The association between undetected small‐for‐gestational age and abnormal admission cardiotocography: A registry‐based study

To assess the association between undetected small‐for‐gestational age (SGA) fetuses and abnormal admission cardiotocography (admCTG) in a low‐risk population.


| I N TRODUC TION
Admission cardiotocography (admCTG) was introduced during the 1980s, and was stimulated in part by a study by Ingemarsson et al. 1 The rationale of admCTG was to detect 'fetal distress' that was already present upon arrival in the labour ward, thus avoiding an unnecessary delay in intervention. However, the use of routine admCTG in low-risk pregnancies has been questioned, and some national guidelines actively advise against its use. 2,3 Their current recommendation is to auscultate the fetal heart rate for at least 1 minute after one contraction at admission. However, in high-risk pregnancies, there is wide agreement that using admCTG is advisable. 2,4 Randomised controlled trials (RCTs) have not shown any neonatal benefit when comparing admCTG with auscultation alone in low-risk pregnancies. Studies have suggested that admCTG results in an increased risk of continuous cardiotocography (CTG) during labour, fetal blood sampling and caesarean delivery. [5][6][7][8] However, the proportion of abnormal admCTG in these studies was high, with a rate of 22-32%. 5,6 CTG was performed following medical history taking and examination (even after iatrogenic rupture of membranes), and 43% of the women were excluded after admission. 5 Moreover, these studies investigated neonatal outcomes occurring several hours after admission, which was not the original reason for performing admCTG, but rather to detect ongoing fetal compromise at admission.
In Sweden, 93-95% of all admCTG are normal. 9 An ad-mCTG is performed immediately upon arrival before history taking and examination. 1 If the first 20-30 minutes of a CTG are normal, a low-risk fetus is monitored by intermittent auscultation during labour. 10 Therefore, after 30 min of a normal admCTG, the fetal surveillance is similar to the guidelines not using universal admCTG (e.g. National Institute of Health and Care Excellence guidelines). 2 Detection of small-for-gestational age (SGA) fetuses is poor. In one Swedish study, only 34% of SGA fetuses were identified before delivery with standard care including serial symphysis-fundus height and ultrasound if indicated. 11 Undetected SGA fetuses are more prone to hypoxic complications, especially during vaginal delivery. 12 One of the main objectives of admCTG is to identify pregnancies that have an increased risk of asphyxia complications, such as undetected SGA fetuses.
We hypothesise that undetected SGA is more common in fetuses with an abnormal admCTG than in fetuses with a normal admCTG.

| M ET HODS
The study population included singleton births from the Stockholm-Gotland perinatal cohort, Sweden, between 1 February 2012 and 15 June 2020. This cohort contains all electronic medical records from maternity, delivery and postnatal care units of the Stockholm and Gotland regions in Sweden. These records are linked to the Swedish Neonatal Quality Register to retrieve neonatal care data using the unique personal identification number assigned to each Swedish resident at birth.
All exclusion criteria were determined before the data were accessed. Multiple gestations, elective caesarean deliveries, induction of labour, non-vertex presentation, placenta previa and preterm births before 34 weeks of gestation were excluded. The study population consisted of women with the symptoms of spontaneous labour (including women admitted in the latent phase who were then induced because of an abnormal admCTG: code O363 in the International Classification Diagnosis, tenth revision code system, 'signs of fetal hypoxia, e.g. antenatal pathological CTG'), cephalic presentation, singleton pregnancy and at least 34 weeks of gestation.
We sub-grouped deliveries into low-risk and high-risk groups. The high-risk group was defined as a pregnancy duration of less than 37 +0 and more than 41 +6 gestational weeks, pre-gestational maternal morbidity (hypertension, diabetes, systemic lupus and chronic kidney disease), fetal demise in a previous pregnancy, body mass index (BMI) more than 35 kg/m 2 , nulliparous at more than 40 years of age, previous caesarean delivery, gestational hypertension, pre-eclampsia, HELLP (haemolysis, elevated liver enzymes and low platelet count), eclampsia, diabetes, diagnosed fetal growth restriction (FGR), oligohydramnios, diagnosed macrosomia, prolonged amniotic fluid leakage for more than 18 hours, meconium staining of the amniotic fluid on admission, chorioamnionitis and placental abruption. All diagnoses were established before admission, except for chorioamnionitis and placental abruption.
Information on admCTG results was extracted from the medical record of when women were in labour where the responsible midwife assessed admCTG as normal, abnormal or not performed. The CTG classification system in Sweden was revised and implemented nationwide in 2017. Before 2017, a normal intrapartum CTG registration included a fetal heart rate baseline frequency of 110-150 beats/min (bpm), a variability of 5-25 bpm, no decelerations (except early or variable decelerations with a duration of <30 seconds and with an amplitude of <60 bpm) and the presence of two or more accelerations/60 minutes. Since 2017, a normal intrapartum CTG has been defined as a baseline frequency of 110-160 bpm, a variability of 5-25 bpm and no repetitive decelerations (occurring with >50% of contractions), except for repetitive early and repetitive variable decelerations with a duration of less than 60 seconds. Accelerations do not have to be present.
Birthweight was measured and classified according to Marŝál et al. 13 as follows: f(x) = −0.000002278843x 4 + 0.001402168x 3 -0.2008726x 2 + 9.284121x − 41.25956, where f(x) corresponds to the expected fetal weight (grams) at a given gestational age (days), x. The birthweight deviation expressed as the percentage was calculated as follows: (birthweight − expected fetal weight)/expected fetal weight (i.e. without sex or ethnic adjustment).

| Outcomes
The main outcome was the proportion of neonates that were SGA, comparing fetuses that had an abnormal admCTG with fetuses that had a normal admCTG. Undetected SGA was categorised as a birthweight at or below the tenth centile and at or below the third centile (corresponding to ≤15% and ≤22% birthweight deviation from the expected fetal weight for gestational age), respectively. 13 These fetuses were regarded as undetected SGA because they had not been identified before delivery by ultrasound assessment.
The secondary outcomes were the risk of neonatal complications (combined into composite outcomes) in the lowrisk group. We compared undetected SGA fetuses with a birthweight at or below the tenth centile and abnormal ad-mCTG with appropriate-for-gestational age (AGA)/largefor-gestational age (LGA) fetuses with normal admCTG. These outcomes were assessed for neonates born less than 6 hours after admCTG because we wanted to focus on the key concept of admCTG -to assess neonatal compromise at admission.
The composite outcomes were defined as follows: Severe adverse neonatal outcome: This outcome comprised an Apgar score less than 4 at 5 minutes, hypoxic-ischemic encephalopathy (HIE) grade 2-3, neonatal seizures, neonatal death (death up to 28 days of age), and intrapartum death.
Moderate adverse neonatal outcome: This outcome comprised an Apgar score less than 7 at 5 minutes, umbilical artery pH less than 7.0, metabolic acidosis defined as umbilical artery pH less than 7.05 and a base deficit of 12 mmol/L or more, and advanced neonatal resuscitation (one of the following: correction of acidosis, adrenaline use, continuous positive airway pressure, extra oxygen supplement and heart compressions). Therefore, outcome groups were not mutually exclusive (e.g. newborns included in the moderate adverse group might have Apgar scores below 4 at 5 minutes).
Operative delivery: This outcome was emergency caesarean or instrumental vaginal delivery due to suspected fetal hypoxia.
No patients or the public were involved in the development of the research study. No core outcome set has been used as an outcome in this study.

| Statistical analysis
Descriptive characteristics are shown as numbers and proportions for categorical data, and as the mean ± standard deviation, median and interquartile range for continuous variables. Logistic regression was used to calculate odds ratios (OR) with 95% CI for neonatal outcomes in the different admCTG groups. IBM SPSS Statistics, version 28 (IBM, Armonk, NY, USA) was used for the statistical analysis.
In the revised Swedish CTG classification, more fetal heart rate patterns have been included in the normal group. Therefore, we assume that a greater proportion of ad-mCTG were classified as normal after 2017. Consequently, we performed adjusted analyses with this covariate before/ after January 2017. A multiple logistic regression analysis was performed to control for other potential confounders. Confounders that we determined to be of clinically relevant interest were maternal age (continuous), parity (nulliparous or parous), BMI (continuous), gestational length (<38 weeks,

| R E SU LTS
We included 192 314 women from the Stockholm-Gotland Perinatal cohort in the study. Of the women who had labour induced, 39 642 were excluded from the analysis because their labour was induced for pre-existing medical indications, while 705 were induced because of suspected fetal hypoxia and were included in the analysis. Therefore, the eligible study population consisted of 153 016 deliveries. In the whole group, including high-and low-risk pregnancies, 16 242 neonates were born SGA at or below the tenth centile (10.6%) and 4163 (2.7%) were born SGA at or below the third centile. Of these, 25 555 pregnancies were defined as high risk and excluded and 127 461 were defined as low risk. In the low-risk group, 13 080 (10.3%) neonates were born SGA at or below the tenth centile and 3165 (2.5%) were born SGA at or below the third centile. Missing information on ad-mCTG was present in 13.1% (n = 16 780) of the low-risk pregnancies. Of the remaining 110 681 births, 4.9% of admCTG tracings in the low-risk group were classified as abnormal. In 1.2% of women, admCTG was not performed ( Figure S1). In all, 102 fetuses had missing data on birthweight and were excluded from the analysis.
The characteristics of the low-risk population were stratified for admCTG outcome and SGA versus AGA/LGA (Table 1). Other admCTG categories included 'not performed' and 'missing data'.
The main outcome is shown in Table 2. The proportion of neonates born SGA at or below the tenth centile in low-risk deliveries (n = 127 461) in the abnormal admCTG group was twice that in the normal admCTG group (18.6% [n = 1010] versus 9.7% [n = 10 057]: OR 2.1, 95% CI 1.9-2.3). The corresponding proportion for SGA at or below the third centile in the abnormal admCTG group was three times that in the normal admCTG group (6.5% [n = 353] versus 2.2% [n = 2294]: OR 3.1, 95% CI 2.7-3.5).
The risk of a composite severe adverse neonatal outcome in the low-risk population was 0.081%. Of the 54 432 neonates born less than 6 hours after admission for labour and with admCTG information, 9.2% (n = 5027) were born SGA at or below the tenth centile. Of these, 13 (0.3%) suffered from a composite severe adverse neonatal outcome compared with 31/51 467 (0.06%) of AGA/LGA neonates (OR 4.1, 95% CI 2.2-7.9). Women with missing information on BMI (n = 1999) were excluded from the adjusted analysis.
The risk of unfavourable neonatal outcomes within 6 hours was significantly higher when admCTG was abnormal and the neonate was born SGA (at or below the tenth centile) than that in those with a normal admCTG and AGA or LGA (Table 3). Of the SGA newborns with abnormal admCTG, 1.2% had a severe neonatal outcome compared with 0.05% who had a normal admCTG and were AGA/LGA (adjusted OR [aOR] 23.7, 95% CI 9.8-57.3; Table 3, Figure 1, Figure S2).
The risk of an operative delivery (emergency caesarean or instrumental vaginal delivery) was more common in neonates with abnormal admCTG/SGA. A total of 23.2% of   Figure 1, Figure S2). The risk of caesarean delivery due to suspected fetal hypoxia was 15.8% in neonates with abnormal ad-mCTG/SGA, whereas the risk was 0.2% in those with normal admCTG/AGA/LGA (aOR 62.7, 95% CI 45.7-86.1).
The number of neonates born less than 6 hours after admission with admCTG patterns that could not be classified was 12 291 (not performed [n = 1247], missing [n = 11 044]). Of these, 0.1% had a composite severe neonatal outcome, and 1.2% had a composite moderate neonatal outcome.
In the group of women who underwent induction of labour owing to suspected fetal hypoxia, only 30 were assessed as low-risk and delivered within 6 hours after admission. Two of these experienced a severe adverse TA B L E 3 Fetal and maternal outcome among women assessed as low risk and delivered within 6 hours from admission, crude odds ratio, n = 54 432.
When we assessed the possible effect of the new revised CTG classification system in Sweden, the proportion of abnormal admCTG in the low-risk group decreased from 5.9% before this system to 3.5% after its implementation in 2017 (p < 0.001). The corresponding rate of a composite severe adverse neonatal outcome was 0.16% (before 2017) versus 0.20% (after 2017) (p = 0.11) and that of a composite moderate adverse neonatal outcome was 1.4% versus 1.6% (p < 0.001). The risk of an operative delivery due to suspected fetal hypoxia decreased after the revised CTG classification was launched (from 4.7% to 3.5%, p < 0.001) ( Table 4).

| Main findings
The main finding in this study was that, in low-risk pregnancies, the proportion of undetected SGA was almost twice as high in fetuses who had abnormal admCTG (18.6%) compared with that in fetuses who had normal admCTG (9.7%). Moreover, fetuses with undetected SGA during pregnancy and abnormal admCTG had a substantially higher risk of moderate and severe intrapartum and neonatal complications. In a normal-sized fetus with normal admCTG, there was a low likelihood of a severe adverse neonatal outcome (1/2000) or being delivered by emergency caesarean section owing to suspected fetal hypoxia (1/500) among those delivered within 6 hours after admission. F I G U R E 1 Odds ratios for neonatal outcomes among fetuses with abnormal admCTG/SGA compared to normal admCTG/AGA or LGA, adjusted for BMI, parity and calendar year.

T A B L E 4
Comparison fetal and maternal outcome among women assessed as low risk and delivered within 6 hours from admission, abnormal admCTG/SGA at or below the tenth centile versus normal admCTG/AGA or LGA, adjusted odds ratios, n = 47568 a . Note: Adjusted for body mass index, parity and calendar year before/after 2017.

Normal admCTG/AGA/LGA
Abbreviations: admCTG, cardiotocogram at admission; AGA, appropriate for gestational age; aOR, adjusted odds ratio; BD, base deficit; LGA, large for gestational age; OR, odds ratio; SGA, small for gestational age. a 1999 women excluded, due to missing information on body mass index.

| Strengths and limitations
Our population-based data were obtained from a large proportion of the births in Sweden. Our data represent routine labour care as opposed to previous RCTs with inclusion criteria, such as early amniotomy and a high proportion of abnormal admCTG. Additionally, these RCTs failed to assess the outcome at admission, as intended. We assessed neonatal outcome restricted to the first 6 hours after admission. There was a high proportion of missing data for umbilical cord samples. We speculate that, in cases of suspected fetal hypoxia, umbilical sample collection is prioritised. Therefore, the missing data were probably from cases of uncompromised neonates.
The exact date of diagnosing maternal disease (e.g. preeclampsia) was dichotomised into before or after the delivery date. Bias could have been introduced when excluding a diagnosis set on the day of delivery or after. Among SGA fetuses with abnormal admCTG who were diagnosed with hypertensive disease at or after the delivery date, none had a severe adverse neonatal outcome and only two had a moderate adverse neonatal outcome. Therefore, we believe that this particular risk of bias was low.
We do not have data on the exact duration of the admCTG, but the routine time is 20-30 minutes. If the admCTG is not reactive, fetal stimulation is recommended. We speculate that the reasons for the missing admCTG data (13%) are as follows. The admCTG could have been difficult to classify and so be more likely to be abnormal. We observed that the risk of SGA was marginally higher in the group with missing admCTG data than in the normal admCTG group, and there was also a higher risk of poor neonatal outcomes.
The reason why we included women with induction of labour owing to suspected fetal distress is that the diagnosis is set after admission (i.e. the women were considered low-risk at admission). In Sweden, we do not use the sexspecific algorithm of Marŝál et al. 13 Therefore, boys are less often classified as SGA in Sweden. We analysed the potential confounding effect of this factor in a regression analysis. However, when adjusting for sex, the estimates for an adverse neonatal outcome did not change.

| Interpretation
Detecting SGA fetuses during pregnancy is difficult and most such fetuses are diagnosed with SGA only after birth, 11 Although the proportion of diagnosed SGA is slightly higher when late ultrasound biometry is used, many SGA fetuses are still missed. 12 Therefore, among presumed low-risk pregnancies, there is a substantial group of SGA fetuses (10.3%) who are at a higher risk of intrapartum compromise. Our study supports such findings. Steer et al. 14 investigated the interaction of different risk factors (fetal heart rate abnormalities, FGR, meconium and tachysystole) in a nested case-control study and observed that FGR was the best discriminator of adverse outcome versus no adverse outcome. Our data are in agreement with this finding that SGA is a strong discriminator of an adverse outcome. However, abnormal admCTG is an even stronger discriminator of a severe adverse neonatal outcome.
Lovers et al. investigated features of admCTG and neonatal outcome in 27 927 women in labour. 15 Compromised neonates were more often born SGA below the third centile than uncompromised neonates. Severe compromise was associated with a baseline fetal heart rate of more than 150 bpm, non-reactive tracing and prolonged decelerations. Importantly, some of these CTG features are difficult to auscultate. There is a question of whether the abnormal admCTG patterns in our study could have been detected by auscultation alone. Our experience is that the growthrestricted fetus frequently presents with late shallow decelerations and reduced variability early in labour, which are most likely not able to be auscultated. 16 A Cochrane report on admCTG versus auscultation stated that 'no trial or metaanalysis will be adequately powered to detect differences in perinatal mortality'. 17 A Swedish observational study concluded that admCTG was of benefit for 74% of women who required emergency caesarean delivery within 1 hour of arrival at the hospital. 16 The main argument against universal admission CTG in low-risk pregnancies is that some studies suggested an increased risk of continuous CTG, fetal blood sampling and caesarean delivery, without any neonatal benefit. 3,5,6,17 We believe that the two former complications are not of such a degree that they justify the abandonment of admCTG. The priority should remain the reduction of neonatal morbidity and mortality.
A Cochrane report from 2017 17 showed that in a metaanalysis on 11 338 women, there was a 20% increased rate of caesarean delivery (risk ratio 1.20, 95% CI 1.00-1.44). A subsequent RCT including 3034 deliveries showed no higher risk, but instead showed a lower risk of caesarean delivery in the admCTG group than in the auscultation group (6.9% versus 8.6%, although not statistically significant), 18 hence reducing the previous belief that admCTG increases the risk of caesarean delivery ( Figure S3).
An appropriate risk assessment is important. A retrospective study that evaluated labour complicated by stillbirth, neonatal death and severe brain injury showed that adverse outcomes were assessed as avoidable in two-thirds of cases, and inappropriate risk assessment was present in 40-60%. 19 Furthermore, studies that assessed whether national guidelines of intermittent auscultation in low-risk pregnancies are followed showed that the majority of lowrisk pregnancies were subjected to admCTG in the UK and Norway, contrary to guidelines. 17,20 A comparison between the only in-hospital birth care unit in Sweden not using ad-mCTG and routine care showed an 18-fold increased risk of peripartum fetal death in the unit not using admCTG (0.9 versus 0.05/1000). 16 Therefore, to determine whether auscultation alone is as safe as admCTG in combination with auscultation in lowrisk pregnancies, large RCTs would need to be performed.
With the routine use of universal admCTG in Sweden, we consider that initiating such a study would be unethical. The reason for this belief is that a substantial proportion of our low-risk population is undetected SGA fetuses with a higher risk of a severe neonatal outcome, especially in combination with abnormal admCTG.

| CONCLUSIONS
In a low-risk population in Sweden, where suspected SGA was excluded, there was a non-negligible group of undetected SGA fetuses who were more prone to having abnormal admCTG. This group of fetuses also had a substantially higher risk of a severe or moderate adverse neonatal outcome than AGA/LGA fetuses. Our findings support the value of admCTG as a screening tool in a lowrisk population.

AU T HOR C ON T R I BU T ION S
MH initiated the study. All authors participated in the study design. EG, MH and KJ participated in the acquisition of data and analysis. All authors participated in interpretation of data. EG wrote the first manuscript draft, and all authors contributed to the intellectual content, revised the manuscript, and approved the final version for submission.

AC K NO W L E D GE M E N T S
We thank Ellen Knapp from Edanz (https://edanz.com/ac) for editing a draft of this manuscript.

F U N DI NG I N FOR M AT ION
The study was funded by the Units of Research, Education and Development Region of Jämtland Härjedalen and Region Stockholm.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared. Completed disclosure of interests form available to view online as supporting information.

DATA AVA I L A BI L I T Y S TAT E M E N T
The data used in this study were available from the Stockholm-Gotland Perinatal Cohort. Restrictions apply to the availability of these data, which are used via virtual private network after approval.

E T H IC S A PPROVA L
The study received ethics approval by the Regional Ethics Committee in Stockholm (Dnr 2009/275-31) and the Swedish Ethical Review Authority (Dnr 2019-02818).