Agreement and disagreement between diagnostic criteria for gestational diabetes and implications for clinical practice: A retrospective observational study

Abstract

Aims

To evaluate agreement/disagreement between eleven gestational diabetes (GDM) diagnostic criteria, including five used in current clinical practice globally.

Methods

Records of 353 pregnant women with oral glucose tolerance test performed after 20 weeks of gestation were retrospectively reviewed. The diagnosis of GDM was compared using the IADPSG, DIPSI, WHO 1999, CDA 2003 and 2013, NICE 2015, JSOG 1984, ADIPS 1998, ADA 2004, NZSSD 2004 and EASD 1996 criteria. The agreement between criteria was expressed as Cohen's kappa coefficient (k; 0.4–0.6, moderate; 0.6–0.8, good; 0.8–1.0, very good) and disagreement as percentage (d). IADPSG criteria were used as a reference for comparison.

Results

The prevalence of GDM varied from 7.4 % (95 % CI, 4.9–10.4 %) by CDA 2003 criteria to 23.8 % (95 % CI, 19.5–28.4 %) by IADPSG criteria. Of the 55 pair-wise criteria comparisons, 29 (52.7 %) showed moderate, 16 (29.1 %) good, and 10 (18.2 %) very good agreement. Among the currently used criteria, the CDA 2013 (k = 0.811; d = 6.2 %) agreed the most, the DIPSI/WHO 1999 (k = 0.456) agreed the least, and the NICE 2015 (k = 0.580) criteria showed an intermediate agreement with the IADPSG criteria.

Conclusions

There is a marked variation in the prevalence of GDM, with a significant degree of disagreement between different diagnostic criteria. The study findings should be interpreted in the context of its retrospective nature and non-consecutive recruitment, which introduce a potential for selection bias.