Diabetes was common among influenza deaths long before COVID-19, analysis finds

A new brief analysis argues that the statistic deserves far more context than it received. Among older adults who died in hospital after COVID-19, the researchers found that 46.6 percent had diabetes before infection. Among older adults who died in hospital after influenza in the years before the pandemic, the proportion was even higher: 61.2 percent crudely, and 63.8 percent after statistical adjustment for age. In other words, a very large share of people dying from respiratory viral infections having diabetes is not a phenomenon that arrived with SARS-CoV-2.

The analysis, published as a brief report in Cardiovascular Diabetology – Endocrinology Reports on 15 September 2026, was led by Neha V. Reddy and Carolyn T. Bramante along with colleagues at the University of Minnesota and collaborating institutions. Its stated purpose was not to measure whether diabetes raises the risk of dying from a viral infection, a question that other studies have already addressed, but to quantify how common pre-existing diabetes was among people who died from COVID-19 compared with people who died from influenza. The authors chose influenza as the comparator because it shares similar indications for testing and similar risk factors for severe infection, hospitalization, and death.

The motivation, according to the report, came from a discrepancy between what the popular media reported and what the underlying studies actually showed. In 2022, several news outlets headlined that 30 to 40 percent of individuals dying from COVID-19 in the United States had diabetes. That proportion is similar to the approximate prevalence of diabetes among older adults, which the authors cite as 28.8 percent, the group most susceptible to mortality from COVID-19. When the researchers reviewed the references cited in support of the media reports, they found a complicated picture: a CDC report noting that underlying conditions were unknown for 54.9 percent of COVID-19 decedents; a meta-analysis of 186 observational studies attributing 8 percent of COVID-19 deaths to diabetes; a ten-hospital cohort study reporting mortality of 27 percent among patients with diabetes and 15 percent among those with prediabetes; and a fourth study reporting similar COVID-19 mortality across several diabetes and prediabetes categories.

To build a fairer comparison, the team drew on two large, complementary data sources. For COVID-19, they used the National COVID Cohort Collaborative, or N3C, data enclave, a centralized and harmonized database of de-identified electronic health record data drawn from 78 healthcare systems across the United States. For influenza, they used Medicare data, and they deliberately restricted the N3C sample to people older than 65 years so that the two populations would align with Medicare eligibility. Death due to COVID-19 in the N3C was defined as inpatient mortality during the hospital encounter in which someone was admitted for COVID-19, covering March 2020 through February 2022. Death due to influenza in Medicare was defined using the destination code from the hospital encounter in which someone was admitted for influenza, covering January 2017 through December 2019, with influenza identified by ICD-10 diagnosis codes J09, J10, or J11 on the hospitalization record.

Defining diabetes consistently across the two databases was a central methodological concern. In both databases, individuals were considered to have pre-infection diabetes if they had a diabetes diagnosis or prescription record before the index event. In the N3C, diabetes was defined as at least two outpatient diagnoses, whether office visits or telehealth visits, or one inpatient diagnosis of type 2 diabetes, using ICD-10 codes E11 or E13, before the index event, or ever having a hemoglobin A1C level of 6.5 percent or higher prior to the index date. In both databases, diabetes was also defined using ICD code 250.xx or ICD-10 codes E10.xx or E11.xx, or at least one prescription for an antidiabetic medication with any available look-back period. The analysis was approved by the University of Minnesota institutional review board, which provided a waiver of consent.

Because age is such a strong risk factor for mortality from respiratory infections, the researchers could not simply compare the raw percentages from the two databases. The age distribution of people dying from COVID-19 differs from the age distribution of people dying from influenza, and any difference in diabetes prevalence between the two groups of decedents could reflect age rather than the virus. To address this, the team calculated the age distribution of COVID-19 deaths in the N3C within five-year age strata from 65 to 90 years and older. These age-specific weights were then applied to the stratum-specific percentages of influenza decedents with diabetes in Medicare, a technique known as direct standardization. The result estimates the percentage of people dying of influenza who had diabetes if those decedents had the same age distribution as the COVID-19 decedents. Confidence intervals were calculated two-tailed using an alpha of 0.05 and a z-score of 1.96 as the critical value.

The headline numbers emerged from this design. Among older adults with inpatient mortality due to COVID-19 in the N3C, 46.6 percent had diabetes, with a 95 percent confidence interval of 46.1 to 47.0 percent. Among older adults with inpatient mortality from influenza in Medicare, the crude percentage with diabetes was 61.2 percent. After age-standardizing the Medicare data to match the age distribution of the N3C COVID-19 decedents, the percentage of influenza decedents with diabetes was 63.8 percent, with a 95 percent confidence interval of 61.5 to 66.1 percent. The authors summarize the finding bluntly: among older adults with inpatient mortality from respiratory viruses, a very large proportion had diabetes before infection, roughly 63 percent of influenza decedents and 47 percent of COVID-19 decedents. A high proportion of decedents having diabetes, they conclude, is not new and not unique to COVID-19.

The report is careful about what the analysis does and does not show. The authors explicitly state that they did not assess diabetes as an independent risk factor for mortality in patients with COVID-19 compared with patients with influenza, and that the objective of the work was not to conduct an analysis of the risk of death associated with diabetes. They note that several meta-analyses have shown diabetes to be an independent risk factor for mortality from COVID-19, and that their study reported only the proportions of decedents from each virus who had diabetes. The discussion also points toward the potential value of diabetes management: a meta-analysis of adults with type 2 diabetes reported two medications with high-certainty evidence for inverse associations with COVID-19 mortality, with a summary relative risk of 0.69 for metformin and 0.83 for glucagon-like-peptide-1 receptor agonists, and another study of adults with diabetes reported that metformin was associated with better survival, with an odds ratio of 0.65.

The limitations are substantial and acknowledged in detail. The analysis did not account for potential confounding factors, including other comorbid conditions such as obesity, cardiovascular disease, and chronic kidney disease, or socioeconomic factors, in either database, and the authors stress that it should not be considered a causal analysis. Diabetes was defined through ICD codes or prescriptions for diabetes medications, an approach that could produce misclassification bias, although the authors note that such bias might be expected to be similar in both databases. The analysis was limited to inpatient deaths in both databases, which may limit generalizability. The authors also flag that pandemic-related strain on the healthcare system affected inpatient care during 2020 through 2023 compared with 2017 through 2019, and that older adults with diabetes were negatively affected by that strain, a factor that would be expected to increase the proportion of COVID-19 decedents with diabetes.

The broader conclusion the authors draw is about communication as much as epidemiology. Popular media articles implied that the proportion of COVID-19 decedents who had diabetes was elevated, yet the analysis demonstrates that among older adults with inpatient mortality after respiratory infection, the proportion of COVID-19 decedents who had diabetes was not more than the proportion among influenza decedents in prior years. Diabetes affects approximately 29.2 percent of adults aged 65 and older in the United States and is a serious risk factor for mortality from viral infections, so it is unsurprising that it appears frequently among the oldest and sickest patients who die from either virus. The findings, the authors argue, highlight the value of using available data to contextualize health communication to the public, so that statistics about who dies from COVID-19 are interpreted against the backdrop of who was already living with diabetes and dying from other respiratory infections before the pandemic began.

Subject of Research: Pre-existing diabetes among older adults who died in hospital from COVID-19 compared with influenza

Article Title: Among individuals who die of COVID-19, is the percentage who had diabetes actually higher than in those dying of other viral infections?

Article References: Reddy, N. V., Pate, V., Stürmer, T., Wong, R., Harper, J., Reusch, J. E., Wilkins, K. J., Tronieri, J., Buse, J. B., Johnson, S. G., & Bramante, C. T. (2026). Among individuals who die of COVID-19, is the percentage who had diabetes actually higher than in those dying of other viral infections?. Cardiovascular Diabetology – Endocrinology Reports, 12(1), Article 61. https://doi.org/10.1186/s40842-026-00325-0

Image Credits: AI Generated

DOI: 10.1186/s40842-026-00325-0

Keywords: COVID-19, influenza, diabetes, inpatient mortality, older adults, N3C, Medicare, age standardization, electronic health records, health communication, comorbidity, respiratory infections