Risk factors for poor neonatal outcome in pregnancies with decreased fetal movements

The incidence of Swedish stillbirths has varied little in the past 40 years, with a reported frequency of 400‐450 stillbirths/y (approximately 4‰), despite increased information about fetal movement in the media and awareness among healthcare providers. The objectives of this project were to describe the outcome of pregnancies with reduced fetal movement in a Swedish context and to investigate factors associated with poor neonatal outcome in this group.


| INTRODUC TI ON
One of the most feared complications of pregnancy is stillbirth.
Despite efforts to prevent this complication, the incidence of stillbirths remains high, with an estimated 2.6 million deaths globally each year. 1 According to the World Health Organization, Sweden is ranked 12th highest of the high-income countries for stillbirth incidences (and the highest among the Scandinavian countries). The incidence of Swedish stillbirths has varied little in the past 40 years, with a reported frequency of approximately 4‰ (400-450 stillbirths/y of 115 000 births) despite increased information about fetal movement in the media and awareness among healthcare providers. 2 Fetal movement monitoring is used as a reassurance of fetal wellbeing during pregnancy. Self-reported fetal movements vary throughout pregnancy, with a gradual increase from week 16 to week 36 and a small decrease during the last month of pregnancy, but only 33%-88% of the movements that were seen on ultrasound were perceived by the mothers. 3,4 The movements vary during pregnancies and maternal perception of fetal movements may be affected by the quantity of amniotic fluid, position of the fetus, maternal medication and wellbeing of both mother and fetus. 5 The perception of fetal movements is also affected by maternal stress and anxiety, 6 medication, smoking, 7 localization of placenta 8 and maternal position. 9 Reduced fetal movements (RFM) might also be a sign of pregnancy complications such as preterm birth, oligohydramnios, fetal growth restriction and stillbirth. 10,11 Study of women who experienced RFM during pregnancy showed that they had an increased risk of late stillbirth compared with women that experienced normal movements. [11][12][13] It is of great importance to differentiate the women at risk for poor fetal outcome from the heterogeneous group of pregnancies with RFM, as most of these pregnancies and deliveries are uncomplicated. An accurate risk assessment for this group will help to find a balance between unwanted interventions/over-investigation and maintaining good neonatal outcome. The aim of this study was to describe the outcome of pregnancies with RFM in a Swedish context and to investigate factors associated with poor neonatal outcome in this group.

| MATERIAL AND ME THODS
This retrospective cohort study was performed at Soder Hospital, Stockholm, Sweden. This hospital is the second largest maternity clinic in Sweden and a secondary referral center in the center of Stockholm, with nearly 8000 deliveries per year. All single pregnancies at the hospital from January 2016 to December 2017 presenting with RFM after 22 gestational weeks were included in the cohort.
Each visit in the department is given a diagnostic code and the cases were identified after the code for RFM.
Pregnant woman presenting with RFM at the clinic were managed in accordance with local clinical guidelines. A cardiotocography (CTG) registration and an ultrasound to assess the amount of amniotic fluid and fetal movement were routinely performed. In cases where the CTG assessment (computerized analyses according to Dawnes-Redman criteria until 32 weeks of gestation and human analysis after that) was normal but no fetal movements were identified sonographically after repeated examinations, an additional ultrasound for fetal biometry or induction of labor was offered depending on gestational age (commonly after 40 weeks of gestation) of the pregnancy.
Data on maternal characteristics, such as body mass index (weight in kg/height m 2 ) at the beginning of pregnancy, parity, age, previous cesarean section, past illnesses, complications of pregnancy and data on pregnancy outcome, were collected from the maternal medical records. Information regarding the newborn, such as gender, birthweight, Apgar scores at delivery, admission to neonatal ward and umbilical cord pH were retrieved from the delivery charts (Obstetrix, Cerner, Sverige AB, Lund, Sweden). The women included in the study were delivered during the study period or at the beginning of 2018. The majority were delivered at our clinic, but a small percentage were delivered in other hospitals in Stockholm.
A composite for poor neonatal outcome was constructed and described as one or more of following: 5-minute APGAR score ≤7, arterial pH in the umbilical cord ≤7.10, transfer to neonatal ward for further care, intrauterine fetal death (IUFD).
All Swedish pregnancies are dated by an ultrasound examination in the first or second trimester using the biparietal diameter and femur length. The definition of small for gestational age (SGA) and large for gestational age were based on the Swedish reference curve and defined as 2 SD below or above the expected weight for gestational age and gender. 14 The intrauterine expected weight was calculated according to the formula of Marsal et al as previously described by Lindqvist et a.l 15,16 Stillbirth is defined based on the Swedish national guidelines as intrauterine death at ≥22 weeks of gestation.

| Statistical analyses
Histograms were initially used to assess data distribution. As some of the data were not normally distributed and there were no differences in clinical significance between the mean and median values, we chose to present all continuous variables as medians with min and max values. Other maternal and fetal characteristics are presented as frequencies. Maternal, pregnancy and outcome characteristics were compared between the groups (women with RFM and

Key message
The highest risk factors for poor neonatal outcome in the group with reduced fetal movements were small for gestational age and in vitro fertilization. Recurrent reduced fetal movements did not carry an increased risk for poor neonatal outcome. women with no RFM) using the Chi-square test. For the outcomes the relative risk with 95% confidence interval (CI) was calculated. factor with a poor neonatal outcome was studied as a crude association. One additional multivariable model was built in which the statistically significant factors from the univariable analysis were used.
The interactions between the statistically significant variables were tested in a three-step analysis, with each interaction tested individually in the final multivariable model. The interaction was significant if P < .05. Due to an interaction between parity and IVF, a new dummy variable was constructed: no IVF multipara (reference), no IVF nulliparous, IVF nullipara and IVF multipara. The results of the analyses are presented as odds ratios (ORs) with 95% CIs. OR was used as a proxy for relative risk.

| Ethical approval
The study was approved by the independent regional Research Ethics Committee, Karolinska Institutet, Stockholm, Sweden

| RE SULTS
During the study period, there were 4417 visits for RFM at the clinic.
Of these, 377 visits were made by women who later delivered in other counties/abroad or for whom there were no delivery data available for the analysis. Additionally, 42 visits were multiple pregnancies and therefore excluded. A total of 3243 women with RFM were included in the analysis, with 3998 visits to the department due to RFM.
During the same period, the obstetrical clinic at Soder Hospital had 14 815 singleton deliveries with 34 IUFDs (2.3/1000). Baseline data are presented in Table 1. Newborn children of women with RFM had a higher frequency of low Apgar score at 5 minutes and low pH in umbilical cord blood, and the labor was induced more often ( The classification tree analysis shows that the occurrence of composite poor neonatal outcome ranged from 6.2% to 18.4% within the different groups ( Figure 1). The results showed that the highest risk for poor neonatal outcome (18.4%) was found in the group with an SGA fetus. Another high-risk group for composite poor fetal outcome (12.8%) was women with normal birthweight/ macrosomia fetuses and an in vitro fertilization pregnancy.  This may give a misleading reassurance to pregnant women. An improved ability to identify at-risk pregnancies among the normal ones must be regarded as a priority in future research.

| D ISCUSS I ON
In the project, a composite fetal outcome was constructed to help us investigate whether fetuses in the group with RFM were born with a worse outcome, even if not stillborn. The data showed that there is an increased risk of having a poor fetal outcome in pregnancies where the mother seeks help due to RFM compared with those with normal fetal movements. The group with no RFM had a higher rate of admissions to the neonatal ward, which can be explained partially by a higher premature delivery rate (5.2% vs 3.8%, P < .01). Two high-risk combinations could be identified: women with SGA fetuses and women with an IVF pregnancy. The incidence of presentation with RFM at the hospital was 21.4%, which is similar to newly published data from the UK 17 but considerably higher than previous studies. 18,19 This may be due to increased awareness both in pregnant women and in the healthcare givers.
Similar to our findings, a recent published study from UK showed no association between recurrent RFM episodes and increased adverse neonatal outcome, but an association with higher induction rates. 17 Furthermore, a liberal induction policy starting from 37 weeks for the women with recurrent RFM showed no benefit. 20 Several large projects such as "Each Baby Counts" and the AFFIRM study are trying to improve information and knowledge about IUFD but have not been successful in preventing it. 21,22 A Norwegian project is described as one of the most successful regarding RFM and stillbirth. The project reduced the incidence of IUFD by 50% (from 4.2‰ to 2.4‰) among the RFM group and from 3.0‰ to 2.0‰ in the overall population. Strict standardized information is given to the women regarding RFM. The project performed CTG and ultrasound to assess fetal well-being; ultrasound was the most important (86%) in detecting abnormalities in fetal health (11.6%). 23 Late routine ultrasound screening has been shown to triple detection of SGA fetuses (to 57%), a group which is at an 18-fold increased risk of newborn morbidity. 24 A Swedish cohort study reported a sixfold reduced risk of IUFD among identified vs non-identified SGA pregnancies. 15 In agreement with those finding, a Cochrane report on third trimester ultrasound reported large difference in IUFD risk among those with vs those without 34-36 weeks' ultrasound (0/1447 vs 9/1455, OR = 0.05, 95% CI 0.0-0.9). 25 Thus signs of growth restriction seem to be a major risk factor for stillbirth and other severe adverse outcomes in women with and without RFM. 10,26,27 Ultrasound fetometry seem to increase the in utero identification of SGA. A late routine ultrasound with the aim to identify signs of growth restriction would thus probably be an effective measure to reduce the stillbirth rate.
Following fetal growth during pregnancy is important, and symphysis-fundal height (SFH) is the metric used in every Swedish maternity clinic for identifying SGA fetuses. However, it has been shown already in 1995 that ultrasound is superior. 28  One of the strengths of this study is that all single pregnancies with reported RFM at the clinic during the 2-year period were included. The sample size was large, and only 10% were lost to follow up.
There are also some limitations of this study. Soder Hospital is a large city hospital in Stockholm and the women included in the study were older and probably better educated than the national average in Sweden, which may have some influence on the findings.
Furthermore, the retrospective study design used is prone to selection bias and the quality of the data is dependent on the accuracy of the medical data registered. Another limitation is that our study included only one hospital, which could make it difficult to generalize the results. However, our results can be applied in similar settings with similar routines.

| CON CLUS ION
The low incidence of adverse neonatal outcome in this study may be explained by the easy access and proximity to obstetric care in Stockholm, well-informed patients, good compliance of caregivers with the local guidelines, and the provision of free medical care for all pregnant women. There was no difference in the neonatal outcome between the group with recurrent RFM and the that with single episodes. We found the highest incidence of poor neonatal outcome among nulliparous and multiparous women with RFM whose fetuses were diagnosed as SGA. This suggests that a routine ultrasound assessment for fetal growth in late pregnancy could be beneficial, especially in women with RFM. By better selection of atrisk patients within the group with RFM, we can hopefully minimize the intervention rate while improving the neonatal outcome.

ACK N OWLED G M ENTS
We would like to acknowledge the assistance of Hans Pettersson with the statistical analysis.

CO N FLI C T O F I NTE R E S T
The authors have stated explicitly that there are no conflicts of interest in connection with this article.