To evaluate insulin resistance (IR) among outpatients with bipolar disorder (BD) in order to determine clinical correlates of IR in this patient population.

We performed a cross-sectional study in sixty-five DSM-IV-TR BD patients consecutively assessed from January to August 2007 at the Bipolar Disorder Program, Hospital de Clínicas de Porto Alegre, Brazil. IR was diagnosed using the homeostatic model assessment - insulin resistance (HOMA-IR). Metabolic syndrome (MS) diagnosis and metabolic variables were assessed using three definitions: National Cholesterol Educational Program - Adult Treatment Panel III (NCEP-ATP III); NCEP-ATP III modified criteria and International Diabetes Federation (IDF).

IR was present in 43.1% of the sample (women 40%, men 44.4%). The prevalence of MS defined by the NCEP-ATP III criteria was 32.3%, NCEP-ATP III modified was 40% and IDF was 41.5%. NCEP-ATP III modified criteria showed the best trade-off between sensitivity (78.6%) and specificity (89.2%) to detect insulin resistance. Waist circumference was the best clinical parameter associated with IR in the linear regression model (B=0.014, SE 0.002, t=6.18, p< 0.001). Areas under ROC curves were similar for waist circumference and different MS definitions (chi2=2.98, df=3, p=0.39).

Currently MS criteria may provide reasonable sensitivity and specificity for the detection of insulin resistance in patients with bipolar disorder. Waist circumference may be a simple and inexpensive means to predict insulin resistance in this population.