Researchers in Israel developed a COVID calculator to predict anyone’s chance of getting hospitalized with COVID-19 or dying from it once infected. Dr. Ariel Israel MD, Ph.D., et. al. created the calculator, the basis for which is published as a preprint study here.
Dr. Israel is a prestigious researcher and the head of Leumit Research Institute in Israel, which is associated with Leumit Health Services, one of Israel’s national healthcare providers. He is a family doctor by training and specializes in data mining, analytics, bioinformatics, and machine learning. He has published several studies related to COVID-19, including one related to the decline of antibody titers in the aftermath of COVID vaccination vs decline in the aftermath of recovering from a COVID infection. He discussed those findings on the TrialSite News Interview Series here. He also conducted research on early treatment drugs and supplements that were associated with a lower risk of COVID-19 infection and severity and did a live webinar discussing that research with TrialSite News here.
For the COVID calculator study, Dr. Israel and his team analyzed electronic health records from 101,039 individuals infected with COVID-19 since the beginning of the pandemic through 30 November, ’21. They then built models to estimate the risk of hospitalization and death based on number of COVID vaccines received (BNT162b2 mRNA vaccine), and the following risk factors: age, sex, BMI, hemoglobin A1C, kidney function, high blood pressure, pulmonary disease, and malignancy.
Results of the analysis drive home the point that COVID-19 is a disease that significantly hits the elderly far more than the younger population and those people who are obese or underweight. Key results with statistical significance include:
Age is significantly associated with a higher risk of hospitalization and death. Each year of age increases the odds of hospitalization by a multiplicative factor of 1.061. This means that if a 20-year-old has the same underlying risk factors as an 80-year-old, the 80-year-old is 34 more times likely to get hospitalized. For death, each year of age increases the odds of death by a factor of 1.1, meaning that an 80-year-old compared to a 20-year-old with similar risks is 393 times more likely to die.
Females were associated with 34% reduced odds of hospitalization compared to males and about a 50% reduced odds of death.
While the BMI is not a perfect measure of body fat, it is a cost-effective tool to use at the population level. In the study, when it came to hospitalizations and compared to a normal BMI, BMIs greater than 35 had an odds ratio of 2.35; BMIs between 30-35 had an odds ratio of 1.7 and BMIs between 25 and 30 had an odds ratio of 1.3. For death and when compared to a normal BMI, the odds ratio for a BMI over 35 was 1.9 and the odds ratio for a BMI below 18.5 was 2.1. This means that being obese or being underweight, both of which may have other comorbidities that play a role, significantly increases one’s chance of death.
Hemoglobin A1C is a measure of how well an individual is controlling his/her Diabetes mellitus. For hospitalizations and compared to Hemoglobin A1C values in the normal range, A1C values above 10% had an odds ratio of 3.0; values between 8- 10% had an odds ratio of 1.9; and values between 6.5 -8% had an odds ratio of 1.5.
The GFR (Glomerular Filtration Rate) is a measure of kidney function. For hospitalizations and compared to the reference value of GFR greater than 90, a GFR below 30 had an odds ratio of 4; a GFR between 30-44 had an odds ratio of 2.8, and a GFR between 45-59 had an odds ratio of 1.6.
For hospitalizations and compared to unvaccinated individuals, the odds ratio for 1 vaccine dose was .8; for 2 vaccine doses was .6, and the odds ratio for 3 vaccines was .3. For death and compared to unvaccinated individuals, a 2 dose series followed by a booster decreased one’s chance of death by 78% (Odds Ratio =.223).
How Accurate are These Models?
The Area Under the Curve, or AUC, is a value that tells you how accurate a model is at predicting a particular outcome. It can range from 0 to 1, with 1 meaning a perfect prediction. The models in this study were shown to be highly accurate for predicting whether an infected individual would be hospitalized or die. For hospitalizations, the AUC value was .889 and for death, the AUC value was .96.
Can you try the COVID calculator?
Yes! As of now, the COVID calculator is available to the public as a web-based app. You can enter your various values into the model and learn your risk of hospitalization or death from COVID-19. Keep in mind that this study is a preprint and hasn’t yet been peer-reviewed, so the COVID calculator is not set in stone. If you decide to use it, read the disclaimers first. It is available here for use: The COVID-19 Risk Calculator.
The study was supported by the Intramural Research Program, National Institutes of Health, National Cancer Institute, Center for Cancer Research.