Fetal blood sampling in addition to intrapartum ST-analysis of the fetal electrocardiogram: evaluation of the recommendations in the Dutch STAN® trial

Authors


Dr JH Becker, Department of Obstetrics and Gynaecology, University Medical Centre Utrecht, Location WKZ, KE 04.123.1, PO Box 85090, 3508 AB Utrecht, the Netherlands. Email j.h.becker@umcutrecht.nl

Abstract

Please cite this paper as: Becker J, Westerhuis M, Sterrenburg K, van den Akker E, van Beek E, Bolte A, van Dessel T, Drogtrop A, van Geijn H, Graziosi G, van Lith J, Mol B, Moons K, Nijhuis J, Oei S, Oosterbaan H, Porath M, Rijnders R, Schuitemaker N, Wijnberger L, Willekes C, Visser G, Kwee A. Fetal blood sampling in addition to intrapartum ST-analysis of the fetal electrocardiogram: evaluation of the recommendations in the Dutch STAN® trial. BJOG 2011; DOI: 10.1111/j.1471-0528.2011.03027.x.

Objectives  To evaluate the recommendations for additional fetal blood sampling (FBS) when using ST-analysis of the fetal electrocardiogram.

Design  Prospective cohort study.

Setting  Three academic and six non-academic teaching hospitals in the Netherlands.

Population  Labouring women with a high-risk singleton pregnancy in cephalic position beyond 36 weeks of gestation.

Methods  In labouring women allocated to the STAN® arm of a previously published randomised controlled trial who underwent one or more FBS during delivery, we assessed whether FBS was performed according to the trial protocol and how fetal acidosis, defined as an FBS pH < 7.20, was related to ST-waveform analysis.

Main outcome measures  The number of FBS showing fetal acidosis, related to the different STAN® criteria where additional FBS is recommended.

Results  Among 2827 women monitored with STAN®, 297 underwent FBS, of whom 171 (57.6%) were performed according to the predefined criteria and 126 were performed in absence of these criteria. In the first group, rates of fetal acidosis (pH < 7.20) were two of 18, none of nine, 12 of 111 and three of 33 when FBS was taken for abnormal cardiotocogram (CTG) at the start, intermediary CTG at the start, abnormal CTG >60 minutes, and poor electrocardiogram quality, respectively. When the predefined criteria were not met and ST-analysis showed no ST-events, only two incidents of fetal acidosis were seen.

Conclusions  The performance of FBS is valuable in the advised STAN® criteria. When these criteria are not met, performance of FBS does not seem helpful in the detection of fetal acidosis.

Introduction

Intrapartum fetal monitoring aims to identify fetuses at risk for neonatal and long-term injury caused by asphyxia. To serve this purpose, cardiotocography (CTG) was introduced in the 1960s. Although CTG is widely applied, it is neither very sensitive nor specific for the detection of fetal acidosis, but comes at the cost of a significant increase of the operative delivery rate.1–8 The addition of fetal blood sampling (FBS) to the CTG produces slightly better results at a lower intervention rate.6 However, performance of FBS is invasive, has to be repeated when CTG abnormalities persist and some studies report up to 21% failures because of technical problems.9 As with any other form of intrapartum surveillance, it still does not guarantee prevention of asphyxia.10

In recent years, ST-analysis of the fetal electrocardiogram (ECG; STAN®, Neoventa Medical AB, Mölndal, Sweden) has been introduced, combining standard CTG with computerised ST-waveform analysis of the fetal ECG during labour. Several studies have identified changes in the ST-segment and T-wave of the fetal ECG in response to fetal hypoxia.11–15 These changes are interpreted together with the CTG, according to the STAN® clinical guidelines.

Five randomised trials have been performed to compare intrapartum fetal monitoring by CTG plus ST-analysis of the fetal ECG with CTG only.16–20 Two large studies have shown that monitoring by CTG plus ST-analysis decreases the incidence of metabolic acidosis and the number of instrumental deliveries for fetal distress, as compared with CTG alone.16,17 However, in both studies the FBS was equally performed in the two study groups. In two smaller trials, the improvement in neonatal outcome as reported in the two largest trials could not be confirmed, but in both trials the incidence of FBS in the CTG plus ST-analysis group was significantly lower.18,19 Recently, we reported that fetal monitoring by CTG plus ST-analysis of the fetal ECG, compared with CTG only, using a strict protocol for performance of FBS,20 reduced the number of newborns with (metabolic) acidosis, with a reduction of FBS from 20 to 10%.21

An important issue in intrapartum fetal monitoring concerns false-negative test results, leading to the birth of an infant with adverse outcome. In the use of the STAN® technique, such false negatives have also been reported.22–24 Following these reports, in January 2007 European experts on ST-waveform analysis formulated consensus-based criteria for CTG classification and ST-waveform interpretation. Existing STAN® clinical guidelines were adjusted, now including recommendations for situations in which additional fetal information, such as by FBS, is needed to prevent false-negative results.25

As these STAN® criteria are not based on empirical data, we evaluated all women monitored by ST-analysis of the fetal ECG in the previously mentioned Dutch trial20 in which at least one FBS had been performed. This study aims to evaluate the current recommendations for additional FBS when using ST-analysis of the fetal ECG.

Methods

Population and measurements

We used data from women monitored in the STAN® arm of a previously published multicentre randomised controlled trial20,21 among women in labour with a high-risk singleton pregnancy in cephalic position at term (ISRCTN95732366). From January 2006 until July 2008 participants had been randomly assigned to monitoring by CTG combined with ST-analysis of the fetal ECG (index group) or CTG without ST-analysis (control group). The study was approved by the Institutional Review Board of the University Medical Centre Utrecht and had local approval from all other participating hospitals. The present study focused on the women randomised to the index group in whom FBS was undertaken.

In women in the index group, a scalp electrode was applied to the fetal head and connected to a STAN® S21 or S31 fetal heart monitor (Neoventa Medical, Gothenburg, Sweden). Both monitors used the same ST-analysis algorithms. Clinical management was guided by the STAN® clinical guidelines (see Appendix S1). In the study protocol FBS was advised in three situations: (1) start of STAN® registration with an intermediary or abnormal CTG trace; (2) abnormal CTG trace for more than 60 minutes without ST-events; and (3) poor ECG signal quality in the presence of an intermediary or abnormal CTG trace. Poor signal quality was defined as absence of ST-information for more than 4 minutes or less than one average ECG-complex per minute within a period of 10 minutes. Except for the advice to perform FBS in the case a STAN® recording started with an intermediary CTG, the study protocol was identical to the revised STAN® guidelines of 2007.23 Immediate delivery was recommended if FBS showed a pH < 7.20. If the pH was between 7.20 and 7.25 the advice was to repeat FBS after 30 minutes. If the pH was >7.25, the fetal condition was considered well enough to continue labour. Apart from the FBS criteria, immediate delivery was also indicated if STAN® abnormalities occurred as defined in the protocol.

Analysis of women undergoing fetal blood sampling

ECG recordings

For the present analysis, two observers (JB and KS) independently assessed all STAN® recordings of women in the index group in which at least one FBS was performed. They examined whether or not additional FBS was performed according to the trial protocol. In the case of discordant observations, the opinion of a third observer (MW) was decisive.

The STAN® recordings were presented digitally (STAN®viewer (R21) version 2.0.2484.603; Neoventa Medical AB, Mölndal, Sweden) to the observers using a paper speed of 2 cm/minute as this is the paper speed routinely used in the Netherlands. The observers were only provided with information on the timing of FBS, without knowledge of its result, other clinical parameters obtained during labour, or the neonatal outcome.

For each FBS the following items had to be scored: (1) classification of the CTG as normal, intermediary, abnormal or (pre)terminal within a 60-minute period before performance of FBS; (2) duration of an intermediary, abnormal or (pre)terminal CTG in minutes; (3) interpretation of any ST-events; and (4) judgement of whether FBS was performed according to the randomised controlled trial protocol. If so, the reason for FBS had to be defined based on one of the situations in the protocol.

Evaluation of assessment by observers

For the FBS performed according to the trial protocol, the relation between pH result measured by FBS and the reason to perform FBS was described. In the samples that were not performed according to the randomised controlled trial protocol (i.e. protocol violation), the relation between pH results of FBS and ST-waveform interpretation regarding fetal indications to intervene, was evaluated. Fetal acidosis was defined as an FBS pH < 7.20.

Women were classified as being treated ‘not according to trial protocol’ if at least one of the FBS was not performed according to the trial protocol. This classification at patient level instead of FBS level allowed the description of observer results in relation to neonatal outcome. For neonatal outcome, metabolic acidosis was defined as an umbilical cord artery pH < 7.05 and base deficit in extracellular fluid (BDecf) >12 mmol/l.

Results

In the trial, we randomised 5681 women, 2827 to the index group and 2840 to the control group, whereas 14 did not meet the inclusion criteria (Figure 1). Of the 2827 women allocated to the index group of the trial, at least one FBS was performed during 301 (10.6%) deliveries. STAN® recordings for assessment by the observers were complete for 224 (74.4%) of these deliveries. The remaining 77 women (25.6%) were excluded from analysis for the following reasons: FBS was performed before the start of a STAN® recording (n = 27), STAN® recordings were missing or incomplete because of technical failures (n = 32) or STAN® recordings were interrupted before FBS was performed (n = 18). Among the 77 women excluded from analysis there were no incidents of neonatal metabolic acidosis.

Figure 1.

 Flowchart.

In the 224 available women, 314 FBS were performed. As a result of the partly incomplete STAN® recordings in women with multiple FBS, 17 (5.4%) samples could not be analysed. In total, 297 FBS were available for analysis (Figure 1).

After protocol adherence was evaluated by two observers, 47 (15.8%) of 297 observations were discordant and therefore the opinion of a third observer was decisive. According to the observers, 171 (57.6%) FBS were performed according to the trial protocol; the remaining 126 (42.4%) samples were not performed according to protocol (i.e. protocol violation). We were not informed of the reasons for FBS performed not according to the trial protocol as this was not documented. Failed FBS procedures with missing pH results occurred in 4.7% (Table 1).

Table 1.   FBS in deliveries monitored by ST-analysis of the fetal ECG related to the trial protocol
 TotalAccording to trial protocol*Not according to trial protocol**
  1. *In the study protocol FBS was advised in three situations: (1) start of STAN®-registration with an intermediary or abnormal CTG trace; (2) abnormal CTG trace for more than 60 minutes without ST-events; (3) poor ECG signal quality in the presence of an intermediary or abnormal CTG trace.

  2. **FBS performed on indication of the clinician although the above-mentioned conditions were not met (protocol violation).

Number of FBS297171126
FBS result, n (%)
pH > 7.25208 (70.0)112 (65.5)96 (76.2)
pH 7.20–7.2548 (16.2)33 (19.3)15 (11.9)
pH < 7.2027 (9.1)17 (9.9)10 (7.9)
Missing pH14 (4.7)9 (5.3)5 (4.0)

Fetal blood sampling according to trial protocol

In 17 of the 171 (9.9%) FBS performed according to the trial protocol, fetal acidosis was present (Table 1). This occurred in 12 of 111 (10.8%) samples performed because of an abnormal CTG beyond 60 minutes without ST-events, in two of 18 (11.1%) samples because of an abnormal CTG at the start of a STAN® recording and in three of 33 (9.1%) samples because of poor ECG signal quality combined with an abnormal CTG (Table 2).

Table 2.    FBS in deliveries monitored by ST-analysis of the fetal ECG related to reasons according to the trial protocol
 Reason to perform FBS according to trial protocol
TotalAbnormal CTG at startIntermediary CTG at startAbnormal CTG > 60 min without ST-eventPoor ECG signal quality
No. of FBS (%)171 (100)18 (10.5)9 (5.3)111 (64.9)33 (19.3)
FBS result, n (%)
pH > 7.25112 (65.5)9 (50.0)9 (100)69 (62.2)25 (75.8)
pH 7.20–7.2533 (19.3)5 (27.8)0 (0.0)24 (21.6)4 (12.1)
pH < 7.2017 (9.9)2 (11.1)0 (0.0)12 (10.8)3 (9.1)
Missing pH9 (5.3)2 (11.1)0 (0.0)6 (5.4)1 (3.0)

Fetal blood sampling not according to trial protocol

In 10 of the 126 scalp samples not performed according to the trial protocol fetal acidosis was found (7.9%) (Table 1). In eight of these women, intervention based on a significant ST-event was indicated before FBS was performed. In the remaining two women FBS indicated fetal acidosis in the absence of a significant ST-event. In both women an abnormal CTG for <60 minutes with rapid deterioration of fetal heart rate variability had been present. As a secondary finding, we found that in one of these two women a significant ST-event occurred 25 minutes after FBS showed a pH of 7.13 during the first stage of labour. In the other woman, there were no ST-events and FBS was performed in the second stage of labour, and showed a pH of 7.18; part of that recording during the second stage of labour is shown in Figure 2. For both women, the neonatal outcome was favourable. For the FBS that were taken not according to the protocol, the relation between the presence of significant ST-events or a preterminal CTG pattern within 60 minutes before the FBS and the result of the FBS (i.e. normal, intermediate or abnormal) is shown in Table 3.

Figure 2.

 Recording from a woman in labour showing rapid deterioration of the fetal heart rate variability without ST-events.

Table 3.   Relation of presence or absence of significant ST-events and preterminal CTG with results of FBS not taken according to protocol (n = 126)
 pH < 7.20pH 7.20–7.25pH > 7.25Missing valueTotal
  1. *Although there was no indication to intervene at the time of FBS, a significant ST-event occurred 25 minutes after FBS.

Indication to intervene (at least one significant ST-event), n (%)8 (23.5)5 (14.7)19 (55.9)2 (5.9)34
No indication to intervene, n (%)2* (2.2)10 (10.9)77 (83.7)3 (3.2)92
Preterminal CTG, n (%)1 (100)0001

Neonatal outcome

Of 224 deliveries available for analysis, 123 (54.9%) were treated according to the trial protocol and 101 (45.1%) were not. Among the women where FBS was taken according to the trial protocol, three delivered a neonate with metabolic acidosis at birth. In one of these women the CTG had been abnormal for only 36 minutes in combination with poor ECG signal quality before FBS, with pH 7.19, was performed. In the other two women FBS was performed because of CTG abnormalities for more than 60 minutes and these were normal, but CTG abnormalities persisted between this normal FBS and birth. In one woman the time between FBS and birth was only 20 minutes with poor signal quality (result FBS pH 7.24, umbilical cord artery pH 7.00), whereas in the other the time between FBS and birth was 9 hours with an abnormal CTG for the last 115 minutes (FBS pH 7.32, umbilical cord artery pH 6.93).

In the group where FBS was performed not according to the trial protocol, metabolic acidosis at birth was found in three infants as well. In all three women earlier intervention was recommended based on significant ST-events. In one of these women multiple FBS were performed because of an abnormal CTG-pattern (pH 7.38, 7.33, 7.31, 7.28 and 7.28). The final two FBS were both preceded by a significant ST-event. Abnormalities on CTG persisted hereafter and ST-analysis showed one more significant ST-event 76 minutes after the last FBS, during the second stage of labour. Time between the last FBS and delivery was 114 minutes. After a failed vacuum extraction, the baby was born by a caesarean section that showed a uterine rupture. The baby was born with an umbilical cord pH of 6.96 and BDecf 15 mmol/l and died of severe asphyxia and encephalopathy.

Discussion

This study provides empirical data with respect to intrapartum situations that require additional assessments in women monitored by ST-analysis of the fetal ECG. When additional FBS was applied according to additional criteria it revealed acidosis in almost 10% of women (17/171). When FBS was performed in the absence both of these criteria and of significant ST-events, only two incidents of acidosis were detected. These results largely support the recently adjusted STAN® guidelines.25 This is also illustrated by Table 3 which shows that in FBS taken not according to the protocol and where ST-analysis showed no significant events, acidosis was seen in just 2.2% (2/92), whereas acidosis was seen in as much as 23.5% (8/34) in the presence of significant ST-events.

However, Table 3 also shows that the majority of FBS, taken in the presence of significant ST-events, show a pH > 7.25. As a consequence, the question arises whether performance of FBS may prevent the clinician from conducting delivery unnecessarily when significant ST-events are present during the first stage of labour and when there are no possible causes of fetal hypoxia that could be alleviated. However, because the performance of FBS in this situation was not recommended, it is likely that there was bias in the clinicians’ decision to perform FBS in these situations, and as a result this finding has to be interpreted with caution.

There are some limitations of this study that need to be addressed. First, we had to exclude a relatively large number of women in whom at least one FBS was performed (25%) for different reasons. However, there was no metabolic acidosis among these exclusions. Second, we only analysed women in which FBS was performed. It is therefore unknown to what extent criteria with respect to performing FBS were indeed followed in women without FBS. Although we speculate that the need to perform FBS in addition to ST-analysis could be further reduced when criteria are followed more strictly, this cannot be thoroughly concluded without analysing all deliveries in which no FBS was performed. Furthermore, we defined a threshold for fetal acidosis to occur at a scalp pH < 7.20. Although this threshold represents the lower limit of normal fetal intrapartum capillary blood values as described by Beard et al.26 in the late 1960s, it is questionable whether this can be correlated to ST-analysis of the fetal ECG. As the latter informs us about the ability of the fetal myocardium to respond to hypoxia, and FBS informs us about acidosis, the results of both techniques are not completely comparable. However, although a scalp pH is not a standard measure, in clinical practice it is used as a reference test with pH of 7.20 as the lowest acceptable value for continuation of expectant management. In addition, as mentioned in the Methods section, when STAN® recordings were evaluated the observers were only provided with information on the timing of FBS, without knowledge of its result, other clinical parameters obtained during labour, and of the neonatal outcome. The observers, however, might be indirectly informed of the result of the FBS because a low pH would probably have led to delivery (and therefore the end of the registration) and an intermediate pH would probably have led to another FBS. The nature of this study made this unavoidable and we do not believe that this has led to significant bias. Lastly, we would like to stress the fact that the evaluation was performed using data from a randomised controlled trial. Although unavoidable for solid research purposes, it is a situation far from everyday practice conditions and the results should be judged appropriately.

The addition of the category ‘abnormal CTG for more than 60 minutes with normal ST’, which was added to the latest STAN® guidelines as a criterion that may require additional fetal information, seems valuable, because 12 neonates with fetal acidosis were detected in this way. However, 62.2% of scalp samples performed because of an abnormal CTG trace for more than 60 minutes without ST-events, showed a pH > 7.25. One may wonder whether, in cases of an abnormal CTG without ST-events, the threshold for assessment of fetal wellbeing in addition to ST-analysis might be raised to 90 minutes. Expectant management for 90 instead of 60 minutes of abnormal CTG without ST-changes would, in our study, have led to four missed incidents of fetal acidosis (which is an additional 1.8%), assuming that no ST-events would have appeared in the additional 30 minutes (data not shown). Consequently, we feel that a period of 60 minutes of abnormal CTG without ST-events is the maximal period.

In our trial protocol the category ‘intermediary CTG at the start of registration’ for assessment of fetal condition at the start of a STAN® recording was incorporated to evaluate whether in these cases ST-waveform analysis would be reliable. This extension of the STAN® criteria has not been proved to be useful in detecting fetal hypoxia. Although based on only nine women, we do not advise addition of this category to the STAN® clinical criteria as a recommended situation for the assessment of fetal wellbeing.

The results of our study showed that in the group of FBS that was not performed according to the trial protocol, fetal acidosis was present in 8%, of which eight out of ten were detected by significant ST-events before FBS was taken and one shortly after. The remaining neonate without ST-events may be considered a ‘false-negative’. In this case, an abnormal CTG for <60 minutes with rapid deterioration of fetal heart rate variability had occurred.

Although the existing criteria do mention rapid deterioration of fetal heart rate pattern in combination with an abnormal CTG and normal ST as an indication to perform FBS, heart rate variability is not mentioned. It is not entirely clear which criterion would fulfil the definition of ‘rapid deterioration of the fetal heart rate pattern’. Although the interpretation of a CTG will always be prone to subjectivity and debate among clinicians, we feel that the fetal heart rate variability is a more objective parameter to assess a deterioration of the CTG. Therefore, we would encourage addition of the category ‘abnormal CTG <60 minutes with rapid deterioration of fetal heart rate variability with normal ST-waveform analysis’ to the STAN® clinical guidelines as a required situation for qualified assessment and checking for non-deteriorating fetal condition. This may make STAN® users more alert to the assessment of fetal heart rate variability and further reduce the false-negative rate.

Overall, in 10% of women in the index group of our large randomised trial FBS was performed.21 Given the fact that in more than 40% of women FBS was not performed according to the STAN® criteria, we speculate that this incidence may further be reduced when STAN® clinical criteria are followed more strictly. The reason for this large proportion of FBS performed not according to protocol could not be studied, but lack of trust in a new method of fetal surveillance might have played a role. Of all 224 women with FBS analysed in this study, only six had metabolic acidosis at birth, of which three had been identified by ST-waveform analysis before the moment of FBS. In previous studies, evaluating the relation between scalp pH and ST-waveform analysis, it was also found that a substantial number of FBS were preceded by ST-events and so were performed to verify the significance of CTG plus ST abnormalities.27,28 As FBS can be a time-consuming procedure, hampered by technical and logistic problems and failed attempts, this may cause a serious delay in delivery, especially during the second stage of labour and so should no longer be performed in the presence of significant ST-events during the second stage of labour.

Whereas in most western European countries the performance of FBS is incorporated in daily obstetric practice, in the USA it is not. The above-mentioned drawbacks of FBS could well be the explanation for this major transatlantic difference. The National Institutes of Health are in the preparation phase of a randomised controlled trial comparing intrapartum ST-analysis of the fetal ECG in combination with CTG with CTG alone as a multicentre trial across the USA (NCT01131260). It will be very interesting to compare these results with the European trials conducted so far in the light of these transatlantic differences regarding the performance of FBS.

Finally, to increase the external validity of these results, we encourage a comparable analysis in other trials.

Conclusion

In our study the consensus-based recommendation in the STAN® criteria to additionally check fetal condition by FBS in the case of an abnormal CTG beyond 60 minutes without ST-changes, has been effective for early identification of fetal acidosis. Consequent adherence to the STAN® criteria may further decrease the necessity for FBS in addition to ST-analysis of the fetal ECG.

Disclosure of interests

All authors declare that they had no potential conflicts of interest relevant to this article.

Contribution to authorship

AK initiated and coordinated the original study. All authors actively participated in the interpretation of the results and revision of the original study, as well as the current paper, which was written by JB, MW, GV, KM, BWM and AK. Statistical analysis has been conducted by JB and KS.

Details of ethics approval

The original study was approved by the Institutional Review Board of the University Medical Centre Utrecht on 5 August 2005, approval number: 05-157, and had local approval from all other participating hospitals.

Funding

The original Dutch STAN® trial was funded by a grant from ZonMW, the Dutch Organization for Health Research and Development (grant number: 945-06-557).

Acknowledgements

We thank all research nurses and midwives of the Dutch Obstetric Consortium, as well as the staff of the labour wards of the participating centres for their invaluable contributions to the original study.

Ancillary