Older patients with diabetes mellitus who are obese have a better long-term prognosis: an analysis of outcome in 12,025 diabetic patients

Abstract

Background: Obesity is a key trigger for insulin resistance leading to type-2 diabetes mellitus (T2DM). However, recent evidence suggests that obese patients with T2DM may have lower morbidity and mortality compared to patients of normal weight. These reports are limited by statistical power and confounders. We investigated the relationship between Body Mass Index (BMI), mortality and cardiovascular (CV) morbidity in a long-term large cohort of patients with T2DM. Methods: Between 1995 and 2011, weight (BMI), blood pressure, dyslipidemia, smoking and comorbidities was collected in patients with T2DM without known CV disease. Patients were followed prospectively. Total mortality and hospital admissions for acute coronary syndrome (ACS), cerebrovascular accidents (CVA) and heart failure (HF) were gathered. Subjects were divided according to BMI quartiles and in age tertiles. ANOVA was used to compare covariates amongst the BMI groups, Chi square and multivariate Cox-Regression analysis were used to assess the prognostic impact of BMI and confounders on the above-defined events. Sensitivity analysis was performed accounting for cancer, BMI<18.5 and gender. Results: Of 12025 patients (54% men, mean age 60+15 years), followed for a mean of 10+4 years), 4125 (34%)died. In the first age tertile (42+10 years), there was a U-shaped relationship between BMI and outcome; those with BMI 25-28 had the lowest mortality (X2 15.2 P<0.01). In the second age tertile (62+6 years), there was a similar mortality across BMI quartiles (X2 5.1 P=0.14). In the oldest age tertile (75+5 years), mortality was inversely related to BMI; those with BMI 27-30 had the lowest mortality X2 33.0 P<0.0001. Excluding patients with cancer, BMI <18.5 or adjusting for sex did not significantly affect these results. In a multi-variable Cox Regression model, including age, sex, smoking, blood pressure, cancer and diabetes duration, higher BMI was still associated with a lower mortality. However, patients in higher BMI quartiles had a higher incidence of ACS and HF in all age tertiles (all X2 P<0.05) and CVA showed a similar trend. In multi-variable Cox regression models, the association between higher BMI and CV morbidity remained after adjusting for other variables. Conclusion: In this analysis, although being overweight was associated with an increased risk of CV events in patients with T2DM, higher BMIs were associated with a survival benefit in older patients. Slim patients with T2DM may have a more severe metabolic disorder than patients in whom insulin resistance is primarily due to obesity.