Johns Hopkins study of 1.9 million COVID-19 cases links severity with blood types + more

Researchers from Johns Hopkins University School of Medicine and Tufts Medical Center conducted a study using artificial intelligence to do a deep dive into EHR notes of patients who tested positive for COVID-19 in a nationwide data lake, according to a July 13 report published in JAMA Network Open.

The study retrospectively examined 1,926,526 adults in the United States who were tested for COVID-19 at one of 34 medical centers nationwide, between Jan. 1 and Dec. 7, 2020. Patients were examined using a severity scale from the World Health Organization.

The study was conducted by researchers from Baltimore-based Johns Hopkins University School of Medicine, the National Institutes of Health, Boston-based Tufts Medical Center, University of Texas Medical Branch in Galveston, the University of Rochester (N.Y.) Medical Center and more. 

Six study findings:

1. The study found that of 174,568 adults who contracted COVID-19, 18.6 percent were hospitalized and 20.2 percent of those patients were severely ill. Of the hospitalized patients, 14 percent received invasive respiratory support, 8.5 percent received medications for cardiovascular support and 3.2 percent received dialysis or renal replacement therapy.
   
   
2. The study found that machine learning models used on the first day frequently predicted clinical severity with accuracy. The AI programs that boasted the highest accuracy are XGBoost and the random forest method, with accuracy rates ranging from 88 percent to 95 percent.
   
   
3. Mortality decreased from 16.4 percent in March and April 2020 to 8.6 percent in September and October.
   
   
4. Consistent with other study findings, men, older adults, those who are obese and Black people were associated with greater clinical severity. However, social determinants of health may have greatly affected how COVID-19 severity and race are connected.
   
   
5. Patients with liver disease and dementia were also more likely to experience higher COVID-19 severity.
   
   
6. Some studies have inconsistently reported an increased risk of intubation and death for patients whose blood types are AB, A, B and O. However, patients who had AB blood type had a protective association and were less likely to experience intubation or death, the study found.