Abstract
Background
Engaging in unhealthy behaviours [poor diet, insufficient physical activity (PA)] increases risk for recurrent stroke and can be compounded by obesity and diabetes, but the association of obesity and diabetes with poor diet and insufficient PA in stroke survivors is unknown.
Objective
The purpose of this study was to compare prevalences of low fruit and vegetable consumption (low FV consumption, <1 fruit and <1 vegetable daily) and low physical activity (low PA, <150 minutes of weekly moderate-intensity PA) in stroke survivors, stratified by obesity–diabetes status (neither condition, obesity only, diabetes only, both conditions).
Methods
Cross-sectional data from 32 876 non-institutionalized, US stroke survivors aged ≥45 years from the 2015 and 2017 Behavioral Risk Factor Surveillance System were examined. Weighted, age-adjusted prevalence estimates and adjusted odds ratios (AORs) of the investigated unhealthy behaviours (adjusted for sex, age, race, income, education and marital status) and 95% confidence intervals (CIs) were calculated.
Results
Prevalences of low FV consumption and low PA exceeded 50% across all obesity–diabetes categories. Compared with respondents with neither obesity nor diabetes, AORs for low PA were increased for respondents with both obesity and diabetes (2.02, 95% CI: 1.72–2.37) and respondents with obesity only (1.31, 1.13–1.53); AORs for low FV consumption did not differ across obesity–diabetes categories.
Conclusions
Results indicated a joint effect of obesity and diabetes with low PA among stroke survivors. Regardless of obesity–diabetes status, however, prevalence of low FV consumption and low PA exceeded 50%. Targeted interventions that modify these unhealthy behaviours among stroke survivors should be explored.
Keywords: Diabetes, diet, obesity, physical activity, primary care, stroke
Key Messages.
More than 50% of US stroke survivors report low fruit and vegetable consumption.
More than 50% of US stroke survivors also report low physical activity.
Obesity and diabetes are jointly associated with low physical activity.
Introduction
Stroke is a prevalent condition and a leading cause of disability worldwide, affecting an estimated 104.2 million people and contributing to approximately 132 million disability-adjusted life years lost in 2017 (1). In the USA, approximately 23% of annual strokes are recurrent events, and the 10-year cumulative risk of stroke recurrence is 25% (2). An estimated 74% of global stroke risk can be attributed to behavioural risk factors, including smoking, physical inactivity and an unhealthy diet (3). These unhealthy behaviours increase risk for high blood pressure and high cholesterol, insulin resistance and glucose intolerance and obesity, which in turn increase risk for cardiometabolic complications such as cardiovascular disease, diabetes and stroke (2).
Avoiding physical inactivity and adopting a healthy diet are important for preventing recurrent stroke (4). Guidelines for secondary stroke prevention such as the American Heart Association’s (AHA) Guidelines for the Prevention of Stroke in Patients with Stroke and Transient Ischemic Attack recommend lifestyle management of behavioral risk factors (4), which encompasses self-managing health behaviours [e.g. physical activity (PA) and dietary intake] and comorbidities (e.g. obesity and diabetes). PA of approximately 150 minutes of weekly moderate-intensity activity and following a healthy dietary pattern, including consumption of fruits, vegetables, whole grains and low-fat dairy, are specifically recommended. Self-managing glycemic control and cardiovascular risk, and screening for obesity, are also recommended.
It is concerning, therefore, that prevalences of unhealthy behaviours among US stroke survivors are high, exceeding those observed for US adults without stroke for low fruit and vegetable (FV) consumption (51.7% versus 46.0%) and insufficient PA (56.5% versus 49.5%). Similarly, prevalences for overweight/obesity (70.2% versus 64.5%) and diabetes (22.6% versus 9.3%) are also higher in stroke survivors compared with adults without stroke (5). These data suggest that stroke survivors experience marginal and moderate disparities in health behaviours that lead to disparities in cardiometabolic conditions, and are in need of health behaviour intervention. Furthermore, in the general population obesity and diabetes are independently associated with unhealthy behaviours (6,7). In combination, however, obesity and diabetes may interact, resulting in a joint association with unhealthy behaviours that is stronger than would exist with either condition individually. This is concerning because multimorbidity is associated with poor health and all-cause mortality in the general population (8), and likely generalizes to stroke survivors as well.
There is a lack of research on the joint association of obesity and diabetes with unhealthy behaviours among stroke survivors. Such knowledge could be important to identify stroke subpopulations at greatest risk for unhealthy behaviours and in greatest need of health behaviour intervention. Furthermore, it could suggest that different intervention approaches are needed for survivors with different combinations of comorbidity (e.g. obesity only versus diabetes only versus both obesity and diabetes). The purpose of this study, therefore, was to examine the prevalence of low FV consumption and low PA among stroke survivors, and to examine the joint association of obesity and diabetes on these unhealthy behaviours.
Methods
Data sources
Data for this cross-sectional study were obtained from the Behavioral Risk Factor Surveillance System (BRFSS), a national telephone survey that collects health-related data, including risk behaviours, chronic conditions and use of preventive services, among non-institutionalized adults aged ≥18 years in the USA. Data were collected in all 50 US states, the District of Columbia and 3 US territories and are representative of the US population. BRFSS uses random-phone digit dialing to contact potential respondents. For this study, data from the 2015 and 2017 BRFSS surveys were combined to increase the analytical sample size. There were 441 456 respondents in the 2015 survey and 450 016 respondents in the 2017 survey, which together corresponds to 0.36% of the adult US population. Because respondents are chosen at random, overlapping respondent participation across years is unlikely. Additional data about the BRFSS can be accessed online (https://www.cdc.gov/brfss/). This study was exempt from Institutional Review Board approval because all data were de-identified.
Study sample
Data from adults with stroke aged ≥45 years at time of the BRFSS survey were eligible for inclusion in this study. History of stroke was determined by an affirmative response to the question, ‘Has a health professional ever told you that you had a stroke?’ The age threshold was chosen because ≥45 years is a risk factor for type 2 diabetes. Exclusion criteria included missing data for self-reported height, self-reported weight or missing data for the diabetes-related question, ‘Has a health professional ever told you that you have diabetes?’
Variables of interest
Study variables were derived from BRFSS data, and included low FV consumption, low PA and obesity–diabetes status. Demographic characteristics, including sex, age, ethnicity, marital status, education and annual household income (see Table 1 for categorical levels), were included to describe the study sample and to control for potential confounders during statistical analyses.
Table 1.
Characteristic | Number of respondentsa | Neither conditionb | Obesity onlyb | Diabetes onlyb | Both conditionsb |
---|---|---|---|---|---|
% (95% CI) | % (95% CI) | % (95% CI) | % (95% CI) | ||
n = 16 137 | n = 6284 | n = 5147 | n = 5308 | ||
Total sample | 32 876 | 46.1 (44.8, 47.4) | 21.3 (20.2, 22.4) | 15.2 (14.2, 16.1) | 17.5 (16.5, 18.4) |
Sex | |||||
Male | 15 316 | 50.3 (48.3, 52.4) | 44.5 (41.9, 47.2) | 54.4 (50.9, 57.9) | 51.3 (48.5, 54.1) |
Female | 19 505 | 49.7 (47.6, 51.7) | 55.5 (52.9, 58.2) | 45.6 (42.1, 49.1) | 48.7 (45.9, 51.6) |
Age, years | |||||
45–64 | 11 301 | 39.7 (38.0, 41.3) | 53.7 (51.3, 56.1) | 36.3 (33.4, 39.1) | 51.8 (49.3, 54.2) |
≥65 | 21 575 | 60.3 (58.7, 62.0) | 46.3 (43.9, 48.7) | 63.7 (60.9, 66.6) | 48.2 (45.8, 50.7) |
Ethnicity | |||||
Non-Hispanic White | 26 186 | 72.4 (70.4, 74.5) | 66.6 (63.9, 69.4) | 57.4 (53.7, 61.0) | 62.8 (60.0, 65.6) |
Non-Hispanic Black | 3945 | 14.2 (12.6, 15.9) | 19.3 (17.0, 21.6) | 21.7 (18.4, 25.1) | 20.3 (18.0, 22.5) |
Hispanic | 1629 | 7.0 (5.7, 8.4) | 9.2 (7.1, 11.3) | 12.5 (9.7, 15.2) | 11.6 (9.4, 13.8) |
Other | 2404 | 6.3 (5.2, 7.4) | 4.9 (3.7, 6.1) | 8.4 (6.3, 10.6) | 5.3 (4.3, 6.4) |
Marital status | |||||
Married or coupled | 14 737 | 45.6 (43.6, 47.6) | 49.8 (47.1, 52.5) | 48.2 (44.6, 51.8) | 47.1 (44.3, 50.0) |
Previously married | 16 914 | 44.9 (42.9, 46.9) | 40.1 (37.5, 42.7) | 41.9 (38.6, 45.2) | 41.3 (38.6, 44.1) |
Never married | 3034 | 9.5 (8.4, 10.7) | 10.1 (8.4, 11.9) | 9.9 (7.0, 12.9) | 11.6 (9.7, 13.5) |
Education | |||||
Didn’t finish high school | 4811 | 21.7 (20.0, 23.4) | 23.4 (20.8, 26.0) | 24.7 (21.7, 27.8) | 25.2 (22.6, 27.9) |
Graduated high school | 11 680 | 32.7 (30.9, 34.6) | 33.9 (31.4, 36.5) | 32.3 (29.1, 35.5) | 34.4 (31.7, 37.1) |
College, <4 years | 9846 | 29.6 (27.6, 31.6) | 28.3 (26.0, 30.6) | 30.0 (26.5, 33.5) | 29.2 (26.7, 31.8) |
College, ≥4 years | 8405 | 16.0 (14.8, 17.1) | 14.4 (12.9, 15.9) | 13.0 (10.8, 15.2) | 11.2 (9.8, 12.6) |
Annual household income, $ | |||||
<15 000 | 6286 | 22.6 (20.7, 24.5) | 23.0 (20.3, 25.6) | 26.9 (23.5, 30.3) | 27.8 (25.1, 30.4) |
15 000 to <25 000 | 7643 | 26.0 (24.1, 27.9) | 24.8 (22.5, 27.1) | 29.8 (26.1, 33.6) | 30.0 (27.2, 32.9) |
25 000 to <35 000 | 3779 | 12.0 (10.7, 13.4) | 13.1 (11.0, 15.2) | 10.8 (9.1, 12.5) | 11.7 (9.7, 13.7) |
35 000 to <50 000 | 3782 | 12.5 (10.9, 14.2) | 12.8 (11.1, 14.6) | 10.8 (8.8, 12.8) | 12.0 (9.8, 14.3) |
≥50 000 | 7119 | 26.8 (24.9, 28.8) | 26.3 (23.7, 28.8) | 21.7 (18.1, 25.4) | 18.5 (16.1, 20.9) |
aUnweighted number of respondents. Number of respondents within each demographic characteristic may not sum to sample total because of missing data. Number of respondents with missing data for each respective sociodemographic characteristic: sex, n = 13; age, n = 1958; ethnicity, n = 670; marital status, n = 149; education, n = 92; annual household income, n = 6225.
bEstimates are weighted and age-adjusted to the 2000 US standard population except age, for which group-specific percentages are displayed. Percentages may not sum to 100 due to missing data.
Low FV consumption was defined as consuming <1 fruit and <1 vegetable daily, and was derived from BRFSS questions about the frequency of fruit, 100% fruit juice, bean and vegetable consumption during the previous month. This variable was chosen as an index of nutritional risk because the 2015–2020 Dietary Guidelines for Americans recommends consumption of approximately 2.0 cup-equivalent of fruit and 2.5 cup-equivalent of vegetables per 2000 calories (9); thus, consuming <1 fruit and <1 vegetable daily would fall below this recommendation.
Low PA was defined as performing <150 minutes of moderate-intensity, <75 minutes of vigorous-intensity or an equivalent combination of moderate- and vigorous-intensity PA, per week, as recommended by the 2018 Physical Activity Guidelines for Americans (10), and was derived from BRFSS questions about time spent in leisure-, household- and exercise-based activities during the previous month.
Obesity–diabetes status is a four-level categorical variable that reflects the presence or absence of obesity and diabetes (i.e. neither condition, obesity only, diabetes only, both conditions). Obesity was defined as a body mass index (BMI) ≥30 kg/m2 and was calculated from self-reported height and weight. Diabetes was determined by an affirmative response to the BRFSS question, ‘Has a health professional ever told you that you have diabetes?’ and excluded women who only reported gestational diabetes.
Data analysis
All analyses were performed using SAS 9.4 (SAS Institute, Cary, NC). Data were weighted to account for complex sampling design and to adjust for selection probability and non-response bias; data were also age-adjusted to the 2000 US standard population (11). Linear regression was used to compute prevalence estimates and 95% confidence intervals (CIs) of demographic variables and unhealthy behaviours, stratified by obesity–diabetes status. Logistic regression was used to compute adjusted odds ratios (AORs, adjusted for all demographic characteristics) and 95% CIs for unhealthy behaviours. For AORs, the ‘neither condition’ obesity–diabetes category served as the reference group. P-values were not reported for analyses due to the large sample sizes obtained from weighting data, but significance can be inferred by examining 95% CIs (12). For unhealthy behaviours with AORs that differed across obesity–diabetes strata, the relative excess risk due to interaction (RERI) was calculated to assess interaction of obesity and diabetes on an additive scale (13). A RERI value of 0 indicates that no additive interaction is present and a value greater than 0 indicates positive additive interaction.
Results
Most of the 891 472 respondents in the 2015 and 2017 BRFSS surveys were excluded from analysis because they had one or more of the following: no reported history of stroke (n = 851 607), were younger than 45 years of age (n = 243 302) or had missing data for height, weight or history of diabetes (n = 2391). Thus, the resulting analytical sample consisted of 32 876 stroke survivors. Obesity and diabetes were present in 35.3% (n = 11 592) and 31.8% (n = 10 455) of the sample, respectively.
Weighted, age-adjusted prevalence estimates of sociodemographic characteristics are displayed in Table 1. A majority of respondents were aged ≥65 years and non-Hispanic white, with the remaining sociodemographic characteristics distributed across categorical levels. Additionally, 54% of respondents reported at least one chronic condition (i.e. obesity only, diabetes only or both conditions).
Estimated prevalences of unhealthy behaviours are displayed in Table 2. Prevalences for both unhealthy behaviours exceeded 50% across all obesity–diabetes categories (range: 51.8–71.7%), but prevalences for low PA were higher than prevalences for low FV consumption. Furthermore, prevalences for low PA were lowest for respondents with neither obesity nor diabetes and highest for respondents with both conditions, whereas prevalences for low FV consumption were similar across obesity–diabetes categories.
Table 2.
Unhealthy behaviour | Total samplea | Neither conditiona | Obesity onlya | Diabetes onlya | Both conditionsa |
---|---|---|---|---|---|
% (95% CI) | % (95% CI) | % (95% CI) | % (95% CI) | % (95% CI) | |
n = 32 876 | n = 16 137 | n = 6284 | n = 5147 | n = 5308 | |
Low FV consumptionb | 52.3 (50.9, 53.7) | 51.8 (49.7, 53.9) | 52.7 (49.9, 55.6) | 53.9 (49.6, 57.1) | 53.0 (49.9, 56.0) |
Low PAb | 61.0 (59.6, 62.3) | 56.0 (53.9, 58.1) | 63.0 (60.2, 65.8) | 61.0 (57.1, 64.9) | 71.7 (69.1, 74.3) |
aEstimates are weighted and age-adjusted to the 2000 US standard population.
bNumber of respondents with missing data for unhealthy behaviours: low FV consumption, n = 4590; low PA, n = 3448.
AORs for unhealthy behaviours, stratified by sociodemographic characteristics and obesity–diabetes status, are displayed in Table 3. Several sociodemographic characteristics, including sex, age, ethnicity, education and annual household income, were independently associated with the investigated unhealthy behaviours. Females were less likely to report low FV consumption and more likely to report low PA. Respondents ≥65 years, those with higher education and those with higher annual household income were also less likely to report engaging in the investigated unhealthy behaviours.
Table 3.
Sociodemographic characteristic | Low FV consumptiona | Low PAa |
---|---|---|
AOR (95% CI) | AOR (95% CI) | |
Sex | ||
Male | Referent | Referent |
Female | 0.68 (0.61, 0.76) | 1.30 (1.16, 1.46) |
Age, years | ||
45–64 | Referent | Referent |
≥65 | 0.75 (0.66, 0.84) | 0.85 (0.75, 0.96) |
Ethnicity | ||
Non-Hispanic White | Referent | Referent |
Non-Hispanic Black | 1.19 (0.99, 1.42) | 1.31 (1.09, 1.58) |
Hispanic | 0.79 (0.60, 1.03) | 0.87 (0.66, 1.16) |
Other | 0.93 (0.72, 1.19) | 0.88 (0.69, 1.13) |
Marital status | ||
Married or coupled | Referent | Referent |
Previously married | 1.06 (0.93, 1.22) | 1.02 (0.89, 1.17) |
Never married | 1.05 (0.82, 1.35) | 0.88 (0.69, 1.13) |
Education | ||
Did not finish high school | Referent | Referent |
Graduated high school | 0.75 (0.63, 0.89) | 0.74 (0.62, 0.89) |
College, <4 years | 0.67 (0.56, 0.80) | 0.60 (0.50, 0.72) |
College, ≥4 years | 0.37 (0.31, 0.45) | 0.47 (0.39, 0.58) |
Annual household income, $ | ||
<15 000 | Referent | Referent |
15 000 to <25 000 | 0.96 (0.82, 1.13) | 0.89 (0.75, 1.06) |
25 000 to <35 000 | 0.91 (0.74, 1.11) | 0.92 (0.75, 1.14) |
35 000 to <50 000 | 0.91 (0.74, 1.13) | 0.79 (0.63, 0.98) |
≥50 000 | 0.76 (0.62, 0.94) | 0.57 (0.46, 0.71) |
Obesity–diabetes status | ||
Neither condition | Referent | Referent |
Obesity only | 1.04 (0.90, 1.21) | 1.31 (1.13, 1.53) |
Diabetes only | 1.00 (0.85, 1.18) | 1.18 (0.99, 1.40) |
Both conditions | 1.03 (0.88, 1.21) | 2.02 (1.72, 2.37) |
Note: AORs whose CIs do not span the value 1.0 are printed in bold.
aNumber of respondents with missing data for each outcome variable: low FV consumption, n = 4590; low PA, n = 3448.
Obesity and diabetes were jointly associated with low PA after controlling for all sociodemographic characteristics. Compared with respondents with neither obesity nor diabetes, the AOR for low PA was elevated for respondents with both conditions (AOR: 2.02, 95% CI: 1.72–2.37). The RERI was significantly elevated at 0.52 (95% CI: 0.24–0.81), indicating that the joint effect of obesity and diabetes on low PA was 0.52 higher than the risk associated with summing their separate effects. Low PA was also elevated in respondents with obesity only (AOR: 1.31, 95% CI: 1.13–1.53), but not in respondents with diabetes only. In contrast, comparison of AORs for low FV consumption across obesity–diabetes categories revealed no association; thus, the RERI was not calculated for this variable.
Discussion
Our results indicated that adjusted odds for low PA were 2.02 times higher for stroke survivors with both obesity and diabetes and 1.31 times higher for survivors with obesity only compared with survivors with neither condition. The presence of low PA among respondents with additional comorbidities mirrors findings observed among individuals with diabetes and metabolic syndrome in other studies. Sullivan et al. (14) reported that physically inactivity was 2.0 times higher in individuals with diabetes and 1.6 times higher in individuals with obesity compared with individuals with neither condition; and Ford et al. (15) reported that the prevalence of physical inactivity was 27% higher among individuals with metabolic syndrome compared with individuals without. Considering these reports, it is unsurprising that our findings demonstrated low PA among stroke survivors with obesity and diabetes. However, the high prevalence of low PA in our study sample—ranging from 56% to 71.7% across obesity–diabetes categories—and the 52% excess risk due to the interaction of obesity and diabetes suggests that different health behaviour interventions may be required for stroke survivors based on their obesity–diabetes status.
One possible explanation for the high prevalence of low PA observed in stroke survivors could be stroke-related motor impairment (e.g. hemiparesis, spasticity), which is further complicated by changes in muscle composition of the affected side (e.g. muscle wasting, fat infiltration, switch from slow-twitch to fast-twitch muscle fibres) (16). Indeed, prevalence of mobility impairment is 13% in the general adult US population but 50.8% in stroke (17,18). Stroke, therefore, increases the likelihood of functional disability, though does not guarantee it. Furthermore, obesity and diabetes are independent predictors of disability in the general population (19,20), and are associated with functional disability in stroke survivors (18). Specifically, odds of functional disability are 2.62 higher among survivors with both obesity and diabetes compared with survivors with neither condition, with respective prevalence estimates for functional disability of 70.3% and 45.4% (18).
It is also important to acknowledge that although our results yielded a significant additive interaction for obesity and diabetes with PA, causation cannot be determined due to the cross-sectional nature of the data. It is plausible that diabetes- and obesity-related disability contributes to reduced PA among stroke survivors. Alternatively, obesity and diabetes may contribute to stroke occurrence, which increases the likelihood of experiencing stroke-related disability that is associated with reduced PA. Rather than implying causation, the additive interaction observed in this study indicates that risk for low PA among stroke survivors with both obesity and diabetes is higher than the risk expected were their separate effects to be summed. Prospective studies are needed to disentangle the separate and interactive effects of stroke, obesity and diabetes on PA to inform development of tailored PA interventions for stroke survivors who experience obesity, diabetes or both conditions.
In contrast, we found no association of obesity–diabetes status with low FV consumption. In population-based studies, FV consumption is lower in individuals with chronic conditions (e.g. diabetes, metabolic syndrome) compared with individuals without chronic conditions. Ford and Mokdad (21) reported age-adjusted odds of 0.69 for consuming one to five servings of FVs among diabetics compared with non-diabetics, and Tian et al. (22) reported that adjusted odds of metabolic syndrome were 0.76 for high versus low FV consumption in a meta-analysis of nine observational studies. Given similarities in risk factors and cardiometabolic dysregulation across stroke, diabetes and the metabolic syndrome, we expected that FV consumption among stroke survivors with obesity and diabetes would mirror findings reported for individuals with diabetes and the metabolic syndrome. However, our results indicate that low FV consumption does not differ based on obesity–diabetes status in stroke survivors. The lack of difference across obesity–diabetes categories and the high prevalence (52.3%) among the total sample of stroke survivors suggests that health behaviour interventions for increasing FV consumption among stroke survivors without regard to obesity–diabetes status may be warranted.
From an intervention standpoint, our results indicate that most US stroke survivors are not adherent to PA and dietary behaviour guidelines; thus, clinical interventions aimed at improving health behaviours and managing diabetes and obesity among stroke survivors are needed. Unfortunately, primary care-based interventions for improving unhealthy behaviours among survivors are few. In a recent Australian multi-site trial, survivors were randomized to an intervention group in which a stroke coordinator facilitated discussion between survivors and general practitioners during follow-up appointments every 3 months for 1 year following hospital discharge. Compared with participants in the usual care group, participants in the intervention group reported greater mobility and less functional disability; experienced greater reductions in systolic blood pressure, total cholesterol and triglycerides; and less weight gain (i.e. weight maintenance) (23). Although this study did not address dietary behaviour, it does provide evidence of improvement in health behaviours among stroke survivors in response to primary care intervention. Observational data from the USA further support the efficacy of primary care-based intervention for encouraging healthy behaviours among stroke survivors. Data from the 1999 BRFSS survey demonstrated that a higher percentage of survivors reported engaging in exercise (76.5% versus 38.8%) and practicing healthy dietary behaviours (85.4% versus 56.0%) in response to physician advice compared with survivors who did not receive such advice (24). These studies are encouraging because they suggest that many stroke survivors respond to physician-led lifestyle management intervention. However, there is an unmet need to develop and rigorously evaluate primary care-based health behaviour interventions for stroke survivors, including individualization of advice for stroke-related disability and obesity–diabetes status.
Finally, it is important to note that several sociodemographic characteristics (i.e. sex, age, ethnicity, education and annual household income) were independently associated with the unhealthy behaviours investigated in this study. In the general US population, female sex, older age, higher education and higher income are positively associated with FV consumption; and male sex, younger age, white non-Hispanic ethnicity, higher education and higher income are associated with PA (25–29). Our results mirror these associations, confirming that the associations observed in the general population also exist among stroke survivors. One exception observed was among adults aged ≥65 years, who reported greater PA than did adults aged 45–64 years; further investigation is required to determine why younger age was more associated with low PA than lower age among stroke survivors. Despite the associations between select sociodemographic characteristics and PA observed in our study, we found an excess risk of low PA among survivors with obesity and diabetes that persisted after inclusion of the independent sociodemographic characteristics in our analyses. Thus, tailored clinical interventions focussed on increasing PA and exercise are needed specifically for this high-risk stroke subgroup in addition to those needed for stroke survivors generally.
Strengths and limitations
The present study is the first to examine the joint association of obesity and diabetes with low FV consumption and low PA among US community-dwelling stroke survivors. Data from the BRFSS national public health survey were combined across years, resulting in a large analytical sample size that allowed for stratification across obesity–diabetes status categories and to control for the potential modifying effects of sociodemographic characteristics in the statistical models.
Despite these strengths, there are several limitations inherent to the use of public health survey data. First, as previously noted, data were cross-sectional, therefore causal relationships among investigated factors could not be determined. Second, all data were self-reported and thus subject to self-report bias. Third, respondents were non-institutionalized stroke survivors; thus, data from institutionalized adults—individuals likely engaged in more unhealthy behaviours—were not included in the analysis. As a result, the estimated prevalence of unhealthy behaviours among stroke survivors in the USA may actually be higher. Finally, the BRFSS survey was not specifically intended for stroke survivors; thus, data on stroke-related motor, cognitive and psychosocial deficits were not available for exploration of how these factors might influence the reported associations among diabetes, obesity and the investigated unhealthy behaviours. Similarly, duration since stroke, and the possible additive effects of multiple strokes over time are not captured in the BRFSS dataset. Despite this limitation, the study’s findings resulted from a robust sample, and future studies should examine the moderating effect of stroke-related deficits on obesity, diabetes and lifestyle behaviours.
Conclusions
Prevalences of low FV consumption and low PA exceed 50% in US stroke survivors, and survivors with both chronic conditions experience 52% excess risk for low PA due to the dual presence of obesity and diabetes. Stroke survivors experience high risk for poor health, disability and mortality due to the presence of multiple comorbidities and unhealthy behaviours and require health behaviour intervention. Observational and clinical studies suggest that primary care-based health behaviour interventions can be effective for stroke survivors. However, prospective research is necessary to identify best practices for engaging and supporting stroke survivors in successful lifestyle intervention.
Declaration
Funding: none.
Ethical approval: not applicable.
Conflict of interest: Dr Bailey has no conflicts of interest to report. Dr Majersik is funded through NINDS as PI for UT StrokeNet, is an Associate Editor of Stroke and provides stroke consultation to Foldax, Inc. Dr Singleton has received funds from NIDDK and NINDS to explore exercise and health behaviours intervention effects on diabetic neuropathy.
Data availability
Behavioral Risk Factor Surveillance System (BRFSS) data are freely available from the Centers for Disease Control and Prevention and can be accessed online (https://www.cdc.gov/brfss/).
References
- 1.Global Burden of Disease Study 2017 (GBD 2017). Institute for Health Metrics and Evaluation. http://ghdx.healthdata.org/gbd-results-tool (accessed on 1 April 2020). [Google Scholar]
- 2.Virani SS , Alonso A, Benjamin EJet al. Heart disease and stroke statistics 2020 update: a report from the American Heart Association. Circulation 2020; 141 (9):e139–596. [DOI] [PubMed] [Google Scholar]
- 3.Feigin VL , Roth GA, Naghavi Met al. Global burden of stroke and risk factors in 188 countries, during 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet Neurol 2016; 15 (9):913–24. [DOI] [PubMed] [Google Scholar]
- 4.Kernan WN , Ovbiagele B, Black HRet al. Guidelines for the prevention of stroke in patients with stroke and transient ischemic attack. Stroke 2014; 45: 2160–236. [DOI] [PubMed] [Google Scholar]
- 5.Bailey RR , Phad A, McGrath R, Haire-Joshu D. Prevalence of five lifestyle risk factors among U.S. adults with and without stroke. Disabil Health J 2019; 12 (2):323–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hruby A , Manson JE, Qi Let al. Determinants and consequences of obesity. Am J Public Health 2016; 106 (9):1656–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.van Dam RM . The epidemiology of lifestyle and risk for type 2 diabetes. Eur J Epidemiol 2003; 18 (12):1115–25. [DOI] [PubMed] [Google Scholar]
- 8.Koroukian SM , Warner DF, Owusu C, Given CW. Multimorbidity redefined: prospective health outcomes and the cumulative effect of co-occurring conditions. Prev Chronic Dis 2015; 12: 140478. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.U.S. Department of Health and Human Services and U.S. Department of Agriculture. 2015–2020 Dietary Guidelines for Americans.8th edn. 2015.http://health.gov/dietaryguidelines/2015/guidelines/ (accessed on 24 April 2020). [Google Scholar]
- 10.U.S. Department of Health and Human Services. Physical Activity Guidelines for Americans. 2nd edn. Washington, DC: U.S. Department of Health and Human Services, 2018. [Google Scholar]
- 11.Klein RJ , Schoenborn CA. Age adjustment using the 2000 projected U.S. population. Healthy People Statistical Notes, no. 20. Hyattsville, MD: National Center for Health Statistics, January 2001. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.138.2225&rep=rep1&type=pdf (accessed on 12 September 2020). [Google Scholar]
- 12.Tryon WW . Evaluating statistical difference, equivalence, and indeterminacy using inferential confidence intervals: an integrated alternative method of conducting null hypothesis statistical tests. Psychol Methods 2001; 6 (4):371–86. [PubMed] [Google Scholar]
- 13.de Mutsert R , Jager KJ, Zoccali C, Dekker FW. The effect of joint exposures: examining the presence of interaction. Kidney Int 2009; 75 (7):677–81. [DOI] [PubMed] [Google Scholar]
- 14.Sullivan PW , Morrato EH, Ghushchyan V, Wyatt HR, Hill JO. Obesity, inactivity, and the prevalence of diabetes and diabetes-related cardiovascular comorbidities in the U.S., 2000–2002. Diabetes Care 2005; 28 (7):1599–603. [DOI] [PubMed] [Google Scholar]
- 15.Ford ES , Kohl HW III, Mokdad AH, Ajani UA. Sedentary behavior, physical activity, and the metabolic syndrome among U.S. adults. Obes Res 2005; 13 (3):608–14. [DOI] [PubMed] [Google Scholar]
- 16.Hafer-Macko CE , Ryan AS, Ivey FM, Macko RF. Skeletal muscle changes after hemiparetic stroke and potential beneficial effects of exercise intervention strategies. J Rehabil Res Dev 2008; 45 (2):261–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Courtney-Long EA , Carroll DD, Zhang QCet al. Prevalence of disability and disability type among adults—United States, 2013. MMWR Morb Mortal Wkly Rep 2015; 64 (29):777–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bailey RR , Serra MC, McGrath RP. Obesity and diabetes are jointly associated with functional disability in stroke survivors. Disabil Health J 2020; 13 (3):100914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wong E , Backholer K, Gearon Eet al. Diabetes and risk of physical disability in adults: a systematic review and meta-analysis. Lancet Diabetes Endocrinol 2013; 1 (2):106–14. [DOI] [PubMed] [Google Scholar]
- 20.Alley DE , Chang VW. The changing relationship of obesity and disability, 1988–2004. JAMA 2007; 298 (17):2020–7. [DOI] [PubMed] [Google Scholar]
- 21.Ford ES , Mokdad AH. Fruit and vegetable consumption and diabetes mellitus incidence among U.S. adults. Prev Med 2001; 32 (1):33–9. [DOI] [PubMed] [Google Scholar]
- 22.Tian Y , Su L, Wang J, Duan X, Jiang X. Fruit and vegetable consumption and risk of the metabolic syndrome: a meta-analysis. Public Health Nutr 2018; 21 (4):756–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Joubert J , Davis SM, Donnan GAet al. ICARUSS: an effective model for risk factor management in stroke survivors. Int J Stroke 2020; 15 (4):438–53. [DOI] [PubMed] [Google Scholar]
- 24.Greenlund KJ , Giles WH, Keenan NL, Croft JB, Mensah GA. Physician advice, patient actions, and health-related quality of life in secondary prevention of stroke through diet and exercise. Stroke 2002; 33 (2):565–70. [DOI] [PubMed] [Google Scholar]
- 25.Trudeau E , Kristal AR, Li S, Patterson RE. Demographic and psychosocial predictors of fruit and vegetable intakes differ: implications for dietary interventions. J Am Diet Assoc 1998; 98 (12):1412–7. [DOI] [PubMed] [Google Scholar]
- 26.Giskes K , Turrell G, Patterson C, Newman B. Socio-economic differences in fruit and vegetable consumption among Australian adolescents and adults. Public Health Nutr 2002; 5 (5):663–9. [DOI] [PubMed] [Google Scholar]
- 27.Kruger J , Yore MM, Solera M, Moeti R. Prevalence of fruit and vegetable consumption and physical activity by race/ethnicity—United States, 2005. MMWR Morb Mortal Wkly Rep 2007; 56 (13):301–4. [PubMed] [Google Scholar]
- 28.Trost SG , Owen N, Bauman AE, Sallis JF, Brown W. Correlates of adults’ participation in physical activity: review and update. Med Sci Sports Exerc 2002; 34 (12):1996–2001. [DOI] [PubMed] [Google Scholar]
- 29.Kaplan MS , Newsom JT, McFarland BH, Lu L. Demographic and psychosocial correlates of physical activity in late life. Am J Prev Med 2001; 21 (4):306–12. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Behavioral Risk Factor Surveillance System (BRFSS) data are freely available from the Centers for Disease Control and Prevention and can be accessed online (https://www.cdc.gov/brfss/).