The Obesity Paradox in Type 2 Diabetes Mellitus: Relationship of Body Mass Index to Prognosis

Editors' Notes Context Obesity confers a survival advantage in patients with established cardiovascular disease (CVD). It remains unclear whether obesity provides a similar benefit for type 2 diabetes mellitus. Contribution This longitudinal study involved patients with type 2 diabetes without baseline CVD. Investigators collected information on CVD events and all-cause mortality during a median of 10.6 years of follow-up. Caution Information was not available on patient fitness levels, medication use, or cause of death. Implication Overweight and obese patients had an increased risk for CVD events. A survival advantage was found in overweight but not obese patients. Type 2 diabetes mellitus and obesity are common, growing, and related problems (1). Obesity, which promotes insulin resistance, may account for 80% of the population-attributable risk for type 2 diabetes (2). According to estimates from the World Health Organization, in 2005 a total of 1.6 billion adults worldwide were overweight and at least 400 million were obesenumbers that are expected to reach 2.3 billion and 700 million, respectively, by 2015 (3). If these predictions come true, then the prevalence of type 2 diabetes is also likely to increase (4). The association between obesity and increased risk for cardiovascular disease (CVD) is well-established in the general population (5, 6). However, once CVD occurs, paradoxically, obesity seems to confer a survival advantage. There is growing evidence that overweight patients with CVD survive longer than their normal-weight counterparts, an effect called the obesity paradox (7). Although obesity accounts for much of the risk for type 2 diabetes, a similar obesity paradox might exist after type 2 diabetes has developed. However, results conflict, with studies reporting both positive and negative associations between higher body mass index (BMI) or other weight indices and CVD (822) (Table 1). Population selection, inadequate study power, and incomplete adjustment for age and comorbid conditions may account for inconsistent results. Table 1. Previous Studies Investigating the Relationship Between BMI and Survival Outcomes in Patients With Type 2 Diabetes Mellitus We investigated the relationship among obesity, CVD, and mortality in a large cohort of persons with type 2 diabetes followed prospectively since 1995 by a single clinical service. Methods Study Population Patients with a known diagnosis of type 2 diabetes that attended the outpatient clinic service for diabetes in Kingston upon Hull, which serves a population of approximately 600000 persons, were enrolled in a registry between 1995 and 2005. Data were collected by medical and nursing staff and entered into a specifically designed electronic database (Angoss [Westman Medical Software]). More than 99% of patients had no known history of CVD (ischemic heart disease, cerebrovascular disease, heart failure [HF], or peripheral vascular disease). Data on age, duration of diabetes, smoking history, height, weight, and blood pressure were collected at the initial visit. Information on comorbid conditions (cancer, chronic obstructive pulmonary disease [COPD], and chronic kidney disease [CKD]) was collected at baseline (Table 2). The cohort was followed for clinical events until December 2011. The study was approved by a research ethics committee. Research ethical approval was granted by National Research Ethics Service (reference number 13/SW/0168). Table 2. Baseline Characteristics Outcomes The primary outcome of the analysis was all-cause mortality. A national register informed the hospital of the death of any patient under the hospital's care regardless of whether the patient left the region; information on cause of death was not given. Secondary outcomes were hospitalizations for cardiovascular events, including the acute coronary syndrome (ACS), cerebrovascular accident (CVA), or HF. Information on hospitalizations, coded using International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), and mortality was collected through the Patient Information Service of the Hull and East Yorkshire Hospitals National Health Service Trust, the sole hospital provider of emergency medical services in the region. Hospitalizations occurring when the patient was not a resident in the region would be missed, but the rates of emigration of the adult population in this area are low. Statistical Analysis Results are presented as medians and interquartile ranges (IQRs) for continuous variables. The primary analysis of interest was the relationship between BMI, expressed in clinical categories according to the World Health Organization (25), and either cardiovascular morbidity or all-cause mortality. The interaction between age (tertiles) and BMI categories was also explored. Tertiles rather than quartiles of age were chosen to maintain the size and statistical power of subgroups while a clear separation between younger and older patients was maintained. KruskalWallis tests for nonparametric data and chi-square tests were used to compare continuous and dichotomous covariates between BMI groups, respectively. Proportional hazard assumption was tested with Schoenfeld residuals. KaplanMeier survival curves were constructed, and log-rank chi-square testing was used to assess the time to cardiovascular event (ACS, CVA, or HF) and all-cause mortality. If a patient had more than 1 admission for a given cause, only the time to first admission was analyzed. We constructed a multivariable Cox regression model for all-cause mortality, adjusting for age, sex, duration of diabetes, smoking history, systolic blood pressure, COPD, cancer, CKD, and previous CVD. To assess the interaction among age, BMI, and outcomes, a Cox regression analysis was done by dividing the population in age tertiles and BMI categories. Patients with COPD, cancer, or CKD may have had an increased mortality risk. Therefore, we assessed the effect of excluding these patients. Patients who died in the first 2 years of follow-up, who might have had a preexisting serious disease leading to weight loss, were excluded in the sensitivity analysis. We also assessed the shape of the association between BMI and survival at different lengths of follow-up (2, 5, or 10 years). The interaction between age (as a continuous variable) and BMI categories was also investigated using a logistic regression analysis (26, 27). We repeated the analysis using body surface area instead of BMI. A 2-tailed P value less than 0.05 was considered statistically significant. All analyses were done using SPSS, version 19.0 (SPSS). KaplanMeier curves and logistic regression analysis were produced using Stata, version 11.0 (StataCorp). Role of the Funding Source The National Institute for Health Research, University of Hull, and Imperial College London financially supported the authors of this article. The funding source had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; or decision to submit the manuscript for publication. Results The cohort included 10568 patients (54% men; median age, 63 years [IQR, 55 to 71]) who were followed for a median of 10.6 years (IQR, 7.8 to 13.4). The median baseline BMI was 29.0 kg/m2 (IQR, 26.0 to 32.0). There were many differences among the BMI categories for the characteristics considered (Table 2). Nine hundred twelve patients were admitted for ACS (9%), 760 for CVA (7%), and 598 for HF (6%); 3728 (35%) patients died. Overweight or obese patients (BMI >25 kg/m2) had a higher rate of cardiac events (ACS and HF) than normal-weight persons (BMI, 18.5 to 24.9 kg/m2). The risk for CVA was greater only in obese patients (BMI, 30 to 34.9 kg/m2) (Figures 1 and 2 and Table 2). Figure 1. Unadjusted KaplanMeier estimates of cardiovascular events and all-cause mortality. Patients were followed for a median of 10.6 y (interquartile range, 7.813.4). Admissions for ACS occurred in 912 patients (9%), CVA in 760 patients (7%), and HF in 598 patients (6%); 3728 patients (35%) died. ACS = acute coronary syndrome; BMI = body mass index; CVA = cerebrovascular accident; HF = heart failure. Figure 2. Cox regression analysis, according to BMI categories, for cardiovascular events and all-cause mortality. Adjusted for age, sex, duration of diabetes, systolic blood pressure, smoking, and comorbid conditions (such as cancer, chronic obstructive pulmonary disease, and chronic renal failure). The reference group is the normal BMI category (18.524.9 kg/m2). Squares represent HRs, and bars represent 95% CIs. The y-axis corresponds to an HR of 1. ACS = acute coronary syndrome; BMI = body mass index; CVA = cerebrovascular accident; HF = heart failure; HR = hazard ratio. Obesity was associated with a higher rate of ACS in the youngest tertile of patients (aged <57 years), with a similar trend in the middle tertile (aged 57 to 67 years) but not among the oldest tertile. The risk for CVA was higher in obese patients only in the middle tertile. The risk for HF was higher in obese patients in all age tertiles (Appendix Figure 1). Appendix Figure 1. Cox regression analysis, according to age tertiles and BMI categories, for cardiovascular events and all-cause mortality. Adjusted for age, sex, diabetes duration, systolic blood pressure, smoking, and comorbid conditions (such as cancer, chronic obstructive pulmonary disease, and chronic renal failure). There were 3522 patients in each tertile. The reference is the normal-weight BMI category (18.524.9 kg/m2). Squares represent HRs, and bars represent 95% CIs. The y-axis corresponds to an HR of 1. ACS = acute coronary syndrome; BMI = body mass index; CVA = cerebrovascular accident; HF = heart failure; HR = hazard ratio; NA = not applicable. Although the risk for cardiovascular events was higher in patients who were overweight or obese, mortality risk was not. Uncorrected KaplanMeier estimates suggested a surv