The DESiGN trial (DEtection of Small for Gestational age Neonate), evaluating the effect of the Growth Assessment Protocol (GAP): study protocol for a randomised controlled trial

The rate of stillbirth in England and Wales has declined recently from 5.7/1000 births in 2003 to 4.7/1000 in 2013 [1] but the United Kingdom (UK) rate remains amongst the highest in developed countries [2, 3]. Reducing stillbirth is a national priority [4‐6]. It has been estimated that up to 57% of babies who die in utero are small for gestational age (SGA) and 9% have placental insufficiency. However, rates are dependent upon which perinatal classification system is used (32.6–57%) [7‐9].

The use of routine screening achieves antenatal detection of SGA in approximately 20–25% of cases [10‐13]. In the UK, routine screening involves fundal-height measurements in low-risk pregnancies and serial growth scans in high-risk pregnancies; however, there is variation in risk factors considered and frequency of assessment. Since the publication of the National Health Service (NHS) England ‘Saving Babies’ Lives Care Bundle’ in 2017, there has been a move towards consistency on risk factor stratification [5]. Other strategies, such as universal third-trimester ultrasound, could increase the antenatal detection of SGA to 57% for nulliparous women with singleton pregnancies [13]. A population-based study has reported that antenatal detection of SGA may halve the risk of stillbirth through appropriate antenatal surveillance and timely delivery [10]. This suggests that improvements in the detection of SGA infants could potentially result in further reduction in the incidence of stillbirths.

SGA is traditionally defined as weight below the 10th centile for gestational age and sex according to population references [14, 15]. The development of population-based centiles is an improvement over the use of estimated fetal weight or absolute birthweight to characterise abnormalities of fetal and infant growth. This is exemplified by a 2000-g neonate which will be considered small at term (37–42 completed gestational weeks) but normally grown at 32 weeks. SGA by population birthweight references account only for the physiological effects of gestational age and fetal sex, and it has been hypothesised that there are additional physiological maternal and fetal characteristics that may affect fetal growth [16]. Some foetuses, currently defined as SGA, may be appropriately grown for maternal constitution (i.e. the constitutionally small foetus). Other foetuses with impaired growth may still fall within the normal range by population references, particularly if genetically predetermined to be at higher centiles.

The concept of customised fetal growth standards attempts to address these issues and is based on three principles: (1) individualised (adjusted for physiological factors that affect birth weight – maternal height, weight, ethnicity, parity, fetal sex and gestation at delivery); (2) optimised growth potential (excluding pathological factors affecting the weight standard such as smoking and diabetes) and (3) use of fetal growth curves, derived from normal pregnancies [17]. International studies have demonstrated that the use of customised standards identifies additional SGA foetuses, which were not otherwise identified by population standards [18‐22]. Infants who are SGA only by customised standards are at increased risk of adverse outcomes, including stillbirth [18]. These studies have also suggested that foetuses that are considered SGA by population references, but not by customised standards, have outcomes similar to those of appropriately grown foetuses [18‐22]. A non-randomised study of the use of customised charts for fundal-height measurement and estimated fetal weight demonstrated an increase in antenatal detection of small babies (47.9% vs. 29.2%, odds ratio 2.23; 95%CI 1.12–4.45) when compared to routine care, but did not demonstrate differences in other key perinatal outcomes, including for stillbirth (five vs. four stillbirths: odds ratio 1.14 (95%CI 0.30–4.25) [23]. Implementation of customised charts has also been explored observationally in Australia where a doubling in the detection rate of SGA was demonstrated in nulliparous women, when compared with historical controls [24].

The Growth Assessment Protocol (GAP) is a programme developed by the Perinatal Institute that consists of the use of gestation-related optimal weight (GROW) customised charts, alongside a schedule of antenatal risk assessment for SGA, management protocols for suspected SGA foetuses, audit tools and training [17]. GROW utilises a systematic method of measurement, achieved through standardised training and accreditation, with measurements plotted onto customised fundal-height charts. Estimated fetal weights from ultrasound assessment are also plotted onto these customised fetal growth charts when growth scans are performed. The risk assessment and management protocols were adapted from the Royal College of Obstetricians and Gynaecologists (RCOG) Green-top Guideline on Investigation and Management of the Small-for-Gestational-Age Foetus [25]. The audit tool is aimed at identifying missed cases and assessing the reasons for failure to antenatally recognise the SGA foetus.

The use of the GAP/GROW programme has expanded since its development and is now implemented in 121 (76%) UK Trusts [26]. A study in the UK compared regions with high uptake to regions with low GAP uptake during the period 2008 to 2012, using stillbirth data from the Office for National Statistics’ records [1]. Overall, there was a 22% lower stillbirth rate in the high-uptake regions during the period analysed, compared to static rates in the low-uptake areas, although it should be noted that the low-uptake regions had lower rates to begin with (4.86 stillbirths/1000 births vs. 5.63/1000 births in the high-uptake regions) and ecological comparisons such as these cannot be assumed to reflect an effect of GAP and may instead reflect a number of changes over time [23]. For example, the Mothers and Babies: Reducing Risk through Audits and Confidential Enquiries across the UK (MBRRACE-UK) Perinatal Mortality Surveillance Report (2013) confirms a gradual decline in stillbirth rates across the UK over a similar period (the decade up to 2013), and attributes this to Sands (the Stillbirth and Neonatal Death charity) raising awareness of a number of initiatives designed to reduce stillbirth rates, and/or to new guidance from the UK Royal College of Obstetricians and Gynaecologists in redefining which stillborn babies require registration (i.e. not those known to have died in utero prior to the end of the 24th gestational week) [3].

It is widely recognised that the highest level of evidence for intervention effectiveness is obtained from randomised controlled trials (RCTs), and a Cochrane review has highlighted that these are currently lacking in this area [27]. A RCT to accurately assess the GAP programme is, therefore, imperative and timely. Furthermore, there is a paucity of data on the effect of GAP on other maternal and neonatal outcomes such as caesarean section rates, induction of labour, gestational age at delivery or on the use of clinical resources including ultrasound scan appointments, neonatal intensive care unit (NICU) admission and lengths of maternal or neonatal stay in hospital. These outcomes may have a significant health-economic impact which must also be balanced against any improvement in perinatal outcomes (stillbirth, early neonatal death, serious neonatal morbidity).

We propose a cluster RCT to evaluate the GAP programme as a strategy for improving the antenatal detection of SGA, including an evaluation of the GAP implementation and a health economic assessment. This protocol (version 7.0, dated 18 January 2018) has been written and reported according to the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guidance and Checklist [28] (see Additional file 1: Table S1-S2).

This study will provide the first evidence from a RCT into the effect of implementing the GAP programme, compared to standard care, on the identification of SGA measured by ultrasound scan. The study is powered for detection of SGA babies, and not for stillbirth rates. Although a study to show differences in stillbirth is ultimately desirable, it is unfortunately not feasible to power such a trial [41] due to the infrequency of this event and hence very large sample size required. It is widely recognised that antenatal detection of SGA is associated with reduction in stillbirth rates [10].

The primary outcome for this trial has been chosen following consensus agreement from the co-investigators and independent reviewers. It is recognised that there is an overlap between SGA defined by customised and population standards. The group of infants meeting criteria for both definitions are at the highest risk of morbidity/mortality [19]. Detection of these babies is crucial. Focussing on this overlapping group addresses the issue that GAP was not developed to detect SGA by population standards, and vice versa.

Customised standards have been criticised because some factors (namely maternal weight and ethnicity) might contribute to differences in birth weights due to pathological, rather than physiological, mechanisms [32, 42, 43]. The Perinatal Institute recommends the adjustment for maternal weight based on their report that, in women with a Body Mass Index (BMI) within the normal range (20–25 kg/m²), variation in maternal weight was not associated with increased risk of perinatal mortality. In women with BMI > 25 kg/m², the rate of perinatal mortality was correlated with SGA using customised standards but not population references [44]. However, cause-specific perinatal mortality may differ by maternal BMI. The influence of ethnicity is more complex due to the known association between ethnicity and socioeconomic deprivation [45‐47], which is associated with antepartum stillbirth risk [48, 49]. The INTERGROWTH-21st study recruited low-risk, well-nourished women of optimal health, education and socioeconomic status in urban sites with low levels of pollutants [50] and showed similarities in fetal growth (as assessed by head circumference) and newborn length in those pregnancies studied across eight international and multi-ethnic sites. In contrast, work using the Canadian Stillbirth Registry reported that when using population standards, women of Chinese and South Asian ethnicity have higher rates of SGA, when compared to native ethnicities, but lower rates of perinatal mortality. In the same group, the rate of SGA was lower and concordant with the prevalence of perinatal mortality when neonates were assessed using customised, rather than population growth, standards [51]. More recently, the World Health Organisation has published new fetal growth charts that do acknowledge differences in fetal growth between women of different ethnicities, heights, weights, ages and parity [52]. According to this evidence, adjustment for ethnicity may improve the detection of SGA infants at risk of morbidity and mortality, irrespective of any indirect adjustment for the pathological influences such as socioeconomic conditions.

In the UK, there is a high-level ambition to reduce the rate of stillbirths, arising from the RCOG and the Department for Health [4, 6]. Maternity clinical networks and local maternity systems are currently recommending several interventions aimed at reducing rates of stillbirth in the UK [5]. One theme of the national stillbirth recommendations is reduction of SGA-related stillbirth by improved detection of SGA. The GAP programme is the most widely adopted of the available potential solutions, leaving few available sites for recruitment into this trial. We therefore also face the last opportunity to conduct a prospective randomised trial of customised growth charts and the GAP protocol in the UK. Finally, other national initiatives to reduce stillbirth include policies to support smoking cessation, improve intrapartum fetal monitoring and increase reporting of reduced fetal movements [5]. Uptake of smoking cessation support is anticipated to impact upon rates of SGA infants, which could affect analysis of our primary outcome, and all initiatives may impact upon several of the secondary outcomes in this trial. Hospitals recruited to both arms of the trial are under external pressure to adopt some of these policies. We are collecting information on all policies introduced which are independent of the intervention during the trial period to evaluate any potential effect of contamination.

Regarding the health-economic analyses, we have considered the usefulness of reporting the costs from a Trust or health service perspective. The current UK payment system uses tariffs, according to a low-/high-risk stratification of individual patients, rather than crude costs for work done. Whilst this will allow us to calculate the cost of the GAP programme to the health service (by estimating any changes to the likely tariff distribution because of GAP), it does not necessarily translate into financial cost to the hospitals themselves from providing resources to implement GAP, or from the implications of implementing GAP on the patients/resources (by affecting rates of SGA diagnosis). This justifies our approach of reporting the health-economic analysis from a Trust perspective.

Regarding the implementation analysis, we anticipate variation between clusters in GAP implementation, including local adaptations to suit available resources (e.g. number of sonographer lists, and the financial resources required to implement the programme). Such variations will need to be placed in context when assumptions are made about external validity. Nevertheless, the implementation study should produce documented evidence of such differences, and by examining these in relation to the logic model produced for the study, we hope to extend understanding of how variations in implementation might affect the impact of GAP in real-world situations.

Overall, we aim to provide high-quality evidence, through a RCT, of the performance and service provision requirements of the GAP protocol. This may have the potential to inform national policy decisions on methods to reduce the rate of stillbirth.

The first eight cluster sites were randomised to either the intervention or control arm on 3 November 2016. A further three sites were randomised in December 2016 and the final two sites randomised in July 2017. Baseline audit and training procedures were then required prior to the use of the intervention in the active clusters. The first cluster sites began to use the intervention in October 2017 and we anticipate that the intervention will continue to be applied to women attending for maternity care at those sites, with the last women to be included in the data collection likely to have the intervention applied in early July 2018. Data will be collected on all babies being born between 1 June and 31 November 2018.