[Skip to Navigation]
Sign In
Figure.  New Diagnoses Before and After COVID-19 Infection
New Diagnoses Before and After COVID-19 Infection

The frequency of new diagnoses (diabetes, hyperlipidemia, hypertension, and benchmark conditions) occurring 90 days before and after COVID-19 infection is displayed as color bars in the overall cohort of 23 709 patients (A), in the subset of 14 856 patients who were unvaccinated prior to infection (B), and in the subset of 8853 patients who were vaccinated prior to infection (C).

Table.  Multivariable-Adjusted Risk for New Cardiometabolic Diagnosis After COVID-19 Infectiona
Multivariable-Adjusted Risk for New Cardiometabolic Diagnosis After COVID-19 Infectiona
1.
Zhang  T, Mei  Q, Zhang  Z,  et al.  Risk for newly diagnosed diabetes after COVID-19: a systematic review and meta-analysis.   BMC Med. 2022;20(1):444. doi:10.1186/s12916-022-02656-yPubMedGoogle ScholarCrossref
2.
Xie  Y, Xu  E, Bowe  B, Al-Aly  Z.  Long-term cardiovascular outcomes of COVID-19.   Nat Med. 2022;28(3):583-590. doi:10.1038/s41591-022-01689-3PubMedGoogle ScholarCrossref
3.
Al-Aly  Z, Xie  Y, Bowe  B.  High-dimensional characterization of post-acute sequelae of COVID-19.   Nature. 2021;594(7862):259-264. doi:10.1038/s41586-021-03553-9PubMedGoogle ScholarCrossref
4.
Karim  SSA, Karim  QA.  Omicron SARS-CoV-2 variant: a new chapter in the COVID-19 pandemic.   Lancet. 2021;398(10317):2126-2128. doi:10.1016/S0140-6736(21)02758-6PubMedGoogle ScholarCrossref
5.
Wei  W-Q, Bastarache  LA, Carroll  RJ,  et al.  Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record.   PLoS One. 2017;12(7):e0175508. doi:10.1371/journal.pone.0175508PubMedGoogle ScholarCrossref
6.
Redelmeier  DA.  The exposure-crossover design is a new method for studying sustained changes in recurrent events.   J Clin Epidemiol. 2013;66(9):955-963. doi:10.1016/j.jclinepi.2013.05.003PubMedGoogle ScholarCrossref
2 Comments for this article
EXPAND ALL
Caution needed in interpreting causation and results
Kyle Sheldrick, MBBS, PhD | UNSW
This is an interesting finding, however great caution is needed in interpreting the strength of the results, and drawing causal inference.

The potential for ascertainment bias here is very high. Of particular note the negative control diagnoses (in this paper referred to as "benchmark" diagnoses) are fundamentally dissimilar to the diagnoses of interest.

All the diagnoses of interest are diagnosed from investigation results or physical findings (DM, HTN and dyslipidemia), all of them will often be diagnosed incidentally during routine workup and do not necessarily require any symptomatic complaint.

The benchmark diagnoses such as GERD by contrast
are symptomatic conditions diagnosed primarily based on a patient complaint, and so are less likely to be diagnosed incidentally during the workup of another condition.

Because of this fundamental difference in the way the diagnoses of interest vs the benchmark diagnoses are usually diagnosed, it is not possible to say that this finding is not simply a result of the workup, examination and investigation undergone by patients as a result of a COVID diagnosis, rather than genuine sequelae. Likewise the differential rates based on vaccination status could well accord with prior healthcare engagement being correlated with vaccine uptake.

It is still possible that there is a genuine causal pathway between COVID-19 and diabetes, however care needs to be taken in assessing the strength of evidence provided by this paper, given the results could be partially or wholly driven by ascertainment bias.
CONFLICT OF INTEREST: None Reported
READ MORE
No evidence of bias
Shawn Grimm |
The previous comment on potential ascertainment bias is addressed in the study: “There was no evidence of interaction by age, sex, or preexisting cardiovascular risk factors, including hypertension or hyperlipidemia. Age, sex, and timing of index infection regarding the Omicron variant were not associated with an increased risk of a new cardiometabolic diagnosis before or after COVID-19 infection in any model (Table).” The table suggests no evidence of benchmark bias.
CONFLICT OF INTEREST: None Reported
Research Letter
Infectious Diseases
February 14, 2023

Association of COVID-19 Vaccination With Risk for Incident Diabetes After COVID-19 Infection

Author Affiliations
  • 1Smidt Heart Institute, Department of Cardiology, Cedars Sinai Medical Center Los Angeles, California
  • 2Cardiovascular Division, Brigham and Women’s Hospital, Boston, Massachusetts
JAMA Netw Open. 2023;6(2):e2255965. doi:10.1001/jamanetworkopen.2022.55965
Introduction

In early phases of the COVID-19 pandemic, persons who recovered from infection had increased risks for new-onset cardiometabolic diseases, including diabetes, hypertension, and hyperlipidemia.1-3 In the current pandemic phase, which is dominated by less virulent Omicron variants,4 it remains unclear whether risks of cardiometabolic disease after COVID-19 infection persist or have become attenuated and whether vaccination status is associated with these risks.

Methods

This large cohort study of adult patients with 1 or more COVID-19 infections treated within the Cedars-Sinai Health System in Los Angeles, California from March 2020 to June 2022 used International Classification of Diseases, Ninth Revision and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes (eTable in Supplement 1) to identify cardiometabolic diagnoses (hypertension, hyperlipidemia, and diabetes) newly reported before or after a patient’s first COVID-19 infection.5 A self-controlled exposure-crossover design6 was used to estimate odds of a new cardiometabolic diagnosis occurring 90 days after vs 90 days before COVID-19 infection. To account for temporal confounders arising from disruptions in health care use during the pandemic, we compared the odds of a new cardiometabolic diagnosis with those of a new benchmark diagnosis (ie, urinary tract infection and gastroesophageal reflux), representing a marker of health care engagement unrelated to COVID-19. In multivariable logistic regression models, we estimated the odds ratio (OR) for a new cardiometabolic vs new benchmark diagnosis occurring 90 days after vs before infection while adjusting for age, sex, timing of infection (before vs after emergence of Omicron variant), and COVID-19 vaccination status (eMethods in Supplement 1). The study was approved by the Cedars Sinai Medical Center Institutional Review Board, which waived the informed consent requirement given the study’s retrospective nature. We followed the STROBE reporting guideline.

Data were analyzed using R, version 4.2.1 (R Foundation for Statistical Computing). The threshold for statistical significance was a 2-tailed P <.05.

Results

The cohort of 23 709 patients (mean [SD] age, 47.4 [19.3] years) included 12 706 females (54%) and 10 981 males (46%) (22 patients of unknown sex) with 1 or more COVID infection. Rates of new-onset diabetes, hypertension, hyperlipidemia, and benchmark diagnoses occurring in the 90 days after COVID-19 infection were higher than those before infection (Figure). The highest odds postinfection were for diabetes (2.35; 95% CI, 1.94-2.89; P < .001), followed by hypertension (1.54; 95% CI, 1.35-1.76; P < .001), benchmark diagnoses (1.42; 95% CI, 1.25-1.61; P < .001), and hyperlipidemia (1.22; 95% CI, 1.03-1.47; P = .03). In adjusted multivariable models, risk of new-onset diabetes (vs benchmark) diagnosis occurring after vs before COVID-19 infection was significantly elevated (OR, 1.58; 95% CI, 1.24-2.02; P < .001); however, risks of hypertension and hyperlipidemia vs benchmark diagnoses were not (Table). Although the diabetes risk after infection was higher among unvaccinated (OR, 1.78; 95% CI, 1.35-2.37; P < .001) than vaccinated (OR, 1.07; 95% CI, 0.64-1.77; P = .80) patients, the interaction term between vaccination status and diabetes diagnosis was not statistically significant (OR, 0.59; 95% CI, 0.34-1.06; P = .08). There was no evidence of interaction by age, sex, or preexisting cardiovascular risk factors, including hypertension or hyperlipidemia. Age, sex, and timing of index infection regarding the Omicron variant were not associated with an increased risk of a new cardiometabolic diagnosis before or after COVID-19 infection in any model (Table).

Discussion

In this cohort study, COVID-19 infection was associated with increased risk of diabetes, consistent findings of a meta-analysis.1 Our results suggest that this risk persisted as the Omicron variant became predominant, and the association remained even after accounting for temporal confounders. Diabetes risk after COVID-19 infection was higher in unvaccinated than vaccinated patients, suggesting a benefit of vaccination. Mechanisms contributing to postinfection diabetes risk remain unclear, although persistent inflammation contributing to insulin resistance is a proposed pathway. Study limitations include reliance on diagnostic coding, unaccounted confounders (infection severity indices), and insufficient sample size and statistical power for testing multiple interactions. Additional studies are needed to understand cardiometabolic sequelae of COVID-19 and whether COVID-19 vaccination attenuates risk of cardiometabolic disease.

Back to top
Article Information

Accepted for Publication: December 26, 2022.

Published: February 14, 2023. doi:10.1001/jamanetworkopen.2022.55965

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2023 Kwan AC et al. JAMA Network Open.

Corresponding Author: Alan C. Kwan, MD, MSc, 127 S San Vicente Blvd, AHSP A3600, Los Angeles, CA 90048 (alan.kwan@cshs.org).

Author Contributions: Alan Kwan and Susan Cheng had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kwan, Cheng.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Kwan, Cheng.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Kwan, Claggett.

Obtained funding: Cheng.

Administrative, technical, or material support: Kwan, Ebinger, Botting, Cheng.

Supervision: Cheng.

Conflict of Interest Disclosures: Dr Kwan reported receiving grants from the Doris Duke Charitable Foundation during the conduct of the study. Dr Claggett reported receiving personal fees from consulting for Cardurion, Corvia, Cytokinetics, Intellia, and Novartis outside the submitted work. Dr Cheng reported receiving grants from the National Institutes of Health (NIH) and personal fees from Zogenix outside the submitted work. No other disclosures were reported.

Funding/Support: This work was funded by the Erika J. Glazer Family Foundation, the Doris Duke Charitable Foundation (grant 2020059; Dr Kwan), and NIH grants R01-HL151828 (Dr Cheng), R01-HL131532 (Dr Cheng), and K23-HL153888 (Dr Ebinger).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.

References
1.
Zhang  T, Mei  Q, Zhang  Z,  et al.  Risk for newly diagnosed diabetes after COVID-19: a systematic review and meta-analysis.   BMC Med. 2022;20(1):444. doi:10.1186/s12916-022-02656-yPubMedGoogle ScholarCrossref
2.
Xie  Y, Xu  E, Bowe  B, Al-Aly  Z.  Long-term cardiovascular outcomes of COVID-19.   Nat Med. 2022;28(3):583-590. doi:10.1038/s41591-022-01689-3PubMedGoogle ScholarCrossref
3.
Al-Aly  Z, Xie  Y, Bowe  B.  High-dimensional characterization of post-acute sequelae of COVID-19.   Nature. 2021;594(7862):259-264. doi:10.1038/s41586-021-03553-9PubMedGoogle ScholarCrossref
4.
Karim  SSA, Karim  QA.  Omicron SARS-CoV-2 variant: a new chapter in the COVID-19 pandemic.   Lancet. 2021;398(10317):2126-2128. doi:10.1016/S0140-6736(21)02758-6PubMedGoogle ScholarCrossref
5.
Wei  W-Q, Bastarache  LA, Carroll  RJ,  et al.  Evaluating phecodes, clinical classification software, and ICD-9-CM codes for phenome-wide association studies in the electronic health record.   PLoS One. 2017;12(7):e0175508. doi:10.1371/journal.pone.0175508PubMedGoogle ScholarCrossref
6.
Redelmeier  DA.  The exposure-crossover design is a new method for studying sustained changes in recurrent events.   J Clin Epidemiol. 2013;66(9):955-963. doi:10.1016/j.jclinepi.2013.05.003PubMedGoogle ScholarCrossref
×