The treatment of booking gestational diabetes mellitus (TOBOGM) pilot randomised controlled trial

Gestational diabetes mellitus (GDM) increases the risk of obstetric and neonatal complications, with future risks to mother and baby [1‐3]. Two randomised controlled trials (RCTs) of GDM management from 24 to 28 weeks gestation, among women without pre-existing diabetes, showed significant reductions in the risk of both macrosomia and pre-eclampsia/Pregnancy Induced Hypertension [4, 5].

It has been recommended that women with possible undiagnosed diabetes are identified early in pregnancy and treated [6, 7]. However, during the screening process, a significant proportion of women are identified as below the biochemical criteria for ‘Diabetes in Pregnancy’, yet fulfilling criteria for GDM at 24–28 weeks gestation. Fasting glucose can be higher at the beginning of pregnancy [8], with one Chinese study reporting that many women with ‘early GDM’ did not have GDM when the oral glucose tolerance test (OGTT) was repeated at 24–28 weeks [8]. This has led to calls not to use the 24–28 week GDM criteria earlier in pregnancy [9]. However, this has left a void, with no recommendations on the next management steps for women with lesser degrees of hyperglycaemia early in pregnancy.

The Treatment of BOoking Gestational diabetes Mellitus (ToBOGM) Study is a randomised controlled trial (RCT) developed to test whether women with ‘hyperglycaemia’ at booking (Booking GDM), should be treated to avoid GDM complications, or whether earlier treatment increases other pregnancy complications. We now report on the findings of the ToBOGM pilot RCT investigating recruitment, proportion with ‘GDM’ at booking, blinding procedures, uptake of heel-prick glucose and outcome measures to facilitate power calculations.

The primary aim of this pilot study was to test the protocol for a larger scale RCT of either immediate or deferred treatment for Booking GDM. Clinic midwives at the Campbelltown Hospital Book-in clinic (i.e. first antenatal clinic attendance) assessed the need of all pregnant women for early testing for Diabetes in Pregnancy, based upon GDM Risk factors [10]. Women are referred for a 75 g, 3 point, 2 h oral glucose tolerance test (OGTT) as per local policy. Between July 20th 2015 and April 7th 2016, consecutive pregnant women at 4 to 19 + 6 weeks gestation, with a singleton pregnancy, aged ≥18 years and referred for an OGTT were invited to participate. Exclusions were inability to understand English, or a presence of a major active medical disorder. Consenting women completed questionnaires at baseline, 24–28 and 34–38 weeks and attended a study OGTT appointment. Other data were extracted from the clinical notes.

OGTT results were sent to one investigator (DS). Women with Diabetes In Pregnancy (fasting ≥7.0 and/or 2 h ≥ 11.1 mmol/l) were referred immediately to the clinic for treatment. Women below the 24–28 week GDM [4] criteria (fasting ≥5.1 and/or 1 h ≥ 10.0 mmol/l and/or 2 h ≥ 8.5 mmol/l), were advised that they did not require referral to the clinic (‘decoys’). Such decoys blinded the obstetric team as to whether women involved in ToBOGM were controls or women without ‘hyperglycaemia’. Women fulfilling criteria for GDM were randomised to either referral to the clinic (Treated) or were advised (using identical letters to the ‘decoys’) that they did not require referral to the clinic (No Treatment). The participant, as well as midwifery, obstetric, diabetes clinic, and research staff were kept blinded to all numeric results and only knew if a woman had been referred for GDM treatment. Randomisation to either immediate or deferred treatment was undertaken using an electronic randomiser (SPSS Version 22.0, IBM, USA), stratified as either low risk (fasting 5.1–5.2 mmol/l, 1 h 10–10.5 mmol/l and /or 2 h 8.5–8.9 mmol/l) or high risk (based upon the IADPSG 2 fold excess risk of adverse outcomes at 24–28 weeks gestation [6]: fasting 5.3–6.9 mmol/l, 1 h ≥ 10.6 mmol/l and /or 2 h 9.0–11.0 mmol/l).

All women receiving treatment for GDM received group education, were taught to self-monitor blood glucose and saw a dietitian. Fasting and 2 h post-prandial glucose targets were < 5.3 mmol/l and < 6.8 mmol/l respectively [5]. Where glucose thresholds were exceeded on more than two occasions with no obvious cause, women were offered medication (metformin or insulin treatment). Only women not referred to clinic attended a 24–28 week OGTT, and were referred for treatment if GDM was diagnosed.

All women were asked for their baby to provide a heel prick glucose 1–2 h after birth. Venous umbilical cord blood was drawn into EDTA tubes for assessing a range of metabolites and assessing the adipo-insular axis. Glucose, triglyceride and 3-beta hydroxy butyrate were measured in a single International Organization for Standardization accredited laboratory. Plasma insulin and C-peptide levels were measured using a sandwich chemiluminescence immunoassay (Liaison XL, Diasorin, Saluggia, Italy). Leptin and adiponectin measurements were determined using a radioimmunoassay kit (Merck Millipore, Darmstadt Germany).

Figure 1 includes the numbers invited (607), consented (100), and randomised (22). One woman randomised to No Treatment decided to withdraw from the trial and was referred to the GDM clinic and treated. Of the 128 eligible women, 100 (78.1%) consented and 79 (61.7%) entered the study. The randomisation process was adhered to, 88% of babies had the heel prick test and other outcomes were available for 96% of women/babies.

Table 1 compares the Treated and No Treatment groups, and ‘any booking GDM’ with the decoys. Compared with decoys, women with booking GDM had higher baseline BMI, serum insulin, leptin and 3 beta hydroxyl butyrate concentrations and lower adiponectin levels. At baseline, 4/21 (19%) with booking GDM, were diagnosed at the low and 17/21 (81%) at the higher glycaemic stratum; 17/21 (81%) had an elevated fasting glucose level with or without a higher post load glucose. Treated women had a higher systolic blood pressure than No Treatment women at baseline. Among the remaining women randomised to No Treatment, 8/9 (89%) had GDM on the 24–28 week OGTT while 11/56 (19.6%) had developed GDM in the decoy group. There were no differences between the Treated and No Treatment group for gestational age at delivery. Babies of women in the No Treatment group were more likely to have a large for gestational age (LGA) baby (p = 0.03), but babies of women in the Treated group were more likely to be admitted to NICU (p = 0.043: three small gestational age (SGA), one of whom was < 37 weeks). The only stillbirth was in the No Treatment group with placental abruption (not considered to be due to delayed treatment). Compared with women with booking GDM, decoys had more gestational weight gain (p = 0.001) and were less likely to be induced (p = 0.001).

The characteristics of midwives and women participating in the survey and focus groups are shown in Additional file 1: Table S1 and Additional file 2: Table S2. Of the women, one was a participant in the pilot study, one was currently pregnant but did not have GDM, one had recently had a baby and did not have GDM, and one had GDM in a previous pregnancy. All were familiar with the OGTT testing in pregnancy. Tables 2 and 3 describe the qualitative responses of midwives and antenatal women respectively from the open ended survey questions and focus groups. Overall, the midwives’ and the mothers responded positively to the study procedures. Respondents gave feedback on the importance of informed consent throughout all stages of the study. Most women reported preferring face-to-face initial contact. There were some concerns around the ethics of delaying treatment. The heelprick glucose was widely accepted. In the focus group, the midwives talked about the need to have a researcher dedicated to the study in the clinic, so that the study did not impact on their workload.

We have shown that our study protocol was feasible, requiring few changes. Randomisation and each step of the protocol were implemented with no significant problems. Both the mothers interviewed and the midwives were generally positive about the procedures, although there was some concern over the delay in treatment for women allocated to the No Treatment condition. In spite of this, most eligible women consented to the RCT and were able to enter the study.

Our overall capture rate for the heel prick was over 80%, higher than the 71.25% in HAPO [14]. Those not captured included three premature births that occurred at a neighbouring hospital, a stillbirth, and the remainder not collected due to participants being discharged early and lost to follow-up. We hope to improve the heel-prick capture rate through providing more information to midwives and by providing participants with leaflets/bags with the study logo so that midwives are aware of their involvement in the study and the importance of capturing the heel-prick.

Although small, our study is the first to show that early GDM treatment could be associated with a play off between a reduced LGA rate but an increased NICU admission rate (largely associated with SGA for those treated from booking). SGA/fetal undernutrition can be a consequence of overtreatment [15], or insufficient gestational weight gain [16], with putative long term consequences [17]. Hypertension and smoking (and potentially metformin treatment [18]) can also contribute to reduced fetal growth and in the main trial, it will be important to assess any interaction between these characteristics, fetal gender and outcomes from early GDM treatment. Our study reminds us that we do not know the optimal glycaemic targets early in pregnancy and that we should be cautious. Our study is also the first to challenge the finding that 33–66% of women early in pregnancy might not have GDM on repeat fasting glucose testing [8]. As 89% of untreated women in our study had GDM at both 18 weeks and 24–28 weeks gestation, there is a case for using the 24–28 week criteria from the middle of the second trimester.

Our study has the strength of obstetrician blinding and the use of decoys, to avoid confounding of the NICU admission and LSCS rates. The use of quantitative and qualitative measures is also a strength. The study only tests the use of the IADPSG/WHO criteria [6, 7] and not other criteria for GDM. Weaknesses include the higher baseline systolic blood pressure and the small sample size. The latter prevents the identification of the impact of any specific management issues such as the use of metformin or treatment compliance. The late clinic booking of women was a major impediment to recruitment and we have tested the benefits/harms with a mid-second, not a first trimester GDM diagnosis. Most (48%) of the exclusions were due to this late booking, which was due to local clinic processes (now rectified). For the main study, we have negotiated with the antenatal clinic to expedite the booking of women with risk factors for GDM, to avoid delays and hopefully recruit before 12–14 weeks gestation. Such earlier recruitment is more likely to directly answer the question about the benefits/risks of treatment in the first trimester raised by both Zhu et al. [8] and Sweeting et al. [19]. The latter showed that the group with the poorest pregnancy outcome were diagnosed before 12 weeks. Women diagnosed with GDM between 13 and 23 weeks had outcomes more similar to those diagnosed > 24 weeks. The study would have recruited the target number of women quicker if an early OGTT had been requested universally. However, the study followed the national guideline to use risk factor screening to assess who should proceed to an early OGTT.

In conclusion, we have completed the first RCT of whether GDM, defined by 24–28 week criteria, should be treated at booking. Our surveys and interviews of mothers and midwives, along with the successful completion of the study highlights its feasibility. Although a small study, the findings suggest that there may be a play off between reducing macrosomia and increasing fetal undernutrition, a finding warranting a larger RCT. A full trial is urgently required.