Association of T-wave electrocardiogram changes and type 2 diabetes: a cross-sectional sub-analysis of the MASHAD cohort population using the Minnesota coding system

The current cross-sectional study was conducted on the population of Mashhad stroke and heart atherosclerotic disorder (MASHAD) cohort study [24]. A total of 9704 individuals aged 35 to 65 years were enrolled into this cohort study. A checklist containing participants’ demographic data including age, sex, educational level, and marital status was recorded. Patients whose systolic blood pressure levels were at or above 140 mmHg and/or diastolic blood pressure were at or beyond 90 mmHg—measured using a mercury sphygmomanometer- were considered hypertensive [25]. A fasting blood glucose (FBG) > 126 mg/dl, or being under anti-hyperglycemic medication was defined as diabetic patients [26]. The FBG was measured in a peripheral blood sample following 14 h of fasting [27]. The study was approved by the Human Research ethics committee of Mashhad University of Medical Sciences, Mashhad, Iran, and all participants provided informed consent prior to data collection.

A standard resting 12-lead ECG was taken from each participant of the study. These ECGs were interpreted by instructed medical students in accordance with Minnesota coding system [28]. Five percent of all ECGs were also read by certified cardiologists. Among the 9704 participants, the ECGs of 9035 participants were available and readable according to Minnesota coding system [28].

Four different t-wave abnormalities were described within the coding system including codes 5–1, 5–2, 5–3 and 5–4. The code 5–1 was defined as T amplitude negative 5.0 mm or more in either of leads I, V6, or in lead aVL when R amplitude is ≥ 5.0 mm and code 5–2 was defined as T amplitude negative or diphasic (positive–negative or negative–positive type) with negative phase at least 1.0 mm but not as deep as 5.0 mm in lead I or V6, or in lead aVL when R amplitude is ≥ 5.0 mm. Code 5–3 was described as flat, negative or diphasic t-wave with less than 1 mm negative phase in any leads of I, II or V3 to V6 or in lead aVL when the R amplitude is ≥ 5.0 mm. Lastly, code 5–4 was defined as a positive T amplitude and a T/R amplitude ratio < 1:20 in any of leads I, II, aVL, or V3 through V6. The R-wave amplitude must be ≥ 10.0 mm [28].

Qualitative and quantitative variables were summarized as Mean \(\pm\) SD and frequency (%), respectively. An independent t-test was used in order to compare the mean of quantitative variables between the two groups. In addition, evaluating the association between qualitative variables was performed using Chi-square and Fisher's exact test. Further analyses were performed in order to investigate the association between T wave impairments and T2DM after adjusting the effect of potential confounders (variables with P < 0.25 in the univariate logistic regression model) using the multiple logistic regression (LR) model. Furthermore, receiver operating characteristic (ROC) curves were used to evaluate the ability of the multiple LR model to predict the occurrence of TWA and T2DM. All statistical analyses were carried out using SPSS version 20 and the statistical significance level was considered at 0.05.

A total of 9035 subjects were included into this study, including 1273 diabetic patients and 7762 non-diabetic individuals (Fig. 1). The average age was 47.45 ± 8.17 years and 51.77 ± 7.73 years in non-diabetic and diabetic patients which differed significant (p < 0.001). Diabetic patients were found to have higher body mass index (BMI), as well as higher rates of hypertension (50.3 vs 27.9%, p < 0.001). Marital status and educational levels also showed a significant different distribution between the two diabetic and non-diabetic groups with married being the most prevalent status among studied groups (P < 0.001). Table 1 presents patients’ demographic data distributions.

A total of 1246 T wave abnormalities were reported among the study sample population, approximately 13.79% of all participants. The most frequent TWA among both groups were code 5–2 (4.9% in diabetics and 3.6% in the control group) and major T-wave abnormalities (5% in diabetics and 3.7% in the control group). Different TWA yielded varying associations with T2DM. While T-wave abnormalities code 5–1 and 5–4 failed to show a significantly different distribution among diabetic and non-diabetic participants (P = 0.24 and 0.92 respectively), code 5–2 and 5–3 were shown to be significantly higher among diabetic patients compared to the non-diabetic individuals (P = 0.02 and 0.01, respectively). Overall, both major and minor T-wave abnormalities were significantly more frequent among patients with T2DM compared to the control group, (p = 0.02 and 0.008, respectively). Figures 2 and 3 compare T wave impairments and T2DM distribution.

Results from the multiple logistic regression models indicated a significant association between age (OR = 1.05, 95%CI = 1.04–1.05) and BMI (OR = 1.03, 95%CI = 1.02–1.05) with having T2DM. Gender, marital status, and educational level did not show a significant association with having T2DM (all P > 0.05) (Tables 2 and 3). Hypertension has been reported to increase the odds of T2DM by 86 percent (95%CI = 1.63–2.12, p < 0.001). According to Tables 2 and 3, only major and minor T wave impairments as well as impairments code 5–2 and 5–3 were reported to be higher among diabetic patients and thus only these items were further analyzed by inclusion in the multiple LR model. A model analyzing T-wave abnormality code 5–2 and 5–3 showed that the odds of having T2DM among patients with T-wave code 5–2 and 5–3 abnormalities were 1.07 and 1.31 times as those without these abnormalities, respectively. This observed difference between patients with and without T-wave abnormalities regarding having T2DM failed to yield statistical significance (P = 0.63 and 0.12, respectively). The area under the ROC curve (AUC) of the final multiple LR model was 0.6847, which indicates a good predictive power of the final model, as shown in Fig. 4A. Also, the results of our model for major and minor T-wave abnormality revealed that the odds of having T2DM in patients who had T major and T minor abnormalities were 1.06 and 1.30 times than those without ischemia abnormalities, respectively. However, this difference did not show a significant difference within the logistic regression model (P = 0.65 and 0.11 respectively). The AUC for this model was 0.6846 which suggests a good predictive power of the final model, as shown in Fig. 4B. Tables 2 and 3 presents the results of the regression model analyses.