Association between Body Mass Index and Stroke Risk Among Patients with Type 2 Diabetes Free

Abstract

Context

Very few studies focused on the association between body mass index (BMI) and stroke risk among patients with diabetes.

Objective

We aimed to investigate the association between BMI and stroke risk in patients with type 2 diabetes.

Design

Demographic, anthropometric, laboratory, and medication information were extracted from the National Patient-Centered Clinical Research Network common data model.

Participants

We performed a retrospective cohort study of 67 086 patients with type 2 diabetes.

Main Outcome Measures

Incident stroke including both ischemic and hemorrhagic stroke were defined.

Results

During a mean follow up of 3.74 years. 8918 incident stroke events occurred. Multivariable-adjusted hazard ratios across different categories of BMI at baseline (18.5–24.9 [reference group], 25.0–29.9, 30.0–34.9, 35.0–39.9, and ≥40 kg/m²) were 1.00, 0.92, 0.85, 0.74, and 0.63 (P_trend <0.001) for total stroke; 1.00, 0.93, 0.88, 0.77, and 0.65 (P_trend <0.001) for ischemic stroke; and 1.00, 0.79, 0.50, 0.50, and 0.41 (P_trend <0.001) for hemorrhagic stroke, respectively. When we used an updated mean value of BMI, the graded inverse association of body mass index with stroke risk did not change. This linear association was consistent among patients of different subgroups. Further sensitivity analysis excluding patients who were diagnosed stroke within 6 months after first diagnosis of type 2 diabetes or including non-smokers only also confirmed our findings.

Conclusion

The present study found an inverse association between BMI and the risk of total, ischemic, and hemorrhagic stroke among patients with type 2 diabetes. More clinical and molecular insights are still needed in explaining these findings.

Obesity and diabetes are 2 major epidemics in the United States, especially in Louisiana (1). Both conditions have been shown to be associated with an increased risk of developing cardiovascular diseases. Notably, diabetes is known to be an independent risk factor of cardiovascular diseases and related mortality, while for obesity the strength of the association remains controversial. Results demonstrating a positive association (2, 3), a U-shaped association (4, 5), or no association (6) of body mass index (BMI) with cardiovascular mortality or all-cause mortality have been reported in several studies.

Stroke, as one of the most threatening cardiovascular conditions, ranked 5th among leading causes of death in 2016 in the United States (7). However, studies with stroke as the primary outcome are limited. Among these studies, an obesity paradox was noted in some of them such that people with higher BMI were less likely to suffer stroke and die compared with those who with lower BMI (8, 9). However, results from some studies (10, 11) did not support this obesity paradox. Most studies have drawn their conclusions from studies in the general population, while very few studies focused on the association among patients with type 2 diabetes. Given the heavy burden and severe consequences of stroke, it is important to determine the association between obesity and stroke in this specific population. Therefore, we aimed to investigate the association between BMI and the risk of stroke among patients with type 2 diabetes in a large healthcare system-based study.

Materials and Methods

Study participants

Data on patients with type 2 diabetes in the Louisiana Experiment Assessing Diabetes outcomes (LEAD) cohort study were obtained through the Research Action for Health Network (REACHnet) (12, 13). The data set included electronic health record data for the study cohort between January 1, 2013 and July 31, 2018. For the present study, data from 3 REACHnet partner health systems were included in the final pooled analysis. A unique global identifier was used to link records across the 3 health systems to avoid duplication of individual patients in the pooled dataset. In total, 18 706 patients (out of 203 701 records) were identified as duplicates across the 3 partner health systems.

The definition of type 2 diabetes in the present study was formulated according to the SUPREME-DM (14) criteria as follows: (a) 1 or more of the International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) codes and Tenth Revision, Clinical Modification (ICD-10-CM) codes for type 2 diabetes associated with in-patient encounters; (b) 2 or more ICD codes associated with outpatient encounters on different days within 2 years; (c) combination of 2 or more of the following associated with outpatient encounters on different days within 2 years: (i) ICD codes associated with outpatient encounters; (ii) fasting glucose level ≥126 mg/dL; (iii) 2-hour glucose level ≥200 mg/dL; (iv) random glucose ≥200 mg/dL; (v) hemoglobin A1C (HbA1c) ≥ 6.5%; and (vi) prescription for an antidiabetic medication. A total of 107 562 patients between the ages of 30 and 94 years were identified. After the exclusion of patients with incomplete data, the present study included 67 544 patients with type 2 diabetes (40 431 whites and 27 113 African Americans). Compared with patients with type 2 diabetes excluded from the present study, the patients included had similar ages (66.5 ± 12.1 versus 66.3 ± 12.5 years of age) with more African Americans than whites (40.1% vs. 36.2%) and slightly fewer men than women (47.5% vs.49.1%). Of the 67 544 patients with type 2 diabetes, 458 with BMI lower than 18.5kg/m² were excluded because these patients may have other severe comorbidities. The study and analysis plan were approved by the Pennington Biomedical Research Center (2016-064-PBRC), Tulane University (906810) and Ochsner Health System Institutional Review Boards (Ochsner acknowledged Tulane’s approval). We used an electronic data set compiled from medical records but not containing personally identifiable information except for the date of birth; thus, we did not obtain written informed consent from patients in this observational study cohort.

Baseline measurements

The National Patient-Centered Clinical Research Network (PCORnet) common data model is a specification that defines a standard organization and representation of data for the PCORnet distributed research network (15). Patients’ data extracted from this common data model for the present study included date of birth, age at diabetes diagnosis, race/ethnicity, sex, encounter dates, weight, height, BMI, blood pressure, tobacco use, diagnoses of various diseases and dates of diagnoses, laboratory test dates, total cholesterol, triglycerides, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, glycosylated hemoglobin (HbA1c), estimated glomerular filtration rate (eGFR), and medication prescriptions such as antihypertensive drugs, glucose-lowering drugs, and lipid-lowering drugs. These data elements were collected starting from the date of diabetes diagnosis (baseline). Using smoking status reported at each clinical visit, we classified the patients into 3 groups: current smokers, ever smokers, and never smokers. The eGFR was estimated using the Modification of Diet in Renal Disease (16).

Follow-up

We created the follow-up database in electronic form by using the number assigned to each patient who visited the health system with a unique patient identifier. The updated mean value of BMI was calculated for each participant from baseline to each year of follow-up. For example, after 1 year, the updated mean was the average of the baseline and 1-year values, and after 3 years, it was the average of baseline, 1-, 2-, and 3-year values. In the case of an event occurring during follow-up, the period for estimating the updated mean value was from baseline to the year before the event occurred. The average number of BMI measurements per patient during the follow-up period was 20.3. Stroke (ischemic or hemorrhagic) was the primary outcome in the present analysis. ICD-9-CM and ICD-10-CM codes were used to identify hemorrhagic stroke (ICD-9-CM codes 430–432 and ICD-10-CM codes I60 –I62), ischemic stroke (ICD-9-CM codes 433–436; ICD-10-CM codes I63-I66), and any stroke (ICD-9-CM codes 430–436; ICD-10-CM codes I60-I66) events. The distributions of all ICD-9 and ICD-10 codes were 430 (0.4%), 431 (1.0%), 432 (1.3%), 433 (26.8%), 434 (19.0%), 435 (8.0%), 436 (0.9%), I60 (0.6%), I61 (1.0%), I62 (1.6%), I63 (17.2%), I65 (22.2%), and I66 (0.0%). These diagnoses were recorded in the course of routine patient care by the patients’ treating clinicians. The duration of follow-up for each cohort member (person-years) was tabulated from the date of the first documented diabetes diagnosis to the date of diagnosis of the outcome, death of inpatients or July 31, 2018. Diagnosis of stroke events could be made in outpatient, inpatient, or emergency encounters. Encounter types including ambulatory visits were considered as outpatient encounters, while encounter types including inpatient, emergency department, emergency admission to inpatient, institutional stay, observation stay, and institutional consult were considered as either inpatient or emergency encounters.

Statistical analyses

Cox proportional hazards regression was used to estimate hazard ratios (HRs) for incident stroke according to different levels of BMI. BMI was evaluated in the following 2 ways: (1) as categories (18.5–24.9 [reference group], 25.0–29.9, 30.034.9, 35.0–39.9, and ≥40.0 kg/m²) and (2) as a continuous variable. BMI data were included in the models as a dummy variable, and the significance of the trend across categories of BMI was tested in the same models by giving an ordinal numeric value for each dummy variable. The proportional hazards assumption in the Cox model was assessed with graphical methods and with models including time-by-covariate interactions. In general, all proportionality assumptions were appropriate. All analyses were carried out in 2 models. Model 1 adjusted for age, sex, and race; Model 2 adjusted for variables of age, sex, race, health insurance, smoking, systolic blood pressure, HbA1c, low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, eGFR, use of antihypertensive drugs, use of thiazolidinediones, use of lipid-lowering agents, and use of antiplatelet or anticoagulant agents. Sensitivity analyses were performed excluding patients diagnosed stroke within 6 months after first diagnosis of type 2 diabetes and further excluding past or current smokers. Statistical significance was considered to be P < 0.05. All statistical analyses were performed by using IBM SPSS Statistics for Windows, version 24.0 (IBM Corp., Armonk, NY, USA).

Results

There are 67 086 patients with type 2 diabetes included into the final analysis. Baseline characteristics of the study population are listed in Table 1. African American patients with type 2 diabetes had higher BMI. Patients with higher BMI were younger and had higher blood pressure, HbA1c level, and lipid panels except HDL cholesterol. They also had better renal function, were less likely to be smokers, and were more likely to be users of lipid-lowering, antihypertensive, and glucose-lowering medications including thiazolidinediones. Patients with higher BMI used antiplatelet agents less frequently.

During a mean follow-up of 3.74 ± 1.69 years, a total of 8918 incident stroke events were recorded including 8375 ischemic cases and 543 hemorrhagic cases. Multivariable-adjusted HRs across different categories of BMI at baseline (18.5–24.9 [reference group], 25.0–29.9, 30.0–34.9, 35.0–39.9, and ≥40.0 kg/m²) were 1.00, 0.92, 0.85, 0.74, and 0.63 (P_trend <0.001) for total stroke (Table 2); 1.00, 0.93, 0.88, 0.77, and 0.65 (P_trend <0.001) for ischemic stroke (Table 3); and 1.00, 0.79, 0.50, 0.50, and 0.41 (P_trend <0.001) for hemorrhagic stroke (Table 3). When we used an updated mean value of BMI, the graded inverse association of BMI with the risk of stroke and its subtypes did not change.

When BMI at baseline was considered as a continuous variable, the multivariable-adjusted HRs for each 1 kg/m² increase in BMI were 0.980 (95% confidence interval [CI] 0.977–0.983) for total stroke, 0.957 (95% CI 0.943–0.971) for ischemic stroke, and 0.981 (95% CI 0.978–0.985) for hemorrhagic stroke, respectively (Tables 2 and 3). Similar results could be found when we used an updated mean value of BMI. In addition, age, systolic blood pressure, HbA1c, and LDL cholesterol at baseline were positively associated with the risk of total stroke, while HDL cholesterol and eGFR were inversely associated with the risk of total stroke. White women seemed to have the lowest risk of total stroke among different sex and race combinations. Notably, patients using thiazolidinediones had a lower risk of total stroke than nonusers, while patients with other medications had a higher risk of total stroke than non-users.

When stratified analyses were utilized, the graded inverse association between BMI and the risk of total stroke was consistent among patients of different ages, races, sexes, HbA1c, eGFR, never smokers, and past or current smokers (Table 4). This inverse association was also observed among patients taking antidiabetic, lipid-lowering, antihypertensive, antiplatelet, and anticoagulant medications or not with a P-value for interaction <0.01 for lipid-lowering, antihypertensive, and antiplatelet subgroups.

In sensitivity analysis 1 (Table 5), we excluded patients who were diagnosed with stroke within 6 months after their first diagnosis of type 2 diabetes. The overall number of incident stroke events reduced to 6332 including 5908 ischemic cases and 424 hemorrhagic cases. Multivariable-adjusted HRs across categories of BMI at baseline (18.5–24.9, 25.0–29.9, 30.034.9, 35.0–39.9, and ≥40.0 kg/m²) were 1.00, 0.95, 0.90, 0.80, and 0.70 (P_trend <0.001) for total stroke (Table 5). When we used an updated mean value of BMI, the graded inverse association of BMI with the risk of still did not change. In addition, when patients with past or current smoking status were also excluded (Table 6) in sensitivity analysis 2, similar results were found.

Discussion

In this large retrospective cohort using electronic medical records from 3 healthcare systems, a graded inverse association of BMI with total stroke risk as well as its subtypes was found. Our findings were consistent with the “obesity paradox” and thus extended in a specific population with type 2 diabetes where patients with higher BMI both at baseline and during follow-up would have a lower risk for future stroke risks, including both ischemic stroke and hemorrhagic stroke.

Obesity is a demonstrated risk factor for both cardiovascular (coronary artery disease and stroke) and noncardiovascular (obstructive sleep apnea, osteoarthritis, etc.) comorbidities (17). However, several studies have shown that obese patients may have a lower risk of all-cause or cardiovascular mortality compared with those with normal BMI, which has been termed the “obesity paradox” (18, 19). It has also been shown among patients with heart failure that patients with higher BMI had a lower risk of all-cause mortality (20). Narrowly speaking, the obesity paradox was only applicable in predicting the prognosis of patients who had already had an onset of coronary artery disease or heart failure. There is some selection bias in the analysis of associations between BMI and outcomes. Explanations may be that patients will always have a great weight loss when they suffer several severe conditions or comorbidities. These patients tend to be older, frailer, and more likely to have complications, which might result in higher risk of mortality (21). In addition, smoking status is an important confounding factor during the analysis (22). Current smokers tend to have lower BMI, and smoking is closely associated with cardiovascular diseases. To further confirm the results considering the impact of these confounding factors, we performed sensitivity analysis excluding patients who were diagnosed with stroke within 6 months after first diagnosis of type 2 diabetes, as well as patients with past or current smoking status. Similarly, the observed graded inverse association did not change. Patients with higher BMI still had a lower risk of incident stroke events.

Very few studies have analyzed the association between BMI and stroke among patients with type 2 diabetes. In a previous paper by our group among underinsured patients with type 2 diabetes, we found an inverse association of BMI with the risks of total, ischemic, and hemorrhagic stroke.(8) Another Japanese cohort found that low BMI was a risk factor for all-stroke and cerebral infarction in men, while high BMI was a risk factor for all-stroke incidence in women (11). However, we did not detect this sex difference in our analysis. For intracerebral hemorrhage, a national inpatient sample analysis also found an obesity paradox such that obesity and morbid obesity appeared to protect against mortality (23). Another Australian national study focusing on the association between BMI and fatal hemorrhagic stroke in young adults, however, did not support this obesity paradox as overweight and obese cases were prominent among young fatalities of stroke (24). The study populations also varied among these studies and may contribute to the different results. For ischemic stroke, a Chinese cohort including only newly diagnosed type 2 diabetes patients found a lower ischemic stroke risk among patients with increasing BMI categories (9). Inconsistent with their study, a Swedish cohort reported a higher risk of nonfatal total stroke with each increase of 5 kg/m² in BMI (HR, 1.10 [1.2–1.09], P = 0.019) among patients with type 2 diabetes (10). The average age of patients in the Chinese cohort are similar with those in our cohort, while it is older than those in the Swedish cohort (average age around 65 years old vs. 60 years old). Despite findings from these studies, the underlying mechanism of the obesity paradox is still unclear. In our analysis, patients with severe obesity were much younger than those with normal weight (mean age 60.2 vs. 73.0 years old). Meanwhile, patients with severe obesity were more likely to use lipid-lowering agents including statins to regulate their lipid panels, which may help improve the outcomes. These may be 2 possible explanations for our findings. A recently published study also found the obesity paradox for mortality and functional outcome in insulin-resistant patients but not in insulin-sensitive patients. Insulin resistance may be one of the mechanisms underlying the obesity paradox of the outcome in patients with ischemic stroke (25). This paper could strongly support our findings on the lower risk of stroke among obese patients with type 2 diabetes. However, more clinical and molecular insights are still needed in explaining the obesity paradox. More prospective studies are warranted to reveal the exact causal relationship between BMI and stroke, as well as the benefit of weight loss among these patients.

A major strength of this study was the large sample size, which allowed for high statistical power and the ability to perform stratified analyses. Further, the relatively rich clinical data and numerous events also make the results robust. The data we used were derived from administrative databases, avoiding the problem of differential recall bias. Data in this study were extracted from 3 partners of REACHnet, which minimizes the influence of low accessibility of health care. There are also several limitations in this study. First, some socioeconomic variables were missing in the EMR data including education level, family income, etc. Second, the stroke diagnoses in the present study were based on physician diagnosis, and no chart review was performed. However, most American and European cohort studies, such as the Framingham Study (26), the Kaiser Permanente Medical Care Program (27), and the Atherosclerosis Risk in Communities Study (28) shared the same method used in our study to diagnose stroke events. Finally, our analyses adjusted for some confounding factors; however, unmeasured factors such as family history of diabetes, other related chronic diseases, dietary factors and physical activity status could not be evaluated.

In conclusion, based on this large healthcare system based cohort study, we were able to demonstrate a graded inverse association between BMI and risks of total stroke as well as its subtypes among patients with type 2 diabetes. More prospective studies are warranted to reveal the exact causal relationship between BMI and stroke, as well as the benefit of weight loss among these patients.

Additional Information

Disclosure Summary: The authors have no conflict of interest to disclose.

Data Availability: Restrictions apply to the availability of data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will, on request, detail the restrictions and any conditions under which access to some data may be provided.

Acknowledgments

The LEAD Study would like to acknowledge the contributions of our partners. The success of this study depended on their ongoing support and expertise. These partners include Ochsner Health System and the Ochsner Patient Research Advisory Board; Tulane Medical Center; University Medical Center New Orleans; Research Action for Health Network (REACHnet, a PCORnet CDRN) and their multi-stakeholder Diabetes Advisory Groups; Pennington Biomedical Research Center; Blue Cross and Blue Shield of Louisiana; and our patient and community partners Patricia Dominick, Catherine Glover, and Peggy Malone.

Financial Support: This work was supported by a Patient-Centered Outcomes Research Institute (PCORI) cooperative agreement (NEN-1508–32257) as part of Natural Experiments for Translation in Diabetes 2.0 (NEXT-D2). All statements in this manuscript, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of PCORI, its Board of Governors, or Methodology Committee. Drs. Shen, Hu and Katzmarzyk were partly supported by a grant from the National Institute of General Medical Sciences (U54GM104940) of the National Institutes of Health.

Author Contributions: YS and GH wrote the manuscript. LS, EN, PTK, EPH, ANB, and SN reviewed and edited the manuscript. GH is the guarantor of this work and, as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.

References