Age, sex, comorbidities impact outcomes after COVID-19 hospitalization
Click Here to Manage Email Alerts
In a national private health care database, age, male sex and comorbidities increased risk for death in patients hospitalized with COVID-19, according to data presented at the virtual American Heart Association Scientific Sessions.
The findings were mostly consistent with data from the AHA’s COVID-19 CVD registry, also presented at the meeting.
The data set of patients with COVID-19 was created by Cerner Corp. and Amazon Web Services, Cardiology Today Next Gen Innovator Ann Marie Navar, MD, PhD, associate professor of internal medicine and of population and data sciences at University of Texas Southwestern Medical Center, said during a presentation.
“We need to understand who is most at risk, particularly as we are deploying immunization strategies,” she said. “We also need to understand risk factors so that people can understand their own risk of disease and make appropriately informed choices. Among people who are hospitalized with COVID-19, it’s critical that we understand risk factors for worse outcomes, as we have to have important informed conversations with patients and their families about their prognosis.”
The data set, which included visits through July 1, was created from electronic health record data from 52 health systems, she said. To be included, a patient had to have a positive COVID-19 test within 2 weeks or during hospitalization or a diagnosis code for COVID-19 during hospitalization.
The analysis included 19,584 patients with COVID-19 (median age, 52 years; 47% women; 29.4% Hispanic) who died or were discharged to home during the study period.
Among the cohort, 31.1% had diabetes, 50.4% had hypertension, 14.3% had HF, 18% had CAD and 5.6% had end-stage renal disease, Navar said.
The rate of in-hospital mortality was 20.7%, she said, noting 32.6% required mechanical ventilation, 5% had MI, 2% had pulmonary embolism and 1.5% had stroke.
“Mortality within those complications was quite high,” she said, noting that 74.6% of those with mechanical ventilation, 55.5% of those with MI, 26.5% of those with PE and 56% of those with stroke died.
Regarding age, “what’s striking is that starting about the age of 55, the risk of death among patients who are hospitalized increases almost linearly,” Navar said.
Hispanic patients had lower rates of mortality compared with non-Hispanic patients (12.7% vs. 25%), she said. Among those whose ethnicity was unknown, the mortality rate was 19.7%.
Mortality rates by race were as follows: white, 20.8%; Black, 22.7%; Asian/Pacific Islander, 19.7%; American Indian/Alaska Native, 24.1%; other, 15.5%; and unknown, 26.6%.
Mortality rates by comorbidity were as follows: CAD, 28.8%; end-stage renal disease, 28.7%; HF, 32.4%; hypertension, 20.4%; and diabetes, 21.5%.
In a restricted cubic spline analysis, “although we do see an increase in mortality in BMI starting about 30 kg/m2, we actually saw the most dramatic increases among those who were underweight,” Navar said. “This is after patients are hospitalized, and we did see a disproportionate number of patients who were obese who were hospitalized to begin with. What this suggests to us is that once hospitalized, being underweight may be as much or worse of a risk factor than being overweight.”
After multivariable adjustment, the following were predictors of mortality after COVID-19 hospitalization: male sex (OR = 1.46; 95% CI, 1.31-1.62), Medicare insurance (OR = 1.77; 95% CI, 1.51-2.08), Medicaid insurance (OR = 1.62; 95% CI, 1.31-2), diabetes (OR = 1.27; 95% CI, 1.13-1.42), HF (OR = 1.29; 95% CI, 1.13-1.47) chronic kidney disease (OR = 1.5; 95% CI, 1.32-1.69), Navar said, noting Hispanic ethnicity was protective against mortality (OR = 0.71; 95% CI, 0.59-0.86), there were differences by age and BMI along the restricted cubic spline, and there were no significant differences by race.
“It is a little bit challenging because of the correlation between these different risk factors to understand which are really driving the risk vs. which are just more statistically significant in the model,” Navar said.