Investigate the characteristics of neurological manifestations of COVID-19 infection in T2D patients.

There is mounting evidence suggesting that type 2 diabetes (T2D) and neurological manifestations are associated with severe SARS-CoV-2 infection.

Using the Cerner COVID-19 de-identified dataset, we identified patients with COVID-19 infection, T2D, and neurological complications using ICD-10-CM. We investigated and compared the incidence, demographics, and characteristics of neurological manifestations in the following groups: COVID-19 patients with/without neurological complications (COVID-19N/COVID-19WN) with and without T2D.

Among 25,427 COVID-19 patients, 23.75% had T2D (mean age 62.8 years; 50% male), and among them, 2,013 patients (33.4%) developed a neurological complication (mean age 63.6 years, 54% male). The mean length of stay in days (LoS) and mortality rate of COVID-19N with T2D were significantly higher than COVID-19N without T2D, COVID-19WN with T2D, and COVID-19WN without T2D, respectively 9.8, 16%; 6.2, 10%; 7.5, 12.2%; 3.9, 6.1% (p<0.05). Home discharges were significantly higher in COVID-19WN without T2D patients compared to COVID-19WN with T2D, COVID-19N without T2D, and COVID-19N with T2D, respectively 71.8%; 53%; 57.8%; 39.1% (p<0.05). Skilled nursing facilities (SNF) and hospice discharges in COVID-19N with T2D were significantly higher than COVID-19N without T2D and COVID-19WN, with and without T2D, respectively, 15.8%, 6.7%; 8.6%, 4.2%; 8.8%, 4.3%; 4.7%, 2.2% (p<0.05). T2D in COVID-19 patients was associated with increased mortality, discharge to SNF, and discharge to hospice than neurological manifestations in COVID-19 patients, respectively 12%, 10%; 4.3%, 4.2%; 8.8%, 8.6% (p<0.05).

Neurological manifestations and T2D in COVID-19 patients were associated with synergistically increased mortality rate and discharge to SNF and hospice compared to COVID-19 patients without neurological complications or T2D. Work is in progress to develop and implement risk stratification machine learning-based models using clinical data (T2D severity and neurological conditions) to predict patient outcomes like death, ICU admission, and discharge status.