NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Women's and Children's Health (UK). Intrapartum Care: Care of Healthy Women and Their Babies During Childbirth. London: National Institute for Health and Care Excellence (UK); 2014 Dec. (NICE Clinical Guidelines, No. 190.)

Cover of Intrapartum Care

Intrapartum Care: Care of Healthy Women and Their Babies During Childbirth.

Show details

10Monitoring during labour

10.1. Cardiotocography compared with intermittent auscultation during established labour

10.1.1. Review question

What is the effectiveness of electronic fetal monitoring compared with intermittent auscultation during established labour?

For further details on the evidence review protocol, please see appendix E.

10.1.2. Description of included studies

Six studies were included in this review (Grant et al., 1989; Kelso et al., 1978; Leveno et al., 1986; MacDonald et al., 1985; Vintzileos et al., 1993; Wood et al., 1981).

Five of the included studies reported 4 randomised controlled trials that compared continuous electronic fetal monitoring (EFM) using cardiotocography (CTG) with intermittent auscultation during labour (Grant et al., 1989 followed up children whose mothers had participated in MacDonald et al., 1985). The sixth included study was a quasi-randomised trial that allocated women to selective or universal CTG in alternating months, a- - thnd this generated data for the comparison of interest (Leveno et al., 1986).

Two of the included studies only included women with low risk pregnancies (Wood et al., 1981) or reported data separately for women with low risk pregnancies (Leveno et al., 1986). In the other 4 studies, the majority of women had low risk pregnancies, but 20–30% of women were giving birth before term, underwent induction of labour or had antepartum risk factors (more details of specific inclusion and exclusion criteria can be found in the evidence tables in appendix I).

In 1 study, EFM was performed externally unless the CTG trace quality became unsatisfactory, in which case monitoring was done internally using a fetal scalp electrode (Vintzileos et al., 1993) whereas in another study, monitoring was external until membranes ruptured and then internal (Wood et al., 1981). In 3 studies, monitoring was done internally (Grant et al., 1989; Kelso et al., 1978; MacDonald et al., 1985). In 1 study it was not reported whether monitoring was internal or external (Leveno et al., 1986).

10.1.3. Evidence profile

A fixed effect model was used for these analyses, with the exception of 2 outcomes (instrumental vaginal birth for any indication and neonatal acidosis), for which a random effects model was used due to the high heterogeneity (I2>60%).

Table 61. Summary GRADE profile for comparison of electronic fetal monitoring using CTG compared with intermittent auscultation during established labour.

Table 61

Summary GRADE profile for comparison of electronic fetal monitoring using CTG compared with intermittent auscultation during established labour.

10.1.4. Evidence statements

There was evidence that women monitored with CTG had lower rates of spontaneous vaginal birth (n=2859) and higher rates of instrumental vaginal birth and caesarean section for fetal distress (n=15,823) than women monitored with intermittent auscultation. There was evidence of a higher risk of seizures (n=16,099) in babies born to women monitored with intermittent auscultation, but no evidence of a difference in other neonatal outcomes, including: mortality (n=30,561); cerebral palsy (n=13,079); hypoxic ischaemic encephalopathy (n=1428); intraventricular haemorrhage (n=1428); respiratory distress (n=1428); abnormal neurologic symptoms or signs (n=11,571); admission to neonatal intensive care unit (NICU) (n=30,491); and low umbilical artery or venous pH at birth (n=2494). The evidence was of high to very low quality.

10.1.5. Health economics profile

No published economic evaluations were identified for this question.

10.1.6. Evidence to recommendations

10.1.6.1. Relative value placed on the outcomes considered

In this review, the guideline development group hoped to find whether the use of continuous CTG in labour was any more effective than intermittent auscultation in identifying babies who are at greater risk due to developing acidosis during labour and who might require additional care or expedited birth. The key outcomes of interest were: the mode of birth; the rates of fetal and neonatal death; and the rates of more serious morbidities such as cerebral palsy and hypoxic ischaemic encephalopathy.

10.1.6.2. Consideration of clinical benefits and harms

The evidence showed that there were significantly more spontaneous vaginal births in the group that received intermittent auscultation compared with the group that received continuous CTG. There was also a significantly greater number of instrumental vaginal births (both for any indication and specifically for fetal distress) in the CTG group. CTG was also associated with a statistically significant increase in the number of caesarean sections for fetal distress (5 more per 1000 births). These findings seemed to suggest that the use of CTG in labour results in an increase in interventions. However, for the majority of neonatal morbidities, there were no statistically significant findings between the 2 groups. The only statistically significant difference in neonatal morbidity was in seizures, with a lower incidence in the CTG group than the auscultation group, but although this was a significant finding, the absolute reduction was very low, with a rate of 1 fewer per 1000 babies.

The guideline development group concluded that the evidence seemed to suggest that the use of CTG in labour leads to an increase in the number of interventions without a concomitant increase in improved neonatal outcomes. The group noted that in a low risk population, major adverse outcomes are going to be very rare, and thus a large number of women would have to undergo CTG in order to prevent these outcomes. The group did not feel that this was a proven and clinically beneficial trade-off, so they endorsed the recommendation from the previous edition of the guideline that CTG should not be used in established labour unless there was a specific indication suggesting increased risk for the wellbeing of the unborn baby which would justify switching from intermittent auscultation.

The group discussed the appropriate method for conducting auscultation. They noted that the previous version of the guideline had indicated that the fetal heart rate heard should be recorded as an average. It was felt that this wording was not clear about whether this should be written as a range or a single figure. The group agreed that the fetal heart rate should be counted for 1 minute and the result should be written as a single figure, and the wording of the recommendation was changed accordingly. The need to auscultate the fetal heart for 1 minute immediately after a contraction in order to detect any late decelerations was also noted as important and included in the recommendations. The group also noted that if any fetal heart rate abnormality is suspected during labour when conducting intermittent auscultation, or if there is any uncertainty about the origin of the heart sounds being heard, then the maternal pulse should be palpated to differentiate between the woman's and the unborn baby's heart rates.

10.1.6.3. Consideration of health benefits and resource uses

The clinical evidence showed that the use of CTG rather than intermittent auscultation in established labour would lead to an increase in the number of interventions such as caesarean section and instrumental vaginal birth (as well as their associated morbidities both for the woman and the baby, including the potential interruption of physiological maturation). The perceived benefits from CTG monitoring are that there will be fewer babies born with severe fetal acidosis or, at least, its impact may be ameliorated. However, the increase in caesarean sections and instrumental births did not appear to be associated with the reduction in poor outcomes required to ensure that using continuous CTG could be considered cost effective. As a result, the guideline development group felt that not recommending the use of continuous CTG could lead to potential health benefits from fewer unnecessary birth interventions. Reducing the use of continuous CTG could lead to cost savings if less CTG equipment is required in the labour ward, with its associated maintenance costs, and also reduced use of ancillary resources such as pH monitoring.

10.1.6.4. Quality of evidence

The previous version of this guideline included a Cochrane systematic review as evidence for this review question. However, for this updated review, due to the variability in the study populations among the included studies, the guideline development group agreed that it would be more appropriate to appraise them for inclusion as individual trials. This meant that it was possible to remove some papers because they were not applicable to the population of the guideline, for example if the study population only included women at higher risk. This ensured that the evidence for the updated review was more directly applicable to a low risk population than was the case for the original version of the guideline.

The quality of the evidence ranged from high to very low for different outcomes. In general though, despite the reasons for downgrading the evidence, the group felt sufficiently confident in the findings to make its recommendations

10.1.6.5. Other considerations

The guideline development group was aware that there was some concern among clinicians about not using CTG in established labour (this is similar to the concern about monitoring on admission). The group felt that too often clinicians were using CTG and the resultant fetal heart tracings for reassurance, rather than for clearly defined reasons. As a result, the group was confident in recommending that continuous CTG not be used in established labour for women at low risk of complications.

As with monitoring on admission to labour, the group agreed that accelerations or decelerations should be recorded if they are heard on intermittent auscultation.

The group agreed that it would be helpful to be more specific about the potential reasons why clinicians should consider changing from intermittent auscultation to continuous CTG. After considering the list of reasons presented in the original guideline, the group added some additional reasons based on their clinical experience. The group acknowledged that for most of these factors there is no evidence of improved outcomes with the use of CTG compared with intermittent auscultation, but they believed that the continuous surveillance afforded by CTG, along with its ability to pick up variability in the fetal heart rate, suggested it may be of benefit in circumstances where the unborn baby is at increased risk of developing acidosis. They felt it appropriate to divide indications for CTG into those where CTG should be considered and those where CTG should be advised; this was done based on the degree of risk of fetal acidosis thought to be associated with each factor and is reflected in the recommendations. Findings from the evidence for continuous CTG when there is significant meconium show an increase in intrapartum interventions but fewer admissions to neonatal intensive care (see section 10.2). This led the group to continue to recommend that CTG should be offered where there is significant meconium present. From their clinical experience, the group members felt it appropriate to differentiate between non-significant meconium and significant meconium. The group agreed that non-significant meconium alone was not a reason to advise continuous CTG, but that instead it should prompt a full risk assessment and continuous CTG should be advised if other risk factors were then found to be present.

The group also discussed a woman's request for CTG as it was included in the original guideline. They felt that it was unusual for a woman to request switching to CTG without any other clinical indication. In addition, there is evidence that use of CTG in low risk women is associated with increased intrapartum interventions and therefore also their associated morbidities, with no evidence of improved outcomes for the baby. Consequently, the group felt it inappropriate to include women's choice as a standard intrapartum indication for switching to CTG, with the understanding that intrapartum care is always underpinned by the principles of choice and control as outlined in chapter 4 (Care throughout labour).

Finally, the group acknowledged that CTG does not measure the fetal pulse itself but an electrical signal (generated by either a Doppler recording of the fetal heart movements or an electrical signal from the fetal ECG with a fetal scalp electrode). They noted that there are limitations to the extent to which fetal heart rate is a surrogate for fetal hypoxia and acidosis. Fetal heart rate can be affected by factors other than fetal hypoxia, such as fetal behavioural state, maternal analgesia and pyrexia.

10.1.7. Recommendations

99.

Offer intermittent auscultation of the fetal heart rate to low-risk women in established first stage of labour in all birth settings:

  • Use either a Pinard stethoscope or Doppler ultrasound.
  • Carry out intermittent auscultation immediately after a contraction for at least 1 minute, at least every 15 minutes, and record it as a single rate.
  • Record accelerations and decelerations if heard.
  • Palpate the maternal pulse if a fetal heart rate abnormality is suspected, to differentiate between the two heart rates. [new 2014]
100.

Do not perform cardiotocography for low-risk women in established labour. [new 2014]

101.

Advise continuous cardiotocography if any of the following risk factors are present or arise during labour:

  • suspected chorioamnionitis or sepsis, or a temperature of 38°C or above
  • severe hypertension (160/110 mmHg above [see Hypertension in pregnancy (NICE clinical guideline 107)]).
  • the presence of significant meconium (see recommendation 164)
  • fresh vaginal bleeding that develops in labour. [new 2014]
102.

If any one of the following risk factors is present or arises during labour, perform a full assessment of all factors listed in recommendation 163.

  • prolonged period since rupture of membranes (24 hours or more) (see recommendations 59 to 64)
  • moderate hypertension (150/100 to 159/109 mmHg [see Hypertension in pregnancy (NICE clinical guideline 107)])
  • confirmed delay in the first or second stage of labour (see recommendations 175, 195 and 199)
  • the presence of non-significant meconium.

Advise continuous cardiotocography if 2 or more of the above risk factors are present, or any other risk factor in recommendation 163 is present with 1 of these. [new 2014]

103.

Do not regard amniotomy alone for suspected delay in the established first stage of labour as an indication to start continuous cardiotocography. [2007, amended 2014]

104.

Address any concerns that the woman has about continuous cardiotocography, and give her the following information:

  • Explain that continuous cardiotocography is used to monitor the baby's heartbeat and the labour contractions.
  • Give details of the types of findings that may occur. Explain that a normal trace is reassuring and indicates that the baby is coping well with labour, but if the trace is not normal there is less certainty about the condition of the baby and further continuous monitoring will be advised.
  • Explain that decisions about whether to take any further action will be based on an assessment of several factors, including the findings from cardiotocography. [new 2014]
105.

If continuous cardiotocography has been used because of concerns arising from intermittent auscultation but there are no non-reassuring or abnormal features (see table 92) on the cardiotocograph trace after 20 minutes, remove the cardiotocograph and return to intermittent auscultation. [new 2014]

10.1.8. Research recommendations

16. What are the natural frequencies of the avoidable harms that cardiotocography is intended to prevent for women who are assessed as being at low risk of complications at the start of labour? Does using cardiotocography in labours where complications develop confer a net benefit compared with intermittent auscultation?

Why this is important

Cardiotocography is used in current practice to monitor the fetal heart rate when there is a concern that fetal hypoxia may develop. It is regarded as unethical, in most circumstances, to conduct clinical research where women whose labour is categorised as ‘high risk’ are not offered cardiotocography. There is therefore no high-quality evidence about the size of the benefit or harm derived from the use of cardiotocography compared with intermittent auscultation, either in individual cases or across a whole population. Further analysis is needed to evaluate the actual (or probable) benefits and harms associated with this screening test. This would be based on analysis and modelling using data and assumptions derived from existing evidence from a range of countries, comprising data from any studies and/or historic data sets that record the natural frequencies of avoidable damage caused by intrapartum events. These data could then be used to ascertain both the natural frequencies of adverse events and whether widespread use of cardiotocography reduces these. Primary outcomes would be intrapartum fetal death, neonatal encephalopathy, cerebral palsy or other significant neurodevelopmental injury, and maternal morbidity. Other outcomes might include long-term physical and psychological outcomes (health across whole of life), health and social care costs, implications for informed decision-making, and analysis of ethical considerations.

17. A randomised controlled trial of intermittent auscultation vs. continuous cardiotocography in otherwise low risk pregnancies complicated by meconium stained liquor

Population: women assessed at the onset of labour as being at low risk of developing intrapartum complications who go on to have meconium stained liquor

Intervention: continuous cardiotocography

Comparator: intermittent auscultation

Primary outcome: neonatal mortality or developmental delay at 2 years

Secondary outcomes: caesarean section, woman's experience of labour; neonatal unit admission, requirement for respiratory ventilation, neonatal encephalopathy

Study design: Randomised controlled trial

Why this is important

Women at low risk of intrapartum complications have lower rates of intervention (e.g. caesarean section) and no difference in neonatal outcomes when the fetus is monitored using intermittent auscultation rather than continuous cardiotocography. The studies that demonstrated this involved conversion from intermittent auscultation to cardiotocography if a fetal heart rate abnormality was detected on intermittent auscultation or if risk factors developed such as meconium stained liquor. However, it may be that in the presence of meconium-stained liquor with no other concerns intermittent auscultation would have been as effective from the fetal point of view but with the benefit of a reduced risk of intervention. A randomised controlled trial with sufficient power to consider long term neonatal outcomes is required to determine whether intermittent auscultation could be used to reduce intervention in labour whilst maintaining the safety for the fetus where there is meconium-stained liquor.

10.2. Fetal heart rate monitoring for meconium-stained liquor

10.2.1. Review question

What is the effectiveness of continuous electronic fetal monitoring compared with intermittent auscultation when there is meconium-stained liquor?

For further details on the evidence review protocol, please see appendix E.

10.2.2. Description of included studies

One study was included in this review (Alfirevic et al., 2008). The included study is a systematic review of randomised control trials with 12 component trials from a variety of countries. Two of these trials were considered for this review.

All included trials within the systematic review evaluated the effectiveness of continuous electronic fetal monitoring (EFM) using cardiotocography (CTG) compared with intermittent auscultation of the fetal heart rate. Ten of the included studies within the systematic review had a small proportion of women with meconium stained liquor included but no subgroup analysis performed for that group, and so therefore could not be used to contribute to this review. The 2 remaining studies included a higher percentage of women with meconium stained liquor and are reported for this review. The studies were conducted in Pakistan and Melbourne. All women in the trial in Pakistan had meconium-stained liquor, but this was true for only 40% women in the Melbourne trial. Both studies were carried out more than 23 years ago and have substantial limitations. The effectiveness of continuous CTG monitoring compared with intermittent auscultation was not reported in the original intrapartum care guideline.

10.2.3. Evidence profile

The effectiveness of continuous CTG compared with intermittent auscultation when there is meconium-stained liquor is reported here in 1 GRADE profile.

Table 62. Summary GRADE profile for comparison of continuous CTG with intermittent auscultation.

Table 62

Summary GRADE profile for comparison of continuous CTG with intermittent auscultation.

10.2.4. Evidence statements

Evidence from 2 studies (n=550) showed that women with meconium-stained liquor who received continuous CTG during labour were less likely to have a spontaneous vaginal birth than those who received intermittent auscultation, with this difference being explained by a higher caesarean rate in the continuous CTG group. In terms of neonatal outcomes, there were no significant differences observed between the 2 groups in perinatal mortality (n=550) and neonatal seizure rate (n=350), but the rate of NICU admission (n=350) was higher in the intermittent auscultation group when compared with the continuous CTG group.

10.2.5. Health economics profile

No published economic evaluations were identified for this question.

10.2.6. Evidence to recommendations

See section 10.1.6.

10.3. Interpretation of an electronic fetal heart rate trace

10.3.1. Review question

What are the appropriate definitions and interpretation of the features of an electronic fetal heart rate (FHR) trace?

For further details on the evidence review protocol, please see appendix E.

10.3.2. Introduction

Babies in the uterus derive oxygen from the mother via the placenta and umbilical cord. During contractions of the uterus in labour this oxygen exchange can be intermittently interrupted. During normal labour, babies who are well are not adversely affected by this. However, this is not always the case and fetal hypoxia and then acidosis can occur. Fortunately, these are relatively rare events in normal pregnancies. The Birthplace study (Birthplace in England Collaborative Group, 2011), for example, reported that intrapartum stillbirths, early neonatal deaths and cases of neonatal encephalopathy – a proportion of which will have been due to intrapartum fetal hypoxia/acidosis – occurred in less than 4 in 1000 births in women at low risk of intrapartum complications.

Surveillance for fetal hypoxia in labour is undertaken by fetal heart rate monitoring either by intermittent auscultation or by a continuous recording by a cardiotocograph. The aim of using a cardiotocograph is to provide a visual continuous record of fetal heart rate and uterine contractions. There are features that can indicate the baby is well and responding normally to the events of labour (for example slowing of the fetal heart rate during a contraction). There are other features that may indicate a serious emergency (for example development of a persistent bradycardia following cord prolapse or placental abruption).

The four features of the fetal heart rate that are scrutinised in a cardiotocograph are:

  • baseline heart rate
  • baseline variability
  • presence or absence of decelerations
  • presence of accelerations

All of these have been examined in relation to the development of fetal hypoxia-acidosis.

10.3.3. Description of included studies

This review includes 38 studies (Cahill et al., 2013; Maso et al., 2012; Salim et al., 2010; Roy et al., 2008; Larma et al., 2007 ; Giannubilo et al., 2007; Menihan et al., 2006; Sameshima et al., 2005; Williams and Galerneau 2004; Williams and Galerneau 2003; Williams and Galerneau, 2002; Hadar et al., 2001; Low et al., 2001; Sheiner et al., 2001; Dellinger et al., 2000; Heinrich, 1982; Honjo and Yamaguchi, 2001; Low et al., 1999; Berkus et al., 1999; Ozden and Demirci, 1999; Spencer et al., 1997; Nelson et al., 1996; Cardoso et al., 1995; Gaffney et al., 1994; Cibils, 1993; Samueloff et al., 1994; Ellison et al., 1991; Murphy et al., 1991; Gilstrap et al., 1987; Spencer et al., 1986; Gilstrap et al., 1984; Krebs et al., 1982; Cibils, 1980; Low et al., 1980; Powell et al.1979; Cibils, 1978; Low et al., 1977; Cibils, 1975).

Fifteen included studies are from the USA (Cahill et al., 2013; Larma et al., 2007; Menihan et al., 2006; Dellinger et al., 2000; Berkus et al., 1999; Nelson et al., 1996; Gilstrap et al., 1984; Samueloff et al., 1994; Cibils, 1993; Gilstrap et al., 1987; Krebs et al., 1982; Cibils, 1980; Powell et al.1979; Cibils, 1978; Cibils, 1975). Seven studies are from Canada (Williams and Galerneau 2004; Williams and Galerneau 2003; Williams and Galerneau 2002; Low et al., 2001; Low et al., 1999; Low et al., 1980; Low et al. 1977), 3 from the UK (Gaffney et al., 1994; Murphy et al., 1991; Spencer et al. 1986), 3 from Israel (Salim et al., 2010; Hadar et al., 2001; Sheiner et al., 2001), 2 from Italy (Giannubilo et al., 2007; Maso et al., 2012), 2 from Japan (Sameshima et al., 2005; Honjo and Yamaguchi, 2001) and 1 each from India (Roy et al., 2008), Australia (Spencer et al., 1997), Germany (Heinrich, 1982), Turkey (Ozden and Demirci, 1999), Portugal (Cardoso et al., 1995) and Ireland (Ellison et al., 1991).

All included studies are observational studies (either retrospective or prospective cohort studies, case–control studies or consecutive or non-consecutive case series). All included studies evaluated the predictive value of fetal heart rate features for neonatal adverse outcomes including cerebral palsy, seizure, neonatal acidemia, encephalopathy, sudden infant death syndrome and birth asphyxia.

The predictive value and association of tachycardia and bradycardia for neonatal adverse outcomes were assessed in 13 studies (Maso et al., 2012; Salim et al., 2010; Roy et al., 2008; Giannubilo et al., 2007; Williams and Galerneau 2004; Sheiner et al., 2001; Honjo and Yamaguchi, 2001; Berkus et al., 1999; Ozden and Demirci, 1999; Nelson et al., 1996; Gilstrap et al., 1984; Ellison et al., 1991; Gilstrap et al., 1987).

The relation between FHR baseline variability and neonatal encephalopathy, sudden infant death and/or metabolic acidosis was evaluated in 9 studies (Roy et al., 2008; Larma et al., 2007; Menihan et al., 2006; Sheiner et al., 2001; Berkus et al., 1999; Spencer et al., 1997; Nelson et al., 1996; Samueloff et al., 1994; Ellison et al., 1991).

The predictive value of accelerations and decelerations for neonatal adverse outcomes was assessed in 18 studies (Cahill et al., 2013; Roy et al., 2008; Giannubilo et al., 2007; Sameshima et al., 2005; Williams and Galerneau, 2002; Williams and Galerneau 2004; Williams and Galerneau 2003; Hadar et al., 2001; Sheiner et al., 2001; Berkus et al., 1999; Ozden and Demirci, 1999; Spencer et al., 1997; Nelson et al., 1996; Samueloff et al., 1994; Cibils, 1993; Ellison et al., 1991; Krebs et al., 1982; Powell et al.1979; Low et al. 1977).

The ability of defined FHR classification systems to predict early adverse neonatal outcomes was assessed in 13 studies (Williams and Galerneau, 2004; Hadar et al., 2001; Sheiner et al., 2001; Dellinger et al., 2000; Heinrich, 1982; Low et al., 2001; Low et al., 1999; Ozden and Demirci, 1999; Spencer et al., 1997; Cardoso et al., 1995; Gaffney et al., 1994; Murphy et al., 1991; Gilstrap et al., 1987).

All included studies consisted of predominantly low risk or mixed populations apart from 5 (Cibils, 1993; Cibils, 1980; Low et al., 1980; Cibils, 1978; Cibils, 1975). The findings for the 5 studies with high risk populations are reported in separate GRADE profiles.

10.3.4. Evidence profile

Data is reported in GRADE profiles for the following FHR parameters:

  • fetal heart rate (bradycardia, tachycardia)
  • baseline variability
  • accelerations
  • decelerations
    • early decelerations
    • late decelerations
    • variable decelerations
  • categorisation/classification of FHR patterns and traces.

The grading of evidence from prospective comparative observational studies or prospective consecutive case series started at high quality and was then downgraded if there were any issues identified that would undermine the trustworthiness of the findings. Evidence from retrospective comparative observational studies or retrospective consecutive case series started at moderate quality and was then downgraded if there were any issues. Evidence from non-consecutive case series started at low quality and was then downgraded if there were any issues.

Published classifications of FHR traces are detailed in appendix N.

10.3.5. Predictive accuracy and correlation data

In the following tables, predictive accuracy of CTG trace features are reported for different test findings (such as pH and base deficit) and for different neonatal outcomes (such as encephalopathy). The specific CTG feature and the thresholds used (for example more than 160 beats per minute for tachycardia) are presented in the rows of the GRADE table and the outcomes that they predict are detailed in the ‘definition of outcome’ column. The measures of diagnostic accuracy in each row represent the specific values for that test at the defined threshold in relation to the specified outcome.

10.3.6. Low risk and mixed populations of women

10.3.6.1. Fetal heart rate (bradycardia and tachycardia)

Table 63. Predictive value of bradycardia and tachycardia for adverse neonatal outcomes.

Table 63

Predictive value of bradycardia and tachycardia for adverse neonatal outcomes.

Table 64. Umbilical arterial pH and base excess in babies with intrapartum tachycardia or bradycardia.

Table 64

Umbilical arterial pH and base excess in babies with intrapartum tachycardia or bradycardia.

Table 65. Association between FHR (bradycardia and tachycardia) and umbilical artery blood gas values or adverse neonatal outcomes.

Table 65

Association between FHR (bradycardia and tachycardia) and umbilical artery blood gas values or adverse neonatal outcomes.

Table 66. Baseline fetal heart rate in babies born with umbilical cord blood acidemia compared with those born without academia.

Table 66

Baseline fetal heart rate in babies born with umbilical cord blood acidemia compared with those born without academia.

Table 67. Correlation of marked tachycardia to neonatal convulsions.

Table 67

Correlation of marked tachycardia to neonatal convulsions.

10.3.6.2. Baseline variability

Table 68. Predictive value of fetal heart rate baseline variability for neonatal adverse outcomes.

Table 68

Predictive value of fetal heart rate baseline variability for neonatal adverse outcomes.

Table 69. Association between fetal heart rate variability and neonatal adverse outcomes or umbilical artery blood gas values.

Table 69

Association between fetal heart rate variability and neonatal adverse outcomes or umbilical artery blood gas values.

Table 70. Association between variability (with or without accelerations or decelerations) and umbilical artery blood gas values.

Table 70

Association between variability (with or without accelerations or decelerations) and umbilical artery blood gas values.

10.3.6.3. Accelerations

Table 71. Predictive value of lack of fetal heart rate accelerations for adverse neonatal outcomes.

Table 71

Predictive value of lack of fetal heart rate accelerations for adverse neonatal outcomes.

Table 72. Association of sporadic accelerations and perinatal mortality.

Table 72

Association of sporadic accelerations and perinatal mortality.

10.3.6.4. Early decelerations

Table 73. Correlation of fetal heart rate early decelerations with neonatal convulsions.

Table 73

Correlation of fetal heart rate early decelerations with neonatal convulsions.

10.3.6.5. Late decelerations

Table 74. Predictive value of fetal heart rate late decelerations for adverse neonatal outcomes.

Table 74

Predictive value of fetal heart rate late decelerations for adverse neonatal outcomes.

Table 75. Association between fetal heart rate late decelerations and adverse neonatal outcome.

Table 75

Association between fetal heart rate late decelerations and adverse neonatal outcome.

Table 76. Correlation of fetal heart rate late decelerations with neonatal convulsions.

Table 76

Correlation of fetal heart rate late decelerations with neonatal convulsions.

10.3.6.6. Variable decelerations

Table 77. Predictive value of variable fetal heart rate decelerations for adverse neonatal outcome.

Table 77

Predictive value of variable fetal heart rate decelerations for adverse neonatal outcome.

Table 78. Association between variable fetal heart rate decelerations and adverse neonatal outcome.

Table 78

Association between variable fetal heart rate decelerations and adverse neonatal outcome.

Table 79. Association between variable fetal heart rate decelerations and maternal outcome.

Table 79

Association between variable fetal heart rate decelerations and maternal outcome.

Table 80. Number of fetal heart rate decelerations (>15 bpm/15 seconds) and association with fetal academia.

Table 80

Number of fetal heart rate decelerations (>15 bpm/15 seconds) and association with fetal academia.

Table 81. Correlation of fetal heart rate decelerations and neonatal convulsions.

Table 81

Correlation of fetal heart rate decelerations and neonatal convulsions.

10.3.6.8. Categorisation of fetal heart rate traces

Table 82. Predictive value of published categorisation of fetal heart rate traces for adverse neonatal outcomes.

Table 82

Predictive value of published categorisation of fetal heart rate traces for adverse neonatal outcomes.

Table 83. Predictive value of published categorisations of fetal heart rate traces for mode of birth.

Table 83

Predictive value of published categorisations of fetal heart rate traces for mode of birth.

Table 84. Association between categorisation of fetal heart rate traces and adverse neonatal outcomes.

Table 84

Association between categorisation of fetal heart rate traces and adverse neonatal outcomes.

Table 85. Umbilical cord arterial pH in women with ‘normal’ and ‘abnormal’ fetal heart rate tracing.

Table 85

Umbilical cord arterial pH in women with ‘normal’ and ‘abnormal’ fetal heart rate tracing.

10.3.7. High risk population

10.3.7.1. Accelerations

Table 86. Association between absence of, or decreased, fetal heart rate accelerations and fetal metabolic acidosis.

Table 86

Association between absence of, or decreased, fetal heart rate accelerations and fetal metabolic acidosis.

10.3.7.2. Decelerations

Table 87. Association between no decelerations/early decelerations and adverse neonatal outcomes.

Table 87

Association between no decelerations/early decelerations and adverse neonatal outcomes.

Table 88. Association between no decelerations /variable decelerations and adverse neonatal outcomes.

Table 88

Association between no decelerations /variable decelerations and adverse neonatal outcomes.

Table 89. Association between no decelerations/late decelerations and adverse neonatal outcomes.

Table 89

Association between no decelerations/late decelerations and adverse neonatal outcomes.

Table 90. Association between marked patterns of total decelerations, moderate/marked pattern of late decelerations and fetal asphyxia.

Table 90

Association between marked patterns of total decelerations, moderate/marked pattern of late decelerations and fetal asphyxia.

Table 91. Predictive value of fetal heart rate decelerations for adverse neonatal outcomes in prolonged pregnancy (>42 gestational weeks).

Table 91

Predictive value of fetal heart rate decelerations for adverse neonatal outcomes in prolonged pregnancy (>42 gestational weeks).

10.3.8. Evidence statements

It is intended that the evidence statements are read in conjunction with the evidence profiles, providing a synthesis of the key points from the evidence.

10.3.8.1. Studies with low risk or mixed population

10.3.8.1.1. Neonatal outcomes
Fetal heart rate, bradycardia, tachycardia

Fetal baseline tachycardia or bradycardia has moderately useful to not useful positive and negative likelihood ratios for poor neonatal outcome or neonatal acidosis at birth, with most studies showing a moderate or high specificity and low sensitivity. Evidence from five studies (n=15,500 approximately) showed a varied association between bradycardia and neonatal acidosis at birth and no association with poor neonatal outcome unless the bradycardia is described as moderate or severe in either the first or second stage of labour, in which case there is evidence of some association with poor neonatal outcomes. There is a low degree of association between tachycardia and poor neonatal outcome or neonatal acidosis. The evidence was of moderate to very low quality.

Baseline variability

Evidence from 9 studies (n=6685) showed that reduced baseline variability has moderate or high specificity for poor neonatal outcome and neonatal acidosis. It also has not useful positive and negative likelihood ratios for neonatal acidosis. Other diagnostic parameters are low for both poor neonatal outcome and neonatal acidosis. A non-reactive trace has high sensitivity, not useful positive likelihood ratio and moderately useful negative likelihood ratio for poor neonatal outcome. There is a low or moderate association between reduced variability with other specific features (no accelerations; late decelerations; late or variable decelerations and no accelerations; or bradycardia with recovery) with neonatal acidosis. There is a high degree of association between decreased variability and no recovery from bradycardia and neonatal acidosis. There is no evidence of an association between reduced variability and poor neonatal outcome. There are no studies investigating the association between reduced variability plus other FHR trace features and poor neonatal outcome. The evidence was of moderate to very low quality.

Accelerations

Evidence from 5 studies (n=6314) showed that the diagnostic value of absence of accelerations is low for poor neonatal outcomes (across all diagnostic parameters). Fewer than 3 sporadic accelerations in the first 30 minutes of electronic fetal monitoring may be associated with neonatal mortality in women at high risk of adverse neonatal outcomes. The evidence was of very low quality.

Late decelerations

Findings for the diagnostic value of late decelerations (n=8636) are conflicting for specificity for poor neonatal outcomes (ranging from high to low). Moderately useful positive and negative likelihood ratios are reported in 1 study, but all other diagnostic values are low for both neonatal outcomes and neonatal acidosis. There is a moderate or low degree of association between late decelerations and neonatal acidosis, but a low degree of association with poor neonatal outcome. The evidence was of moderate to very low quality.

Variable decelerations

Evidence from 10 studies (n=7100) showed that variable decelerations have mostly low diagnostic value for neonatal acidosis (3 types of variable deceleration have moderate specificity but all other diagnostic parameters are low). Variable decelerations classified as severe or with a minimum below 70 bpm have varied association with neonatal acidosis (ranging from no association to moderate association) but there is no association with poor neonatal outcome. Variable decelerations not defined as severe or with a minimum of 70 bpm or higher are not associated with neonatal acidosis or poor neonatal outcome. The evidence was of moderate to very low quality.

Categorisation/classification of fetal heart rate traces

Findings from 14 studies (n=4030) for the diagnostic value of a range of different categorisations of fetal heart rate tracings are conflicting (ranging from high to low across all diagnostic parameters). Different classification systems have different strengths, with 10 reporting high specificity for poor neonatal outcome, but only 4 classification systems have high sensitivity and very useful negative likelihood ratios for poor neonatal outcome. Fetal heart rate patterns categorised as ‘stressed’ or ‘distressed’ using the Dellinger classification have high sensitivity and specificity and also a very useful positive and negative likelihood ratios for neonatal acidosis and high specificity for NICU admission. There is a high or moderate degree of association between tracings categorised as abnormal (however defined) and neonatal acidosis and a low association with poor neonatal outcomes. The evidence was of moderate to very low quality.

10.3.8.1.2. Maternal outcomes

Only 3 studies (n=1829) reported maternal outcomes, reporting the association between variable decelerations and classification of FHR traces and mode of birth.

Variable decelerations

Severe variable decelerations have a high degree of association with caesarean section and a moderate degree of association with ventouse birth. Variable decelerations classified as non-significant are not associated with caesarean section and ventouse birth. The evidence was of moderate quality.

Categorisation/classification of fetal heart rate traces

The association between classification of FHR traces and caesarean section varies from low to high, with one classification system having a moderate specificity and negative predictive value. All other diagnostic parameters have a low degree of association with caesarean section. The evidence was of moderate to very low quality.

10.3.8.2. Studies with high risk population

10.3.8.2.1. Neonatal outcomes
Accelerations

Findings from 1 study (n=400) indicated that the absence of, or a decrease in, FHR accelerations was not associated with fetal acidosis at birth. The evidence was of moderate quality.

Late decelerations

Evidence from one study (n=400) showed a moderate to low degree of association between late decelerations in the last hour before to birth and neonatal acidosis. The evidence was of moderate and low quality.

Variable decelerations

Variable decelerations have mostly low diagnostic values for neonatal acidosis. Findings from 1 study (n=400) indicated that variable decelerations in first stage of labour and in the last 20 minutes prior to birth had a high degree of association with neonatal acidosis. The evidence was of moderate and low quality.

10.3.9. Health economic profile

No published economic evaluations were identified for this question.

10.3.10. Evidence to recommendations

10.3.10.1. Relative value placed on the outcomes considered

The guideline development group agreed that the consequences of intrapartum fetal acidosis were the main outcomes for this question. However, the fetal heart rate is only a surrogate for fetal oxygenation and potential associated acidosis. Furthermore, other factors can influence the fetal heart rate (for example maternal temperature). Therefore the group felt it was important to assess how effective CTG is at identifying babies with fetal hypoxia that may lead to acidosis, both in terms of identifying true positives and ruling out false negatives.

10.3.10.2. Consideration of clinical benefits and harms

There are two types of hypoxia in labour – acute and chronic.

Acute hypoxia develops because there is a sudden, almost total, interruption of the oxygenation of the baby. This can be caused by maternal collapse, complete placental abruption, uterine rupture, cord prolapse or complete cord compression. Acute profound hypoxia can occasionally occur as an end stage event following chronic compromise. These are sudden events and require immediate action if prolonged severe acidosis leading to irreparable fetal injury is to be avoided.

Chronic partial hypoxia leading to acidosis develops over a period of hours rather than minutes. Most babies benefit from the normal intermittent relative hypoxia of labour associated with uterine contractions. However, chronic hypoxia followed by acidosis develops in some babies, for example as a result of long labours, or where there is repeated cord compression with contractions, or where there are excessive contractions (either spontaneous or stimulated). In these cases, a more gradual change occurs in the characteristics of fetal heart rate.

CTG only records 2 parameters: the fetal heart rate and contractions. Continuous monitoring of both allows a number of features to be considered. In addition, CTG produces an automated continuous record that can show trends over a period of time. These are the purported advantages CTG has over intermittent auscultation of the fetal heart rate. Intermittent auscultation is used to record the fetal heart rate over a period of 1 minute immediately after a contraction once every 15 minutes during the first stage of labour, and after every contraction in the second stage. It can be used to detect decelerations that occur in that minute but it does not identify decelerations at other times or baseline variability. For this reason, CTG is used when there are risk factors for fetal hypoxia, including abnormalities detected with intermittent auscultation.

Disadvantages of CTG use include the increased likelihood that women may be left alone, mobility may be reduced and women may be more frightened as they hear changes in the fetal heart rate. Clinical staff may focus on the recording rather than the woman. This may translate into a lack of support for the woman. Staff may also derive a sense of false reassurance and fail to act promptly in the face of abnormality, or over-react in the face of normal physiological fetal heart rate changes which may in turn lead to increase in the rate of interventions. CTG is sometimes incorrectly used in place of continuous supportive one-to-one care.

CTG is used in current practice to monitor the fetal heart rate when there is a concern that fetal hypoxia leading to acidosis may develop, although there is no high quality evidence about the size of the risks and benefits derived from its use. There are no other forms of alternative monitoring available that could replace CTG, although there are adjuncts to CTG that have either been considered or are being assessed (and are reviewed elsewhere in this guideline).

It is important to remember that a cardiotocograph is a screening tool, not a diagnostic test or a treatment. It is common to see abnormalities in cardiotocographs in labour and interventions based on such abnormalities occur in 10–20% of monitored labours. In contrast, severe perinatal asphyxia (causing death or severe neurological impairment) is very rare (see 10.3.2 Introduction). It is difficult to identify what proportion of perinatal asphyxia is ‘avoidable’. The incidence of avoidable death or brain damage that is caused, or exacerbated by, aspects of labour and birth in higher risk labours is not known, neither is the number of interventions (operative births) required to avoid one poor outcome, though it could be high. Nevertheless, the guideline develoment group argued that, because the incidence of avoidable death or brain damage is higher in higher risk labours than in the whole population, CTG should be a more effective screening test than intermittent auscultation in such labours for 2 reasons: first, it records the fetal heart rate continuously rather than intermittently; and second, it provides more information about the fetal heart rate than is possible to determine with intermittent auscultation. However, there is also the argument that CTG can cause more harm in terms of unnecessary intervention due to the high false positive rate without the purported benefit.

CTG has a good negative likelihood ratio; that is, when it is normal there is a very low chance of hypoxia (and therefore acidosis). However, evidence from this review showed that the use of CTG is only moderately useful at best in predicting poor fetal/neonatal outcomes, with the majority of studies showing it to be not useful (not useful positive likelihood ratios). In addition, there are randomised studies (see section 10.1), albeit underpowered, which fail to show that the use of CTG in practice improves neonatal outcomes in a clinically significant way. There is some evidence supporting the usefulness of these features of CTG in predicting neonatal outcome or the surrogate measure of low umbilical cord blood pH:

  • prolonged or severe bradycardia – some evidence of association with poor outcomes/low pH
  • normal variability more than 5 bpm – associated with absence of poor outcomes/low pH
  • decreased variability (NICHD) – some evidence of association with low pH
  • decreased variability (NICHD) with no accelerations – associated with low pH
  • decreased variability (NICHD) – associated with variable or late decelerations and no accelerations - some evidence of association with low pH
  • the presence of accelerations – some evidence of association with good neonatal outcomes
  • recurrent late decelerations with decreased variability – predictive of low pH
  • the presence of late decelerations – associated with low pH and adverse neonatal outcomes
  • the presence of variable decelerations – associated with low pH.

The guidelines development group felt that current practice assumes CTG has greater accuracy than the evidence demonstrates. Individual parameters are probably interpreted with an impression of precision that is not supported by the evidence. It is superficially attractive to suggest each parameter can be defined in terms of its severity and then a classification attached, but the available evidence does not support the assumption that the CTG tracing can be viewed that precisely. The classification presented in the original version of this guideline takes no account of the stage or progress of labour, the presence or absence of meconium or signs of infection, and little account of the contractions or the woman's condition. This can have adverse effects on the care delivered. For example, if a cardiotocograph is only considered abnormal after an arbitrary period of time, a clearly abnormal pattern may be left when in fact intervention is required. The converse example would be the unnecessary intervention that takes place with some CTG patterns in a second stage of labour that is progressing normally. In a rapidly progressing labour, fetal heart rate changes are common. The inclusion in the classification in the original guideline of both ‘suspicious’ and ‘pathological’ has led to the view that there are 2 distinct categories of ‘abnormal’. By definition, a ‘suspicious’ cardiotocograph was intended to be one that required consideration of risk factors and whether any change in management might avoid a worsening change in the future rather than indicating the baby is at risk of compromise at that time. It is for these reasons that the group felt that the classification should be less complex and less rigid.

10.3.10.3. Consideration of health benefits and resource uses

As this question looked at the diagnostic accuracy of different features of fetal heart rate traces, there were no resource use issues to consider.

10.3.10.4. Quality of evidence

The quality of the evidence reviewed varied from moderate to very low. The guideline development group noted several factors that limit the usefulness of the research findings.

First, the outcomes of importance are rare so that a large numbers of cases would be needed to show a difference, if one were to exist, especially in terms of long term neurodevelopment.

Second, there is likely to be a ‘treatment effect’. Because of prior knowledge and experience, many clinicians would feel it inappropriate to not act in the presence of a significant CTG ‘abnormality’ if it is associated with a poor outcome. Thus, if a case of adverse outcome is avoided by intervention before the damage occurs, it might be thought to be a ‘false positive’ when this is not the case. However, this belief has led to the situation where a technology with limitations as a screening test is being widely used without good evidence of benefit.

Third, the fetal heart rate is only a surrogate for fetal hypoxia and arguably not a very good one. Fetal heart rate is influenced by other factors. In an analogous intensive care setting after birth, no one would rely exclusively on a woman's pulse to assess her condition.

Fourth, this guideline recommends the use of CTG only in high risk labours (see section 10.1). However, the majority of the studies reported in this review were in low or moderate risk populations only.

Finally, the cardiotocograph is analysed clinically taking into account multiple factors. It is not just the fetal heart rate which is considered but the underlying risk factors and any other relevant information, such as the progress of labour and/or maternal complications. This means that the performance of individual parameters may not reflect the risks and benefits of using CTG in a clinical setting. Complex tasks of ‘pattern recognition’ together with clinical evaluation may not be captured in simple algorithms and not reflected in the research reviewed.

At present, the evidence base for the use of CTG by itself to monitor high risk labours is not strong. The guideline development group drafted a number of research recommendations in order to address the key gaps in the evidence, including the lack of data regarding the natural frequencies of rare, potentially preventable, serious intrapartum adverse events. Furthermore, the group noted that there are no randomised trials in higher risk women to measure the advantages and harms of CTG monitoring in terms of long term child health and included a research recommendation to look at the use of CTG in labours where there is meconium. The rationale for its use in such labours has been based on both the association of certain abnormal CTG features with adverse fetal/neonatal outcomes and the theoretical reasoning that it provides more information than is available from intermittent auscultation. In addition, no better alternative is available to clinicians.

10.3.10.5. Other considerations

The guideline development group was aware that the reliability of interpretation of CTG recordings, both between different users and when carried out by the same person, has been shown to be variable (see section 10.9). This suggests that there will be differences between healthcare professionals regarding interpretation of cardiotocographs, including baseline variability and categorisation of decelerations. This means care should be taken when interpreting cardiotocographs so that appropriate action is taken where there are signs that cause concern, and so that unnecessary actions and interventions are avoided.

The group noted that medico-legal practice affects custom and practice in clinical care and it would be very difficult to defend a case of intrapartum fetal hypoxia leading to acidosis if a CTG had not been used in the management of a high risk labour. However, the group noted that part of the reason that CTG has become a key point of medical-legal practice is precisely because it has previously been recommended as part of good practice. The group agreed that best medico-legal practice should follow good evidence-based medical practice and so did not feel that it was appropriate to make a recommendation based on current medico-legal practice.

Although the guideline development group considered it would be more appropriate to establish principles of interpretation, they appreciated that they would have to produce practical guidance to influence clinical practice. In developing the recommendations for definition, interpretation and management of women being monitored with a cardiotocograph, the GDG relied on the evidence as far as they could, but they also had to use a consensus process because of the wide variation in the definitions used in the studies reviewed. They noted that this combination of evidence and opinion was a feature of all cardiotocography scoring systems.

Accelerations

There was moderately good evidence that the presence of fetal heart rate accelerations were associated with no adverse fetal/neonatal outcomes.

There was conflicting evidence about whether the absence of fetal heart rate accelerations was associated with or predicted adverse outcomes

Baseline heart rate

There were significant associations with adverse fetal/neonatal outcomes with values above 160 bpm. There was also some evidence that this threshold predicted adverse outcomes. There was some limited evidence that values above 180 bpm were predictive of adverse outcomes. Therefore the group recommended that the upper limit of the normal baseline heart rate should be 160 bpm.

Empirically the group felt that if fetal hypoxia/acidosis was associated with a fetal tachycardia, that risk would be greater at values above 180 bpm than values between 161 bpm and180 bpm, though there was no evidence to confirm that. The group therefore distinguished 2 categories of fetal tachycardia: 161–180 bpm and more than 180 bpm.

There was some limited evidence that a baseline of less than 110 bpm was associated with adverse fetal/neonatal outcomes. However, the evidence was much stronger when the value was less than 100 bpm, with most of the studies in this category actually looking at values below 90 bpm. The group therefore recommended the lower limit of the baseline should be less than 100 bpm. In addition, from their clinical experience, the group members were aware that a stable baseline of 90–99 bpm, with normal variability and no decelerations, may be a normal feature in a few pregnancies, especially those that were past 40 weeks.

Baseline variability

There was good evidence that baseline fetal heart rate variability of 5 bpm or more for up to 30 or 40 minutes was both associated with and predicted good fetal/neonatal outcomes.

Variability of less than 5 bpm lasting for 90 minutes or more was both associated with and predicted adverse outcomes, so the guideline development group agreed that a variability of less than 5 bpm lasting for 90 minutes or more should be considered to be abnormal.

However, there was less evidence for the significance of a baseline variability of less than 5 bpm lasting between 30 or 40 minutes and up to 90 minutes. Nevertheless, the group felt this could not be regarded as normal and decided to call it ‘non-reassuring’. Furthermore, if it was an indicator of fetal hypoxia/acidosis, then it was reasonable to consider that the hypoxic risk was less than if the poor variability lasted for over 90 minutes. In addition, although most of the studies looked at 40 minutes as the upper time limit of ‘normal’ for the duration of low variability, from a practical point of view, and to align with the time limits set for decelerations (see below) and some of the published studies, the group decided to set this at 30 minutes.

There was no evidence relating to the upper limit of baseline variability, so the group was unable to make a recommendation about this.

There was evidence that mild (‘pseudo’) sinusoidal patterns (oscillations of 5–15 bpm) were not associated with adverse outcomes, but there were no data on other sinusoidal patterns and fetal/neonatal outcome. Therefore the group could only make a recommendation about the mild type.

Decelerations

There was good evidence that variable decelerations were both associated with and predicted adverse outcomes. There was no evidence relating the characteristics of the variable decelerations to the outcomes. Nevertheless, the guideline development group argued that, given variable decelerations were associated with a risk of fetal hypoxia/acidosis, then the risk was greater when the decelerations were deeper, the time to recovery was greater and they were present for longer. By consensus, therefore, the guideline development group set 2 thresholds to distinguish severe variable decelerations from the less severe; namely 60 bpm for the depth and 60 seconds for the duration. In addition, the group felt that there should be a time limit for the duration of variable decelerations that would prompt intervention. For the ‘severe’ variable decelerations (decelerations of more than 60 bpm lasting for more than 60 seconds) they felt this should be 30 minutes. For the ‘mild’ variable (decelerations of less than 60 bpm lasting for less than 60 seconds) they felt this should be 90 minutes.

There was good evidence than late decelerations were both associated with and predicted adverse fetal/neonatal outcomes. Again, the group reasoned that the longer the duration of late decelerations, the greater the risk of fetal hypoxia/acidosis, although there was no evidence to confirm this view. Therefore the group empirically set this threshold at 30 minutes.

Although there was very little evidence of the relationship between the number of decelerations to outcome, the group was aware that in practice many interventions occur unnecessarily early; perhaps after only 2 or 3 decelerations. They felt that it was important that decelerations should only be regarded as significant if they occurred with the majority of the contractions. They therefore made a recommendation that decelerations should occur with over 50% of contractions.

The group also noted that a prolonged late deceleration would only be distinguishable from a bradycardia if it recovered. In practice, irrespective of the terminology, a persistent fall in the fetal heart rate would inevitably be associated with fetal hypoxia and acidosis. The group arbitrarily chose 3 minutes as the interval when action should be taken on the basis of other evidence that fetuses could possibly withstand up to 10 minutes of absolute hypoxia without sustaining irreversible neurodevelopmental injury.

Only 1 small study examined atypical decelerations and it failed to show any relationship with adverse outcomes, so the group felt that describing variable decelerations as ‘typical’ and ‘atypical’ was not of value in clinical practice.

Patterns/combinations

There was good evidence that ‘scoring’ systems that took account of more than 1 fetal heart rate feature demonstrated a relationship between an abnormal ‘score’ and adverse fetal/neonatal outcomes, and such systems also predicted adverse outcomes. The guideline development group therefore felt it was important to consider all features of the fetal heart rate when using it to predict fetal health.

10.3.10.6. Key conclusions

The best available evidence to guide interpretation of CTG is limited because:

  • The adverse outcomes are rare, especially in low or moderate risk populations.
  • One principle of use of CTG in practice is for it to be used for monitoring fetuses in high risk pregnancies. However, only a minority of the studies reviewed were of high risk pregnancies. Only late and variable decelerations and accelerations were studied in high risk populations. Baseline, baseline variability and combinations were studied in pregnancies that were low or moderate risk only.
  • There is a ‘treatment paradox’ that intervention will have occurred before the clinically significant adverse outcome takes place – that was the aim of the surveillance. This might be offset, however, by the assertion that without proper testing, the good outcomes associated with an intervention might be wrongly attributed to it and the harm it is causing may go unnoticed.
  • The fetal heart rate is not a good surrogate for hypoxia and acidosis – it can be affected by a number of other factors and may be unaffected with some types of hypoxia.
  • Looking at CTG in isolation is too simplistic and does not take account of the whole clinical picture.

Despite these serious limitations, the guideline development group felt that, on balance, the potential benefits of continuous CTG probably outweigh the risks and the use of CTG in high risk labours should be recommended in the absence of a better alternative.

The guideline development group reasoned that when making the recommendations for the interpretation of CTGs:

  • There are certain pregnancies where there is an increased risk of intrapartum fetal acidosis (‘high risk’ or ‘at risk’ labours) (see section 3.4).
  • The fetal heart rate is the only parameter by which the fetal condition can be continuously assessed and monitored. The role of the fetal ECG was reviewed as part of this guideline update and was not recommended for use in practice (see section 10.8).
  • There is some evidence that the likelihood of adverse outcome from intrapartum fetal acidosis is greater with certain abnormal features of CTG, although the risk of false positives is high with many features.
  • Importantly, the absence of certain abnormal features on a cardiotocograph is very reassuring that fetal acidosis is absent.
  • Given that abnormalities of fetal heart rate are not only due to fetal hypoxia, various conservative actions are recommended in the first instance which will ameliorate some of the non-hypoxic and hypoxic factors. This is discussed in section 11.7 (Intrauterine Resuscitation).
  • Fetal blood sampling is the only single assessment which directly assesses whether the observed fetal heart rate abnormality is due to hypoxia severe enough to cause acidosis. This test is discussed in section 10.5, as is the value of fetal scalp stimulation as an adjunctive test of fetal health in labour (section 10.4). Although the guideline development group could not recommend its routine use in practice, they did feel that the absence of a fetal heart rate acceleration during a fetal blood sampling procedure would be a feature of concern.

10.3.11. Recommendations

106.

Use tables 92 and 93 to define and interpret cardiotocograph traces and to guide the management of labour for women who are having continuous cardiotocography. These tables include and summarise individual recommendations about fetal monitoring (106 to 130), fetal scalp stimulation (134 and 135), fetal blood sampling (136 to 149) and intrauterine resuscitation (132 to 134 and 185) in this guideline. [new 2014]

Table 92. Description of cardiotocograph trace features.

Table 92

Description of cardiotocograph trace features. Overall care Do not make any decision about a woman's care in labour on the basis of cardiotocography (CTG) findings alone. Take into account any antenatal and intrapartum risk factors, the current wellbeing (more...)

Table 93. Management based on interpretation of cardiotocograph traces.

Table 93

Management based on interpretation of cardiotocograph traces.

107.

If continuous cardiotocography is needed:

  • explain to the woman that it will restrict her mobility, particularly if conventional monitoring is used
  • encourage and help the woman to be as mobile as possible and to change position as often as she wishes
  • remain with the woman in order to continue providing one-to-one support
  • monitor the condition of the woman and the baby, and take prompt action if required
  • ensure that the focus of care remains on the woman rather than the cardiotocograph trace
  • ensure that the cardiotocograph trace is of high quality, and think about other options if this is not the case
  • bear in mind it is not possible to categorise or interpret every cardiotocograph trace: senior obstetric input is important in these cases.[new 2014]
108.

Do not make any decision about a woman's care in labour on the basis of cardiotocography findings alone. [new 2014]

109.

Any decision about changes to a woman's care in labour when she is on a cardiotocograph monitor should also take into account the following:

  • the woman's report of how she is feeling
  • the woman's report of the baby's movements
  • assessment of the woman's wellbeing and behaviour
  • the woman's temperature, pulse and blood pressure
  • whether there is meconium or blood in the amniotic fluid
  • any signs of vaginal bleeding
  • any medication the woman is taking
  • the frequency of contractions
  • the stage and progress of labour
  • the woman's parity
  • the results of fetal blood sampling if undertaken (see recommendations 136 to 149)
  • the fetal response to scalp stimulation if performed (see recommendations 134 and 135). [new 2014]
110.

When reviewing the cardiotocograph trace, assess and document all 4 features (baseline fetal heart rate, baseline variability, presence or absence of decelerations, and presence of accelerations). [new 2014]

111.

Supplement ongoing care with a documented systematic assessment of the condition of the woman and unborn baby (including any cardiotocography findings) every hour. If there are concerns about cardiotocography findings, undertake this assessment more frequently. [new 2014]

112.

Be aware that if the cardiotocography parameters of baseline fetal heart rate and baseline variability are normal, the risk of fetal acidosis is low. [new 2014]

Baseline fetal heart rate

113.

Take the following into account when assessing baseline fetal heart rate:

  • this will usually be between 110 and 160 beats/minute
  • a baseline fetal heart rate between 100 and 109 beats/minute (having confirmed that this is not the maternal heart rate) with normal baseline variability and no variable or late decelerations is normal and should not prompt further action
  • a stable baseline fetal heart rate between 90 and 99 beats/minute with normal baseline variability (having confirmed that this is not the maternal heart rate) may be a normal variation; obtain a senior obstetric opinion if uncertain. [new 2014]
114.

If the baseline fetal heart rate is between 161 and 180 beats/minute with no other non-reassuring or abnormal features on the cardiotocograph:

  • think about possible underlying causes (such as infection) and appropriate investigation
  • check the woman's temperature and pulse; if either are raised, offer fluids and paracetomol
  • start one or more conservative measures (see recommendation 132). [new 2014]
115.

If the baseline fetal heart rate is between 161 and 180 beats/minute with no other non-reassuring or abnormal features on the cardiotocograph and the woman's temperature and pulse are normal, continue cardiotocography and normal care, since the risk of fetal acidosis is low. [new 2014]

116.

If the baseline fetal heart rate is between 100 and 109 beats/minute or above 160 beats/minute and there is 1 other non-reassuring feature on the cardiotocograph, start conservative measures (see recommendation 132) to improve fetal wellbeing. [new 2014]

117.

If the baseline fetal heart rate is above 180 beats/minute with no other non-reassuring or abnormal features on the cardiotocograph:

  • think about possible underlying causes (such as infection) and appropriate investigation
  • check the woman's temperature and pulse; if either are raised, offer fluids and paracetamol
  • start one or more conservative measures (see recommendation 132).
  • offer fetal blood sampling to measure lactate or pH (see recommendation 136 to 149) if the rate stays above 180 beats/minute despite conservative measures [new 2014]
118.

If there is a bradycardia or a single prolonged deceleration with the fetal heart rate below 100 beats/minute for 3 minutes or more:

  • start conservative measures (see recommendation 132)
  • urgently seek obstetric help
  • make preparations for urgent birth
  • expedite the birth (see recommendations 220 to 223) if the bradycardia persists for 9 minutes.

If the fetal heart rate recovers at any time up to 9 minutes, reassess any decision to expedite the birth, in discussion with the woman. [new 2014]

Baseline variability

119.

Take the following into account when assessing fetal heart rate baseline variability:

  • baseline variability will usually be 5 beats/minute or more
  • intermittent periods of reduced baseline variability are normal, especially during periods of quiescence (‘sleep’)
  • mild or minor pseudo-sinusoidal patterns (oscillations of amplitude 5-15 beats/minute) are of no significance. [new 2014]
120.

If there is reduced baseline variability of less than 5 beats/minute with a normal baseline fetal heart rate and no variable or late decelerations:

  • start conservative measures (see recommendation 132) if this persists for over 30 minutes and
  • offer fetal blood sampling to measure lactate or pH (see recommendations 136 to 149) if it persists for over 90 minutes. [new 2014]
121.

If there is reduced baseline variability of less than 5 beats/minute for over 30 minutes together with 1 or more of tachycardia (baseline fetal heart rate above 160 beats/minute), a baseline fetal heart rate below 100 beats/minute or variable or late decelerations:

  • start conservative measures (see recommendation 132) and
  • offer fetal blood sampling to measure lactate or pH (see recommendations 136 to 149). [new 2014]

Decelerations

122.

When describing decelerations in fetal heart rate, specify:

  • the depth and duration of the individual decelerations
  • their timing in relation to the peaks of the contractions
  • whether or not the fetal heart rate returns to baseline
  • how long they have been present for
  • whether they occur with over 50% of contractions. [new 2014]
123.

Describe decelerations as ‘early’, ‘variable’ or ‘late’. Do not use the terms ‘typical’ and ‘atypical’ because they can cause confusion. [new 2014]

124.

Take the following into account when assessing decelerations in fetal heart rate:

  • early decelerations are uncommon, benign and usually associated with head compression
  • early decelerations with no non-reassuring or abnormal features on the cardiotocograph trace should not prompt further action. [new 2014]
125.

If variable decelerations are observed that begin with the onset of a contraction:

  • be aware that these are very common, can be a normal feature in an otherwise uncomplicated labour and birth, and are usually a result of cord compression
  • think about asking the woman to change position or mobilise. [new 2014]
126.

Start conservative measures (see recommendation 132) if variable decelerations are observed with a normal baseline fetal heart rate and normal baseline variability that are:

  • dropping from baseline by 60 beats/minute or less and taking 60 seconds or less to recover
  • present for over 90 minutes
  • occurring with over 50% of contractions. [new 2014]
127.

Start conservative measures (see recommendation 132) if variable decelerations are observed with a normal baseline fetal heart rate and normal baseline variability that are:

  • dropping from baseline by more than 60 beats/minute or taking over 60 seconds to recover,
  • present for up to 30 minutes
  • occurring with over 50% of contractions. [new 2014]
128.

Offer fetal blood sampling to measure lactate or pH (see recommendations 136 to 149) if non-reassuring variable decelerations (see recommendation 125 and 126) are:

  • still observed 30 minutes after starting conservative measures or
  • accompanied by tachycardia (baseline fetal heart rate above 160 beats/minute) and/or reduced baseline variability (less than 5 beats/minute). [new 2014]
129.

If late decelerations (decelerations that start after a contraction and often have a slow return to baseline) are observed:

  • start conservative measures (see recommendation 132) if the late decelerations occur with over 50% of contractions
  • offer fetal blood sampling to measure lactate or pH (see recommendations 136 to 149) and/or expedite the birth (see recommendations 220 to 223) if the late decelerations persist for over 30 minutes and occur with over 50% of contractions
  • take action sooner if the late decelerations are accompanied by an abnormal baseline fetal heart rate and/or reduced baseline variability. [new 2014]
130.

Take into account that the longer, the later and the deeper the individual decelerations, the more likely the presence of fetal acidosis (particularly if the decelerations are accompanied by tachycardia and/or reduced baseline variability), and take action sooner than 30 minutes if there is concern about fetal wellbeing. [new 2014]

Accelerations

131.

Take the following into account when assessing accelerations in fetal heart rate:

  • the presence of fetal heart rate accelerations is generally a sign that the baby is healthy
  • the absence of accelerations in an otherwise normal cardiotocograph trace does not indicate acidosis. [new 2014]

Conservative measures

132.

If there are any concerns about the baby's wellbeing, think about the possible underlying causes and start one or more of the following conservative measures based on an assessment of the most likely cause(s):

  • encourage the woman to mobilise or adopt a left-lateral position, and in particular to avoid being supine
  • offer oral or intravenous fluids
  • offer paracetamol if the woman has a raised temperature
  • reduce contraction frequency by:
    • stopping oxytocin if it is being used (the consultant obstetrician should decide whether and when to restart oxytocin) and/or
    • offering a tocolytic drug (a suggested regimen is subcutaneous terbutaline 0.25 mg). [new 2014]
133.

Inform the coordinating midwife and an obstetrician whenever conservative measures are implemented. [new 2014]

134.

Do not use maternal facial oxygen therapy for intrauterine fetal resuscitation, because it may harm the baby (but it can be used where it is administered for maternal indications such as hypoxia or as part of preoxygenation before a potential anaesthetic). [new 2014]

10.4. Predictive value of fetal scalp stimulation

Does the use of fetal stimulation as an adjunct to electronic fetal monitoring improve the predictive value of monitoring and clinical outcomes when compared with:

  • electronic fetal monitoring alone
  • electronic fetal monitoring plus ECG?

10.4.1. Description of included studies

Nineteen studies are included in this review (Arulkumaran et al., 1987; Clark et al., 1982, Clark et al., 1984; Elimian et al., 1997; Lazebnik et al., 1992; Spencer 1991; Trochez et al., 2005; Anyaegbunam et al., 1994; Bartelsmeyer et al., 1995; Chauhan et al., 1999; Ingemarsson and Arulkumaran 1989; Irion et al., 1996; Lin et al., 2001; Polzin et al., 1988; Sarno et al., 1990; Smith et al., 1986; Tannirandorn et al., 1993; Edersheim et al., 1987; Umstad et al., 1992). One of the included studies was a randomised controlled trial (Anyaegbunam et al., 1994), 2 of the studies were prospective comparative observational studies (Smith et al., 1986; Tannirandorn et al., 1993) and the remaining studies were case series. Six of the case series were consecutive, of which 4 were prospective (Elimian et al., 1997; Irion et al., 1996; Sarno et al., 1990; Umstad et al., 1992), and 2 were retrospective (Spencer 1991; Trochez et al., 2005). Two studies were specifically reported as being non-consecutive case series (Chauhan et al., 1999; Polzin et al., 1988), and in the remaining 8 studies it was unclear.

Seven studies investigated fetal scalp stimulation (Arulkumaran et al., 1987; Clark et al., 1982, Clark et al., 1984; Elimian et al., 1997; Lazebnik et al., 1992; Spencer 1991; Trochez et al., 2005), 10 studied vibroacoustic stimulation (Anyaegbunam et al., 1994; Bartelsmeyer et al., 1995; Chauhan et al., 1999; Ingemarsson and Arulkumaran 1989; Irion et al., 1996; Lin et al., 2001; Polzin et al., 1988; Sarno et al., 1990; Smith et al., 1986; Tannirandorn et al., 1993) and 2 studied vibroacoustic stimulation followed by fetal scalp stimulation (Edersheim et al., 1987; Umstad et al., 1992). In the studies where fetal scalp stimulation was performed, 2 used digital stimulation (Elimian et al., 1997; Trochez et al., 2005), 2 used Allis clamp stimulation (Arulkumaran et al., 1987; Clark et al., 1984) and 3 used scalp puncture as the stimulation (Clark et al., 1982; Lazebnik et al., 1992; Spencer 1991).

Studies reported the predictive value of fetal scalp stimulation or vibroacoustic stimulation for the following:

  • fetal scalp pH less than 7.20
  • fetal scalp pH less than 7.25
  • cord pH less than 7.20
  • caesarean section and Apgar less than 7 at 5 minutes.

All studies defined an acceleration as increase in fetal heart rate over baseline of at least 15 bpm for at least 15 seconds (apart from Lazebnik et al., 1992, which defined it as a net difference in heart rate of more than 15 bpm).

No study reported the time elapsed between fetal stimulation and birth. All studies except 1 (Anyaegbunam et al., 1994) were of women whose unborn babies had a cardiotocograph recording which was interpreted as being indicative of the need for a fetal scalp blood sample to be tested for acidemia.

10.4.2. Evidence profile

Data is reported in GRADE profiles below for the following tests.

  • fetal scalp stimulation
    • fetal scalp blood sampling puncture as stimulus
    • digital massage as stimulus
    • allis clamp as stimulus
  • vibroacoustic stimulation.

The majority of included studies used absence of an acceleration following stimulation as a positive predictive test result in order to calculate predictive values. For those studies that used presence of an acceleration as a positive predictive test result, these were reported in the evidence table and the NCC-WCH technical team then calculated predictive values using no acceleration as a positive predictive test result in order to provide consistency of interpretation across all studies.

Similarly, where fetal blood sample pH was the reference test, the majority of included studies defined a positive test result as acidosis (either pH less than 7.20 or pH less than 7.25). For those studies that used no acidosis (either pH greater than or equal to 7.20 or pH greater than or equal to 7.25) as a positive reference test result, these were reported in the evidence table and the NCC-WCH technical team then converted these to predictive values using acidosis as a positive reference test result.

Evidence from randomised controlled trials, prospective comparative observational studies or prospective consecutive case series started at high quality and was then downgraded if there were any issues identified that would undermine the trustworthiness of the findings. Evidence from retrospective comparative observational studies or retrospective consecutive case series started at moderate quality and was then downgraded if there were any quality-related issues. Evidence from non-consecutive case series started at low quality and was then downgraded if there were any issues.

Table 94. Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following fetal scalp blood sampling puncture as stimulus.

Table 94

Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following fetal scalp blood sampling puncture as stimulus.

Table 95. Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following digital massage as stimulus.

Table 95

Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following digital massage as stimulus.

Table 96. Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following Allis clamp as stimulus.

Table 96

Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following Allis clamp as stimulus.

Table 97. Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following 3 or 5 seconds of vibroacoustic stimulation (VAS).

Table 97

Summary GRADE profile for predictive accuracy of no fetal heart rate acceleration following 3 or 5 seconds of vibroacoustic stimulation (VAS).

10.4.3. Evidence statements

10.4.3.1. Fetal scalp stimulation

10.4.3.1.1. Neonatal outcomes

Evidence from 5 studies (n=537) indicated that the lack of an acceleration in fetal heart rate following fetal scalp stimulation (by fetal blood sampling puncture, digital stimulation or Allis clamp) has varied (low to high) sensitivities for fetal scalp pH of 7.20 or less or umbilical cord pH of 7.20 or less, with more studies showing high sensitivity than moderate or low. Most studies also show a useful negative likelihood ratio. Other diagnostic parameters (specificity and positive likelihood ratio) are low. The evidence was of moderate to very low quality.

The lack of fetal heart rate acceleration following fetal scalp stimulation (by fetal blood sampling puncture) has low to moderate sensitivity and specificity for fetal scalp pH less than 7.25, with 1 study (n=60) showing high specificity. Findings for positive and negative likelihood ratios are conflicting. One study (n=200) showed that a lack of fetal heart rate acceleration has high sensitivity and specificity for fetal scalp pH less than 7.21. It also showed useful positive and negative likelihood ratios. The evidence was of moderate to very low quality.

The lack of fetal heart rate acceleration following fetal scalp stimulation (by fetal blood sampling puncture or digital stimulation) has low to high sensitivity but low specificity for Apgar score less than 7 at 5 minutes (n=50). The positive likelihood ratio is not useful, but 1 study showed a useful negative likelihood ratio. The evidence was of very low quality.

10.4.3.1.2. Maternal outcomes

Evidence from 2 studies (n=272) indicated that the lack of fetal heart rate acceleration following fetal scalp stimulation (by Allis clamp) has high specificity and low sensitivity for caesarean section. Positive and negative likelihood ratios are moderately useful. The evidence was of very low quality.

10.4.3.2. Vibroacoustic stimulation

10.4.3.2.1. Neonatal outcomes

Evidence from 7 studies (n=808) indicated that the lack of a fetal heart rate acceleration following vibroacoustic stimulation (for 3 or 5 seconds) has varied (low to high) sensitivity and specificity for fetal scalp pH of 7.20 or less, with more studies showing high sensitivity than moderate or low, and more studies showing low specificity than moderate or high. The values for negative likelihood ratio are conflicting, but the values for positive likelihood ratios are consistently low. One study (n=271) showed low sensitivity and high specificity for umbilical cord pH less than 7.10 and less than 7.00. Positive likelihood ratios were moderately useful and negative likelihood ratios were not useful. The evidence was of moderate to very low quality.

Evidence from 4 studies (n=477) showed that the lack of a fetal heart rate acceleration following vibroacoustic stimulation (for 3 or 5 seconds) has varied findings for sensitivity and low to moderate specificity for fetal scalp pH less than 7.25. Two out of 4 studies (n=124) showed a useful negative likelihood ratio. The values for positive likelihood ratio ranged from moderate to low. The evidence was of moderate to very low quality.

Evidence from 5 studies (n=834) showed that the lack of fetal heart rate acceleration following vibroacoustic stimulation (for 3 or 5 seconds) has low to high sensitivity and specificity for Apgar score less than 7 at 5 minutes, with more studies showing low and moderate sensitivity and specificity than high sensitivity and specificity. The positive likelihood ratio is not useful, but 1 study showed a useful negative likelihood ratio. The evidence was of moderate to very low quality.

10.4.3.2.2. Maternal outcomes

One study (n=471) found the lack of a fetal heart rate acceleration following vibroacoustic stimulation (for 3 seconds) has high specificity but low sensitivity for caesarean section. The positive and negative likelihood ratios are not useful. The evidence was of low quality.

10.4.4. Health economics profile

No published economic evaluations were identified for this question.

10.4.5. Evidence to recommendations

10.4.5.1. Relative value placed on the outcomes considered

The purpose of fetal stimulation is to prompt a fetal heart rate acceleration (which the majority of studies defined as an increase in fetal heart rate over baseline by 15 bpm for at least 15 seconds). The aim of this review was to determine the predictive value of fetal stimulation (either by using some form of scalp stimulation or by using vibroacoustic stimulation) for neonatal outcomes when used as an adjunctive test to CTG. The guideline development group agreed that it was useful to consider both sensitivity and specificity, and positive and negative likelihood ratios when considering the evidence findings.

The group had hoped that the reported outcomes would include both maternal and neonatal ‘patient important outcomes’, including major morbidities such as neonatal seizures or cerebral palsy. However, the majority of the reported outcomes related to fetal scalp pH values and so the group primarily used these in its decision making.

10.4.5.2. Consideration of clinical benefits and harms

The evidence is varied for the usefulness of fetal stimulation for predicting low pH values. The negative likelihood ratios for fetal stimulation ranged from useful to not useful, with no clear evidence one way or the other. Similarly, there was no consistent finding for sensitivity and specificity. This means that if an acceleration is observed, this may indicate that the fetal pH value is not low (a reassuring finding) but this is not a certain finding. Positive likelihood ratios were more often than not found to be not useful for predicting low pH values. Again the sensitivity and specificity were also varied, ranging from high to low with no consistent pattern of findings. This means that if an acceleration is not observed, this cannot be relied upon as an indicator for the fetal pH value. The group recognised that the act of fetal scalp pH sampling is simultaneously an act of scalp stimulation, and thus even if it is not possible to obtain a result from a scalp sample (for example because not enough blood is taken), if an acceleration is observed, this should still be treated as a potentially reassuring feature to take into account when considering the whole clinical picture.

10.4.5.3. Consideration of health benefits and resource uses

There were no specific resource use issues addressed for this question, because fetal scalp stimulation would be carried out during a vaginal examination or when taking a fetal blood sample and so there are unlikely to be any additional resources required. Given the usefulness of the test in providing potential reassurance about babies that are well, the guideline development group felt confident in recommending the use of the test.

10.4.5.4. Quality of evidence

The evidence was of mixed quality, ranging from moderate to very low (with the majority of studies rated as low or very low). The group was concerned about the poor quality of the evidence available and noted that the results of the different studies varied greatly. Many results had wide or very wide confidence interval (95% CI). This led the guideline development group to exercise caution in the wording of the recommendations.

10.4.5.5. Other considerations

The guideline development group did not feel that the evidence provided a clear indication for the effectiveness of fetal scalp stimulation, nor when fetal scalp stimulation should be used as an adjunct to monitoring. As a result, they did not wish to recommend that scalp stimulation should be used in its own right in certain circumstances. Instead, the group recognised that there are occasions when the baby's scalp will be stimulated anyway – such as when performing a vaginal examination or when taking a fetal blood sample – and that on these occasions the healthcare professional should be alert to accelerations as a potential indication of fetal wellbeing.

10.4.6. Recommendations

135.

If fetal scalp stimulation leads to an acceleration in fetal heart rate, regard this as a reassuring feature. Take this into account when reviewing the whole clinical picture (see recommendation 109). [new 2014]

136.

Use the fetal heart rate response after fetal scalp stimulation during a vaginal examination to elicit information about fetal wellbeing if fetal blood sampling is unsuccessful or contraindicated. [new 2014]

10.5. Fetal blood sampling

10.5.1. Fetal blood sampling as an adjunct to electronic fetal monitoring

10.5.1.1. Review question

Does the use of fetal blood sampling (FBS) as an adjunct to electronic fetal monitoring (EFM) improve outcomes, when compared to:

  • Electronic fetal monitoring alone
  • Electronic fetal monitoring plus electrocardiogram (ECG)

10.5.1.2. Description of included studies

Four studies (Alfirevic et al., 2012; Stein et al., 2006; Noren, et al., 2007; Becker et al., 2011) are included in this review. Two studies (Alfirevic et al., 2012; Stein et al., 2006) evaluated the use of fetal blood sampling as an adjunct to CTG when compared to CTG or intermittent auscultation. Two studies (Noren et al., 2007; Becker, et al., 2011) examined the use of fetal blood sampling as an adjunct to CTG plus ECG.

Of the 2 studies that evaluated the use of fetal blood sampling as an adjunct to CTG compared with continuous CTG or intermittent auscultation, 1 is a systematic review (Alfirevic et al., 2012) with 13 component trials from a variety of locations. Of these 13 included trials, none reported data for fetal blood sampling as an adjunct to CTG compared with CTG alone. Eight of the included trials reported a subgroup analysis for women who had fetal blood sampling as an adjunct to CTG compared with intermittent auscultation. An additional observational study conducted in Germany (Stein et al., 2006) compared the impact of CTG alone versus CTG with additional fetal blood sampling in vaginal births complicated by pathologic fetal heart rate.

Of the two studies that evaluated the use of fetal blood sampling as an adjunct to CTG plus ECG (Noren et al., 2007; Becker et al., 2011), 1 was conducted in Norway and 1 in the Netherlands. Both studies are secondary analyses of sub-groups of data from large multi-centre studies. One study (Becker et al., 2011) used data from the experimental arm of a multicentre randomised trial and evaluated the recommendations for additional fetal blood sampling when using ST analysis of the fetal ECG. The other study (Noren et al., 2007) also used data from a European multi-centre study and assessed the relationship between fetal blood sampling and ST analysis in the presence of acidosis. In this case controlled study, out of 911 participants with fetal blood sampling results, 97 cases were identified: 53 had a cord artery pH less than 7.06 and 44 had a cord artery pH ranging from 7.06 to 7.09, categorised as marked acidosis and moderate acidemia respectively. These cases were analysed with 97 controls with a cord artery pH of 7.20 or more.

10.5.1.3. Evidence profile

The findings for the effect of fetal blood sampling as an adjunct to CTG are reported in 5 GRADE profiles. The following comparisons were considered based on whether fetal blood sampling was used as an adjunct to CTG and compared to CTG or intermittent auscultation alone, or fetal blood sampling used as an adjunct to CTG plus ECG (ST waveform analysis):

10.5.1.3.1. Fetal blood sampling as an adjunct to CTG compared with CTG or intermittent auscultation alone
  • CTG plus fetal blood sampling versus CTG or intermittent auscultation alone in labour.
10.5.1.3.2. Fetal blood sampling as an adjunct to CTG plus ECG
  • Distribution of fetal blood sampling and ECG guideline (ST waveform analysis) indication to intervene. Marked acidosis (cord artery pH<7.06) versus control.
  • Distribution of fetal blood sampling and ST guideline indication to intervene. Moderate acidosis (cord artery pH 7.06 −7.09) versus control.
  • Cases with abnormal CTG and their relation to normal and abnormal fetal blood sampling and ST waveform analysis.
  • Additional fetal blood sampling when using ST analysis of fetal electrocardiogram.
Table 98. Summary GRADE profile for comparison CTG plus fetal blood sampling (FBS) with intermittent auscultation (IA) (Alfirevic et al., 2012) or CTG alone in labour (Stein et al., 2006).

Table 98

Summary GRADE profile for comparison CTG plus fetal blood sampling (FBS) with intermittent auscultation (IA) (Alfirevic et al., 2012) or CTG alone in labour (Stein et al., 2006).

10.5.1.3.3. Fetal blood sampling as an adjunct to CTG plus ECG

The data presented in the following GRADE profiles are taken from papers reporting secondary analyses of sub-groups taken from larger studies in order to investigate the role of fetal blood sampling when used as an adjunct to CTG with ECG analysis. These studies were not designed as intervention studies comparing CTG with ECG analysis plus fetal blood sampling versus CTG with ECG analysis without fetal blood sampling.

The first 3 tables present findings from Noren et al. (2007) which is a case controlled study. Cases are defined as babies born with marked acidosis (cord artery pH less than 7.06; n=53) or moderate acidemia (cord artery pH 7.06 to 7.09; n=44); controls are babies with cord artery pH of 7.20 or more.

Table 99. Summary GRADE profile for distribution of fetal blood sampling (FBS) findings and ST guideline indication to intervene: Marked acidemia: cord artery pH<7.06.

Table 99

Summary GRADE profile for distribution of fetal blood sampling (FBS) findings and ST guideline indication to intervene: Marked acidemia: cord artery pH<7.06.

Table 100. Summary GRADE profile for distribution of fetal blood sampling (FBS) and ST guideline indication to intervene-. Moderate acidemia cord artery pH 7.06 – 7.09.

Table 100

Summary GRADE profile for distribution of fetal blood sampling (FBS) and ST guideline indication to intervene-. Moderate acidemia cord artery pH 7.06 – 7.09.

Table 101. Summary GRADE profile for cases with abnormal or intermediary CTG noted at start of ST analysis recording.

Table 101

Summary GRADE profile for cases with abnormal or intermediary CTG noted at start of ST analysis recording.

The following GRADE table presents data from Becker et al. (2011) which represents a secondary analysis of fetal blood sampling findings within the experimental arm of an ST analysis trial. A comparison is made between findings for fetal blood samples taken according to the ST analysis trial protocol with those taken based on clinical judgement not according to the protocol.

Table 102. Summary GRADE profile for additional fetal blood sampling (FBS) when using ST analysis of fetal electrocardiogram.

Table 102

Summary GRADE profile for additional fetal blood sampling (FBS) when using ST analysis of fetal electrocardiogram.

Some neonatal outcomes are described for the Becker et al. (2011) study. Among women where fetal blood samples were taken according to the trial protocol, 3 out of 123 babies were born with metabolic acidosis (cord artery pH less than 7.05 and base deficit in extracellular fluid more than 12 mmol/l). Fetal blood sample findings for these babies were pH 7.19 (time interval to birth not reported), pH 7.24 (20 minutes before birth) and pH 7.32 (9 hours before birth). Among women where an fetal blood sample was performed outside the trial protocol, 3 out of 101 babies were born with metabolic acidosis (no difference between groups; p=0.81). In all 3 cases, ST events (abnormality of the ST segment of the fetal ECG) were present. Fetal blood sample findings are only reported for 1 of these babies, where multiple samples were taken with recordings of pH 7.38, 7.33, 7.31, 7.28 and 7.28. Time before the final fetal blood sample and birth was 114 minutes (caesarean section following failed ventouse). Umbilical cord artery pH was 6.96 and the baby died of severe asphyxia and encephalopathy.

10.5.1.4. Evidence statements

10.5.1.4.1. Fetal blood sampling as an adjunct to CTG compared with CTG or intermittent auscultation alone

Evidence from 6 studies showed that the rates of caesarean section (n=16,001) and instrumental vaginal birth (n=65,315) were higher in women who received CTG plus fetal blood sampling compared with women who received intermittent auscultation only. The rates of resuscitation (n=49,560), neonatal seizure (n=15,004) and Apgar score less than 7 at 5 minutes (n=49,560) were lower in babies born to women who received electronic fetal monitoring plus fetal blood sampling compared with babies born to women who received intermittent auscultation or electronic fetal monitoring only. The rate of cord blood acidosis (n=50,635) was lower in women who received electronic fetal monitoring plus fetal blood sampling compared with women who received electronic fetal monitoring alone, but there was no difference when compared with women who received intermittent auscultation. No difference was found between the 2 groups in the incidence of cerebral palsy (n=13,252). The evidence was of moderate to very low quality.

10.5.1.4.2. Fetal blood sampling as an adjunct to CTG plus fetal ECG
Distribution of fetal blood sampling findings and ST analysis guideline indication to intervene (marked acidosis: cord artery pH less than 7.06)

Evidence from 1 study (n=106) showed that a higher number of babies with marked cord artery acidosis (pH less than 7.06) had abnormal fetal blood sampling and ST analysis indications to intervene compared with the control group (babies with cord artery pH of 7.20 or more). A lower number of babies with marked acidosis (who were adequately monitored) had no ST analysis indications to intervene compared with the control group. The evidence was of very low quality.

Distribution of fetal blood sampling and ST analysis guideline indication to intervene (moderate acidemia: cord artery pH less than 7.06–7.09)

Evidence from 1 study (n=88) showed that a higher number of babies with moderate cord artery acidemia had abnormal fetal blood sampling or ST analysis indications to intervene compared with the control group (babies with cord artery pH of 7.20 or more). A lower number of babies with cord artery moderate acidemia (who were adequately monitored) had no ST analysis indications to intervene compared with the control group. The evidence was of very low quality.

Cases with abnormal CTG noted at start of fetal ECG recording

Evidence from 1 study (n=61) showed that a lower number of babies with marked acidosis and moderate acidemia had normal fetal blood sampling with normal ST analysis compared with the control group (babies with cord artery pH of 7.20 or more). However, a higher number of babies with marked acidosis and moderate acidemia had abnormal fetal blood sampling with abnormal ST analysis compared with the control group. No differences were found in the number of babies with marked acidosis and moderate acidemia who had normal fetal blood sampling with abnormal ST analysis or abnormal fetal blood sampling with normal ST analysis compared with the control group. The evidence was of moderate to very low quality.

ST analysis of fetal electrocardiogram plus fetal blood sampling

Evidence from 1 study (n=297) showed that the number of women with a fetal blood sample pH of more than 7.25 was lower where fetal blood samples were performed according to the ST analysis trial protocol compared with women where fetal blood sampling was not performed according to the ST analysis trial protocol. However, this difference was not observed for women with a fetal blood sample pH of 7.25 or less. The evidence was of very low quality.

10.5.1.5. Health economics profile

No published economic evaluations were identified for this question.

10.5.1.6. Evidence to recommendations

10.5.1.6.1. Relative value placed on the outcomes considered

For this review, the main maternal outcomes of interest were the rates of caesarean section and instrumental birth. The main neonatal outcome of interest was cerebral palsy. These were felt to be clinically significant, with caesarean section and instrumental birth an important component of the woman's experience of birth.

10.5.1.6.2. Quality of evidence

Although the comparison of interest was fetal blood sampling as an adjunct to CTG compared with CTG alone (or CTG plus ECG), only 1 observational study was identified which investigated this specific comparison, and the quality of its findings was very low for each of the relevant outcomes. The decision was made to also include a large systematic review which compared CTG plus fetal blood sampling with intermittent auscultation, as it was felt that this review might contain relevant information for the guideline development group to consider. The group was aware that a majority of the population in the systematic review consisted of women with a high risk pregnancy. In addition, women with preterm pregnancy and multiple pregnancy were also included. Because of the way the data were reported in the individual studies, it was not possible to perform a sub-group analysis for women with a low risk pregnancy, term pregnancy or singleton pregnancy. Furthermore, of 13 trials included in the systematic review, none reported data for fetal blood sampling as an adjunct to CTG compared with CTG alone, which was really the focus of the review question. Eight of the included trials reported a subgroup analysis for women who had fetal blood sampling as an adjunct to CTG compared with intermittent auscultation. Given these problems, the group did not feel that it was appropriate to consider the findings of the large systematic review when developing its recommendations.

One further case controlled study (Noren et al., 2007) was identified which took the findings from the experimental arm of a randomised trial where women received fetal blood sampling as an adjunct to CTG plus ECG, and compared them with a group of controls. This was not the most appropriate study design, and the group also noted that the numbers of women included in the study were very small, making it difficult to extrapolate from. Again, the group did not feel that it was appropriate to consider the findings from this study when developing its recommendations.

10.5.1.6.3. Consideration of clinical benefits and harms

The 1 observational study which looked at the direct comparison of interest showed that there was a statistically significant reduction in the number of instrumental vaginal births in the group which received fetal blood sampling in addition to CTG compared with the group which did not receive fetal blood sampling. The study also showed a statistically significant reduction in the rate of cord blood acidosis, neonatal resuscitation and 5 minute Apgar score of less than 7.

Although the group recognised that the quality of the evidence for all of these outcomes was very low, they felt that the findings matched their clinical experience. They agreed that fetal blood sampling as an adjunctive test helps clinicians to identify those babies where additional intervention may be required, and thereby reduces the rates of poor neonatal outcomes while at the same time reducing the number of women receiving unnecessary interventions.

10.5.1.6.4. Consideration of health benefits and resource uses

No formal cost effectiveness analysis was performed for this review. However, it was agreed that as fetal blood sampling is not an expensive test and does not require a large amount of additional clinician time, its use is likely to be cost effective, given the gains in quality adjusted life years (QALY) to be made by avoiding poor neonatal outcomes and unnecessary interventions.

10.5.1.7. Recommendations

For all fetal blood sampling recommendations, see section 10.5.4.6.

10.5.2. Time from decision to take a fetal blood sample to result

10.5.2.1. Review question

What is the optimum time from the decision to perform a fetal blood sample to having the blood result?

For further details on the evidence review protocol, please see appendix E.

10.5.2.2. Description of included studies

Two studies are included in this review (Annappa et al., 2008; Tuffnell et al., 2006). Both studies were prospective studies conducted in the UK which documented consecutive attempts at fetal blood sampling.

10.5.2.3. Evidence profile

Table 103. Summary GRADE profile for the time from the decision to perform a fetal blood sample to having the scalp pH result.

Table 103

Summary GRADE profile for the time from the decision to perform a fetal blood sample to having the scalp pH result.

10.5.2.4. Evidence statements

One study (n=74) reported that the median time from the decision to perform a fetal blood sample to obtaining the result was 18 minutes and that in 9% of cases the time interval was longer than 30 minutes. Another study (n=72) reported that the median time from the decision to perform a fetal blood sample to obtaining the result was 17 minutes and that in 5% of cases the time interval was longer than 30 minutes. The evidence was of very low quality.

10.5.2.5. Evidence to recommendations

10.5.2.5.1. Relative value placed on the outcomes considered

The guideline development group felt that the most important outcome was the average time from the decision to perform a fetal blood sample to having the result. They agreed that it was useful to have the supplementary information about the proportion of samples where the time from decision to result was longer than 30 minutes.

10.5.2.5.2. Consideration of clinical benefits and harms

The aim of this review was to identify the average time taken from decision to perform a fetal blood sample to having the result. This was in order that clinicians could take this information into account when deciding whether or not they should perform a fetal blood sample. In instances where a clinician was concerned about a baby's condition, it might be felt that 18 minutes would be too long to wait, and thus the baby's birth ought to be expedited sooner.

The evidence available supported the recommendation made in the previous guideline and so the group agreed that no changes were necessary.

10.5.2.5.3. Consideration of health benefits and resource uses

This question addresses the time from the decision to perform a fetal blood sample to having the result to provide information for clinicians. As this is not a comparison of alternatives, no economic analysis was conducted. The review gives information on timing only, so there are no health benefit or resource implications related to this question.

10.5.2.5.4. Quality of evidence

The quality of evidence for this review was very low as it was derived from case series. However, the guideline development group felt that this was an appropriate study methodology for this question and agreed that it was sufficient to underpin the recommendation.

10.5.2.6. Recommendations

For all fetal blood sampling recommendations, see section 10.5.5.2.

10.5.3. Predictive value of fetal blood sampling

10.5.3.1. Review question

What is the predictive value of the following measures, for maternal and neonatal outcomes:

  • fetal blood pH analysis
  • fetal blood lactate analysis
  • fetal acid-base status
  • fetal base deficit

10.5.3.2. Description of included studies

Nine studies are included in this review (Bakr et al., 2005; Brandt-Niebelschutz and Saling, 1994; East et al., 2011; Hon et al., 1969; Kerenyi et al., 1970; Khazin et al., 1969; Kubli, 1968; Wiberg-Itzel et al., 2008; Young et al., 1980).

One of the included studies is a systematic review which included 2 randomised controlled trials, both from Sweden (East et al., 2011). One of the other included studies is a further report of 1 of the trials included in the systematic review, which was included as an individual paper because additional data were reported (Wiberg-Itzel et al., 2008). One of the included studies is a prospective comparative observational study from Egypt (Bakr et al., 2005). Two of the included studies are retrospective consecutive case series from Germany (Brandt-Niebelschutz and Saling, 1994) and Canada (Young et al., 1980). The remaining 4 included studies are case series from the USA in which it is not clear whether the cases were consecutive (Hon et al., 1969; Kerenyi et al., 1970; Khazin et al., 1969; Kubli, 1968).

The systematic review (East et al., 2011) incorporated trials which randomised women to have either the lactate level or the pH of the fetal blood sample measured. Clinical outcomes for both mother and baby are reported for this comparison. The remaining included studies evaluated the predictive value of fetal blood pH, lactate, base deficit or base excess values for neonatal outcomes. For predictive value data, only studies reporting data for samples taken within 1 hour of birth were included. The time interval between fetal blood sampling and birth was up to 60 minutes in 6 studies (Bakr et al., 2005; Brandt-Niebelschutz and Saling, 1994; Hon et al., 1969; Kerenyi et al., 1970; Wiberg-Itzel et al., 2008; Young et al., 1980) and up to 30 minutes in 2 studies (Khazin et al., 1969; Kubli, 1968).

One study (Wiberg-Itzel et al., 2008) reported excluding women with multiple pregnancies or who were in labour prior to 34 weeks, but in the remaining studies, the inclusion/exclusion criteria and characteristics of the study populations are poorly reported, so it is not possible to judge whether women would have been classified as low risk prior to the onset of labour.

10.5.3.3. Evidence profile

Data is reported in GRADE profiles below for the following tests and outcomes:

  • Comparative clinical outcome data for women randomised to fetal blood lactate or pH testing (table 104).
  • Predictive accuracy and correlation data:
    • Composite neonatal outcomes: predictive value of fetal blood pH at different thresholds (table 105).
    • 5 minute Apgar score: predictive value of fetal blood pH, lactate and base deficit at different thresholds (table 106) and correlation of fetal blood pH and base deficit measurements with Apgar score (table 107).
    • Umbilical arterial pH at birth: predictive value of fetal blood pH, lactate and base deficit at different thresholds (table 108) and correlation of fetal blood pH and base-excess measurements with umbilical arterial measurements (table 109).
Table 104. Summary GRADE profile for lactate compared with pH for fetal blood sampling.

Table 104

Summary GRADE profile for lactate compared with pH for fetal blood sampling.

Table 105. Summary GRADE profile for predictive accuracy of fetal blood sampling for composite neonatal outcomes.

Table 105

Summary GRADE profile for predictive accuracy of fetal blood sampling for composite neonatal outcomes.

Table 106. Summary GRADE profile for predictive accuracy of fetal blood sampling for Apgar score at 5 minutes.

Table 106

Summary GRADE profile for predictive accuracy of fetal blood sampling for Apgar score at 5 minutes.

Table 107. Summary GRADE profile for correlation of fetal blood sampling with high and low Apgar scores at 5 minutes.

Table 107

Summary GRADE profile for correlation of fetal blood sampling with high and low Apgar scores at 5 minutes.

Table 108. Summary GRADE profile for predictive accuracy of fetal blood sampling for arterial pH at birth.

Table 108

Summary GRADE profile for predictive accuracy of fetal blood sampling for arterial pH at birth.

Table 109. Summary GRADE profile for correlation of fetal scalp blood sample values with umbilical artery values at time of birth.

Table 109

Summary GRADE profile for correlation of fetal scalp blood sample values with umbilical artery values at time of birth.

Evidence from randomised controlled trials, prospective comparative observational studies or prospective consecutive case series started at high quality and was then downgraded if there were any issues identified that would undermine the trustworthiness of the findings. Evidence from retrospective comparative observational studies or retrospective consecutive case series started at moderate quality and was then downgraded if there were any issues. Evidence from non-consecutive case series started at low quality and was then downgraded if there were any issues.

10.5.3.3.1. Comparative clinical outcome data
10.5.3.3.2. Predictive accuracy and correlation data

In the following tables, predictive accuracy is reported for different tests (such as pH or lactate) and for different outcomes (such as Apgar score). The specific tests and the thresholds used (for example fetal scalp pH less than 7.25) are listed in the rows of the GRADE table and the outcomes that they predict are listed in the ‘definition of outcome’ column. The measures of diagnostic accuracy in each row represent the specific values for that test and threshold for that outcome.

10.5.3.4. Evidence statements

10.5.3.4.1. Comparative clinical outcome data

There was no evidence of a difference in mode of birth (n=3319) for women who were managed with fetal blood sample lactate measurements and women who were managed with pH measurements. There was also no evidence of a difference in risk of the neonatal outcomes reported, including death (n=2992), encephalopathy (n=2992), admission to neonatal intensive care unit (n=2992), Apgar score less than 7 at 5 minutes (n=3319) and various cord blood gas measurements (n=3348) (pH, lactate and base deficit). The evidence was of moderate to very low quality.

10.5.3.4.2. Predictive accuracy of fetal blood sampling for composite neonatal outcomes

A pH of less than 7.25 was found to have a moderate specificity for the composite neonatal outcome (n=96), but all other diagnostic accuracy parameters were low or not useful. There was conflicting evidence around the accuracy of a threshold of 7.20 or 7.21: 1 study (n=150) (using a threshold of pH of 7.21 of less) reported a moderate sensitivity and moderately useful negative likelihood ratio with other parameters classed as low or not useful, whereas another study (n=96) (using a threshold of pH less than 7.20) reported a high specificity and very useful positive likelihood ratio with low sensitivity and not useful negative likelihood ratio. The quality of the evidence ranged from moderate to very low.

10.5.3.4.3. Predictive accuracy of fetal blood sampling for Apgar score at 5 minutes

There was consistent evidence from 2 studies (n=531) that a pH threshold of 7.25 or less or less than 7.21 had low sensitivity, low specificity and not useful likelihood ratios for predicting a low 5 minute Apgar score. A pH threshold of less than 7.10 was found to have high specificity and a very useful positive likelihood ratio for predicting low Apgar score at 5 minutes, but the sample size was very small (n=23) which limited the validity of the findings.

Lactate measurements (using a threshold of 4.2 mmol/l or more, or more than 4.8 mmol/l) were found to have a moderate sensitivity and moderately useful negative likelihood ratio for predicting low 5 minute Apgar score (n=684), with other diagnostic accuracy parameters low or not useful.

The use of base deficit measurements (using thresholds of more than 10 mEq/l or more than 12.5 mEq/l) was found to have moderate to high specificity, but other diagnostic accuracy parameters were low or not useful. However, most of this evidence came from 1 study with a very small sample size (n=19). The evidence across all outcomes was of moderate to very low quality.

10.5.3.4.4. Correlation of fetal blood sampling findings with Apgar score at 5 minutes

Evidence from 1 study (n=41) showed that the correlation of fetal blood sample pH and low Apgar score at 5 minutes was low between 60 and 15 minutes of birth, becoming moderately positively correlated for pH measurements taken within 15 minutes of birth and highly positively correlated for pH measurements taken within 5 minutes of birth. However, the sample size was small, particularly for the group with fetal blood samples taken within 5 minutes of birth (n=8). There was very low or no correlation between pH and high Apgar score at 5 minutes, regardless of the point at which the measurement was taken.

Evidence from 1 study (n=13) showed that base deficit taken within 60 minutes of birth was highly negatively correlated with low Apgar at 5 minutes, regardless of at what point the measurement was taken. However, the study sample size was very small. In contrast, there was very low or no correlation between base excess and high Apgar score at 5 minutes. The quality of the evidence was very low.

10.5.3.4.5. Predictive accuracy of fetal blood sampling for arterial pH at birth

There was evidence from 1 study (n=508) that a pH threshold of either 7.25 or less, or less than 7.21 had a low or not useful level of diagnostic accuracy for poor arterial cord blood gas values at birth, as measured either by a pH of less than 7.00 at birth or the diagnosis of metabolic acidaemia (pH less than 7.05 and base deficit more than 12 mmol/l). Evidence from another study (n=21) was that these same pH thresholds also had a high sensitivity and very useful negative likelihood ratio, but the sample size was very small.

There was evidence from 1 study (n=684) that a lactate threshold of 4.2 mmol/l or more, or more than 4.8 mmol/l had a high sensitivity and moderate negative likelihood ratios, with specificity and positive likelihood ratios all low or not useful.

Base deficit thresholds of more than 10 mEq/l or more than 12.5 mEq/l were found to have a moderate to high specificity, but again the sample size was very small (n=18). The evidence was of moderate to very low quality.

10.5.3.4.6. Correlation of fetal blood sampling with umbilical artery values at birth

There was evidence from 1 study (n=31) that pH and base excess measured within 5 minutes of birth have high correlation with umbilical artery pH at birth, but this evidence was from 1 small study. The evidence was of very low quality.

10.5.4. Health economic profile

No published economic evaluations were identified for this question.

A cost analysis was developed in Excel – further details of the cost inputs can be found in appendix A 5.3.

Lactate levels can be measured on some blood gas analysers, but not all. Therefore it is likely that new lactate test meters will be needed. The blood gas analyser is a standard device in an obstetric unit. Based on their experience, the guideline development group estimated that fetal blood sampling would represent approximately one-tenth of the use of the machine. Therefore the analyser would still be needed even if it was not used for fetal blood sampling. The costs of purchasing a lactate meter and the associated consumables (£1.74 per sample taken) were compared to the consumable costs related to using a blood gas analyser for pH measurements (£0.74 per sample taken).

A capillary sample of the baby's blood is taken from the scalp. The technique is the same regardless of whether lactate or pH is measured. The costs for staff to take a sample were estimated (£13 to £20 for 20 minutes of a specialty trainee or registrar's time).

The success rates reported in the clinical review were used to calculate the mean staff costs for taking a sample (97.8% for lactate tests compared to 89.6% for pH samples). For the base case analysis it was assumed that successful tests would only have 1 sample taken and unsuccessful tests would require 2 samples. This was a conservative assumption as sometimes a successful test can require 2 or more attempts to obtain a sample. This rate will depend on the experience of staff.

Under these assumptions the cost per test was lower for the pH sample when using a blood gas analyser, but as the success rates were lower than for taking a lactate sample this analysis showed lactate sampling was slightly less expensive than pH testing. The difference in cost per test was small (£0.66 less for lactate).

10.5.4.1. Evidence to recommendations

10.5.4.1.1. Relative value placed on the outcomes considered

The aim of this review was to determine the predictive value of various fetal blood sampling measures for neonatal outcomes. Clinically, the aim of performing fetal blood sampling is to identify those babies who are acidotic and whose birth therefore needs to be expedited by either caesarean section or instrumental intervention.

In the study which compared clinical outcome data for pH and lactate measurements, the key outcomes of interest were the mode of birth, neonatal encephalopathy and Apgar score less than 7 at 5 minutes.

In the diagnostic studies which evaluated the diagnostic accuracy of various fetal blood sampling tests and thresholds for identifying either low Apgar scores or composites of poor neonatal outcomes, the guideline development group felt that the most important measures were specificity and negative likelihood ratio (as these indicate that the test is effective at ruling out those babies who are not at risk, thus minimising unnecessary intervention). The group felt that this was appropriate as clinically fetal blood sampling would be performed as an adjunctive test to electronic fetal monitoring which generally has a high sensitivity but low specificity (that is, it has a high false positive rate). The results of the 2 tests would thus be considered together.

The group recognised that there were reasons to treat all of the diagnostic accuracy measures with caution. The first problem is that in some of the studies there was a delay of up to 60 minutes between the time of the blood test being taken and the baby being born. During this time, the baby could develop a new complication or go through a traumatic birth, and therefore be born in poor condition despite having an apparently normal fetal blood sampling result. This would therefore have the effect of lowering the sensitivity and generating worse negative likelihood ratio findings, since it would appear that the test had failed to pick up a baby at risk.

A further issue for the group when considering the diagnostic accuracy measures in the review is that the studies were designed so that if the result of a fetal blood sample was of concern, action was taken by the clinicians to resolve the problem. Consequently, even though a large number of the babies who had a concerning fetal blood sample result were born without poor outcomes, it was not possible to determine if this was because the test gave a false positive result or because the clinical intervention avoided a poor neonatal outcome.

The group did not place much value on the correlation findings, except to note that they confirmed what the group would have expected from their clinical experience, which is that there was an increasingly high correlation between a poor fetal blood sample result and a poor outcome when the interval between the sample being taken and birth got shorter.

10.5.4.1.2. Consideration of clinical benefits and harms

With this topic, the guideline development group wished to strike the right balance between ensuring that babies at genuine risk would be identified and treated accordingly, and ensuring that women weren't asked to undergo caesarean sections unnecessarily. The group noted that the systematic review (with 2 included trials) showing a direct comparison between lactate and pH measurements showed no statistically significant difference between the 2 for any clinical outcomes. In other words, the choice of test did not make a significant difference to the numbers of babies experiencing poor outcomes in either arm of the study. The previous recommendation in the guideline had only made reference to measuring pH. However, given the equivalence of the 2 tests, the group felt that it was appropriate to reference lactate measurements as well.

The group considered the evidence comparing the diagnostic accuracy of the tests. They noted that although the measures were similar for pH and lactate, lactate appeared to be associated with a slightly higher negative likelihood ratio. In addition, in a study that evaluated both tests (Wiberg-Itzel et al., 2008) the use of lactate was associated with higher sensitivities for both low Apgar score and arterial pH. The group members were also aware from their clinical experience that the use of lactate could potentially reduce the time for taking a sample as much less blood needs to be taken and fewer repeat samples are required (although not included in the evidence review as one of the priority outcomes, the Cochrane review [East et al., 2011] reported that lactate had a statistically significantly higher success rate than pH [95% compared with 89%]). As the process of taking a fetal blood sample is invasive, the group felt that it would be a positive step if the time required for this process could be reduced.

Ultimately, the group did not feel that they could recommend that only lactate be used as a diagnostic test. They did not feel that there was strong enough evidence in its favour and, as noted above, its use did not lead to an improvement in clinical outcomes. Furthermore, the group recognised that pH is the standard test used in the UK for this indication and that there was not sufficient justification for a complete change in practice. However, the group agreed that if clinicians did have the means available to test lactate, and they had received sufficient training, it should be used as the first line diagnostic test.

The guideline development group noted that there was also evidence available for the use of base deficit. Although the findings were comparable to those of the other tests, the group did not feel that it was appropriate to recommend its routine use. From their clinical experience, the group members were aware that there can sometimes be difficulty with taking a base deficit sample as the results can be affected by exposure to air as the blood sample is taken. In addition, they noted that the majority of the evidence for base-deficit was based on a small sample of less than 20 women in 1 study (Kerenyi et al., 1970).

The group discussed the practicalities of performing a fetal blood sample and agreed that the procedure was generally easier to perform, and more comfortable for the woman, with the woman in a left lateral position. They also recognised that the procedure was more likely to be successful if the woman's cervix was dilated 4 cm or more. The group also made a number of recommendations about repeat sampling. Although the evidence did not look specifically at the use of repeat samples, the group felt that they formed a key part of standard clinical practice. Although the group members acknowledged that sampling is an invasive procedure, they agreed that performing further samples when indicated by the CTG was preferable to performing unnecessary instrumental or caesarean births. The particular thresholds that the group chose for repeat sampling and the timing of this were derived from their own clinical practice and experience. Appropriate actions following failed sampling or findings of fetal acidosis were also discussed and recommendations made accordingly. It was also noted that there are certain instances where the risk of performing fetal blood sampling would outweigh any potential benefits, for example where there is an increased risk of passing infection to the baby, and added this to the recommendations.

10.5.4.1.3. Consideration of health benefits and resource uses

No formal cost effectiveness modelling was performed for this question but a cost analysis was developed. The guideline development group considered the likely cost impact of its recommendations and agreed that it would be minimal. Although lactate was recommended as a first line diagnostic test, this is only in units where the equipment and training is already available. Otherwise, there would not necessarily be a large change in practice. The group felt that it would be possible to have a clearer understanding of the likely cost impact of using lactate rather than pH measurements once there was better quality outcome data available from UK studies.

10.5.4.1.4. Quality of evidence

The evidence was of mixed quality, ranging from very low to moderate for the various outcomes. The evidence supporting the change in the recommendation in favour of lactate was drawn from a study of moderate quality. However, as the study was from a different setting (Sweden) and was not particularly large, the guideline development group did not feel it was sufficient to make a stronger recommendation.

10.5.4.1.5. Other considerations

The guideline development group discussed the appropriate thresholds to use for interpreting the findings of fetal blood samples. They did not feel that there was any evidence to suggest changing the extant thresholds for pH, and agreed that they should recommend the use of the lactate thresholds as reported in the studies.

The group felt it important that women be fully informed of the nature of the procedure required to obtain a fetal blood sample and its risks and benefits, particularly the risk of a ‘failed’ sample and the possible actions that may be considered once a result is obtained.

10.5.4.1.6. Key conclusions – fetal blood sampling

The guideline development group notes that the use of fetal blood sampling as an adjunct to CTG was associated with significantly more instrumental vaginal deliveries and caesarean sections compared to women monitored with CTG alone. In contrast, there was extensive evidence of benefits to the baby, notably lower incidences of: cord blood acidosis; need for neonatal resuscitation; neonatal seizures; and low Apgar scores. Also the predictive accuracy statistics for fetal blood sample values show very good positive predictive values for adverse neonatal outcome with a pH less than 7.20 and very good positive predictive values and moderately good negative predictive values for a fetal blood sampling pH threshold of 7.10. Finally, there is excellent correlation between fetal blood sample pH values and cord arterial pH values. On balance, the group felt that the evidence of benefit to the baby from using CTG supported by fetal blood sampling outweighed the increased likelihood of an operative delivery in the woman.

10.5.4.2. Recommendations

137.

When offering fetal blood sampling, explain the following to the woman:

  • Why the test is being advised.
  • The blood sample will be used to measure the level of acid in the baby's blood, to see how well the baby is coping with labour.
  • The procedure will require her to have a vaginal examination using a small device similar to a speculum.
  • A sample of blood will be taken from the baby's head by making a small scratch on the baby's scalp. This will heal quickly after birth, but there is a small risk of infection.
  • The procedure can help to reduce the need for further, more serious interventions.
  • What the different outcomes of the test may be (normal, borderline and abnormal) and the actions that will follow each result.
  • There is a small chance that it will not be possible to obtain a blood sample (especially if the cervix is less than 4 cm dilated). If a sample cannot be obtained, a caesarean section or instrumental birth (forceps or ventouse) may be needed because otherwise it is not possible to find out how well the baby is coping. [new 2014]
138.

Do not carry out fetal blood sampling if any contraindications are present, including risk of maternal-to-fetal transmission of infection or risk of fetal bleeding disorders. [new 2014]

139.

Take fetal blood samples with the woman in the left-lateral position. [2014]

140.

Measure either lactate or pH when performing fetal blood sampling. Measure lactate if the necessary equipment and suitably trained staff are available; otherwise measure pH. [new 2014]

141.

Use the classification of fetal blood sample results shown in table 110. [new 2014]

Table 110. Classification of fetal blood sample results.

Table 110

Classification of fetal blood sample results.

142.

Interpret fetal blood sample results taking into account any previous lactate or pH measurement, the rate of progress in labour and the clinical features of the woman and baby. [new 2014]

143.

Inform the consultant obstetrician if any fetal blood sample result is abnormal. [new 2014]

144.

Discuss with the consultant obstetrician if:

  • a fetal blood sample cannot be obtained or
  • a third fetal blood sample is thought to be needed. [new 2014]
145.

If the fetal blood sample result is normal, offer repeat sampling no more than 1 hour later if this is still indicated by the cardiotocograph trace, or sooner if additional non-reassuring or abnormal features are seen. [2014]

146.

If the fetal blood sample result is borderline, offer repeat sampling no more than 30 minutes later if this is still indicated by the cardiotocograph trace, or sooner if additional non-reassuring or abnormal features are seen. [2014]

147.

Take into account the time needed to take a fetal blood sample when planning repeat sampling. [2014]

148.

If the cardiotocograph trace remains unchanged and the fetal blood sample result is stable (that is, lactate or pH is unchanged) after a second test, further samples may be deferred unless additional non-reassuring or abnormal features are seen. [new 2014]

149.

If a fetal blood sample is indicated and the sample cannot be obtained, but the associated scalp stimulation results in fetal heart rate accelerations, decide whether to continue the labour or expedite the birth in light of the clinical circumstances and in discussion with the consultant obstetrician and the woman. [new 2014]

150.

If a fetal blood sample is indicated but a sample cannot be obtained and there is no improvement in the cardiotocograph trace, advise the woman that the birth should be expedited (see recommendations 220 to 223). [new 2014]

10.6. Cardiotocography using telemetry compared with conventional cardiotocography

10.6.1. Review question

What is the effectiveness of cardiotocography using telemetry compared with conventional cardiotocography?

For further details on the evidence review protocol, please see appendix E.

10.6.2. Description of included studies

Seven studies were included in this review (Calvert et al., 1982; Flynn et al., 1978; Frenea et al., 2004; Haukkamaa et al., 1982; Hodnett, 1982; Karraz, 2003; MacLennan et al., 1994).

All of the studies were randomised controlled trials, with 2 conducted in the UK (Calvert et al., 1982; Flynn et al., 1978), 2 in France (Frenea et al., 2004; Karraz, 2003) and 1 each in Finland (Haukkamaa et al., 1982), Canada (Hodnett, 1982) and Australia (MacLennan et al., 1994).

Three of the studies were specifically evaluating the comparison of cardiotocography (CTG) using telemetry with conventional CTG (Calvert et al., 1982; Haukkamaa et al., 1982; Hodnett, 1982). In 2 further studies the comparison of interest was primarily ambulation versus recumbency during labour, but this was implemented by monitoring women with telemetry or conventionally (Flynn et al., 1978; MacLennan et al., 1994). In the last 2 studies, the aim was to evaluate ambulatory anaesthesia and therefore all women had an epidural and then were randomised to ambulation monitored using telemetry or recumbency monitored conventionally (Frenea et al., 2004; Karraz, 2003).

The samples in 4 studies were restricted to low risk women, or women with uncomplicated or ‘uneventful’ pregnancies (Frenea et al., 2004; Haukkamaa et al., 1982; Hodnett, 1982; Karraz, 2003). In 2 studies, the authors did not report restricting the study population to low risk women but did exclude certain categories of higher risk women (Calvert et al., 1982; MacLennan et al., 1994). In 1 study, all of the women were in spontaneous labour but the authors did not report their inclusion and exclusion criteria so it is not clear how closely the study population matches the population of interest (Flynn et al., 1978).

10.6.3. Evidence profile

A fixed effects model was used for the majority of meta-analyses, with the exception of 2 outcomes (use of narcotic analgesia or pethidine; use of no pain relief) where there was high heterogeneity (I2 more than 60%) and so a random effects model was used.

Outcomes relating to pain relief are reported only for studies where ambulatory anaesthesia was not the intervention being investigated.

The findings from the included studies are reported in two separate GRADE tables:

  • Table 111 contains data for clinical outcomes and women's views and experience where comparative data were reported.
  • Table 112 contains qualitative findings about women's experience and details about degree of mobility, in addition to any outcomes for which comparative data was unavailable.
Table 111. Summary GRADE profile for the comparison of cardiotocography using telemetry with conventional cardiotocography.

Table 111

Summary GRADE profile for the comparison of cardiotocography using telemetry with conventional cardiotocography.

Table 112. Further findings for women's experience and mobility outcomes for the comparison of cardiotocography using telemetry with conventional cardiotocography.

Table 112

Further findings for women's experience and mobility outcomes for the comparison of cardiotocography using telemetry with conventional cardiotocography.

10.6.4. Evidence statements

There was evidence of no difference (n=800) in mode of birth for women who were monitored using telemetry compared with women who were monitored using conventional CTG. Women monitored using telemetry had lower rates of epidural use (n=554) than women monitored using conventional CTG, but there was no difference in the use of other forms of pain relief. There was no difference in measures of the length of labour (n=196). There was no evidence of a difference in neonatal outcomes (perinatal death [n=196], admission to neonatal intensive care unit [n=196], umbilical artery pH [n=61]), but each of these outcomes was only reported by 1 small study. The evidence was of moderate to very low quality.

One study (n=30) found that the proportion of women reporting that the monitor had a positive effect on labour experience was increased in women monitored with telemetry, but this study was small and had significant limitations. Most measures of women's experience did not show a difference between the 2 groups. However, 1 study (n=200) found that women who were monitored using telemetry generally felt that they were equally or less restricted and anxious than in a previous labour when they had been monitored using conventional CTG. There was consistent evidence across studies that not all women monitored with telemetry chose to get out of bed and that the time spent mobile varied, but non-comparative data from one study (n=200) found that most women who were monitored using telemetry would choose to be mobile again in a future labour. The evidence was of moderate to very low quality.

10.6.5. Health economics profile

No published economic evaluations were identified for this question.

A simple decision tree model was developed to consider the cost differences between the 2 forms of cardiotocography:

  • conventional cardiotocography
  • cardiotocography using telemetry.

The use of cardiotocography was not recommended for low risk pregnancies in any setting in the previous edition of the guideline. The full report of this analysis can be found in appendix A.

The clinical review found no evidence of a difference in mode of birth, length of labour or neonatal outcomes. Women monitored using telemetry had lower rates of epidural use than women monitored using conventional CTG (see table 113).

Table 113. Relative risk of epidural given method of CTG (meta-analysis of 5 studies, Calvert et al., 1982; Flynn et al., 1978; Haukkamaa et al., 1982; Hodnett, 1982; MacLennan et al., 1994).

Table 113

Relative risk of epidural given method of CTG (meta-analysis of 5 studies, Calvert et al., 1982; Flynn et al., 1978; Haukkamaa et al., 1982; Hodnett, 1982; MacLennan et al., 1994).

Information was obtained about the number of CTG units used on a labour ward and the cost of the different monitoring systems from a commercial distributor's sales team for an obstetric unit in England in 2013. The obstetric unit had around 6000 births per year. The unit has 6 telemetry systems all acquired in the last 2 years, as well as an additional 6 older conventional CTG systems.

The cost of a wireless telemetry system was around £6300 (table 114). In addition, the detached tocography and ultrasound transducers cost £910 each and the ECG transducers cost £804 each. The risk with detached transducers is that they are more likely to be lost (1 birth unit had to replace 3 lost transducers in a year). However, the guideline development group believed that this could be avoided with better education on the use of the equipment.

Table 114. Costs of acquiring conventional and telemetry CTG systems, 2013 prices.

Table 114

Costs of acquiring conventional and telemetry CTG systems, 2013 prices.

The conventional CTG system, including attached tocography and ultrasound transducers, costs around £5400 (table 114). Any additional transducers cost £350. The risk of losing a transducer is far lower than for telemetry because they are attached to the monitor.

As the only outcome identified with a difference was epidurals, a simple cost minimisation analysis was developed. Continuous monitoring is not routinely recommended for low risk women. However, the group believed that in an obstetric unit approximately 50% of all women will be monitored during labour. Only 5% of women in the obstetric unit would be low risk and go on to require continuous monitoring.

This analysis is based on a unit with 6000 births, requiring 12 CTG systems. Of these 6000 women, approximately 50% would require continuous monitoring. Of the 3000 women who require continuous monitoring, approximately 300 would be considered low risk: the analysis will focus on this group of women.

For this population of 300 low risk women requiring continuous monitoring, it is expected that 33 more epidurals would be needed if the women were monitored using a conventional system (110 compared with 77). The difference in the cost of epidurals is £6225 per year (table 115).

Table 115. Annual cost difference of epidurals per cardiotocography system, based on 5% of low risk women requiring CTG monitoring (n=300) in an obstetric unit with 6000 births per year.

Table 115

Annual cost difference of epidurals per cardiotocography system, based on 5% of low risk women requiring CTG monitoring (n=300) in an obstetric unit with 6000 births per year.

Purchasing a CTG system is a capital cost, requiring an up-front payment. The expected lifespan of both types of CTG system would be at least 7 years and could be up to 15 years. The annual equivalent cost of both systems has been calculated in table 116 and this has been used to determine the cost per use.

Table 116. Annual cost per CTG system and cost per use assuming 3000 women are monitored per year in a unit with 6000 live births.

Table 116

Annual cost per CTG system and cost per use assuming 3000 women are monitored per year in a unit with 6000 live births.

The staff cost of an epidural was calculated through consensus among the guideline development group (appendix A). The mean cost was £108 (£69 to £247). The cost of consumables (such as epidural pack, gloves, syringes, IV drip) was taken from the Birthplace study and the prices uplifted to 2013. The total cost of an epidural was calculated as £189 (£149 to £328).

The cost per epidural is much greater than incremental cost per use of the telemetry CTG, and so telemetry is found to be cost saving by £5533 in this analysis (table 117).

Table 117. Annual cost difference based on 5% of low risk women requiring CTG monitoring (n=300) in an obstetric unit with 6000 births per year with 12 CTG monitors.

Table 117

Annual cost difference based on 5% of low risk women requiring CTG monitoring (n=300) in an obstetric unit with 6000 births per year with 12 CTG monitors.

Sensitivity analyses were carried out to test the results and the full analyses are presented in appendix A. If the cost of an epidural is lower than these estimates, then using telemetry remains cost effective when the cost of an epidural is £25 or more.

The outcomes for the analysis were based on clinical evidence from a meta-analysis. Women monitored using telemetry had fewer epidurals than women monitored with conventional CTG (77 compared with 110). When using data from a clinical trial there is likely to be uncertainty as the trial setting is unlikely to match the care received in the real world. If the difference in rate of epidurals was smaller, then the cost saving from using telemetry reduces. However, even if the study has over-estimated the effect, telemetry remains cost saving as long as 4 fewer epidurals are performed (106 compared with 110) when using a telemetry system, if all other assumptions in the model are true.

It was suggested that the equipment lasted longer than the 7 years used in the base case scenario. If the monitoring equipment is expected to last 10 years rather than 7 years then the cost per use will be reduced (conventional system £2.57 compared with £3.50; telemetry system £4.27 compared with £5.81) and the cost savings increase when using the telemetry system. A lifespan of 15 years was also suggested, and this would further increase the savings seen with the telemetry system.

Conventional CTG equipment is being replaced by complete telemetry systems. There are also add-on telemetry kits which can be used to adapt conventional systems. Therefore the question of cost effectiveness relates to when telemetry equipment should be bought by a unit. Where a unit finds epidurals are reduced at a rate reflected by the clinical evidence reported here, it is likely that cost savings would be achieved by switching to telemetry before conventional equipment has reached the end of its lifespan.

Using a telemetry system allows women greater mobility than when using a conventional CTG system. This increased movement translates into fewer epidurals according to the clinical evidence. As the cost of an epidural has been estimated at £189, which is considerably greater than the incremental cost per use of telemetry CTG (£2.30 more than the cost per use of a conventional system) then using telemetry is likely to result in cost savings.

10.6.6. Evidence to recommendations

10.6.6.1. Relative value placed on the outcomes considered

The guideline development group agreed that outcomes relating to mode of birth and length of labour were clinically important outcomes, and that outcomes relating to women's experience of labour, birth and mobility during labour would additionally be priorities for women. They also felt that it was vital to assess the comparative effectiveness of telemetry and conventional monitoring for preventing adverse neonatal outcomes.

10.6.6.2. Consideration of clinical benefits and harms

The guideline development group discussed the fact that monitoring with telemetry did not appear to have an effect on mode of birth, when compared with conventional CTG. The group did note that there was a significant reduction in the proportion of women using epidural in the telemetry group, although this did not translate into a reduction in women using other forms of pain relief or a significant difference in the number of women using no pain relief. The group discussed the evidence around length of labour and concluded that there was evidence of little difference between the 2 types of monitoring. While they noted that Flynn et al. (1978) found a significantly shorter first stage in the telemetry arm, it was a small study which was poorly reported and did not report restricting its study population to low risk women. All of the other studies reporting measures of length of labour did not find a difference between women monitored with telemetry and those monitored conventionally.

The group noted that no differences had been found for the neonatal outcomes reported, but that overall the neonatal outcomes had been quite poorly reported in the studies and the studies were underpowered for rare adverse outcomes. From their clinical experience, the group members agreed that there was no particular reason to suspect that neonatal outcomes would be different between the 2 groups, provided that mobilisation was appropriate.

The group discussed the evidence that was reported about women's experience. They agreed that there was weak evidence that women might prefer being monitored with telemetry, but that overall the quality of the evidence was quite poor. From their clinical experience, they agreed that women tended to prefer telemetry as they valued the opportunity to be more mobile and not be restricted during labour. Anecdotally, they felt that this tended to have a positive effect on progress of labour and might affect women's need for pain relief. The group did acknowledge that even with telemetry women are not fully mobile during labour, but their ability to be mobile is better than with conventional CTG monitoring.

10.6.6.3. Consideration of health benefits and resource uses

Conventional CTG systems can now only be replaced with complete telemetry monitoring systems. Additionally, there is add-on telemetry equipment that can be used to adapt a conventional monitoring system. Therefore, it is likely that obstetric units will be moving towards greater use of telemetry systems.

It has been demonstrated that using a telemetry system for monitoring results in a reduction in the need for epidurals as women have more mobility than with conventional monitoring. Telemetry systems are more expensive than conventional equipment but the increased capital costs are offset by savings due to the reduced rate of epidurals. The cost per use of a telemetry system is far less than the cost per epidural (£5.81 compared with £380). The evidence has demonstrated that the move to telemetry for monitoring is likely to result in cost savings.

Where telemetry is introduced, it is important that staff are educated in the use of the equipment to avoid transducers being lost.

10.6.6.4. Quality of evidence

The evidence was generally of low or very low quality and the guideline development group was disappointed that better quality evidence was not available. The group noted that the studies were generally small, had methodological weaknesses and were quite old. In addition, they noted issues of indirectness in the study populations – in particular, they discussed the fact that Hodnett et al. only included married women with an uncomplicated vaginal birth, and that Flynn et al. did not report its inclusion or exclusion criteria. They agreed that this undermined the applicability of the evidence to low risk women in labour, but when combined with their clinical experience, the group members felt that overall the evidence was sufficient to make recommendations. In order to try and prompt further research in this area, they drafted a research recommendation to directly compare conventional CTG with CTG using telemetry.

10.6.7. Recommendations

151.

Offer telemetry to any woman who needs continuous cardiotocography during labour. [new 2014]

10.6.8. Research recommendations

18. In women that require continuous electronic fetal monitoring during labour, what is the effectiveness of cardiotocography using telemetry compared with conventional cardiotocography?

Population: all women requiring continuous electronic fetal monitoring during labour

Intervention: continuous cardiotocography using telemetry

Comparator: conventional cardiotocography

Primary Outcome: neonatal outcomes (including long term outcomes at 2 years)

Secondary outcomes: length of labour, use of pain relief, women's experiences.

Study design: observational study

Why this is important

The use of telemetry to monitor the fetal heart rate and uterine contractions in labour has the potential to enable women to be more mobile and active than with conventional monitoring. There is very little recent research evidence exploring whether the use of telemetry in labour to continuously monitor the fetal heart rate and uterine contractions has any effect on neonatal outcomes, length of labour or use of pain relief. Women's experiences of telemetry also remain an area for investigation. Both quantitative and qualitative aspects of telemetry use in labour should be explored and the cost effectiveness of telemetry cardiotocography evaluated.

10.7. Women's views and experiences of fetal monitoring

10.7.1. Review question

What are women's views and experiences of fetal monitoring in labour?

For further details on the evidence review protocol, please see appendix E.

10.7.2. Description of included studies

Five studies (Parisaei et al., 2011; Mangesi et al., 2009; Hindley et al., 2008; Hansen, et al., 1985; Shields, 1978) are included in this review. Of the studies, 2 were conducted in the UK (Parisaei et al., 2011; Hindley et al., 2008), 1 in South Africa (Mangesi et al., 2009), 1 in Denmark (Hansen, et al., 1985) and 1 in Canada (Shields, 1978).

Each of the 5 studies looked at different comparisons. A recent descriptive study (Parisaei et al., 2011) evaluated the acceptability of a fetal electrocardiographic (STAN) monitoring system by women at a London Hospital. One study (Shields, 1978) examined women's views and experiences of internal (using a fetal scalp electrode) electronic fetal monitoring during labour. One study (Hindley et al., 2008) surveyed women's preferences of fetal heart rate monitoring methods before and after labour by means of antenatal and postnatal questionnaires. One study (Hansen, et al., 1985) compared women's views of cardiotocography (CTG) with intermittent auscultation. A final study (Mangesi et al., 2009) examined women's preferences regarding 3 methods used to monitor their baby's heart rate: CTG, a fetal stethoscope and a hand-held Doppler ultrasound fetal heart rate monitor. Each method was applied for only 10 minutes and then the women's preference was assessed. For further details of the included studies see the evidence table (appendix I).

All included studies are observational studies with considerable limitations.

10.7.3. Evidence profile

The findings for the women's views and experiences of fetal monitoring in labour are categorised in 2 sections:

  • women's views and experiences of STAN (fetal electrocardiographic monitoring system)
  • women's views and preferences for methods used to monitor fetal heart rate; a fetal stethoscope, Doppler ultrasound fetal heart rate monitor and CTG.
Table 118. Findings for women's views and experiences of fetal monitoring in labour.

Table 118

Findings for women's views and experiences of fetal monitoring in labour.

10.7.4. Evidence statements

One study (n=125) found that the majority of women whose babies had been electronically monitored using ECG analysis found it both acceptable and reassuring and felt that the reasons for its use had been well explained. The quality of this study was very low.

One study (n=100) comparing women's views of fetal monitoring using the fetal stethoscope, Doppler ultrasound device and CTG showed that the Doppler ultrasound device was the most popular first choice. This finding was statistically significant. The evidence was of very low quality.

Two studies (n=718) investigated women's choice and preferences for intrapartum fetal monitoring. One study (n=655) comparing women's antenatal and postnatal preferences for intermittent auscultation compared with CTG showed a fairly even spread of preferences antenatally. The most commonly cited advantages of intermittent auscultation were that it was associated with a more natural childbirth and there was no discomfort compared with that experienced from the sensors and belts used in CTG. No specific disadvantages of intermittent auscultation are reported. The most commonly cited advantages of CTG were that it allowed continuous, precise surveillance and that women were positively influenced by hearing the baby's heartbeat and/or seeing it being traced out. The most commonly cited disadvantages were that it enforced immobility and was associated with a technical medicalisation of birth. The second study (n=63) found that there was no clear preference for mode of intrapartum fetal monitoring expressed antenatally. Although the majority of women reported they had been given information about fetal monitoring antenatally, only a minority felt they had been given an informed choice of type of monitoring during labour. The evidence was of very low quality.

One study (n=30) investigated women's experiences of internal fetal monitoring using a fetal scalp electrode. The majority of women responded positively when asked their views of this type of monitoring. Positive responses were associated with receiving adequate information about the monitoring. The evidence was of very low quality.

10.7.5. Health economic profile

No published economic evaluations were identified for this question.

10.7.6. Evidence to recommendations

10.7.6.1. Relative value placed on the outcomes considered

The guideline development group agreed that it was fundamental to consider women's views of, and satisfaction with, the type of fetal monitoring they receive. Monitoring has the potential to reduce a woman's fear and anxiety and provide reassurance. However, the group was aware that often monitoring can have the opposite effect and increase a woman's anxieties and discomfort. It is therefore important to identify how best to ensure women's satisfaction with the monitoring that they receive and how best to support evidence-based informed choice.

10.7.6.2. Consideration of clinical benefits and harms

The guideline development group recognised that 1 study investigating the use of ST wave analysis as a component of fetal monitoring demonstrated extremely positive findings. However, the group felt the findings from the study did not reflect their experience in practice and questioned the validity of the study. They noted that a large proportion of the study sample was made up of women with some form of risk factor and felt that this had the potential to impact on the findings (see below).

The group recognised that some of the comments from the surveys highlighted the importance of information giving and providing reassurance to women. The group acknowledged the importance of giving women accurate information about what cardiotocography (CTG) can and can't achieve, so that women understand the reasons for considering the use of continuous electronic monitoring and that their expectations are valid. They queried whether it would be appropriate for a midwife to have a discussion with a woman about choice of monitoring when she is in labour. It was agreed that it would be more appropriate if the discussion and provision of information was to happen antenatally, and therefore did not feel it was appropriate to make a recommendation for provision of information regarding women's plans for monitoring in the intrapartum period.

In terms of women's monitoring preference, the group noted a general trend in the evidence in favour of intermittent auscultation. Although the group felt that this should be recognised and supported by healthcare professionals, they did not feel that there was sufficient evidence to support a recommendation.

The group considered women's views on CTG monitors. In 1 qualitative study, women expressed their concerns that the CTG monitor can become the focus of attention in labour rather than the woman. This matched the experience of all of the guideline development group members who reflected that they were all aware of this phenomenon. The group agreed that whatever form of monitoring is used, it is important to ensure that the woman and her baby remain the focus of attention. The group was also aware that the monitor can often be incorrectly used in place of one-to-one care, with women left alone on the monitor. The group agreed that this was poor practice and that it is important that healthcare professionals should stay with the woman in order to provide one-to-one support and to monitor both her condition and the baby's condition.

10.7.6.3. Consideration of health benefits and resource uses

There were no specific considerations relating to resource use for this question.

10.7.6.4. Quality of evidence

The evidence available for this review consisted of retrospective and prospective observational studies. There was general agreement that the quality of the studies was extremely poor. It was noted that the studies had considerable limitations and a high risk of bias in both data collection and analysis. The guideline development group also noted that a number of studies included a significant proportion of high risk women. It was felt that this could potentially impact on the results as women identified as being high risk might be more likely to want the reassurance of electronic fetal monitoring.

The group recognised the difficulty in trying to determine women's preferences for a particular type of monitoring when each individual woman will generally only experience a single type. They noted that 1 study had tried to ensure that women experienced all types of monitoring but expressed frustration that each type of monitoring had only been used for 10 minutes.

10.7.6.5. Other considerations

The guideline development group acknowledged that the evidence base for this question was very poor and that it was a topic that merited further investigation. They also noted that the use of central electronic fetal monitoring systems and telemetry was increasing and felt that little was known about how these might impact upon a woman's experience of birth and the care she receives. In light of this, the group drafted research recommendations to explore women's perceptions and experiences of different types of fetal monitoring, and to evaluate its impact on the communication between the midwife and the woman and the provision of one-to-one care.

10.7.7. Recommendations

See recommendation 108.

10.8. Cardiotocography with fetal electrocardiogram analysis compared with cardiotocography alone

10.8.1. Review question

Does the use of fetal electrocardiogram (ECG) analysis with continuous electronic fetal monitoring (EFM) improve outcomes when compared with continuous EFM alone?

For further details on the evidence review protocol, please see appendix E.

10.8.2. Description of included studies

One study (Neilson, 2013) is included in this review. Neilson (2013) is a systematic review with 6 component trials from a variety of locations. All of the included trials in the systematic review compared the use in labour of continuous electronic fetal monitoring plus fetal electrocardiogram (ECG) with continuous electronic fetal monitoring alone. Five trials of ST waveform analysis and 1 trial of PR interval analysis are included in the systematic review (Neilson, 2013). All women included in the trials were at high risk of developing complications in labour. The duration of the monitoring using continuous electronic fetal monitoring and ECG is not reported in the included trials.

Although the wording of this question refers to electronic fetal monitoring it is apparent that in practice studies are referring to electronic fetal monitoring plus monitoring of contractions. This is more accurately termed cardiotocography (CTG) and therefore this term will be used in this evidence summary and throughout the guideline.

10.8.3. Evidence profile

A fixed effects model was used for these analyses, with the exception of 1 outcome (cord PH less than 7.05 plus base deficit more than 12 mmol/L) for which a random effects model was used due to high heterogeneity (I2 equal to 62%).

Sub-group analysis was performed for:

  • PR analysis
  • ST analysis.
Table 119. Summary GRADE profile for comparison of continuous CTG plus fetal electrocardiogram (ECG) PR interval analysis with continuous CTG alone in labour.

Table 119

Summary GRADE profile for comparison of continuous CTG plus fetal electrocardiogram (ECG) PR interval analysis with continuous CTG alone in labour.

Table 120. Summary GRADE profile for comparison of continuous CTG plus fetal electrocardiogram (ECG) ST waveform analysis (STAN) with continuous CTG alone in labour.

Table 120

Summary GRADE profile for comparison of continuous CTG plus fetal electrocardiogram (ECG) ST waveform analysis (STAN) with continuous CTG alone in labour.

10.8.4. Evidence statements

10.8.4.1. PR analysis

Findings from 1 study (n=957) indicated that there was no evidence of a significant difference in the rate of caesarean section and operative vaginal birth for women and fetal and neonatal death, admission to neonatal intensive care unit (NICU), Apgar score less than 7 at 5 minutes and neonatal intubation for babies born to women who received continuous CTG plus fetal ECG compared with women who received continuous CTG only. The evidence was of low to very low quality.

10.8.4.2. ST analysis

There was evidence from over 15,000 women that the rate of instrumental birth for women and admission to NICU was significantly lower for babies born to women who received continuous CTG plus fetal ECG compared with babies born to women who received continuous CTG only. There was no evidence of a difference in the rate of caesarean section, fetal and neonatal death, Apgar score less than 7 at 5 minutes, cord arterial pH less than 7.05 plus base deficit more than 12, hypoxic ischaemic encephalopathy and neonatal intubation for babies born to women who received continuous electronic fetal monitoring plus fetal CTG compared with women who received continuous CTG only. The evidence was of moderate to very low quality.

10.8.5. Review of published evaluations

The literature search identified 2 cost effectiveness analyses comparing CTG with ST analysis to CTG alone (Heintz et al., 2008; Vijgen et al., 2011). Neither of the analyses was set in the UK and so they were not useful as evidence for this guideline.

10.8.6. New economic evaluation

In the previous guideline a costing analysis was developed for ECG ST analysis. This compared the additional equipment costs in purchasing ST analysis equipment to the potential savings from reduced operative vaginal births and caesarean sections (NICE 2007). The net cost of ECG ST analysis was £3.4 million.

Given the updated clinical evidence now available, it was decided that a new economic evaluation should be developed for this guideline. The purpose of monitoring is to identify hypoxia before it is sufficient to lead to damaging acidosis and long-term neurological adverse outcome for the baby. Monitoring should provide a balance between correctly identifying the babies who require intervention without over-identifying resulting in levels of intervention that are too high. A full description of this analysis can be found in the appendix.

For women where monitoring is indicated, which method of monitoring is most cost effective? The following comparisons are considered:

  • CTG alone
  • CTG plus ECG ST waveform analysis.

Monitoring is necessary to identify babies in distress. In these cases intervention is necessary. Good monitoring will allow accurate identification of these situations and prevent unnecessary intervention where possible.

The number of instrumental vaginal births and admissions to neonatal special care units were statistically significantly lower for CTG plus ECG ST analysis. No other outcomes were found to be statistically significantly different. For PR analysis there was no statistically or clinically significant difference for any of the outcomes reported and so no benefit was demonstrated. Therefore, the model was developed for ST analysis only.

The main cost will be purchase of equipment for ST analysis. The cost of purchasing an ST monitor is approximately £25,000 per unit (see appendix A.3.4.1). The ST monitor is fully automated, but if the ST analysis shows a problem then training would be required to interpret the scan in order to decide whether to intervene. Midwives would be trained to interpret the ST analysis, with obstetricians called if there is a problem.

The clinical review included serious outcomes such as neonatal death and neonatal encephalopathy. This model should include the long-term costs for these outcomes. Identifying good quality inputs for long-term costs of neonatal intubation was a problem for previous economic evaluations in NICE guidelines (NICE 2011, NICE 2012) and for the Birthplace study (Schroeder et al., 2012) and so long-term costs have not been included in this analysis.

As with the costs, long-term outcomes such as life-years lost and reduced quality of life should be included in the model. Again, no good quality evidence of the long-term effects has been identified. Therefore the estimates used in the Caesarean Section guideline (NICE 2011) have been used for this model. The estimate used in the Caesarean Section guideline was for mild cerebral palsy as a proxy for neonatal encephalopathy.

The incremental cost effectiveness results show CTG plus ECG ST is less expensive but has worse health outcomes than CTG alone (table 121). The number of fetal and neonatal deaths was slightly higher in the CTG plus ECG ST group (0.104% compared with 0.065%; the difference was not statistically significant) and this drives the loss of quality adjusted life years (QALYs).

Table 121. Deterministic costs, effects, incremental costs and effects per woman needing monitoring and incremental cost effectiveness ratio for the comparison of CTG monitoring alone and CTG monitoring plus ECG ST analysis.

Table 121

Deterministic costs, effects, incremental costs and effects per woman needing monitoring and incremental cost effectiveness ratio for the comparison of CTG monitoring alone and CTG monitoring plus ECG ST analysis.

If the rate of mortality was the same between the 2 monitoring strategies then CTG plus ECG ST dominates CTG alone, being both less expensive and more effective (table 122).

Table 122. Sensitivity analysis – Rate of fetal and neonatal death is equal in both groups.

Table 122

Sensitivity analysis – Rate of fetal and neonatal death is equal in both groups. Costs, effects, incremental costs and effects per woman needing monitoring and incremental cost effectiveness ratio for the comparison of CTG monitoring alone and (more...)

As the majority of outcomes were not found to be statistically significantly different, the model was run with these outcomes equal for both groups, and only instrumental vaginal births and admissions to special care units included in the analysis. In this analysis, CTG plus ECG ST remains the lowest cost option, but still results in fewer QALYs gained than with CTG alone (table 123).

Table 123. Sensitivity analysis – all outcomes not statistically significantly different are held the same.

Table 123

Sensitivity analysis – all outcomes not statistically significantly different are held the same. Costs, effects, incremental costs and effects per woman needing monitoring and incremental cost effectiveness ratio for the comparison of CTG monitoring (more...)

Adding ECG ST monitoring to CTG monitoring results in cost savings due to fewer instrumental births and caesarean sections, and fewer neonatal intubations. However, in the base case more fetal and neonatal deaths occur when ECG ST monitoring is added to CTG monitoring and so more QALYs are lost.

The following outcome rates showed no statistically significant difference: caesarean section, neonatal encephalopathy, neonatal intubation, and fetal and neonatal death. The other 2 main outcomes – instrumental births and admission to neonatal intensive care – were close to unity. This introduces considerable uncertainty to the results. The results of the probabilistic sensitivity analysis demonstrated that CTG alone would be the most cost-effective option above a threshold of approximately £2000 per QALY, but with a likelihood of approximately 65% (see figure 1). When the mortality rate is equal between the 2 strategies, adding ECG ST analysis will dominate CTG alone.

Figure 1. Threshold analysis of CTG monitoring and CTG plus ECG ST monitoring.

Figure 1

Threshold analysis of CTG monitoring and CTG plus ECG ST monitoring.

Long term costs of neonatal encephalopathy were not included because data on long term outcomes and costs could not be identified. As neonatal encephalopathy was reduced when ECG ST monitoring is added to CTG monitoring, then adding these long term costs and outcomes would strengthen the case for adding ECG ST monitoring.

Other clinical outcomes were not reported in the studies and could impact the cost effectiveness results. ECG analysis requires invasive procedures: amniotomy, which may increase the pain of contractions; and the application of a fetal scalp electrodes, which can be associated with a small increase in the risk of infection in the baby.

This analysis has shown that adding ECG ST monitoring results in cost savings due to fewer interventions in birth, but also reduced effectiveness due to increased fetal and neonatal death (although the clinical results were not statistically significant). However, given the uncertainty in the clinical results due to a number of outcomes not reaching a statistically significant difference, these cost effectiveness results should be used with caution as the study results may not transfer to the real world. Further evidence taken from a UK setting where the outcomes of women can be followed through a pathway of care would help with understanding how monitoring could improve final outcomes.

10.8.7. Evidence to recommendations

10.8.7.1. Relative value placed on the outcomes considered

For this review, the guideline development group prioritised the outcomes of mode of birth and neonatal encephalopathy. These were felt to be both clinically significant and mode of birth also important for the woman's experience of labour and birth. The group did not feel that it was appropriate to identify ‘the need for use of fetal blood sampling’ as a priority outcome because the need for fetal blood sampling is reduced when fetal electrocardiogram (ECG) is used according to the ST analysis protocol.

10.8.7.2. Consideration of clinical benefits and harms

For PR waveform analysis, the guideline development group noted there was no statistically or clinically significant difference for any of the outcomes reported, thus no benefit has been demonstrated for this type of ECG analysis.

The group next considered ST waveform analysis of the fetal ECG. The group considered the evidence from the original IPC guideline for this intervention. It noted that there was new evidence available since the publication of the original guideline, and that meta-analysis of these trials had altered the statistical significance of 2 priority outcomes. The rate of neonatal encephalopathy, which was reported as statistically significantly lower for babies monitored using CTG plus ECG in the previous review, was non-significant in the updated review, with 2 more trials contributing to the findings for this outcome. In contrast, the rate of admission to the neonatal care unit was significantly lower in the CTG plus ECG group, which was reported as non-significant in the previous guideline. The guideline development group noted that, in line with findings from the original review, women in the CTG plus ECG group had a significantly lower incidence of instrumental vaginal birth compared with women monitored with CTG only. The guideline development group also recognised that there was no significant difference observed between the CTG plus ECG group and CTG alone group for other outcomes, including caesarean section rate and neonatal death. However, it was noted that the total number of women in the meta-analyses remains underpowered for rare events such as neonatal death.

The guideline development group recognised that 2 outcomes suggested a benefit of using ST wave analysis of the fetal ECG in conjunction with CTG, 1 maternal (vaginal instrumental birth) and 1 neonatal (admission to a neonatal special care unit). They considered the numbers needed to treat to avoid an admission to a neonatal special care unit. An estimated 101 women would need to be monitored with CTG plus ECG rather than CTG alone to avoid 1 neonatal admission to neonatal intensive care. The number needed to treat to avoid 1 vaginal instrumental vaginal birth was estimated as 60. For the priority outcome of neonatal encephalopathy there were approximately half the number of incidences observed in the CTG plus ECG group compared with the CTG alone group (number needed to treat = 1105), although this difference was not statistically significant. The guideline development group noted that the severity of this encephalopathy was not reported in the paper.

The guideline development group also recognised the potential disadvantages of using ECG analysis in conjunction with CTG. In order to monitor using ECG analysis, the invasive procedures of amniotomy and the insertion of a fetal scalp electrode need to be performed. Amniotomy was felt by some guideline development group members to be associated with an increase in the pain of contractions and the application of a fetal scalp electrode was acknowledged to be associated with a small increase in the risk of trauma to, and infection in, the baby,

10.8.7.3. Consideration of health benefits and resource uses

The guideline development group noted that use of ECG analysis involved the capital cost of purchasing the ST analysis monitors (approximately £25,000 per machine), and investment in training all midwives and obstetricians (those involved in providing intrapartum care in the obstetric unit) how to use the monitors. Although the cost of purchasing the ST analysis monitors is high, the cost per use will be minimal given the lifetime of the machine and number of births requiring monitoring. However, where there were differences in clinical outcomes between the two monitoring strategies, they were small (for instrumental births it was 15.2% using CTG alone compared with 13.6% when using ST monitoring; and for admission to neonatal special care unit it was 9% with CTG alone compared with 8% using ST monitoring) and in most outcomes there was no difference (caesarean section, fetal and neonatal death, neonatal encephalopathy, neonatal intubation). The capital costs may be offset by downstream cost reductions through fewer interventions during birth and reduced admission to neonatal special care, but there is considerable uncertainty as the differences between the monitoring strategies are so small.

10.8.7.4. Quality of evidence

The guideline development group noted that the quality of evidence varied, being moderate or low for the critical outcomes considered priorities for this review. This lack of high quality evidence, particularly the lack of power for the meta-analysis to provide valid findings in relation to rare adverse neonatal outcomes, contributed towards the group's reticence in recommending the use of ECG analysis.

10.8.7.5. Other considerations

The group was aware of a large ongoing trial in the US which was designed to evaluate the use of ST waveform analysis in conjunction with CTG compared with CTG alone. The group felt that this trial was likely to provide pertinent results which had the potential to affect any recommendations that the group made and therefore felt it inappropriate to make recommendations prior to these findings being published.

10.8.7.6. Key conclusions

Given the available evidence, the finding of no significant differences observed for the majority of key outcomes and the poor quality of studies, along with the high degree of uncertainty associated with the health economic analysis, the guideline development group felt that recommending the use of ECG in conjunction with CTG is not justified.

10.9. Computerised systems versus human interpretation

10.9.1. Introduction

A new review of computerised systems in FHR trace interpretation was undertaken.

10.9.2. Previous guideline

The Use of Electronic Fetal Monitoring guideline includes computerised interpretation of FHR tracings.460 The same six studies are included as are reviewed here. The summary of evidence concludes that: The use of computerised systems for FHR analysis improves consistency of interpretation. A research recommendation was for further evaluation of the effectiveness of computerised analysis, or decision analysis programs, in the interpretation of the CTG.

10.9.3. Description of included studies

Six studies were identified for review in this section. Five of these studies compared computerised interpretation of FHR tracings with expert interpretation. All studies included women with pregnancy and/or intrapartum complications.

10.9.4. Review findings

A rigorous multicentre comparative study undertaken in the UK investigated whether a computerised system could obtain a performance in labour management comparable with experts when using FHR tracings, obstetric information and FBS. It also investigated the degree of agreement between experts.495 [EL = II] Seventeen peer-nominated experts were selected from 16 UK maternity units to review 50 complete intrapartum FHR tracings. The 50 tracings were selected to represent a range of possible variables and outcomes and all were obtained from women with high-risk labours. The expert reviewers were also given clinical information pertaining to the progress of labour, and could request findings from FBS to supplement this information. Each expert performed the assessments twice (in a different order), with an interim period of 1 month in order to assess intra-rater reliability. Consistency (intra-rater reliability) of ratings for each reviewer was high, ranging from 73.18% to 89.04% (kappa 0.43 to 0.77). Consistency of ratings for the computerised system was 99.16%. Agreement between reviewers (inter-rater reliability) ranged from 58.17% to 74.27% (kappa 0.12 to 0.46). Agreement between the computerised system and the obstetricians was 67.33% (kappa 0.31). In the 11 cases where the computerised system recommended CS, on average 18/34 (52.9%) of the expert reviews also recommended CS within 15 minutes of the system. An average of 23/34 (67.6%) did so within 30 minutes of the system. Only two reviewers and the computerised system consistently recommended no unnecessary intervention. Twelve examples of poor outcome were included in the sample. Poor outcome fell into one of three categories as follows: birth asphyxia (cord arterial pH < 7.05 and base deficit ≥ 12, Apgar score at 5 minutes ≤ 7 with neonatal morbidity); metabolic acidosis (cord arterial pH < 7.05 and base deficit ≥ 12, Apgar score at 5 minutes > 7 with no neonatal morbidity); acidosis (cord arterial pH < 7.05 and base deficit < 12 with neonatal morbidity). The system detected two of the three incidents of birth asphyxia, two of the four incidents of metabolic acidosis and two of the five incidents of acidosis with no significant metabolic component. This was as good as the majority of experts for birth asphyxia, but fewer than for all reviewers for metabolic acidosis, and fewer than all but one of the reviewers for acidosis.

A small prospective observational study (UK, 2000) compared computerised interpretation of 24 intrapartum FHR tracings with expert ratings.496 [EL = II] Analysis was performed on 25 minute sections of tracing.

Inter-rater reliability between the seven experts was good for baseline FHR (r = 0.93), number of decelerations (r = 0.93) and type of decelerations (r = 0.93). Inter-rater reliability for baseline variability was poor (kappa = 0.27), as it was for accelerations (r = 0.27). Computerised interpretation of the tracings showed good agreement with the experts regarding baseline FHR (r = 0.91 to 0.98) and the number of decelerations (r = 0.82 to 0.91). Intra-class correlations were lower for the number of late decelerations (r = 0.68 to 0.85) and the number of accelerations (r = 0.06 to 0.80). There was no agreement between computerised interpretation and expert interpretation for baseline variability (kappa = 0.00 to 0.34).

A similar observational study conducted in Italy (1996) compared interpretations of 63 FHR tracings made by two experts (obstetric consultants), two non-experts (obstetricians with 1 year of experience) and a computerised system.497 [EL = III] The study population included women with pregnancy complications and preterm labour. ‘Randomly’ selected 25 minute sections of tracing were used for analysis. Reliability between expert and non-expert observers for FHR, baseline variability, number of accelerations and number of decelerations was fair to good (kappa ratings ranging from 0.38 to 0.67). Only 17 tracings included decelerations. Agreement regarding type of deceleration was poor (kappa = 0.05). Agreement between computerised interpretation and observers was fair to good for most ratings of variability (kappa = 0.16 to 0.74), number of accelerations (0.37 to 0.64) and number of decelerations (0.41 to 0.54). Agreement for FHR baseline and type of decelerations was poor (kappa = 0.18 to 0.48 and kappa = 0.01 to 0.25, respectively).

A UK retrospective observational study assessed the ability of a computerised system for FHR tracing analysis to predict fetal acidosis at birth.498 [EL = III] Analysis was undertaken of 73 complete FHR tracings for labours lasting more than 3 hours. An umbilical artery pH of < 7.15 was used to define acidosis at birth. Using this definition, 8/73 babies (11%) were found to have acidosis and 65 (89%) were classified as normal. The computer system classified 50 babies (69%) as normal, of whom 49 (98%) had an umbilical artery pH > 7.15. Of the 23 babies (31%) identified by the computer system as having acidosis, 7 (30%) had a pH < 7.15. The overall accuracy of the computer system was 77%, with a sensitivity of 88% and a specificity of 75%. Similar calculations were performed for base excess, with < −8 mmol/l as the cut-off point. Fifty-six of the 73 babies (77%) had a normal base excess and 17 (23%) were classified as abnormal. The computer system identified 50 (69%) babies as normal, 46 (92%) of whom had a base excess of ≥ −8 mmol/l. Of the 23 babies (31%) classified by the computer system as abnormal, 13 (57%) had a base excess < −8 mmol/l. The overall accuracy was 81% with a sensitivity of 76% and a specificity of 82%.

A retrospective observational study (Denmark, 1988) compared interpretations of FHR tracings made by four experienced obstetricians with those made by a computerised system.499 [EL = III] 50 FHR tracings of the last 30 minutes of the first stage of labour were used for the study. These were classified as either normal or abnormal. The obstetricians were informed of the number of compromised babies within the sample (n = 16), the criterion by which a baby was judged to be compromised and the length of the pregnancies. Babies were considered to be compromised if the 1 minute Apgar score was < 7, the umbilical artery pH was < 7.15 or the standard base excess was < −10 mEq/l, or primary resuscitation was needed. Based on the 30 minute segment of FHR tracing, the computer system was able to indicate whether a baby would be born in a healthy state or compromised with 86% accuracy. However, while the system has a high specificity (94%), positive predictive value (85%) and negative predictive value (86%), its sensitivity is quite low (69%), i.e. it did not identify five of the 16 compromised babies. This was higher than that obtained from the four obstetricians, the best of whom achieved the same degree of sensitivity but only 59% specificity, i.e. correctly identifying 20 of the 34 healthy babies from their FHR tracing.

A retrospective observational study compared FHR tracing interpretations of 12 clinical experts with computerised interpretation (UK/Hong Kong, 1997).500 [EL = III] Sixty 40 minute sections were classified to determine the baseline FHR. There was high concordance between expert ratings and between computer interpretation and that of experts (r > 0.9). The 95% confidence interval for the difference between computer and expert ratings was −12 to 15 bpm compared with −10 to 10 bpm for the difference between experts.

10.9.5. Evidence statement

Computerised systems have not been demonstrated to be superior to expert interpretation of the FHR trace and no comparisons have been undertaken with routine care.

10.9.6. Research recommendation on computerised system

19.

Further study investigating computerised expert systems should be undertaken

10.10. Record keeping for electronic fetal monitoring

10.10.1. Review question

How should record keeping be carried out for electronic fetal monitoring?

For further details on the evidence review protocol, please see appendix E.

10.10.2. Description of included studies

No evidence was identified that addressed this question.

10.10.3. Evidence statements

No evidence was identified that addressed this question.

10.10.4. Health economic profile

No published economic evaluations were identified for this question.

10.10.5. Evidence to recommendations

10.10.5.1. Relative value placed on the outcomes considered

In addressing this question, the guideline development group hoped to identify evidence which would show the most effective method for record keeping when performing electronic fetal monitoring. In particular, the group had hoped to identify the most effective way of keeping comprehensive and useful notes without imposing an unnecessary administrative burden on healthcare professionals which might be detrimental to the woman's care. The group was aware that there was a view amongst some healthcare professionals that the recommendations from the previous version of the guideline had led to unnecessary duplication of notes.

Unfortunately, no studies were available which addressed this topic.

10.10.5.2. Consideration of clinical benefits and harms

The guideline development group recognised that the primary intention of medical records is to ensure that all relevant aspects of the woman's care are recorded in order to ensure that she is provided with optimum care, particularly with respect to handover of care from one member of staff to another. They recognised that the records can also be used in medico-legal cases as a record of all of the decisions that were made, and the time that all actions were performed. They felt that it was this second purpose which had been a driver for more comprehensive record keeping and the subsequent duplication of information on both the woman's notes and the CTG trace. The information recorded during the intrapartum period can be of vital importance in these cases, and so it was felt likely that there was a move to record information in 2 places to ensure that if one document was lost, a second document would contain all of the relevant details.

The group agreed that although this is important, the primary focus should always be the care of the woman at the time at which she is receiving care. They noted that there is a recommendation from the previous guideline which states that the CTG trace should be stored securely with the woman's medical records, and that if this is followed, there should be less of a need to duplicate notes.

From their clinical experience, the group members were aware that there can be a large amount of information which needs to be recorded, even during simple labours. If all of this information has to be recorded on both the CTG trace and in the woman's notes, this can result in a significant amount of time spent focusing on writing, rather than on caring directly for the woman. This can be a greater problem in more complicated labours where the number of events which need to be recorded can mount up very quickly.

It was recognised that the recommendations from the previous guideline had placed an emphasis on recording intrapartum events which may affect the CTG trace, and noting the time at which they occurred. This is to make it easier to identify which events were likely to have caused a change in the fetal heart rate, and which actions were undertaken as a response to a changing fetal heart rate. The group agreed that this remains an important principle of record keeping and should be continued.

Taking all of the above into account, the group ultimately agreed that rather than trying to be prescriptive about the precise method of record keeping that should be undertaken, it was more appropriate to set out some important general principles that should be applied in all cases, and then recommend that individual units develop their own systems for recording notes. It would be at their discretion whether this was done in the woman's notes or on the CTG trace. However, it would then be the responsibility of each unit to ensure that all of its staff used a consistent approach.

10.10.5.3. Consideration of health benefits and resource uses

The guideline development group did not feel that there was a large issue about resource use for this question, although the group noted that duplication of work without an evident clinical benefit was not a good use of resources.

10.10.5.4. Quality of evidence

No studies were identified which addressed this question and so the guideline development group developed recommendations based on their clinical expertise and judgement.

10.10.5.5. Other considerations

The guideline development group noted that the previous recommendation about record keeping in relation to fetal heart rate monitoring did not include mention of the maternal pulse. The group agreed that this was a valuable piece of information to have recorded at the start of monitoring and so amended the recommendations accordingly.

10.10.6. Recommendations

152.

To ensure accurate record keeping for cardiotocography:

  • make sure that date and time clocks on the cardiotocograph monitor are set correctly
  • label traces with the woman's name, date of birth and hospital number or NHS number, the date and the woman's pulse at the start of monitoring. [new 2014]
153.

Individual units should develop a system for recording relevant intrapartum events (for example, vaginal examination, fetal blood sampling and siting of an epidural) in standard notes and/or on the cardiotocograph trace. [new 2014]

10.11. Risk management in monitoring babies in labour

10.11.1. Introduction

Obstetric litigation is expensive because of the number of cases and the costs of each case. The majority of obstetric litigation claims revolve around FHR trace abnormalities and interpretation. Litigation can ensue many years after alleged harm has been suffered. In order to provide a fair assessment of a case for all parties, FHR traces need to be available and as much information as possible obtained about the causes of poor outcome.

10.11.2. Storage of FHR traces

10.11.2.1. Description of included studies

This was reviewed in the EFM guideline.460 No new studies were identified.

10.11.2.2. Evidence statement (from the NICE EFM guideline)

Storage of FHR traces is complicated due to issues of security, retrieval, space and conservation. FHR traces related to an adverse outcome for mother or baby are more likely to go missing. The quality of some FHR traces deteriorates over time. This could be due to a number of factors including poor quality storage, paper, intense heat, light or moisture.

10.11.3. Recommendations on risk management in monitoring babies in labour

154.

Keep cardiotocograph traces for 25 years and, if possible, store them electronically. [2007, amended 2014]

155.

In cases where there is concern that the baby may experience developmental delay, photocopy cardiotocograph traces and store them indefinitely in case of possible adverse outcomes. [2007, amended 2014]

156.

Ensure that tracer systems are available for all cardiotocograph traces if stored separately from the woman's records. [2007, amended 2014]

157.

Develop tracer systems to ensure that cardiotocograph traces removed for any purpose (such as risk management or for teaching purposes) can always be located. [2007, amended 2014]

Copyright © 2014 National Collaborating Centre for Women's and Children's Health.
Bookshelf ID: NBK328255