Association between dynamic obesity and mortality in patients with first-ever ischemic stroke: A hospital-based prospective study.

Although obesity is an established risk factor of primary stroke, the association between obesity and post-stroke mortality remains unclear. The aim of this study was to investigate the association between dynamic obesity status and mortality in survivors of their first stroke in China.Of 775 patients with first-ever ischemic stroke included in a longitudinal study, 754 patients were included in this study and categorized into 4 categories of body mass index (BMI) (underweight, normal weight, overweight, and obese) and 2 categories of waist circumference (WC) (normal WC and abdominal obesity) according to standard Chinese criteria. The mortality information and obesity status were obtained via telephone follow-up every 3 months, beginning in 2010 through 2016. Time-dependent Cox proportional hazards models were used to estimate the unadjusted and adjusted hazard ratios (HRs) for the relationship between all-cause mortality and dynamic obesity status.Of 754 patients, 60.87% were male, and the overall mean age was 61.45 years. After adjusting for possible confounders, significant inverse associations were identified between BMI and WC and all-cause mortality. Compared with those with normal BMI or WC, those with abdominal obesity or overweight had a significantly lower risk of all-cause mortality (HR and 95% confidence intervals [CIs]: .521 [.303-.897] and 0.545 [.352-.845], respectively), whereas patients with underweight had the highest risk and those with obesity had lower risk of mortality, though it was not statistically significant (1.241 [.691-2.226] and .486 [.192-1.231], respectively).Overweight and abdominal obesity were paradoxically associated with reduced risk of mortality in patients who survived their first-ever ischemic stroke in China. Future prospective studies must look at evaluating the role of obesity in different stroke subtypes and devise appropriate weight-management strategies for optimal prognoses in secondary prevention in these survivors.

Introduction

Obesity has become a global epidemic and a major public health concern.[ The prevalences of overweight and obesity in adults older than 18 years of age in 2016 were approximately 39% and 13% worldwide, respectively.[ Epidemiological studies have demonstrated that obesity is a risk factor for hypertension,[ diabetes mellitus (DM),[ atrial fibrillation,[ certain cancers,[ major cardiovascular diseases,[ stroke,[ and total mortality in the general population. Accordingly, the general agreement is to recommend striving for a normal weight as part of primary prevention of stroke.[ However, the latest guidelines for secondary prevention suggest that the beneficial effects of reducing weight remain uncertain.[ Furthermore, the association of obesity with mortality in patients with stroke remains controversial. Some findings suggest that obesity worsens the prognosis in survivors of stroke,[ while others suggest that obesity confers better survival.[ Numerous findings indicate that, in contrast to the general population, an inverse relationship is present between obesity and mortality in survivors of stroke, which is termed as obesity paradox.[ Andersen et al[ analyzed 45,615 patients with acute first-ever stroke (irrespective of the subtype of stroke), and concluded that obesity was not only associated with reduced mortality but also with lower risk of readmission for recurrent stroke relative to patients with normal weight. In the Telemedical Project for Integrative Stroke Care (TEMPiS) study that included data of 4,428 patients with acute stroke or transient ischemic attack, the highest mortality was observed in underweight patients.[ Therefore, patients with overweight and obesity demonstrated significantly better survival than did patients with normal body weight. Therefore, it is unclear if the recommendations of primary prevention should be extended to secondary prevention of cardiovascular diseases as well.

Previous studies on the paradox between obesity and stroke outcomes have limitations. Therefore, the results are yet to be confirmed. It is well known, however, that the assumption that obesity status remains constant during follow-up is unreasonable, which may result in misclassification of the obesity status and possibly biased results. Furthermore, it is noteworthy that numerous studies on the obesity paradox fail to fully consider the confounders associated with all-cause mortality in post-stroke patients—e.g., the stroke severity may not have been quantified and the treatment of stroke was ignored[ —which may provide a false impression that obesity constitutes a survival advantage.

Therefore, the question remains if the phenomena that obesity is associated with better prognosis after stroke is true or is it biased by imperfect adjustments for confounders and research methodology? Is maintaining normal weight still advisable if being overweight or obese confers more favorable prognoses? To our best of knowledge, few studies have simultaneously used body mass index (BMI) and waist circumference (WC) to describe the obesity status in patients with stroke from the general and abdominal obesity aspects, and few studies have considered the dynamic obesity status. Furthermore, limited studies have explored the relationship between obesity and post-stroke mortality in the Chinese population, a subgroup with high prevalence rates of obesity and stroke.[ Therefore, the aim of our study was to investigate the association between dynamic obesity status and mortality in survivors of first-ever ischemic stroke in China.

Patients and methods

Study population

The study was performed in the Department of Neurology, West China Hospital. The diagnosis of stroke was established according to the World Health Organization criteria.[ Survivors of first-ever ischemic stroke were screened by computed tomography or magnetic resonance imaging. Patients who underwent cardiovascular surgeries, such as carotid endarterectomy, cardiac surgeries, or angiography, which may induce stroke, were excluded.

Ethics statements

Written informed consents were obtained from all patients or their designated relatives. The study protocol was approved by the Ethics Committee of the West China Hospital, Sichuan University, Chengdu, China (ethical approval number, “2009 year 50”).

Study design

A semi-structured questionnaire was used to conduct face-to-face interviews in the hospital by professionally trained, qualified, and experienced interviewers. The interviewers consisted of students pursuing Masters in Public Health who studied clinical medicine for 5 years as an undergraduate and had sufficient clinical expertise. The face-to-face interview was supplemented by medical records. The demographic characteristics included sex, age (<45, 45–64, and >65 years), marital status (with spouse, without spouse), occupation (farmer, non-agricultural laborer, retired, and unemployed), education (≤6, 7–9, and ≥10 years), household income (<$3057.4, $3057.4–6114.8, and >$6114.8 per year), and number of family members (<3, 3–4, and >4 persons). Lifestyle parameters included smoking status (never smoked, cessation of smoking, currently smoking, and passively smoking), drinking status (never or current drinking), and pre-stroke exercise duration (<210, 210–420, and >420 minutes per week). Vascular risk factors included family history of stroke, hypertension, DM, hyperlipidemia, heart disease, and peripheral vascular diseases. The aforementioned comorbidities were diagnosed as per the current standards.[

Measure of obesity

Most patients were unable to stand on admission; therefore, it was difficult to directly measure their height, weight, and WC. At baseline, the participants were asked to report their weight, height, and WC by answering the following questions: “what is your weight now?” (in kg to the nearest .1 kg) and “what is your height and waist circumference?” (in cm to the nearest .1 cm).

According to the distribution of fat, obesity is divided into general obesity (i.e., obesity) and abdominal obesity.[ Obesity is assessed according to BMI, and abdominal obesity is commonly assessed according to WC. BMI is calculated by dividing the body weight (kg) by height squared (meters). WC refers to the circumference at a point midway between the lowest rib and the iliac crest in the horizontal plane.

Overweight, general obesity, and abdominal obesity were defined by the standard criteria for the Chinese population (underweight: BMI <18.5 kg/m2; normal weight: BMI, 18.5 to <24 kg/m2; overweight: BMI, 24–28 kg/m2; general obesity: BMI >28 kg/m2[; abdominal obesity: WC ≥85 cm for men and ≥80 cm for women[).

Outcomes and follow-up

Follow-up was routinely performed via telephone interviews every 3 months by trained and qualified investigators after hospital discharge. Before formal participation in the survey, the project organizer held a training meeting for investigators about standardization of the survey language and matters that required attention. The follow-up survey included information regarding the severity of stroke, body weight, WC, adherence to anti-thrombotic medications, and health outcomes. Most patients were in the acute phase at the time of the baseline survey, and stroke severity changes quickly; thus, the modified Rankin Scale (mRS) scores at the time of the baseline survey may not accurately reflect the stroke severity.[ Therefore, the mRS score at the first follow-up was deemed to indicate the stroke severity instead. mRS score <2 was defined as mild disability, whereas mRS score ≥3 was defined as severe/moderate disability.[ Strict adherence of patients to anti-thrombotic medications or to alterations of the pharmacotherapy according to the doctors advice during the follow-up period was termed as adherence. The endpoint of the study was all-cause mortality, which was obtained from the patients relatives during follow-ups. Survival time was calculated from the baseline survey until death due to any cause, withdrawal from the study, or the last follow-up since June 2016.

Statistical analysis

Descriptive statistics of the baseline individual characteristics were stratified based on the baseline obesity status as assessed by BMI and WC using the Chi-Squared test or Fisher exact test. We used change in obesity status, which was updated every 3 months, as a time-varying exposure, and a time-dependent Cox proportional hazard regression model[ was used to estimate the hazards ratio (HR) for all-cause mortality during the follow-up. In this study, both empirical and theoretical strategies were considered for the chosen confounding factors.[ Statistical tests were two-sided, and a P value <.05 was considered statistically significant. All statistical analyses were performed using SAS, version 9.3 (SAS Institute Inc., Cary, NC, USA).

Results

Between January 2010 and June 2016, 775 participants were enrolled. After excluding 21 patients (2.7%) because of duplication and missing data, 754 patients were included in our analysis. The overall mean age was 61.45 years (range, 17–89 years) and 60.88% were male. The included patients were followed up for a mean period of 54.86 months (range, 0–75 months). The baseline demographic and clinical characteristics are summarized in Table 1.

Table 1

Epidemiologic characteristics, co-morbidities, and stroke severity by BMI and WC subgroup (at admission).

At admission, 33.69% of patients were categorized as overweight, 10.08% as obese, and 45.62% to have abdominal obesity; the survival curve differed significantly between the subgroups (Figs. 1 and 2). Results of the univariate analysis are summarized in Table 2.

Figure 1

Survival curves following stroke according to groups of body mass index (BMI; univariate analysis).

Figure 2

Survival curves following stroke according to groups of waist circumference (WC; univariate analysis).

Table 2

Relationship between BMI, WC and related factors with all-cause mortality in unadjusted analysis.

Results of the multivariable-adjusted time-dependent Cox proportional hazards regression analysis of the BMI and WC subgroups are summarized in Tables 3 and 4, respectively. When adjusted for only age, overweight was associated with lower mortality compared with the other BMI categories (HR: .491, 95% confidence interval [CI]: .293–.824); in the univariate analysis, the severity of stroke was strongly associated with post-stroke mortality. After age and stroke severity were simultaneously adjusted, the results of overweight differed slightly (HR: .506; 95% CI: .302–.850). Similarly, after simultaneous adjustments for all socio-demographic and other factors presented in Table 2, an inverse relationship between BMI and all-cause mortality was constantly observed in those who were overweight (HR: .521; 95% CI: .303–.897). Particularly, those were underweight had the worst survival constantly in all the adjusted models; however, they were not statistically significant. Notably, in those who were obese, the statistical significance was lost after adjustments were made for any of the listed confounders. After simultaneous adjustments were made for all of the aforementioned factors, the significant association between mortality and obesity was still not observed (HR: .487; 95% CI: .192–1.235).

Table 3

Effects of adjustment for personal characteristics and various indices of health on the association between all-cause mortality and BMI criteria.

Table 4

Effects of adjustment for personal characteristics and various indices of health on the association between all-cause mortality and WC criteria.

Furthermore, regarding the threshold categories for WC, the inverse association was constantly observed in all of the considered models. Following simultaneous adjustments for all of the characteristics, the mortality risk was significantly lower in those with abdominal obesity than in those with normal WC (HR: 572; 95% CI: 370–883).

Discussion

The main finding of present study is that patients hospitalized for ischemic stroke who were overweight and obese had better survival than those who were normal weight. In contrast, patients who were underweight demonstrated the worst mortality rates. It is noteworthy that the inverse association between body weight and outcome measures remained after adjustments for confounders. Furthermore, the inverse association between abdominal obesity and mortality was also constantly observed.

The results of this study are in line with those of studies that demonstrated that overweight was associated with a significantly lower risk of mortality.[ When omitting different covariates from our models, the risk of mortality in those who were overweight remained low. Similarly, Vemmos et al[ revealed that patients who were overweight had significantly better survival rates than those with normal BMI. In contrast, inconsistent with our findings, Dehlendorff et al[ suggested that being overweight was not associated with decreased risk of all-cause mortality. The possible reason for this difference is that survivors of both ischemic stroke and hemorrhagic stroke were included in their study while we only included the former. Another possible explanation for the distinct results is that our participants were followed up for a mean period of 54.86 months and the endpoint was all-cause mortality; the outcomes in the study by Dehlendorff et al were deaths due to index stroke at 1 week and 1 month.

Furthermore, our findings revealed that obesity confers a lower risk of mortality without statistical significance, which is consistent with those of previous studies.[ A Chinese follow-up study[ found that obesity (BMI, 27.5–32.4 kg/m2) was not associated with increased risk of mortality, which is similar to the South Korean results by Ryu et al.[ However, the TEMPiS study[ revealed that the lowest mortality risk was observed in obese and very obese patients and not in those with normal BMI. The discrepancy in the findings may be due to several reasons. Firstly, the TEMPiS study was a non-randomized intervention study that compared acute stroke treatment without specialized care for stroke in hospitals, while ours was a hospital-based longitudinal study without any interventions. Secondly, the TEMPiS study included individuals with hemorrhagic stroke as well, whereas our study included only those with the first incidence of ischemic stroke. Therefore, the relationship between obesity and all-cause mortality remains to be validated.

A remarkable finding in our study was that abdominal obesity was also associated with better prognoses in patients with stroke, which was in accordance with the results of previous studies.[ However, the participants in the current study were divided into normal WC and abdominal obesity according to sex-specific WC criteria, which cannot distinguish between mild, moderate, and extreme abdominal obesity. However, Gomes et al[ used quartiles of WC to categorize different levels of abdominal obesity; they found that patients in the 4th quartile of WC had significantly better survival, while those in the 3rd and 2nd quartiles had decreased risks of mortality without significance compared with those in the 1st quartile.

Our findings suggest that the beneficial effects of higher BMI and WC on all-cause mortality appear to be present in patients with the first episode of ischemic stroke, which was previously termed as obesity paradox.[ The inverse relationship was also confirmed in different patient populations, such as those with heart failure,[ those on dialysis,[ and those with diabetes,[ certain cancers,[ atrial fibrillation,[ and coronary artery bypass surgery.[ A few explanations have been proposed for the paradox. Firstly, individuals may receive different treatments. As previously noted, obese patients who develop stroke demonstrate a readily identifiable high cardiovascular risk phenotype. Therefore, they may tend to present earlier or receive more aggressive treatment, thus, varying the natural history of their disease[ and confer them with benefits of lead-time bias.[ Similarly, Park et al[ found that obese patients had a higher number of diagnosed medical comorbidities, were taking more medications, and had a higher rate of optimal combination drug treatment than lean patients, which could have contributed the differences in the prognoses. Secondly, obesity may affect the immune response and alter the levels of inflammatory factors.[ Previous studies have reported that obese patients have lower levels of several proinflammatory cytokines, particularly interleukin (IL)-6 and IL-8. Additionally, obese individuals may be better equipped to undergo the initial catabolic phase of acute illness due to the additional energy stores in the form of lipoproteins and adipose tissue, which can neutralize the inflammatory effects.[ Thirdly, obesity can worsen the symptoms, such as dyspnea, and obese individuals may have better prognosis because they present earlier with any specific disease.[ High BMI may also be associated with some protective factors.[ Another possible explanation is that relative to patients with normal weight, overweight/obese patients may have a better metabolic reservoir—an aspect that enables them to deal better with catabolic imbalance and impaired metabolic efficiency induced by stroke.[ Alternatively, obese patients who have survived until their event may somehow be healthier individuals or have metabolically benign obesity.[ Finally, obese patients are encouraged to change their unhealthy lifestyle [ for a better prognosis; however, additional research is required in this regard. Additionally, underweight may be the result of other underlying diseases, unintentional weight loss, reduced mental and physical health, or severe comorbidities that result in poor prognosis.[

Several strengths and limitations of this study should be acknowledged and discussed. The strength is that we considered the association of mortality with not only general obesity but also with abdominal obesity in a long-term Chinese cohort. Furthermore, important confounding factors, such as stroke severity and anti-thrombotic medication adherence, were comprehensively considered, which were often absent in previous studies.[ Importantly, dynamic changes in weight and WC during the follow-up were also considered by recording BMI and WC at every follow-up, which may constitute for stronger conclusions.

Our study does have some limitations. Firstly, the weight, height, and WC were based on self-reporting by patients rather than by using objective tools, which may have resulted in biases.[ However, previous studies have suggested that self-reported weight, height, and WC are satisfactorily accurate in the assessment of obesity status and have near-perfect correlations and agreement with the measured anthropometric values.[ Furthermore, BMI fails in distinguishing between individuals with different types of fat distribution,[ and WC is incapable of distinguishing between persons of varying heights.[ However, directly and accurately quantified methods are impractical and time-consuming.[ Therefore, self-reported data should be generally acceptable. Secondly, we could not include the subtypes of stroke because the necessary stroke etiology and clinical features for classification of the subtypes of stroke were insufficiently collected in first interview. Neither the TOAST (Org 10172 in Acute Stroke Treatment) nor Oxfordshire Community Stroke Project classification of the subtypes of stroke was obtained. Thirdly, index bias may be present in our study because though all possible confounding factors were considered, essentially perfect controlling for confounding factors between obesity and mortality is possible.[ Even though unmeasured confounders are inevitable, index bias alone may be a partial explanation for the obesity paradox, which cannot deny the presence of obesity paradox in stroke survivors.[ Finally, our participants were recruited from only the West China Hospital of Sichuan University, which may limit the generalizability of these findings.

Conclusions

Overweight and abdominal obesity were paradoxically associated with a lower risk of mortality in survivors of their first episode of ischemic stroke in China. Consequently, we suggest that the recommendations regarding body weight management in the secondary stroke pre-strategy should not simply refer to primary prevention. Future prospective studies must look at evaluating the role of obesity in different stroke subtypes and devise appropriate weight-management strategies for optimal prognoses in secondary prevention in these survivors.

Acknowledgments

We thank all participants in this study. We also would like to thank Editage (www.editage.com) for English language editing.

Author contributions

Conceptualization: Muke Zhou, Cairong Zhu.

Data curation: Cairong Zhu.

Formal analysis: Ju Wang, Jijie Li, Mier Li, Lisha Hou, Ping Zhu, Xudong Du.

Funding acquisition: Cairong Zhu.

Investigation: Ju Wang, Jijie Li, Mier Li, Lisha Hou, Ping Zhu, Xudong Du.

Methodology: Ju Wang, Jijie Li.

Project administration: Muke Zhou, Cairong Zhu.

Writing – original draft: Ju Wang, Jijie Li, Muke Zhou, Cairong Zhu.

Writing – review & editing: Ju Wang, Muke Zhou, Cairong Zhu.