A simple nomogram for assessing COVID-19 in-hospital mortality in adults with diabetes: a retrospective cohort study 

Abstract

Background: We aimed to develop a simple nomogram and online calculator for predicting the risk of COVID-19-related mortality in patients with diabetes, and to compare the practicality and cost-effectiveness of the nomogram with previous prediction models.Methods: We did a retrospective study in hospitals designated for 127 patients with diabetes hospitalized for COVID-19 in Tehran. Adult patients with previous diabetes status or compliance with diabetes criteria on admission hospitalized for COVID-19 were enrolled. Based on the least absolute shrinkage and selection operator method and multivariable logistic regression computed, a nomogram was developed using predictors upon admission. C-statistics were validated internally in the cohort by bootstrap and in subtypes analysis. The accuracy and cost-effectiveness between the nomogram and previous models were compared through C-statistics.Results: From February 29 to July 20, 2020, 127 patients were eligible for inclusion in the cohort, of which 29 died for COVID-19. Loss of consciousness, pulse, neutrophil to lymphocyte ratio and albumin were independent risk factors for COVID-19 mortality in diabetes. The nomogram and the free online calculator had a high C-statistic of 0.900 (0.842-0.958) for predicting mortality in diabetes with COVID-19, which performed better discrimination and clinical usefulness than previous models. C-statistic values were stable in internal validation and different subtypes.Conclusions: The nomogram and online calculator were used by emergency physicians and vaccine prevention doctors to identify diabetes with high COVID-19 mortality, which performed better than previous models. These patients might benefit from intensive monitoring, early preventive measures and vaccination.