To construct and validate nomogram models that predict the incidence of delirium in elderly patients with non-severe SARS-CoV-2 infection.
Elderly patients (≥65y) tested positive for SARS-CoV-2 infection at the hospital were included. We used the 3-min diagnostic Confusion Assessment Method for delirium diagnosis. Least absolute shrinkage and selection operator (LASSO) logistical regression analysis was performed to explore potential independent influencing factors of delirium. A predict model visualized by nomogram was constructed based on the confirmed variables. The predictive accuracy and clinical value of the model were evaluated using receiver operating characteristic (ROC) curves.
The data of 311 elderly patients were analyzed, of whom 73 (23.47%) patients were diagnosed with delirium. Three independent influencing factors of delirium were confirmed: age (OR1.16,1.11–1.22), Glomerular filtration rate (OR 0.98,0.97–0.99), platelet-large cell ratio (1.06,1.02–1.10). These parameters were used to create a nomogram to predict the development of delirium, which showed good predictive accuracy confirmed by the ROC curves (AUC 0.82,0.76–0.88).
We construct a credible nomogram to predict the development of delirium in elderly patients with Non-severe SARS-CoV-2 infection. Our finding may be useful to physicians in early prevention and treatment of delirium.