Predictive diagnostic models for newly diagnosed diabetes mellitus in moderate to severe COVID-19: the role of TyG Index, BMI, and inflammatory markers

Background:
The association between COVID-19 and newly diagnosed diabetes mellitus (DM) remains uncertain. This cross-sectional study examines the role of insulin resistance (IR) and selected inflammatory markers in COVID-19 associated newly diagnosed DM.

Research design and methods:
A cross-sectional pilot study was conducted at an academic tertiary hospital and a primary healthcare facility, with COVID-19 patients additionally followed for three months post-discharge. Participants included patients hospitalised with moderate to severe COVID-19 during the third wave of predominantly the delta variant. Diagnostic markers predictive of newly diagnosed DM were assessed using logistic regression analysis. Four predictive diagnostic models were developed, incorporating combinations of triglyceride-glucose index (TyG index), homeostatic model assessment of insulin resistance (HOMA-IR), body mass index (BMI) and inflammatory cytokines. Model performance and optimal cutoff values were determined using Receiver Operating Characteristic (ROC) analysis and the Youden index.

Results:
A total of 127 individuals were evaluated, consisting of 84 patients admitted with moderate to severe COVID-19 and 43 healthy controls. Among the 84 COVID-19 participants, 45 were newly diagnosed with DM, 20 had no DM, and 19 had pre-existing DM. Those with newly diagnosed DM exhibited significantly higher BMI and IR markers (HOMA-IR, and TyG index) compared to those without newly diagnosed DM (p < 0.001, p = 0.05 and p = 0.002, respectively). The predictive diagnostic model for newly diagnosed DM included the TyG index, BMI, IL-10 and IL-1β, achieving an area under the curve (AUC) of 0.91 (95% CI, 0.84–0.98). The TyG index was strongly associated with newly diagnosed DM (Crude Odds Ratio [COR] 11.25 (95% CI, 2.80-76.28; p-value = 0.01); Adjusted Odds Ratio (AOR) 6.83 (95% CI, 1.57, 42.96; p-value = 0.01) and showed improved predictive accuracy when used with BMI (AUC 0.86; 95% CI, 0.77–0.95), compared to the TyG index alone (AUC 0.73; 95% CI, 0.59–0.86). These findings support the potential role of the TyG index as a practical alternative to HOMA-IR in resource-limited settings where insulin measurement may not be feasible.

Conclusions:
In our study population, IR rather than insulin deficiency was more strongly associated with newly diagnosed DM in patients with COVID-19. The TyG index may serve as a practical diagnostic marker for predicting newly diagnosed DM in resource-limited settings, with BMI and inflammatory markers further improving model accuracy. However, given our predominantly Black African study population, validation in larger and more diverse populations is needed.