Antenatal non-medical risk assessment and care pathways to improve pregnancy outcomes: a cluster randomised controlled trial
European Journal of Epidemiology
Antenatal non-medical risk assessment and care pathways to improve pregnancy outcomes: a cluster randomised controlled trial
Jacqueline Lagendijk 0 1 2 3 4 5 6
Amber A. Vos 0 1 2 3 4 5 6
Loes C. M. Bertens 0 1 2 3 4 5 6
Semiha Denktas 0 1 2 3 4 5 6
Gouke J. Bonsel 0 1 2 3 4 5 6
Ewout W. Steyerberg 0 1 2 3 4 5 6
Jasper V. Been 0 1 2 3 4 5 6
Eric A. P. Steegers 0 1 2 3 4 5 6
0 & Jacqueline Lagendijk
1 Department of Obstetrics and Gynaecology, University Medical Centre Utrecht , P.O. Box 85090, 3508 AB Utrecht , The Netherlands
2 Department of Psychology, Education & Child Studies, Faculty of Social & Behavioural Sciences, Erasmus University Rotterdam , Burgemeester Oudlaan 50, 3062 PA Rotterdam , The Netherlands
3 Department of Obstetrics and Gynaecology, Erasmus MC, University Medical Centre , PO Box 2040, 3000 CA Rotterdam , The Netherlands
4 Division of Neonatology, Department of Paediatrics, Erasmus MC - Sophia Children's Hospital , PO Box 2040, 3000 CA Rotterdam , The Netherlands
5 Department of Biomedical Data Sciences, Leiden University Medical Center , PO Box 9600, 2300 RC Leiden , The Netherlands
6 Department of Public Health, Erasmus MC, University Medical Centre , PO Box 2040, 3000 CA Rotterdam , The Netherlands
Social deprivation negatively affects health outcomes but receives little attention in obstetric risk selection. We investigated whether a combination of (1) risk assessment focused on non-medical risk factors, lifestyle factors, and medical risk factors, with (2) subsequent institution of risk-specific care pathways, and (3) multidisciplinary consultation between care providers from the curative and the public health sector reduced adverse pregnancy outcomes among women in selected urban areas in the Netherlands. We conducted a cluster randomised controlled trial in 14 urban municipalities across the Netherlands. Prior to the randomisation, municipalities were ranked and paired according to their expected proportion of pregnant women at risk for adverse outcomes at birth. The primary outcome was delivery of a preterm and/or small for gestational age (SGA) baby, analysed with multilevel mixed-effects logistic regression analysis adjusting for clustering and individual baseline characteristics. A total of 33 community midwife practices and nine hospitals participated throughout the study. Data from 4302 participants was included in the Intention To Treat (ITT) analysis. The intervention had no demonstrable impact on the primary outcome: adjusted odds ratio (aOR) 1.17 (95% CI 0.84-1.63). Among the secondary outcomes, the intervention improved the detection of threatening preterm delivery and fetal growth restriction during pregnancy [aOR 1.27 (95% CI 1.01-1.61)]. Implementation of additional non-medical risk assessment and preventive strategies into general practices is feasible but did not decrease the incidence of preterm and/or SGA birth in the index pregnancy in deprived urban areas. Trial registration Netherlands National Trial Register (NTR-3367).
Pregnancy; Risk assessment; Risk factors; Epidemiology; Prevention
Social deprivation negatively affects health outcomes. This
association is already apparent before birth and extends
into early childhood [
]. In addition to the negative
impact of medical and obstetric risk factors, multiple
studies have shown a strong association between
nonmedical risk factors and adverse pregnancy outcomes. Key
examples of such risk factors include low socioeconomic
status (SES), living in a deprived neighbourhood,
Extended author information available on the last page of the article
ineffective social integration into society, smoking, and
psychosocial stressors [
]. The increased prevalence of
non-medical risk factors and the accumulation of such
factors are responsible for at least part of the
overrepresentation of adverse pregnancy outcomes in deprived urban
areas within high-income countries [
]. Risk assessment
and subsequent implementation of preventive measures in
antenatal health care with the aim to reduce adverse
pregnancy outcomes should, therefore, take both medical
and non-medical risk factors into account. However,
current risk selection during pregnancy mainly focuses on
medical risks, and integration between the curative and
public health sector is scarce .
In the Netherlands, obstetric risk selection is based on
the ‘List of Obstetric Indications’(LOI), which specifies
manifest conditions that define a low, medium, or high-risk
pregnancy. These conditions are single medical or obstetric
risk factors, that indicate whether a patient’s care during
pregnancy or parturition is to be supervised by a
community midwife or an obstetrician [
The R4U scorecard is a comprehensive risk assessment
tool which can, in addition to the LOI, be used by obstetric
care providers to identify psychological, social, lifestyle,
obstetric and non-obstetric care related factors [
total R4U score is strongly associated with adverse
pregnancy outcomes and shows a clear gradient across
categories of SES and ethnicity [
We conducted a cluster randomised controlled trial
(CRCT) to assess the effectiveness of using the R4U
scorecard in conjunction with institution of appropriate care
pathways and multidisciplinary consultations, to reduce the
incidence of adverse pregnancy outcomes. The study was
conducted among pregnant women in selected urban areas
in the Netherlands with an overrepresentation of adverse
pregnancy outcomes [
]. This study is part of the
‘Healthy pregnancy 4 All’ (HP4All) programme, a
nationwide study evaluating strategies to improve
pregnancy outcomes, in particular among deprived populations
We conducted a C-RCT in 14 municipalities in the
Netherlands. The Daily Board of the Medical Ethics
Committee Erasmus MC approved the study (METC
2012-322). The study protocol was peer-reviewed and
published, and was registered at the Netherlands National
Trial Register (NTR-3367) [
]. Municipalities were
selected based on multiple criteria: (1) size (having more
than 70,000 inhabitants), (2) disproportionally high
prevalence of risk factors for adverse pregnancy outcomes
(high and/or low maternal age, primiparity, non-western
ethnicity, and low SES), (3) a high incidence of adverse
pregnancy outcomes (delivery of a small for gestational
age baby (SGA), preterm delivery, and perinatal mortality
[mortality from the 22th week of gestation until 7 days
postnatally)], and (4) a higher than average case-fatality
]. The case-fatality rate is the proportion of
perinatal mortality amongst neonates with a so-called ‘BIG4’
condition: congenital anomalies, preterm birth, SGA, and/
or an Apgar score below seven at five minutes after birth
]. A more detailed description of the selection process
of municipalities has been published before [
community midwife practices and hospitals located in the
participating municipalities were invited to participate in
The 14 selected municipalities were divided into ten
clusters; five municipalities in the northern part of the
Netherlands were merged into one cluster due to the
intended formation of a so called ‘obstetric collaborative
network’ in that area [
]. An obstetric collaborative
network is an inter-professional care system in which
community midwives, obstetricians, and maternity care
providers share local guidelines and protocols. All women
with a singleton pregnancy living in a selected municipality
and booking their first antenatal visit at one of the
participating community midwife practices or at a participating
hospital were eligible for this trial. Exclusion criteria
included an obstetric emergency situation or being in
labour during the initial visit.
In the intervention clusters, participating obstetric care
providers used the R4U scorecard as a risk assessment tool
at the first antepartum visit. They did so in addition to their
conventional risk assessment approach (LOI-based). The
R4U scorecard guided coordination of antepartum care
through systematic risk assessment for medical and
nonmedical risk factors for adverse pregnancy outcomes
(Online Resource 1). To increase uniformity in questioning
within the R4U, a ‘script’ text was formulated for each
separate item as a literal text. A positive response indicated
presence of the risk factor. Risk factors were selected after
a broad literature search and complemented with detailed
epidemiological information of prevalence and risk
estimates derived from well-documented large birth cohort
]. Risk factors were weighted based on their
relative risk for adverse pregnancy outcomes [
for individual risk factors were added up to form a total
score (range R4U 0-98). A predefined cut-off score was
based on data from a pilot study in the Netherlands from
2010 to 2011; a score of 16 points or higher was selected to
identify women in the upper 20% of risk scores [
]. In the
current study, a cumulative R4U score of C 16 points
implied a follow-up action via a case-based discussion in a
multidisciplinary setting. In addition, the institution of
appropriate individual care pathways was guided by
particular single, or a set of multiple, risk factors (Fig. 1).
Case-based multidisciplinary consultations involved
community midwives, obstetricians and other healthcare
professionals, such as paediatricians or social workers.
With this approach optimal linkage was sought between the
public health sector and the curative care sector. The aim of
the meeting was to agree on a customised antepartum
policy for each individual patient [
]. Obstetric care
Fig. 1 Trial design
providers were in addition allowed to discuss participants
in these discussions to their own discretion, independently
of the cumulative R4U score. As a result, participants with
relevant individual risk factors could also be discussed
when the cumulative score was below 16 (Fig. 1).
Prior to start of the study, 28 templates of care pathways
were developed based on the medical and non-medical risk
factors incorporated in the R4U. These templates consisted
of a set of steps a healthcare professional was advised to
take in an attempt to reduce the potential contribution of
one or more risk factors to developing an adverse
pregnancy outcome. As such, each care pathway could, for
example, direct the user to a specific health care provider,
or to a health care organisation, public health care
organisation, or an office for legal or financial support. When
there was existing evidence for interventions to address
modifiable risk factors this was used for the contents of the
pathway. To enhance the efficiency of the care provided,
care pathways are explicit as to which caregiver will be
responsible. To facilitate local adaptation of this new way
of organising antenatal care, details of the care pathways
were discussed during meetings with community
midwives, obstetricians, social workers, and a city council
representative. In these meetings the templates for the
pathways were complemented with the availability of local
facilities and health insurance agreements, and refined
through direct interaction with relevant local health care
providers, and organisations. In addition, the introduction
of organized meetings to customize care pathways induced
a change in the mutual professional relationship between
care providers of different echelons. An example of a care
pathway, in this case for psychosocial risk factors, is added
in the Online Resource 2.
In the control clusters, an existing screening instrument
(LOI), which focuses on identification of single, manifest
obstetric and medical risks, was used, combined with
individual care according to local protocols. The LOI
distinguishes between a low, medium, or a high risk
pregnancy based on anticipated third trimester and labour
]. Low risk indications allow women to
choose the preferred place of delivery (i.e. home birth,
birth centre, hospital birth), which is supervised by an
independent community midwife. Women with medium
risk pregnancies should deliver in a hospital supervised by
a midwife, whereas women with high risk pregnancies are
supervised by an obstetrician during pregnancy and
hospital delivery [
]. If the risk changes from a low or
medium risk to a high risk, the woman is referred from
primary to secondary or tertiary care, even during labour
Baseline characteristics were collected via a questionnaire
that was filled out by participants after the first antepartum
visit, generally around ten to 12 weeks of gestational age.
The following characteristics were collected: maternal age
at inclusion, parity, ethnicity (western versus non-western
based on maternal country of birth and classified according
to Statistics Netherlands), single motherhood, maternal
SES (based on the classification by the Netherlands
Institute of Social Research to all postal code areas, and divided
into three centile groups: low \ 20, medium 20-80, and
high [ 80) [
], maternal BMI prior to conception
(categorised into low \ 20, normal 20–29, and high 30 and
more), maternal education [(highest completed education
categorised into low: primary school, special education,
pre-vocational (secondary) education, junior general
secondary education; middle: senior general secondary
education, pre-university education, senior secondary
vocational education; and high: higher professional
education, university education)], smoking during pregnancy
(yes/no), and risk factors derived from the obstetric history
(previous SGA baby, previous preterm delivery).
Perinatal outcome data was collected by a member of
the research team 6 weeks after a participant gave birth
from medical charts of community midwives and
All predefined outcomes regarding the effectiveness of
this intervention pertain to the participant level. The
primary study outcome was: delivery of a preterm (i.e. before
37 weeks of gestation) and/or SGA baby (birth weight
below the 10th centile adjusted for parity, gestational age,
and gender, based on the Dutch reference curves [
together referred to as ‘BIG2’.
Secondary outcomes were: the detection of fetal growth
restriction during pregnancy (defined as fetal growth below
the 10th centile for gestational age) and/or threatening
preterm birth during pregnancy (defined by the detection
of, and any action taken by an obstetric care provider, after
suggestive symptoms of preterm labour), any referral to
non-obstetric health care providers during pregnancy used
as a proxy for involvement (regardless of referral within
the care pathways), any referral to preventive care
organisations during pregnancy used as a proxy for involvement
(regardless of referral within the care pathways), maternal
mortality, unexpected SGA (babies born SGA under
supervision of a community midwife), unexpected preterm
birth (babies born preterm under supervision of a
community midwife), birth asphyxia (an Apgar score below
seven at 5 min after birth was used as a proxy), neonatal
admission to an intensive care unit, and perinatal mortality
(mortality from the 22th week of gestation until 7 days
Two secondary outcome measures, as defined in the
initial protocol were not analysed, as these outcomes were
not considered to be potentially sensitive to a
postconceptional intervention. These were: ‘prevalence of general
risk factors’ (defined as: pre-existing chronic disease, folic
acid use, and medication use) and ‘congenital anomalies’.
Among a sub-cohort we assessed participants’ and
health care providers’ satisfaction, and efficacy of
implementation of the intervention; these findings are reported
Calculation of sample size was based on the presumed
effect of the intervention on the primary outcome, and a
two-group comparison based on the combined prevalence
of preterm birth and SGA in the Netherlands. The
intervention was implemented at the municipality level
(cluster), while the intervention effect was measured at the
participant level. To account for clustering of participants
within municipalities, the sample size was multiplied by a
Variance Inflation Factor (VIF) of 2.06, calculated with the
formula of Donner et al. [
In the selected clusters the average incidence of the
primary outcome before start of the study (2000–2008) was
16.7%; we hypothesised that the intervention would lead to
a decline towards 13% [
]. At an alpha of 0.05 and 80%
power, we required 700 participants per cluster, or 3500
participants in each arm. The pre-defined stopping rule was
based on the end of the HP4All study period (July 2015).
We randomised at the level of the clusters. Before the
randomisation procedure, municipalities were ranked
according to their expected percentage of pregnant women
at risk for a ‘BIG2’ outcome at birth. Expected proportions
were based on incidence rates from 2000–2008 derived
from the Netherlands Perinatal Registry. Municipalities
were then paired based on this ranking. The random
number generator in R version 2.7.1 was used to assign one
of the municipalities in each pair to the intervention arm.
The other municipality of that pair was then assigned to the
control arm. An independent statistician, who was not
involved in executing the study, carried out the
randomisation process. The randomisation at cluster level, instead
of the randomisation of midwife practices or hospitals, was
necessary to avoid contamination as community midwives
and obstetricians generally work closely together in
obstetric collaborative networks. Obstetric care providers
within each cluster were informed and educated with
knowledge of the outcome of the randomisation process.
Blinding of obstetric care providers was not possible given
the nature of the intervention. Allocation concealment of
participants was set out by exclusively foreseeing in study
information about the situation that was assigned to a
specific cluster. As a result, participants were unaware of
the randomised design of the study.
The impact of the intervention on the primary and
secondary outcomes was analysed using multilevel
mixedeffects logistic regression analysis (with an assumed
random effect for each cluster). Multiple imputation using
chained equations was used to account for missing data in
baseline characteristics. Both predictor and outcome
variables were included to inform the multiple imputation
process, forming 15 datasets. Results across the sets were
combined using Rubin’s Rules [
]. No interim analysis
was performed. Analyses were performed according to the
ITT principle. We included the following covariates in our
models: age, ethnicity, BMI prior to pregnancy, SES,
single motherhood, smoking during pregnancy, and obstetric
history (previous SGA baby and/or previous preterm
delivery). To account for over-fitting we only analysed
secondary outcomes when there were more than ten events
in the two groups. Statistical analysis was performed using
Stata SE (version 14). Statistical significance was accepted
at p \ 0.05 (two-sided).
Per protocol analysis
During the trial not all participants in the intervention
clusters were screened using the R4U. Therefore, a
sensitivity analysis was performed using a per protocol
approach to investigate whether this affected the effect
Values are expressed as numbers (first) and percentage (second). Percentages of categorised values are percentages of non-missing cases.
Missing percentages are percentages of total cases
Enrichment of the control clusters
During the study, there was a substantial lag in participant
recruitment in the control clusters. Following an ad-hoc
study group meeting, a decision was made to ‘enrich’ the
control arm with pregnancies included retrospectively from
participating practices to a total of 700 participants per
cluster. Retrospective pregnancy data was extracted from
digital medical charts in participating community midwife
practices. Inclusion criteria were identical to the
prospective inclusion. Demographic characteristics of the
prospectively and retrospectively included participants in
the control arm were compared and the potential
differences between the groups were explored.
Five clusters, including eight hospitals and 20 community
midwife practices, were included in the intervention arm,
and five clusters (eight hospitals and 16 community
midwife practices) served as controls (Fig. 2). Complete data
regarding baseline characteristics was available for 2486 of
2872 (86.8%) participants in the intervention arm and 2227
of 2424 (91.9%) participants in the control arm (Fig. 2).
We excluded participants who had a miscarriage (125 and
72, for the intervention and control group, respectively).
Primary outcome data was unavailable for 92 (3.2%)
participants in the intervention arm and 122 (5.0%)
participants in the control arm (Fig. 2). Accordingly, 4302
participants were included in the ITT analyses.
In the intervention arm, 77.3% of participants were
actually screened using the R4U. Of all participants
screened 7% had a sum score of 16 or higher, and 50% of
participants with an R4U cut-off score above 16 had a
registered multidisciplinary consultation.
Table 1 presents the maternal and pregnancy
characteristics at the individual level of all prospectively included
participants, by treatment allocation. Online Resource 3,
presents the same characteristics at cluster level.
Participants in the control arm had a higher income per month, a
higher educational level and a higher SES as compared to
those in the intervention arm. Participants who did not have
data on the primary outcome did not differ importantly
from those included in the ITT analysis (Online Resource
Impact of the intervention on primary and secondary outcomes
The combined primary outcome delivery of a preterm and/
or a SGA baby (BIG2) occurred in 16.3% of participants in
the intervention arm and in 13.2% of participants in the
control arm [unadjusted odds ratio OR 1.34 (95% CI
0.92–1.94); Table 2 and Online Resource 5]. The
intervention had no demonstrable impact on the primary
outcome in multivariable analysis: adjusted odds ratio (aOR)
1.17 (95% CI 0.84–1.63) (Table 3). The intervention
improved the detection of threatening preterm delivery and
fetal growth restriction during pregnancy compared to the
control arm: aOR 1.27 (95%CI 1.01–1.61), but had no
Primary and secondary outcomes at individual level, categorised in
primary (delivery of a preterm and/or a SGA baby, referred to as
‘BIG2’) and secondary outcomes (maternal, delivery, and neonatal).
Values are expressed as numbers (first) and percentage (second).
Percentages of categorised values are percentages of non-missing
cases. Missing percentages are percentages of total cases
significant impact on any other secondary outcomes
Demographic characteristics differed significantly between
the prospectively and retrospectively included participants
in the control arm (Online Resource 6). Prospectively
included participants were more often of western ethnic
origin, had a higher educational level, and a higher SES.
Due to this important heterogeneity we decided not to
conduct any additional analyses including data from the
retrospectively included participants.
The effect estimates did not change materially when
performing a per protocol analysis as compared to the
intention to treat analysis (Table 4).
By introducing one single tool for additional risk
assessment in all tiers of the Dutch obstetric care system we
achieved uniformity in risk assessment among 33
community midwife practices and nine hospitals in 14 urban
municipalities in the Netherlands [
]. The combination
with subsequent institution of care pathways and
multidisciplinary consultations further promoted uniformity in a
more proactive and preventive approach regarding medical
and non-medical risk factors during pregnancy. Hereby the
traditional risk assessment during pregnancy, aimed at
recognising primarily medical risk factors for
complications during labour, shifted towards the first trimester and
created a larger window of opportunity for prevention.
However in this C-RCT, this combined intervention had no
demonstrable impact on preterm and/or SGA birth.
Health inequalities depend on a person’s social,
economic, and political environment. These environments are
shaped by policies, which makes them amenable to change
]. Our trial is part of the overall HP4All research
programme designed to evaluate the effectiveness of
interventions, and their associated preventive strategies, in
decreasing health inequalities in pregnancy outcomes [
To accomplish implementation of such a programme,
interventions should contain a flexible approach that allows
for adaptation. Such adaptations stimulate the
implementation process and increase sustainability [
the same flexibility may also have influenced our results.
For example, all participating caregivers, including those
belonging to the control arm, were educated prior to the
start of the program about the importance of non-medical
risk factors in relation to adverse pregnancy outcomes.
Such adaptations may have resulted in an unintended spill
over of intervention effects in the control arm.
This study also has other limitations. Firstly, the
intended inclusion of 7000 participants was not achieved within
the study’s time frame, and there was a wide variation in
sample size among clusters. Secondly, despite the fact that
the HP4All programme was set out in the most deprived
neighbourhoods of the Netherlands, the participants in this
study had an educational level and family income above
the national average, suggesting a substantial degree of
selection bias. As a result, whereas based on previous
research we expected 20% of participants to have an R4U
score of 16 or higher, only 7% fulfilled this criterion in the
final sample. Thirdly, our results show that not all
participants received the intervention as intended. In the
intervention arm, 77.4% of all participants were assessed using
the R4U scorecard. Of participants with a cut-off score
higher than 16, only 50% had a registered multidisciplinary
consultation. In addition, the process evaluation of this
study, based on Saunders’ 7-step method, showed that only
half of the participating municipalities met the criteria for
full implementation of the risk assessment program [
The combination of not achieving the intended sample size,
having fewer participants with a high-risk score according
to the pre-defined cut-off, and the above mentioned dilution
of the intervention, reduced the power of this study to
Fourthly, there were differences in demographic
characteristics between participants in the intervention arm and
participants in the control arm (Table 1). Participants in the
intervention arm had a lower income per month, a lower
educational level and a lower SES as compared to those in
the control arm. This heterogeneity could be explained by a
selection bias, which is a well-recognised phenomenon in
]. Participants were recruited after the
clusters had been randomised. Obstetric care providers had
knowledge of whether participants belonged to the
intervention or the control arm and this could have affected the
types of participants they recruited. Health care providers
in the intervention arm may have included more
participants with a higher risk for adverse pregnancy outcomes, or
in other terms, participants more suitable for ‘active
management’. In the control arm this selection likely led to an
inclusion of participants with a favourable risk profile. This
is further substantiated by the observation that
prospectively included participants were more often of western
ethnic origin, had a higher educational level, and a higher
SES than retrospectively included participants, who were
more likely to represent an unbiased sample (Online
Resource 6). Although in our analysis we adjusted for
known potential confounders, unmeasured confounders
could have been imbalanced too and as such may have
influenced the results of our analysis.
Despite careful theoretical planning, cluster randomised
controlled trials are known to be vulnerable to risk of bias,
specifically, bias in selection of participants [
27, 29, 30
Our experience has implications for designing similar trials
in the future. The observed inclusion bias in this trial is
mostly based on a recruitment bias. Blinding the recruiter
of participants for allocation could potentially have
diminished this bias. In our trial, participants were
recruited by their health care providers, who were also
responsible for subsequent pregnancy care, making blinding
impossible. This may be addressed by separating
participant inclusion from participant care in future studies.
Moreover, researchers of C-RCTs may consider
conducting an interim analysis, which could potentially have
detected the differences in baseline characteristics between
the intervention and the control arm. Such an analysis
would potentially also have been able to detect the
additional issues that eventually caused a dilution of the
intervention effect, allowing these to be addressed during
the course of the trial.
Despite the above-mentioned limitations, our study
shows that implementation of additional non-medical risk
assessment and preventive strategies into general practices
are feasible. It did, however, not decrease the incidence of
adverse perinatal outcomes in the index pregnancy in
deprived urban areas.
Extended screening for populations at risk, together with
improved collaboration between the curative and public
health sector in patient-tailored care, is a start in
establishing equity-oriented strategies during pregnancy.
However, an intensive research programme as HP4All should
ultimately seek to serve pregnant women. Serving in these
interventions means detecting those with the greatest health
needs, and help them to find the power to direct resources
towards those needs. In this perspective, future research in
this field should elucidate what empowers pregnant women
and which specific resources they need to address their
health needs. Effectiveness in this regard, could include
value based outcome measures, rather than focusing merely
on health outcomes.
Acknowledgements The research team has received funding from the
Ministry of Health, Welfare and Sports (Grant Number 318 804) in
order to execute the Healthy Pregnancy 4 All study. The funders had
no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Funding The research team has received funding from the Ministry of
Health, Welfare and Sports in order to execute the Healthy Pregnancy
4 All study. JVB is supported by personal fellowships from the
Netherlands Lung Foundation and Erasmus MC.
Compliance with ethical standards
Conflict of interest All authors have completed the Unified
Competing Interest form (available on request from the corresponding
author). The authors declare that they have no conflict of interest.
Ethical standards The study was reviewed by the Medical Ethical
Review Board of the Erasmus MC. The Board provided a waiver for
the need to obtain consent at the individual level according to Dutch
law as all procedures were essentially accepted care, and data were
analysed anonymously (MEC-2012-322).
Open Access This article is distributed under the terms of the Creative
Commons Attribution 4.0 International License (http://creative
commons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided you give
appropriate credit to the original author(s) and the source, provide a
link to the Creative Commons license, and indicate if changes were
Department of Social and Behavioural Sciences, Erasmus
University College, Faculty of Social & Behavioural
Sciences, Erasmus University Rotterdam, Nieuwemarkt 1A,
3011 HP Rotterdam, The Netherlands
1. Mackenbach JP . The persistence of health inequalities in modern welfare states: the explanation of a paradox . Soc Sci Med . 2012 ; 75 ( 4 ): 761 - 9 .
2. Chauvel L , Leist AK . Socioeconomic hierarchy and health gradient in Europe: the role of income inequality and of social origins . Int J Equity Health . 2015 ; 14 ( 1 ): 132 .
3. Currie J , Lin W. Chipping away at health: more on the relationship between income and child health . Health Aff (Millwood) . 2007 ; 26 ( 2 ): 331 - 44 .
4. Steegers EA , Barker ME , Steegers-Theunissen RP , Williams MA . Societal valorisation of new knowledge to improve perinatal health: time to act . Paediatr Perinat Epidemiol . 2016 ; 30 ( 2 ): 201 - 4 .
5. Gray R , Bonellie SR , Chalmers J , Greer I , Jarvis S , Williams C . Social inequalities in preterm birth in Scotland 1980-2003: findings from an area-based measure of deprivation . BJOG . 2008 ; 115 ( 1 ): 82 - 90 .
6. Timmermans S , Bonsel GJ , Steegers-Theunissen RP , et al. Individual accumulation of heterogeneous risks explains perinatal inequalities within deprived neighbourhoods . Eur J Epidemiol . 2011 ; 26 ( 2 ): 165 - 80 .
7. Vos AA , Posthumus AG , Bonsel GJ , Steegers EA , Denktas S. Deprived neighborhoods and adverse perinatal outcome: a systematic review and meta-analysis . Acta Obstet Gynecol Scand . 2014 ; 93 ( 8 ): 727 - 40 .
8. Mensah FK , Hobcraft J . Childhood deprivation, health and development: associations with adult health in the 1958 and 1970 British prospective birth cohort studies . J Epidemiol Community Health . 2008 ; 62 ( 7 ): 599 - 606 .
9. Weightman AL , Morgan HE , Shepherd MA , Kitcher H , Roberts C , Dunstan FD . Social inequality and infant health in the UK: systematic review and meta-analyses . BMJ Open . 2012 ; 2 ( 3 ): e000964 .
10. Davey MA , Watson L , Rayner JA , Rowlands S . Risk scoring systems for predicting preterm birth with the aim of reducing associated adverse outcomes . Cochrane Database Syst Rev . 2011 ; 11 : CD004902 .
11. van der Kooy J , Poeran J , de Graaf JP , et al. Planned home compared with planned hospital births in the Netherlands: intrapartum and early neonatal death in low-risk pregnancies . Obstet Gynecol . 2011 ; 118 ( 5 ): 1037 - 46 .
12. Vos AA , van Veen MJ , Birnie E , Denktas S , Steegers EA , Bonsel GJ . An instrument for broadened risk assessment in antenatal health care including non-medical issues . Int J Integr Care . 2015 ; 15 : e002 .
13. Posthumus AG , Birnie E , van Veen MJ , Steegers EA , Bonsel GJ . An antenatal prediction model for adverse birth outcomes in an urban population: the contribution of medical and non-medical risks . Midwifery . 2015 ; 38 : 78 - 86 .
14. Denktas S , Poeran J , van Voorst SF , et al. Design and outline of the healthy pregnancy 4 all study . BMC Pregnancy Childbirth . 2014 ; 14 : 253 .
15. de Graaf JP , Ravelli AC , de Haan MA , Steegers EA , Bonsel GJ . Living in deprived urban districts increases perinatal health inequalities . J Matern Fetal Neonatal Med . 2013 ; 26 ( 5 ): 473 - 81 .
16. Poeran J , Denktas S , Birnie E , Bonsel GJ , Steegers EA . Urban perinatal health inequalities . J Matern Fetal Neonatal Med . 2011 ; 24 ( 4 ): 643 - 6 .
17. Vos AA , van Voorst SF , Waelput AJ , et al. Effectiveness of score card-based antenatal risk selection, care pathways, and multidisciplinary consultation in the healthy pregnancy 4 all study (HP4ALL): study protocol for a cluster randomized controlled trial . Trials . 2015 ; 16 : 8 .
18. Bonsel GJBE , Denktas S , Poeran J , Steegers EAP . Lijnen in de Perinatale Sterfte. Signalementstudie Zwangerschap en Geboorte . Rotterdam: Erasmus Medical Center; 2010 .
19. van Veen MJ , Birnie E , Poeran J , Torij HW , Steegers EA , Bonsel GJ . Feasibility and reliability of a newly developed antenatal risk score card in routine care . Midwifery . 2015 ; 31 ( 1 ): 147 - 54 .
20. Deville´ WWT. Herijking stedelijke achterstandsgebieden . 2012 . https://www.nivel.nl/en/ethnic-and -socio-economic-differencesin-health-and-illness . Accessed 10 Feb 2017 .
21. Visser GH , Eilers PH , Elferink-Stinkens PM , Merkus HM , Wit JM . New Dutch reference curves for birthweight by gestational age . Early Hum Dev . 2009 ; 85 ( 12 ): 737 - 44 .
22. Vos AA , van Voorst SF , Posthumus AG , Waelput AJM , Denktas S , Steegers EAP . Process evaluation of the implementation of scorecard-based antenatal risk assessment, care pathways and interdisciplinary consultation: the healthy pregnancy 4 all study . Public Health . 2017 ; 150 : 112 - 20 .
23. Donner A , Birkett N , Buck C . Randomization by cluster. Sample size requirements and analysis . Am J Epidemiol . 1981 ; 114 ( 6 ): 906 - 14 .
24. Roysten P. Multiple imputation of missing values . Stata J . 2004 ; 3 : 227 - 41 .
25. Waelput AJMRA , van Veen-van Belle DW , Lagendijk J , Sijpkens MK , van Voorst, SF , Vos AA , de Jong-Potjer LC , Been JV , Steegers EAP . Aanpak babysterfte in Nederlandse gemeenten . Ontwikkeling en implementatie van zorgexperimenten . 2016 .
26. CSDH. Closing the gap in a generation: health equity through action on the social determinants of health . Final Report of the Commission on Social Determinants of Health. Geneva , World Health Organization. 2008 . http://www.who.int/social_determi nants/final_report/csdh_finalreport_ 2008 .pdf. Accessed 15 May 2017 .
27. Hahn S , Puffer S , Torgerson DJ , Watson J . Methodological bias in cluster randomised trials . BMC Med Res Methodol . 2005 ; 5 : 10 .
28. Brierley G , Brabyn S , Torgerson D , Watson J . Bias in recruitment to cluster randomized trials: a review of recent publications . J Eval Clin Pract . 2012 ; 18 ( 4 ): 878 - 86 .
29. Siebenhofer A , Paulitsch MA , Pregartner G , Berghold A , Jeitler K , Muth C , Engler J . Cluster-randomised controlled trials evaluating complex interventions in general practices are mostly ineffective: a systematic review . J Clin Epidemiol . 2018 ; 94 : 85 - 96 .
30. Puffer S , Torgerson D , Watson J . Evidence for risk of bias in cluster randomised trials: review of recent trials published in three general medical journals . BMJ . 2003 ; 327 ( 7418 ): 785 - 9 .