Medical history may help predict COVID-19 risk in people with diabetes
A prediction model factoring in recent hospitalizations, comorbidities and drug exposure may help determine the individual risk for COVID-19 critical care admission and mortality for people with diabetes, according to study data.
Helen M. Colhoun, FRCP, AXA chair in medical informatics and life course epidemiology at the University of Edinburgh, Scotland, and colleagues wrote that the risks for people with diabetes varied, and a model including medical history factors and demographic information may be a better predictor than a model that factors only demographics and the patient’s diabetes type.
“We have shown that, among those with diabetes, the risk of severe disease varies widely and is predictable,” the researchers wrote. “This insight should inform shielding policies and vaccine prioritization strategies.”
Researchers conducted a population cohort study in Scotland from March 1 to July 31, 2020. There were 5,463,300 residents in Scotland 3 weeks before the start of the study period, of whom 319,349 had diabetes. Researchers collected data on all COVID-19 cases, critical care admissions and deaths in Scotland during the study period. Critical care included all admissions to an ICU or high dependency unit. Data were linked to the national diabetes register to identify people with diabetes who contracted COVID-19.
COVID-19 risk factors in diabetes
Of those with diabetes in Scotland, 0.9% tested positive for COVID-19 during the first 5 months of the pandemic, and 0.3% died or were treated in a critical care unit. Of those who died or were admitted to critical care, 89.9% were aged 60 years or older. In those without diabetes, 0.1% died or were treated with critical care for COVID-19.
After adjusting for age and sex, people with diabetes had an increased risk for being admitted to critical care or dying of COVID-19 (OR = 1.4; 95% CI, 1.3-1.49; P < .0001). The likelihood for severe COVID outcomes was similar for men and women. Compared with people without diabetes, those with type 1 diabetes (OR = 2.4; 95% CI, 1.82-3.16; P < .0001) and type 2 diabetes (OR = 1.37; 95% CI, 1.28-1.47; P < .0001) had higher odds for COVID-19 critical care or mortality.
Older age, male sex and a longer diabetes duration were associated with an increased risk for COVID-19 critical care admission or mortality. Those with diabetes who lived in a residential care home had a very high increased risk for COVID-19 critical care treatment or mortality (OR = 16.57; 95% CI, 14.33-19.17; P < .0001).
“More than a third of people with diabetes who developed fatal or critical care unit-treated COVID-19 lived in residential care homes, emphasizing the crucial importance of protecting such vulnerable individuals during the remainder of the pandemic,” the researchers wrote.
The risk for COVID-19 critical care treatment or death was higher for people with diabetes who were admitted to the hospital in the past 5 years for any reason (OR = 3.31; 95% CI, 2.79-3.92; P < .0001). There was also an increased risk for people with a high HbA1c (OR = 1.01; 95% CI, 1.006-1.014; P < .0001) or low systolic blood pressure (OR = 0.986; 95% CI, 0.982-0.99; P < .0001). Being on antihypertensive medication was associated with a lower risk for COVID-19 critical care or mortality (OR = 0.8; 95% CI, 0.71-0.91; P = .0006), whereas there was an increased risk for people who took NSAIDs (OR = 1.85; 95% CI, 1.63-2.1; P < .0001), proton pump inhibitors (OR = 1.41; 95% CI, 1.25-1.59; P < .0001) and anticoagulants (OR = 1.66; 95% CI, 1.47-1.89; P < .0001). The risk for COVID-19 critical care treatment or mortality was also higher for people exposed to more types of drugs other than those used for diabetes in the past 3 years (OR = 1.14; 95% CI, 1.08-1.2; P < .0001).
Predicting individual COVID-19 risk
Researchers used the findings to create a cross-validated COVID-19 outcome prediction model that included age; sex; diabetes duration and type; comorbidities; clinical measures, such as HbA1c, BMI, estimated glomerular filtration rate and systolic BP; and drug exposures. This model had a higher C-statistic (0.85; 95% CI, 0.83-0.86) than a baseline model that included only age, sex and diabetes type and duration (C-statistic = 0.76; 95% CI, 0.75-0.77).
“This level of predictive accuracy disproves the notion that all people with diabetes have similar risk,” the researchers wrote. “The variables retained in the model are those that are the most predictive and not necessarily causal; some of the most valuable predictors include the number of hospital admissions in the past 5 years and number of diabetes and non-diabetes drugs, which were not evaluated in other diabetes COVID-19 studies.”
Researchers used the prediction model to produce the Shiny app, which converts the absolute risk score in the prediction model and converts it to a COVID-age, which is the age of a person without diabetes who has the same absolute risk.
“The Shiny app has been provided for illustrative purposes only, to allow a greater understanding of how a prediction model broadly translates into COVID-age in individuals with diabetes,” the researchers wrote. “External validation, regulatory approval and appropriate licensing would be required before this app could be used in clinical practice.”