Post‐natal assessment of gestational diabetes: fasting glucose or full glucose tolerance test?

Aims To determine the performance of a fasting glucose sample compared with a full oral glucose tolerance test for the detection of glucose abnormalities in a diverse ethnic population after gestational diabetes. Methods Oral glucose tolerance test results for women attending post‐natal testing over a 10‐year (2003–2013) period at St Mary's Hospital, Manchester, UK were reviewed. Demographic data were also extracted from the hospital maternity database. Results The average attendance for a post‐natal oral glucose tolerance test was approximately 45% over the study period. The prevalence of diabetes was 4.8% (30/629), with a higher rate in women of Asian ethnicity compared with other groups (6.6% vs. 3.5%). The sensitivity for a fasting plasma glucose of ≥ 6.1 mmol/l was 90% (95%  CI 74.4–96.5%) for the detection of diabetes and 61% (49.1–71.5%) for the detection of diabetes and/or impaired glucose tolerance, with specificities of 91% (88.8–93.3%) and 93% (91.0–95.2%), respectively. The positive and negative likelihood ratios for diabetes and impaired glucose tolerance were 10.4 (7.8–13.8), 0.11 (0.03–0.32) and 9.2 (6.4–13.3), 0.42 (0.31–0.56), respectively. A fasting plasma glucose threshold of 5.6 mmol/l improved the sensitivity for impaired glucose tolerance (from 61% to 77%), but made no difference to the sensitivity for diabetes. Conclusions The current study has demonstrated that detection of diabetes after gestational diabetes, in an ethnically diverse population using a fasting plasma glucose only, was approximately 90%. Compliance with post‐natal screening might improve if women attended for a fasting plasma glucose only; this strategy would identify approximately 90% of cases of diabetes and 40% of cases of impaired glucose tolerance.