Association of COVID-19 with diabetes: a systematic review and meta-analysis

Abstract

Emerging evidence suggests that coronavirus disease-2019 (COVID-19) may lead to a wide range of post-acute sequelae outcomes, including new onset of diabetes. The aim of this meta-analysis was to estimate the incidence of newly diagnosed diabetes in survivors of COVID-19. We searched MEDLINE, Scopus, Cochrane Central Register of Controlled Trials and the World Health Organization Global Literature on Coronavirus Disease and clinical trial registries for studies reporting the association of COVID-19 and diabetes. Search dates were December 2019-October 16, 2022. Two investigators independently assessed studies for inclusion. Risk of bias was assessed using the Newcastle-Ottawa Scale. We estimated the effect of COVID-19 on incident diabetes by random-effects meta-analyses using the generic inverse variance method. We identified 8 eligible studies consisting of 4,270,747 COVID-19 patients and 43,203,759 controls. Median age was 43 years (interquartile range, IQR 35-49), and 50% were female. COVID-19 was associated with a 66% higher risk of incident diabetes (risk ratio, 1.66; 95% CI 1.38; 2.00). The risk was not modified by age, sex, or study quality. The median risk of bias assessment was 7. In this systematic review and meta-analysis, COVID-19 was associated with higher risk for developing new onset diabetes among survivors. Active monitoring of glucose dysregulation after recovery from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection is warranted.

Figure 1

PRISMA flow chart of a systematic review of diabetes incidence in survivors of COVID-19. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Figure 2

Forest plot for the overall pooled estimate for the association of COVID-19 and incident diabetes. Effect size values represent risk ratio and corresponding 95% CI. Blue squares and their corresponding lines are the point estimates of each study and 95% confidence intervals (95% CI). Maroon diamonds represent the pooled estimate (width denotes 95% CI). Heterogeneity (I² = 94%, p for heterogeneity < 0.0001; 8 studies).

Figure 3

Forest plot of studies stratified by geographic regions.

Figure 4

Funnel plots to assess potential for small-study publication bias. Symmetrical inverted funnel plot suggested absence of publication bias.

Figure 5

Funnel plots from trim and fill analysis. Duval & Tweedie trim and fill analytical method suggests that the adjusted effect estimates would fall in the range of 1.21 to 1.88, and 2 studies were added.

Figure 6

Influence and outlier (leave-one-out meta-analysis) analysis for the association of COVID-19 and incident diabetes. The results of our outlier and influence analysis show the recalculated pooled point estimate ranged from 1.55 to 1.72 when one study was omitted each time.