely in patients with cancer. This is likely because the well-known association between smoking and cancer provides a strong motivation to quit. Second, our analysis did not suggest that the average life expectancy of a given cancer was strongly associated with the underutilization of preventive measures, suggesting that this underutilization may affect people who live long enough to benefit from preventive measures.

Limitations

Several limitations of our study warrant mention. First, the BRFSS uses a cross-sectional telephone survey design. Thus, it is reliant on answers by survey respondents who may exaggerate their utilization of certain interventions. Second, it is possible that some of the underutilization of preventive measures in patients with cancer is related to a limited life expectancy in the case of end-stage cancer disease. However, given that our regression analysis showed no significant association between a cancer’s average life expectancy and use of pharmacological and non-pharmacological therapies, limited life expectancy likely does not fully account for our findings. Third, because the BRFSS is a telephone survey that does not collect detailed data on participants, it was not possible to perform an assessment of differences in terms of blood pressure control or lipid control attributable to the underutilization of preventive agents. Accordingly, it is possible that the lower utilization of anti-hypertensive agents in patients with cancer may be partially attributable to blood pressure levels that are adequately controlled. Fourth, it is possible some preventive agents were contraindicated in some patients due to their cancer treatments; however, as we did not have access to the chemotherapeutic regimens received by respondents, our analysis cannot elucidate these nuances. Fifth, changes in guideline recommendations from 2011 to 2022 regarding the use of preventive agents may affect the degree to which they are used in different patient populations, including those with cancer. Sixth, although we attempted to isolate the influence of cancer by controlling for potential confounders, the observational design of our study does not imply causality.

Conclusion

Cancer survivors face a high risk for subsequent CVD. Several pharmacological and non-pharmacological preventive cardiovascular measures are underutilized in patients with cancer, including blood pressure and lipid-lowering therapies, aspirin, cardiac rehabilitation, and physical exercise. Pharmacological agents appear particularly underutilized in patients with both cancer and CVD, a worrisome finding considering the high risk of recurrent CVD in this patient population. Interventions aimed at improving patient and provider awareness about CVD risk and increasing the uptake of appropriate preventive cardiovascular measures targeting patients with cancer are needed.

Supplementary material

Supplementary material is available at European Journal of Preventive Cardiology.

Funding

S.S.V. is supported by grants from the Department of Veterans Affairs, the National Institute of Health, the Tahir and Jooma Family, and has received Honoraria from the American College of Cardiology (Associate Editor for Innovations, ACC.org). T.G.N. has received grant funding from AstraZeneca and Bristol Myers Squibb.

Author contributions

All authors have made significant intellectual contributions to this work. A.S. conceived the study and performed the analysis. S.V. and D.A. contributed critical feedback on the conception and analysis of this work. A.S. and M.M drafted the initial version of the manuscript. D.A., A.B., A.A., and S.D. contributed critical feedback to the interpretation of the analysis. S.V., A.A., and D.A. significantly refined the methodological explanations in the paper. D.A., A.A., S.D., T.N., A.B., M.F., A.N., K.M., and S.V. provided substantive critical feedback and revisions to the manuscript. All authors have revised and approved the manuscript for submission.

Data availability

This analysis utilized publicly available data that are freely accessible to researchers and can be accessed through the following link: https://www.cdc.gov/brfss/annual_data/annual_data.htm.