A dynamic online nomogram predicting severe vitamin D deficiency at ICU admission

Introduction: Although prevalent and associated with worsened outcomes, vitamin D severe deficiency is not systematically searched among intensive care unit (ICU) admissions and waiting time for measurement results range from hours to few days. Hence, we developed and internally validated a simple nomogram for predicting severe vitamin D deficiency at ICU admission. Patients and methods: Data of 3338 ICU admissions from an observational prospective cohort registered between January 2017 and December 2019 were analyzed. Demographic data as well as severity scores and season of admission were obtained. After splitting the population into training and test sets, a least absolute shrinkage (LASSO) regression model was used to select factors and construct the nomogram. Calibration and discrimination were used to assess the nomogram performance. Clinical use was evaluated by a decision curve analysis. Results: Age, gender, Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiology Score III (SAPS3) and season of admission were identified by the LASSO regression analysis as significant predictors of vitamin D severe deficiency at ICU admission. The nomogram model showed good discrimination with a 1000 bootstrap analysis and good calibration with a C-index of 0.64. The decision curve analysis showed that at a threshold probability between 30% and 50%, using the nomogram adds more benefit that considering that all patients are severely deficient or non-severely deficient. Conclusions: This easy-to-use dynamic nomogram can help physicians to select patients that could benefit the most from vitamin D supplementation at ICU admission. External validation is needed to verify the generalizability of this nomogram.