Introduction
Stroke occurs when there is lack of blood supply to a part of the brain due to either blockage of a vessel from a blood clot (ischemic stroke) or bleeding from a ruptured vessel (hemorrhagic stroke) [
1]. It is the 5th leading cause of death in the US and more than 600,000 people in the country experience first time stroke each year [
2,
3]. It is also the leading cause of long-term serious disability and costs $34 billion each year in the US [
3].
Both prediabetes and diabetes are major risk factors of stroke. Individuals with diabetes have two times higher risk of experiencing a stroke than those that do not have the condition [
4]. Prediabetes, on the other hand, is a modest risk factor for first time stroke but doubles the risk of recurrent stroke [
5,
6]. More than 30 million people in the US have diabetes [
2]; the prevalence of the condition has doubled over the last 20 years and is projected to double or triple by 2050. Similarly, the prevalence of prediabetes in the US has been increasing. Currently, almost a third of the US adult population (88 million) have prediabetes [
7].
There is evidence that about 23–53% of stroke patients have prediabetes, while 14–46% have diabetes [
6,
8]. The increasing trends in prediabetes and diabetes prevalence will likely result in higher stroke burden in the future. Given the changes in prediabetes and diabetes landscape, the American Heart Association (AHA) and American Diabetes Association (ADA) have jointly acknowledged the need to better understand the epidemiology of stroke among individuals that have prediabetes and diabetes so as to better guide stroke prevention and control programs [
9‐
11]. However, not much is known about the epidemiology of stroke in these populations because previous studies have mainly focused on investigating the epidemiology of stroke in the general population [
12]. A study conducted in Europe investigated risk factors of stroke among individuals with diabetes, but it did not include the US population [
13]. To our knowledge, no study has investigated risk factors of stroke among individuals with prediabetes.
Florida is one of the states in the diabetes belt, an area with a higher burden of the condition than the rest of the country [
14]. The burden of diabetes has been increasing in Florida as evidenced by the fact that diabetes prevalence has increased from 5.2% in 1995 to 12.6% in 2018 [
15]. This increase will likely result in an increase of Florida’s stroke burden. Understanding stroke epidemiology among individuals with prediabetes and diabetes is important for guiding evidence-based health planning and service provision to control stroke burden and improve quality of care for stroke patients. Therefore, the objectives of this study were to estimate the prevalence and identify predictors of stroke among persons with prediabetes and diabetes in Florida.
Results
Prevalence estimates
The study included a total of 16,959 survey respondents, of whom 9.5% (1608) and 15.8% (2680) had been told by a doctor that they had prediabetes and diabetes, respectively. Table
1 shows the demographic, health, and lifestyle characteristics of respondents that had prediabetes and diabetes. Most of the respondents with prediabetes were between 45 and 65 years old (44.7%), female (54.2%), non-Hispanic White (59.6%), and married (55.0%). The majority of them had an annual income of ≥$50,000 (42.2%), some college education (35.4%), and health care coverage (80.5%). Regarding risk behaviors, respondents were mostly non-smokers (53.6%), did not have depression (74.8%), had heavy drinking habits (93.8%), and reported consuming fruits (66.0%) and vegetables (77%) at least once a day. However, a lower percentage reported being highly physically active (37.3%). Almost half of the respondents with prediabetes reported being obese (48.5%), having hypercholesterolemia (49.4%) and hypertension (53.7%). Thirty-eight percent of the respondents had arthritis, while a few had chronic kidney diseases (5.8%).
Respondents with diabetes were predominantly ≥65 years old (53%), male (52.5%), non-Hispanic White (54.7%), and married (52.7%). Thirty-one percent of the respondents had high school education and 31.9% reported earning ≥$50,000 annually. Unlike respondents with prediabetes, most of those with diabetes had hypercholesterolemia (59.7%), hypertension (72.3%), and were physically inactive (38.4%). Similar to the respondents with prediabetes, the majority of those with diabetes were obese (46.6%), non-smokers (51.0%), did not have heavy drinking habits (97.0%), had healthcare coverage (84.1%), and consumed fruits (60.1%) and vegetables (75.0%) at least once a day. The percentage of stroke among respondents that had prediabetes and diabetes was 7.8 and 11.2%, respectively.
Univariable associations
Table
2 shows the univariable (unadjusted) associations of each of the predictors with stroke among respondents with prediabetes and diabetes in Florida. The following variables had potentially significant (
p ≤ 0.20) simple/univariable associations with stroke among individuals with prediabetes: age, sex, income, depression, physical activity, vegetable consumption, hypercholesterolemia, hypertension, BMI, arthritis, and chronic kidney disease. All the above variables, except age, sex, and vegetable consumption, also had potentially significant (
p ≤ 0.20) associations with stroke among individuals with diabetes. Additionally, race and ever smoking cigarettes were significantly associated with stroke among individuals with diabetes but not among those with prediabetes.
Table 2
Univariable associations of potential predictors of stroke among adults with prediabetes and diabetes in Florida
Age | 65 years or over | 5.57 (1.52, 20.41) | < 0.001 | 1.56 (0.55, 4.40) | 0.418 |
45 to 65 years | 12.61 (3.68, 43.24) | | 1.99 (0.66, 5.97) | |
18 to 44 years | Reference | | Reference | |
Sex | Male | 1.56 (0.79, 3.10) | 0.200 | 1.32 (0.81, 2.14) | 0.266 |
Female | Reference | | Reference | |
Race | Other races (non-Hispanic) | 2.08 (0.69, 6.24) | | 0.49 (0.24, 1.04) | 0.027 |
Hispanic | 0.66 (0.21, 2.08) | 0.320 | 1.17 (0.54, 2.56) | |
Black (non-Hispanic) | 0.50 (0.12, 2.10) | | 1.87 (1.04, 3.38) | |
White (non-Hispanic) | Reference | | Reference | |
Ever smoked cigarettes | Yes | 1.42 (0.69, 2.91) | 0.340 | 1.55 (0.92, 2.63) | 0.103 |
No | Reference | | Reference | |
Heavy drinking habit | Yes | 0.56 (0.13, 2.38) | 0.430 | 0.75 (0.19, 3.00) | 0.687 |
No | Reference | | Reference | |
Income level | <$15,000 | 0.78 (0.28, 2.21) | | 2.26 (0.91, 5.62) | 0.038 |
$15,000- < $25,000 | 1.01 (0.37, 2.76) | | 1.98 (1.01, 3.89) | |
$25,000- < $35,000 | 3.64 (1.22, 10.86) | 0.069 | 2.05 (0.72, 5.87) | |
$35,000- < $50,000 | 1.87 (0.64, 5.49) | | 0.78 (0.35, 1.77) | |
≥$50,000 | Reference | | Reference | |
Education | College | 0.41 (0.12, 1.34) | | 0.71 (0.31, 1.60) | |
Some college | 0.39 (0.14, 1.12) | 0.260 | 0.65 (0.30, 1.42) | 0.751 |
High school | 0.38 (0.14, 1.04) | | 0.68 (0.31, 1.49) | |
<High school | Reference | | Reference | |
Depression | Yes | 2.22 (1.08, 4.59) | 0.030 | 2.07 (1.18, 3.63) | 0.012 |
No | Reference | | Reference | |
Level of physical activity | Inactive | 1.49 (0.64, 3.45) | | 1.22 (0.67, 2.24) | 0.183 |
Insufficiently active | 0.37 (0.10, 1.33) | 0.178 | 0.54 (0.24, 1.21) | |
Active | 0.98 (0.30, 3.23) | | 1.09 (0.38, 3.13) | |
Highly active | Reference | | Reference | |
Consume fruit(s) | ≥1 per day | 0.71 (0.34, 1.50) | 0.366 | 1.21 (0.71, 2.05) | 0.481 |
< 1 per day | Reference | | Reference | |
Consume vegetable(s) | ≥1 per day | 1.94 (0.81, 4.62) | 0.136 | 0.85 (0.45, 1.62) | 0.622 |
< 1 per day | Reference | | Reference | |
Insurance | Yes | 0.70 (0.29, 1.67) | 0.417 | 1.31 (0.37, 4.67) | 0.675 |
No | Reference | | Reference | |
Hypercholesterolemia | Yes | 5.40 (2.57, 11.35) | < 0.001 | 1.82 (1.04, 3.16) | 0.035 |
No | Reference | | Reference | |
Hypertension | Yes | 7.33 (3.64, 14.78) | < 0.001 | 3.78 (2.15, 6.67) | < 0.001 |
No | Reference | | Reference | |
Body Mass Index (BMI) (kg/m2) | Under weight (< 18.5) | 0.73 (0.15, 3.58) | | 8.08 (1.27, 51.40) | 0.072 |
Normal (18.5–24.9) | Reference | 0.181 | Reference | |
Overweight (25–29.9) | 2.02 (0.70, 5.83) | | 0.75 (0.37, 1.53) | |
Obese (≥30) | 0.94 (0.33, 2.73) | | 0.98 (0.48, 2.02) | |
Marital Status | Never married | 0.87 (0.28, 2.72) | 0.379 | 1.03 (0.46, 2.31) | 0.954 |
Separated/divorced/widowed | 1.62 (0.77, 3.42) | | 1.09 (0.63, 1.89) | |
Married | Reference | | Reference | |
Arthritis | Yes | 2.38 (1.17, 4.83) | 0.017 | 1.58 (0.95, 2.64) | 0.078 |
No | Reference | | Reference | |
Chronic kidney disease | Yes | 3.10 (0.89, 10.75) | 0.075 | 2.36 (1.14, 4.87) | 0.020 |
No | Reference | | Reference | |
Predictors of stroke among respondents with prediabetes and diabetes
The results of the final multivariable logistic regression models are presented in Table
3. Significant predictors of the odds of stroke among respondents with prediabetes were age, hypertension, and hypercholesterolemia. The odds of stroke among individuals 45–65 years old with prediabetes were 2.82 times [95% Confidence Interval (CI): 0.74, 10.69] higher than among those who were 18–44 years old while that among individuals ≥65 years with prediabetes were even higher [Odds Ratio (OR) = 4.90; 95% CI: 1.38, 17.45]. The odds of stroke among individuals with prediabetes that also had hypertension were also higher (OR = 5.86; 95% CI: 2.90, 11.84) than those among individuals that had prediabetes but not hypertension. Similarly, the odds of stroke among individuals that had prediabetes as well as hypercholesterolemia were 3.93 (95% CI: 1.84, 8.40) times higher than among individuals that had prediabetes but not hypercholesterolemia.
Table 3
Final multivariable logistic regression models showing predictors of stroke among adults with prediabetes and diabetes
Age | 65 years or over | 4.90 (1.38, 17.45) | 0.032 | 1.68 (0.56, 5.05) | 0.352 |
45 to 65 years | 2.82 (0.74, 10.69) | | 2.19 (0.69, 6.97) | |
18 to 44 years | Reference | | Reference | |
Hypertension | Yes | 5.86 (2.90, 11.84) | < 0.001 | 3.56 (1.87, 6.78) | < 0.001 |
No | Reference | | Reference | |
Hypercholesterolemia | Yes | 3.93 (1.84, 8.40) | < 0.001 | | |
No | Reference | | | |
Race | Hispanic | | | 1.23 (0.56, 2.70) | 0.025 |
Black (non-Hispanic) | | 1.79 (1.01, 3.19) | |
Other races (non-Hispanic) | | 0.42 (0.18, 0.96) | |
White (non-Hispanic) | | Reference | |
Depression | Yes | | | 2.02 (1.14, 3.59) | 0.016 |
No | | | Reference | |
None of the significant predictors of stroke among individuals with prediabetes, except hypertension, was significantly associated with stroke among individuals with diabetes. The odds of stroke among individuals with diabetes that also had hypertension were 3.56 times (95% CI: 1.87, 6.78) higher than the odds of stroke among individuals that had diabetes but normal blood pressure. In addition, non-Hispanic Black (OR = 1.79, 95%CI: 1.01, 3.19) and depressed (OR = 2.02, 95% CI: 1.14, 3.59) individuals with diabetes had higher odds of stroke compared to non-Hispanic White and non-depressed individuals with diabetes, respectively. Although age was not a significant predictor of stroke among individuals with diabetes, it was retained in the final model because of a priori knowledge that it is a confounder. It is worth noting that non-Hispanic individuals of other races (OR = 0.42, 95% CI: 0.18, 0.96) had significantly lower odds of stroke than non-Hispanic White among individuals with diabetes.
Discussion
This study investigated the prevalence and predictors of stroke among adults who reported having either prediabetes or diabetes in Florida. The percentages of stroke among adults with prediabetes (7.8%) and diabetes (11.2%) were higher than among the general population of Florida (3.6%) in 2019 [
21]. No previous studies have investigated predictors of stroke among adults with prediabetes and diabetes and yet this information is critical for guiding health programs aimed at reducing stroke burden in Florida.
The identification of hypertension as a common predictor of stroke among adults with prediabetes and diabetes in this study is consistent with the findings from previous studies [
22‐
24]. According to a report by the AHA, hypertension increases risk of stroke by weakening arteries and weakened arteries are more likely to burst or clog resulting in hemorrhagic or ischemic stroke, respectively [
24]. Evidence suggests that persons with prediabetes and diabetes have damaged blood vessels and compromised functionalities of heart and kidney due to higher than normal blood glucose levels [
25,
26]. Compromised kidney functions increase blood volume and again decrease the stretching capacity of blood vessels [
25,
26]. As a result, adults with prediabetes and diabetes are more likely to experience stroke if they also have hypertension.
Previous studies reported age as a non-modifiable risk factor of stroke among both males and females [
27‐
29]. Similar to these findings, this study identified higher odds of stroke among individuals ≥45 years old who had prediabetes [
27‐
29]. A study by Bushnell et al. reported that the risk of stroke doubles every 10 years after age 55 [
30]. The possible mechanism underlying the effect of age is that arteries naturally became narrower and harder with increasing age due to the change mediated by endothelial dysfunction and impaired cerebral autoregulation [
28]. Moreover, certain stroke risk factors such as diabetes, hypertension, atrial fibrillation, and coronary and peripheral artery diseases steadily increase with age [
28]. However, evidence also suggests that adolescents and younger people aged 15–49 years also have a high risk of stroke due to obesity and high blood pressure [
31]. In contrast to the previous findings, age was not significantly associated with stroke risks among persons with diabetes in this study [
27‐
29]. The reason for this remains unclear. However, two or more comorbidities are quite common among older individuals with diabetes, which could potentially interact with conventional cardiovascular risk factors (i.e., age) to increase the risk of stroke [
28].
Similar to the findings of this study, several studies suggested that high blood cholesterol levels increased risk of stroke [
32,
33]. Higher odds of stroke among individuals with both prediabetes and hypercholesterolemia could be explained by the changes of lipid metabolism among these populations. Adults with prediabetes have distinctive form of dyslipidemia characterized by low levels of High Density Lipoprotein (HDL)-cholesterol and moderately elevated levels of Triglyceride (TG)-rich lipoprotein [
34]. Dysmetabolism of TG-rich lipoprotein increases the level of smaller and denser Low-Density Lipoprotein (LDL) particles. Overall, non-HDL cholesterol levels, including TG and LDL, almost always increase among adults with prediabetes. Moreover, increased non-HDL cholesterol levels due to lipid dysmetabolism and weakened blood vessels due to hyperglycemia among adults with prediabetes increase the risk of atherosclerosis and ischemic stroke [
35]. Surprisingly, hypercholesterolemia was not a significant predictor of stroke among adults with diabetes in this study. This is possibly due to the fact that anti-diabetic medications such as sulfonylurea and insulin can control hypercholesterolemia and, to some extent, reduce the risk of developing stroke [
36,
37].
Risks of stroke among populations with diabetes vary by race. The higher odds of stroke among non-Hispanic Black compared to non-Hispanic White, identified in this study, is consistent with reports from previous studies [
38‐
41]. There is evidence that non-Hispanic Black populations have high risks of stroke because they are more likely to have hypertension and diabetes [
39‐
41]. However, another study by Heyman et al. suggested that even after adjusting for hypertension and diabetes, non-Hispanic Black individuals had consistently higher risk of stroke than non-Hispanic White [
38,
42]. Studies have reported that only half of the excess risk of stroke among non-Hispanic Black could be attributed to traditional risk factors (such as poor diet, obesity, and high salt diet), implying that genetic and biological factors might have potential roles in stroke disparities among non-Hispanic Black population [
43‐
45]. Additionally, non-Hispanic Black adults with diabetes often do not have access to healthcare due to low socioeconomic conditions and tend to have uncontrolled diabetes [
46]. The presence of inherent excess risk of stroke and uncontrolled diabetes may be responsible for higher odds of stroke among non-Hispanic Black individuals with diabetes compared to non-Hispanic White individuals with diabetes. Although Hispanic individuals had seemingly higher odds of stroke than non-Hispanic White individuals, this association was not statistically significant. A study by Rodriguez et al. also reported that age-adjusted prevalence of stroke among Hispanic individuals ≥18 years were similar to stroke prevalence among their non-Hispanic White counterparts [
47]. However, this relationship may vary by geographic region as several studies, conducted in other US states, reported a significantly higher risk of stroke among Hispanic individuals compared to their non-Hispanic White counterparts. Similarly, the risk of stroke among non-Hispanic other races, including Asian and American Indian/Alaskan Native, also vary by geographic location [
48]. This study identified significantly lower risk of stroke among non-Hispanic other races than non-Hispanic White individuals in Florida, while studies in other US states reported the opposite [
49‐
51]. However, it is worth pointing out that non-Hispanic other races represent a small portion of the Florida population. Overall, reasons for identified disparities in stroke risks among minority populations could be genetic and higher prevalence of traditional risk factors such as diabetes, hypertension, low socioeconomic status, and healthcare system challenges [
52]. Surprisingly, race was not a significant predictor of stroke among adults with prediabetes. The reason for this is not apparent but may be due to the fact that other factors such as age, hypertension, hypercholesterolemia are more important predictors of stroke among these populations.
Based on the findings from this study, adults that had both diabetes and depression had two times higher odds of stroke compared to those that had diabetes but no depression and this finding is consistent with those from a meta-analysis of 17 epidemiological prospective studies showing significant positive associations between depression and stroke even after adjusting for diabetes, hypertension, and other risk factors [
53]. Individuals experiencing depression tend to have unhealthy lifestyles, get less exercise, often times do smoke, and are more likely to miss prescribed medication [
54]. Other possible mechanisms linking depression to stroke could be inflammation, atherosclerosis, lesions in cerebral white matter, cardiac arrhythmia, and increased platelet activity [
53,
54].
Since estimates obtained from the BRFSS data are representative of the adult population of Florida, they are generalizable to Florida adults. Thus, the findings from this study are useful for guiding health planning for the state of Florida. However, Florida population could also be representative of similar populations in other high- and middle-income countries with similar socioeconomic structure and quality of life. Moreover, identified predictors of stroke in this study such as hypertension and hypercholesterolemia are common comorbidities among populations all over the world. Therefore, the findings of this study could provide guidance to conduct similar studies among populations of other high- or middle-income countries. However, it is worth noting that differences in healthcare systems, environment, and inherent biological factors among populations of different races and ethnicities in different countries could result in the identification of different sets of predictors of stroke.
Strengths and limitations
To our knowledge, this is the first study that estimated the prevalence and investigated predictors of stroke among populations with prediabetes. The findings are critical for reducing stroke burden considering the fact that populations with prediabetes represent 1/3 of the US adult population. This is also the first study investigating predictors of stroke among populations with diabetes in Florida. Identifying populations that have prediabetes or diabetes with a high risk of stroke will help enhance evidence-based programs targeting those populations in Florida. These findings are important as the Florida Department of Health seeks to implement the new Paul Coverdell National Acute Stroke Program which aims to improve the quality of care for stroke patients. A strength of the BRFSS data is that it is based on statistical survey sampling methodology and therefore provide representative estimates that are generalizable to the population. However, this study is not without limitations. First, since the BRFSS was based on self-reported responses, prevalence estimates calculated from this survey could be affected by reporting or recall biases. For instance, it is possible that people who had not suffered a severely disabling stroke might forget to mention having a stroke (recall bias). However, these will bias the estimates of associations towards the null, implying that the true associations are even stronger than the results shown in this study. Additionally, since the study used secondary data, there was no way to cross-check medical records with self-reported responses in the BRFSS survey. However, studies have shown that BRFSS data are accurate even after considering reporting bias [
55,
56]. Secondly, rates of detection of prediabetes or diabetes might be higher among individuals of higher socioeconomic status than those in low status. Therefore, more affluent individuals might be overrepresented among those reporting prediabetes and diabetes [
57]. In addition, the BRFSS did not collect data on: (a) the duration of prediabetes and diabetes, (b) medications for management of stroke risk factors (i.e., hypertension and hypercholesterolemia), and (c) effectiveness of control of blood sugar, blood pressure, and cholesterol among populations with prediabetes and diabetes. Therefore, the impacts of these on the risk of stroke could not be assessed. Finally, the BRFSS survey did not gather information on types of stroke and hence stroke risks could not be investigated based on types of stroke. These limitations notwithstanding, the findings of this study provide useful information to guide health planning and programs aimed at reducing stroke burden in Florida.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.