Association of SARS-CoV-2 Infection With New-Onset Type 1 Diabetes Among Pediatric Patients From 2020 to 2021

This cohort study assesses the association of COVID-19 with new-onset type 1 diabetes among pediatric patients.


TriNetX Database and Statistical Analysis Description of TriNetX database:
The data used in this study was accessed on April 4, 2022 from the TriNetX COVID-19 Global Collaborative Network. This resource provides access to electronic health records (EHRs) (diagnoses, procedures, medications, laboratory values, genomic information) from over 90 million patients from 74 healthcare organizations, which is deidentified per criteria from the Health Insurance Portability and Accountability Act (HIPAA), Section §164.514(a) of the HIPAA Privacy Rule. MetroHealth System, Cleveland, Ohio, IRB has determined any research using TriNetX, is not Human Subject Research and therefore exempt from IRB review.
The TriNetX platform de-identifies and aggregates EHR data from 74 contributing healthcare systems, most of which are large academic medical institutions with both inpatient and outpatient facilities at multiple locations, across 50 states in the US and 14 countries. Patient EHR data includes information from hospitals, primary care, and specialty treatment providers, covering diverse geographic locations, age groups, racial and ethnic groups, income levels and insurance types including various commercial insurances, governmental insurance (Medicare and Medicaid), self-pay/uninsured, worker compensation insurance, military/VA insurance among others. Race and ethnicity data in TriNetX is derived from selfreports in the clinical EHR systems, which is then mapped to the following categories: (1) Race: Asian, American Indian or Alaskan Native, Black or African American, Native Hawaiian or Other, White, Unknown race; and (2) Ethnicity: Hispanic or Latino, Not Hispanic or Latino, Unknown Ethnicity.

Statistical analysis:
The status of SARS-CoV-2 infection was based on the lab-test confirmed presence or the International Classification of Diseases (ICD-10) diagnosis codes (complete cohort terms defined below). The outcome measure of type 1 diabetes (T1D) was determined by the presence of ICD-10 code E10 ("Type 1 diabetes mellitus"). SARS-CoV-2 infections that were diagnosed between March 11, 2020 and December 1, 2021 were recorded for children under age 18 years. We also established a cohort of children diagnosed with other respiratory infection (cohort terms below) who were never diagnosed with SARS-CoV2 during the same time period. Each cohort was matched with the cohort of children with SARS-CoV2 infection by the TriNetX built-in propensity score matching function (1:1 matching using a nearest neighbor greedy matching algorithm with a caliper of 0.25 times the standard deviation). Risk of new diagnosis of type 1 diabetes following date of infection was then compared for the SARS-CoV2 cohort to the other respiratory infection cohort (or the fracture or routine visit cohorts) using hazard ratios and 95% confidence intervals. Kaplan-Meier analysis was used to estimate the probability of clinical outcomes.
Cox's proportional hazards model was used to compare the two matched cohorts. The proportional hazard assumption was tested using the generalized Schoenfeld approach. For sensitivity analysis, we also created two additional control cohort of children: (1) who had broken bones during this period but were never diagnosed with SARS-CoV2 infection, and (2) who had routine encounters with the health care system but were never diagnosed with SARS-CoV2. Risks for new diagnosis of T1D were compared between children with SARS-CoV-2 infection and these control cohorts as described above. The TriNetX Platform calculates the hazard ratios and associated confidence intervals, using R's Survival package v3.2-3. For generating hazard ratios, TriNetX sets robust=FALSE using the R survival package, but it does not take into account potential clustering of COVID-19 cases within the healthcare organizations or specific geolocations, a potential weakness or confounding factor in the analysis.
The inclusion criteria for the COVID-19 cohorts are found below and include a mix of diagnostic codes and positive laboratory tests. This same list was also used as exclusion criteria for the control cohorts. Cohort terms were adopted from ref 5.