To the Editor: We thank Dr Ikomi and colleagues [1] for their letter and supportive comments. It was interesting to hear of your experience in south west Essex and particularly of your successful treatment of women who ‘fell through the net’ with normalisation of neonatal macrosomia rates.

Our main reason for choosing the study period of 2004–2008 was that we had an existing database covering this period that had been used for a previous service evaluation to assess the value of the 1 h OGTT test and to evaluate our screening strategy for postpartum diabetes [2]. During the study period (2004–2008), it was our normal practice to take blood tests at baseline, 1 and 2 h after the OGTT but the results of the 1 h test were ignored. In 2008 we re-evaluated the 1 h test after reviewing the case of a 30-year-old primigravida with a negative OGTT (5.1, 13.1 and 7.1 mmol/l at 0, 1 and 2 h, respectively) who went on to have polyhydramnios, emergency Caesarean section and pre-term delivery of a large-for-gestational-age infant. Her postpartum OGTT showed frank diabetes (6.9, 16.3 and 11.4 mmol/l) with no elevation in GAD antibodies. In 2009, we introduced the 1 h OGTT test at a threshold of ≥10.0 mmol/l as diagnostic of gestational diabetes (GDM).

Using an older database allowed us to assess the effects of changing diagnostic criteria on our population before the 1 h criterion was used in diagnosis and enabled us to have the results available in a timely manner. Although we appreciate that a more recent cohort might have been advantageous in some ways, there were several barriers to that in this case. First, we changed to criteria similar to the International Association of the Diabetes and Pregnancy Study Groups IADPSG/WHO-2013 criteria during 2012 (75 g OGTT 0 h ≥5.3 mmol/l, 1 h ≥10.0 mmol/l, 2 h ≥8.5 mmol/l) [3, 4]. After this time point, women who ‘fell through the net’ were offered treatment in our institution that would prevent a true estimation of the odds ratios for ‘missed’ diagnoses. A further barrier was that our hospital changed to an Epic ehospital system in 2014 that moved paper records to electronic files, which has made it difficult to access retrospective data in a timely manner. However, the incidence of GDM was 4.88% in 2013 using the same screening criteria. This shows a gradual but not dramatic increase in incidence since 2008 and we therefore believe our retrospective data remain relevant to our population.

Our screening criteria have been criticised as being out of line with current or previous National Institute for Health and Care Excellence (NICE) recommendations [5]. While this is true, the screening criteria used were a modified two-step approach, comparable with that used in the USA for many decades. The risk factor-based screening advocated by NICE has documented flaws [6], which has led us to believe that a universal screening practice offers a more robust way of screening for GDM. For example, the NICE guidelines advocate offering screening to women with a previous macrosomic infant or a previous diagnosis of GDM, which makes women in their first pregnancy less likely to be screened than parous women [7]. As the importance of fasting hyperglycaemia has come to light in this study, we admit that our screening strategy using the 50 g glucose challenge test may miss some women [8], and we are currently reviewing our service data in order to design a better screening protocol. We agree with the closing statements of Ikomi and colleagues [1] about the importance of reviewing contemporary service data to guide ongoing care.

A further criticism has been that rates of delivery by Caesarean section (CS) and macrosomia/large for gestational age were higher in our study compared with the Hyperglycaemia and Adverse Pregnancy Outcomes (HAPO) cohort [9]. This is to be expected, as women with glucose screening prior to recruitment or previous gestational diabetes were excluded from the HAPO study, while our study included all singleton pregnancies. Our institution is a tertiary referral centre for high-risk antenatal and neonatal care and the relatively high CS rate may reflect this.

The diagnosis of GDM is no trivial matter, and unnecessary medicalisation of pregnancy should be avoided. However, the consequences of unidentified GDM and unexpected shoulder dystocia also carry considerable emotional and physical repercussions for both mother and child [10]. Our data suggest similar risk profiles for women above and below the NICE cut-off of 5.6 mmol/l. The recent supreme court ruling on CS delivery in diabetes highlights the principle of discussing ‘what a reasonable patient wants to know’ [11]. Perhaps it is time to ask women what they think.