Unique risk factors for thiamine deficiency in patients without alcohol dependence

Background: Thiamine (vitamin B1) is a crucial component in multiple essential biochemical pathways, and B1 deficiency is increasingly recognized as a treatment target in sepsis and other critical illnesses. Our objective was to determine risk factors for B1 deficiency in severely ill ED patients without a history of alcohol dependence. Methods: This was a multi-pronged, prospective observational study that assessed B1 levels in ED patients with diabetic ketoacidosis, severe sepsis, and oncological emergencies. We excluded patients that had known alcohol dependence or for whom the ED clinicians planned B1 treatment. Investigators collected whole blood levels on all patients in ED, which measure thiamine-diphosphate, the active form of B1. We collected demographic and clinical characteristics that could contribute to nutritional deficiencies. We collected data specific to metformin use as pre-clinical data suggests that metformin may interfere with intestinal B1 transporters. Analysis consisted of univariate comparisons and multivariable logistic regression to assess significant risk factors for the primary outcome of B1 deficiency, defined as a whole blood level below the normal reference range. Results: The study enrolled 342 patients, of whom the average age was 57 (SD 17) years, 47% were female, and 80% African American. In univariate analysis, patients with B1 deficiency had greater mean age (62 vs. 56 years, p = 0.007), lower mean albumin (2.9 vs. 3.5 g/dL, p \u3c 0.001), and similar body mass index (28 vs. 28, p = 0.81). Patients with B1 deficiency were more often female (24% vs. 14%, p = 0.023). Thiamine deficiency occurred most frequently in patients with severe sepsis (27%) and oncological emergencies (28%) compared to all other diagnoses (10%, p=0.002). Diabetics on metformin had significantly higher rates of B1 deficiency compared to those not on metformin (22% vs. 10%, p = 0.03). In multivariate analysis, the following clinical characteristics (OR, 95% CI) were associated with B1 deficiency: female gender 2.0 (1.1-3.7), history of cancer 2.2 (1.2-4.0), metformin use 2.6 (1.0-6.5), and low albumin 2.6 (1.2-5.3). Conclusion: In this study, independent risk factors for B1 deficiency were female gender, cancer, and low albumin. Furthermore, this is the first clinical data to indicate a significant association between metformin use and B1 deficiency