Results
A total of 502 patients, including 161 pregnant women with PGDM, 171 pregnant women with GDM and 170 controls, were evaluated. The PGDM group consisted of 98 (60.8%) women with Type 1 and 63 (39.2%) women with Type 2 DM. Demographic characteristics and ultrasound findings of three groups were showed in Table 1. Patients in PGDM group were significantly older than both control and GDM groups (p=0.002) (Table 1). FPG, 1-h and 2-h blood glucose values of patients in GDM group were found to be significantly higher than control group (p<0.001). HbA1C values were significantly higher in both GDM and PDGM groups compared to control group (p<0.001). As expected, HbA1c values of patients with PGDM were significantly higher than those with GDM (p=0.021). Body mass index was significantly higher in PGDM group compared to control group (p=0.047). In ultrasound measurements, AC and DVP were found to be significantly higher in PGDM and GDM groups compared to control group and also significantly higher in PGDM group compared to GDM group (p<0.001).
The mean of EFT measurements were found to be 1.19 ± 0.049 mm, 1.40 ± 0.082 mm and 1.47 ± 0.083 mm for the control, GDM and PGDM groups, respectively (Table 1). There was a significant difference between all three groups in terms of EFT (p<0.001). After adjusting by Bonferroni correction, it was observed that the mean EFT of control group was lower than the mean values of GDM and PGDM groups. Furthermore, the mean EFT of PGDM group was higher than the mean EFT of GDM group (p<0.001). Comparison of the mean values of EFT using a box plot among three groups were demonstrated in Figure 3a.
Correlation analysis was performed to test the relationship between EFT and demographic characteristics of patients (Table 2). There was a statistically significant weak positive correlation between EFT and maternal age (r=0.184, p<0.001), FPG (r=0.188, p=0.005) and DVP (r=0.332, p<0.001), moderate positive correlation between EFT and fetal AC (r=0.422, p<0.001), HbA1C (r=0.559, p<0.001) and strong positive correlation between EFT and 1st hour (r=0.650, p<0.001) and 2nd hour glucose values (r=0.719, p<0.001). ROC analysis was performed to identify optimal cut-off points in differentiating pregnancies with GDM and pregestational DM from normal pregnancies (Figure 3b,c). The ROC curve of the multivariate logistic regression model for the diagnosis of GDM demonstrated EFT with other variables (FPG, 1hPG, 2hPG, HbA1c, AC and DVP) have an AUC of (1.000, 95% CI 1.000-1.000), which is larger than when using the individual factors of EFT (0.985, 95% CI 0.973-0.997), FPG (0.643, 95% CI 0.565-0.720), 1hPG (0.956, 95% CI 0.927-0.984), 2hPG (0.976, 95% CI 0.953-0.999), HbA1c (0.902, 95% CI 0.861-0.944), AC (0.696, 95% CI 0.628-0.764) and DVP (0.669, 95% CI 0.595-0.742). The ROC curve of the multivariate logistic regression model for the diagnosis of PGDM demonstrated EFT with other variables (HbA1c, AC and DVP) have an AUC of (1.000, 95% CI 1.000-1.000), which is larger than when using the individual factors of EFT (0.999, 95% CI 0.998-1.000), HbA1c (0.934, 95% CI 0.902-0.967), AC (0.858, 95% CI 0.810-0.906) and DVP (0.814, 95% CI 0.756-0.873). Using a cut-off level of 1.27 mm, EFT diagnosed GDM with a sensitivity of 94% and specificity of 95% (AUC:0.95 [95 CI 0.973-0.997]). Furthermore, using a cut-off level of 1.3 mm, EFT diagnosed PGDM with a sensitivity of 97.3% and specificity of 98.2% (AUC:0.999 [95 CI 0.998-1.000]). We also performed a ROC curve analysis for assessing the performance of fetal EFT value in predicting macrosomia. The AUC was  0.945 (95% CI 0.912-0.979) for macrosomia. The optimal fetal EFT threshold for predicting macrosomia was found as 1.36 mm, with a sensitivity of 93.4% and specificity of 95.8%. We did not calculate inter-observer reproducibility for EFT because all measurements were obtained by a single clinician. However, intra-observer analysis demonstrated good repeatability and reproducibility of EFT. The interclass correlation coefficient of intra-observer reproducibility for EFT measurement was 0.755 (95 % CI 0.526-0.984). Intra-observer variability of measurements of EFT was found to be 0.5 ± 0.4 mm.