Predictive Analytics Tool Calculates COVID-19 Hospitalization Risk

Cleveland Clinic researchers have built a predictive analytics model to better understand which patients with COVID-19 are at high risk of hospitalization from the virus.

For more coronavirus updates, visit our resource page, updated twice daily by Xtelligent Healthcare Media.

In a study published in PLOS One, the team described developing and validating the model using retrospective patient data from more than 4,500 patients who tested positive for COVID-19 at Cleveland Clinic locations. Data scientists used statistical algorithms to transform data from registry patients’ EHRs into the risk prediction model.

When researchers compared characteristics between patients who were and were not hospitalized due to COVID-19, they uncovered several previously undefined risk factors. For example, former smokers were more likely to be hospitalized than current smokers, and patients taking Angiotensin Converting Enzyme (ACE) inhibitors or angiotensin II type-I receptor blockers (ARBs) were more likely to be hospitalized than patients not taking those drugs.

The research group noted that additional studies will be necessary to further examine the relationship between ACE inhibitors and ARBs.

"In our study, taking these drugs was only found to confer increased risk for hospitalization when run through univariable analysis, which means the observed association could be the result of other, confounding variables, like a preexisting condition,” said Michael Kattan, PhD, chair of Lerner Research Institute's Department of Quantitative Health Sciences.

Additionally, the team found that African American patients were more likely to be hospitalized than patients of other races, accounting for 33 percent of hospitalizations but only 18 percent of the total population surveyed.

To better understand this disparity, the team stated that future research should focus on patients’ genetic makeup as well as their social determinants of health.

“A deeper exploration of the underlying genetics and biology of race in the defense against and the response to a SARS-CoV-2 infection is needed. This should be paired with a deeper exploration of social influencers of health such as population per square kilometer, and population per household which were also relevant in our nomogram,” researchers said.

The results also showed that patients presenting with symptom complex including shortness of breath, fever, vomiting, and fatigue were more likely to hospitalized than those who did not experience these symptoms.

“The significant association of shortness of breath and diarrhea with hospitalization may reflect the need for inpatient supportive care with these symptoms, regardless of the etiology,” the team wrote.

The study confirmed other findings that have been well-documented in COVID-19 research, including higher risk of hospitalization among older people, men, and those with comorbidities like diabetes and hypertension. Those from lower socioeconomic backgrounds, as measured by zip code, also had a higher risk of hospitalization.

The team stated that it will be important to further examine on a pathogenic level how these risk factors lead to increased hospitalization risk. However, the risk prediction tool could help providers boost patient outcomes and deliver preventive care.

"Hospitalization can be used as an indicator of disease severity," said Lara Jehi, MD, chief research information officer at Cleveland Clinic and corresponding author on the study.

"Understanding which patients are most likely to be admitted to the hospital for COVID-19-related symptoms and complications can help physicians decide not only how to best manage a patient's care from the time of testing, but also how to allocate beds and other resources, like ventilators."

The predictive analytics model is freely available as an online risk calculator, and was shown to offer significantly better predictions than using no risk model at all. The model was also able to perform well in different geographic regions.

The risk model described in the PLOS One study is the second COVID-19-related tool developed by this Cleveland Clinic team. In June 2020, the group developed a predictive analytics model to help determine an individual patient’s likelihood of testing positive for COVID-19, as well as their potential outcomes from the disease.

“The ability to accurately predict whether or not a patient is likely to test positive for COVID-19, as well as potential outcomes including disease severity and hospitalization, will be paramount in effectively managing our resources and triaging care,” Jehi said at the time.

“As we continue to battle this pandemic and prepare for a potential second wave, understanding a person's risk is the first step in potential care and treatment planning.”

These models will prove vital in the ongoing search for solutions and treatments for COVID-19.

"Ultimately, we want to create a suite of tools that physicians can use to help inform personalized care and resource allocation at many time points throughout a patient's experience with COVID-19," said Jehi.