SEARCH

SEARCH BY CITATION

Keywords:

  • STAN ;
  • ST analysis of the fetal ECG;
  • intrapartum fetal monitoring;
  • cardiotocography;
  • fetal electrocardiogram

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. ST-analysis – the clinical perspective
  5. To what extent does ST-analysis fulfil these demands?
  6. References

Fetal electrocardiogram waveform analysis has been studied for many decades, but it is only in the last 20 years that computerization has made real-time analysis practical for clinical use. Changes in the ST segment have been shown to correlate with fetal condition, in particular with acid–base status. Meta-analysis of randomized trials (five in total, four using the computerized system) has shown that use of computerized ST segment analysis (STAN) reduces the need for fetal blood sampling by about 40%. However, although there are trends to lower rates of low Apgar scores and acidosis, the differences are not statistically significant. There is no effect on cesarean section rates. Disadvantages include the need for amniotic membranes to be ruptured so that a fetal scalp electrode can be applied, and the need for STAN values to be interpreted in conjunction with detailed fetal heart rate pattern analysis.

Abbreviations
CTG

cardiotocography

ECG

electrocardiogram

RR

relative risk

Key Message

Automatic ST analysis of the fetal electrocardiogram in labor reduces the need for fetal blood sampling but in randomized trials has no other statistically significant benefit and does not reduce the cesarean section rate.

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. ST-analysis – the clinical perspective
  5. To what extent does ST-analysis fulfil these demands?
  6. References

The first recording of the fetal electrocardiogram (ECG) was made by Cremer in 1906 [1], using a string galvanometer. This was achieved with one electrode on the mother's abdomen, and another in the vagina. However, the fetal signal was tiny compared with that of the mother's heart, and it was not until the early 1950s that a fetal signal large enough to be analyzed was obtained using a silver wire electrode passed into the amniotic sac [2]. However, despite improvements in instrumentation, the detection of the fetal ECG noninvasively using maternal abdominal leads remained difficult [3, 4]. In the 1960s the introduction of fetal scalp electrodes that could be applied via the vagina made recording and analysis of the fetal ECG more acceptable. Malcolm Symonds in Nottingham reported in 1971 a study of the fetal ECG during labor in relation to maternal and fetal acid–base status, in which he was able to make measurements of fetal ECG morphology between contractions of the uterus [5]. However, he was unable to predict fetal condition at birth reliably by measurement of ECG parameters from individually recorded complexes. To improve assessment, his group subsequently developed a real-time computer system to improve the signal-to-noise ratio by averaging complexes over a 20-min period [6]. They commented that “certain changes in the fetal ECG, previously thought to be characteristic of fetal distress, occurred in normal patients”. More specifically, they said that the “ST segment displacement displayed no significant changes”. This conclusion was reinforced in a publication from the same group some 11 years later, authored by Henry Murray, when he reported from a study of 155 women during labor that “the absolute position of the ST segment above or below the isoelectric line did not correlate with any particular fetal condition” [7]. However, he did find that the analysis of the PR/RR interval, when taken in conjunction with the ST segment, could distinguish groups with an umbilical artery pH above and below 7.25.

The early work of Rosen and his group, first published in 1975, was carried out using guinea pigs, cats and sheep [8]. They reported from their animal work, in contrast to the studies in humans by Symonds and his group, that “there are progressive changes in the ST interval as an early sign of hypoxia”. They continued for 10 years to refine their methodology using animals, and their first publication studying the human fetus was in 1985 [9]. Obtaining an interpretable ECG waveform required the use of two fetal scalp electrodes (one spiral and one Copeland), and a silver/silver chloride wire electrode wound around an intrauterine pressure catheter tip. Measurements of the T-wave and QRS complex were expressed as the T:QRS ratio, thereby correcting for changes in signal gain. They studied ST waveforms from 46 fetuses during labor, although there were times when low-frequency shifts in the baseline prevented measurement. In 21 infants, they measured umbilical vein lactate levels. They reported a positive linear correlation between the T/QRS ratio and the lactate values, with an r-value of 0.58, p < 0.01. However, analysis of their published graph reveals only 19 data points, and the significance of the correlation coefficient relies mainly on a single outlier with a very high T:QRS ratio. Re-analysis of their data with coordinates estimated from the published graph shows that removal of this single outlier reduces the r-value to 0.345, and the p-value is no longer significant, 0.16.

The following year, a new research fellow in the Symonds group, Howard Jenkins, published a study comparing the ECG waveform in 14 normal fetuses with that in 10 babies who became acidotic [10]. In contrast to the conclusions of Henry Murray, Jenkins reported a “highly significant correlation between both a long-term increase in ST segment and T-wave height and acidosis”. However, this group continued to emphasize the importance of the PR interval assessment, and commented in a separate publication in 1996 that the “addition of PR interval assessment would potentially reduce the numbers of normal fetal blood samples being carried out from 85.5% to 26.8% and the proportion of cases of missed acidosis at delivery from 8.5% to 4.5%.” [11].

Meanwhile, Rosen had qualified the interpretation of the ST segment by stating that “the T/QRS ratio is only one parameter to be used – equally important is to identify the occurrence of ST depression with biphasic negative T-waves, and to interrelate the CTG [cardiotocography] and the ST waveform as outlined in the clinical guidelines” [12]. Analysis of the ECG waveform by the attending clinician, in conjunction with fetal heart rate pattern interpretation, was subjected to a prospective randomized trial in 2400 labors in Plymouth, reported in 1993 by Westgate et al. [13]. In the ECG group, there was a 46% reduction in operative delivery for fetal distress (p < 0.001), and a trend to fewer babies being born with low Apgar scores or metabolic acidosis (although these latter changes were not statistically significant). A second randomized trial, based in Sweden and using computerized sequential analysis of 30 aggregated ECG complexes (STAN®), was reported in the Lancet in 2001 [14]. The authors commented that “the monitor needed at least 20 min to establish a baseline T/QRS ratio”. An intention-to-treat analysis of the 4966 cases enrolled showed a 53% decrease in acidosis (p = 0.02), and a 27% decrease in operative delivery for fetal distress (p = 0.047). However, this trial was unusual in that after an interim analysis of 1600 cases had revealed protocol variations (recommendations to intervene had been disregarded and babies with cord artery metabolic acidosis were being born), retraining of the staff was undertaken. Moreover, in 2007, it was alleged that there had been irregularities in the study [15]. Errors in the analysis were subsequently confirmed by a committee of the Swedish Research Council [16]. A revised modified intention-to-treat analysis confirmed that after correction for errors at data collection there was still a 52% reduction of metabolic acidosis, but the p-value of the difference was now 0.038 [17].

Following the publication of the Swedish trial, there have been a series of observational reports and further randomized trials on the introduction of STAN into routine clinical practice. An observational study in Gothenburg in 2006 reported consistent improvements in fetal outcome without any increase in operative interventions for fetal distress [18], whereas a study of 530 patients in six centers in the USA reported similar findings [19]. A report from the Netherlands in 2009 suggested that information regarding the fetal ECG morphology was particularly useful in cases of intermediary or abnormal CTG traces, and resulted in a more standardized decision to intervene [20]. However, a randomized controlled trial in Finland reported in 2006, showed that STAN did not improve neonatal outcome or decrease the cesarean section rate (although the need for fetal blood sampling during labour was reduced) [21]. Similar findings from a prospective randomized controlled trial in France were reported in 2007 [22]. Failure to predict three cases of intrapartum metabolic acidosis, highlighting the limitations of ST analysis, was reported in 2007 [23]. An article in the same year from St George's Hospital in London reported that in the first 1052 labors monitored using STAN, there were 14 cases of neonatal encephalopathy, and there had been a significant “ST event” in only half (n = 7) [24]. An enquiry established that STAN was being started when the fetal heart rate tracing was already abnormal, and that a reassuring ST analysis was being used to justify allowing labour to continue even when the fetal heart rate pattern was “preterminal” (such as with a prolonged bradycardia). STAN also failed to identify seven (30%) of 23 cases of metabolic acidosis. An article in 2008 reported that in approximately 5000 labors monitored by STAN, only two of three cases with severe, and 20 of 48 cases with moderate, metabolic acidemia were preceded by ST events coinciding with CTG abnormalities [25]. Vayssiere et al. commented in 2009 that “In cases with abnormal CTG, ST analysis may improve consistency in clinical decision-making and decrease unnecessary interventions, but may also lead on rare occasions to unjustified decisions not to intervene” [26]. In 2010, Ragupathy et al. commented that “the high incidence of false-positive STAN events may have contributed to the failure to act when a significant ST event occurred” [27]. In 2011, Doret et al. commented after using STAN in 3195 labors, that “The study clearly highlights that adherence to the STAN guidelines is one of the main difficulties with STAN monitoring, and can dramatically affect neonatal outcomes” [28]. The same group emphasized this in a follow-up article in 2012, commenting that “STAN guideline deviations contribute to an increased operative delivery rate in patients with suspected fetal distress and normal neonatal outcomes” [29].

In May 2013, Neilson published a Cochrane collaboration systematic review of five prospective randomized trials of STAN [30]. This included data from 14 574 babies. He found no difference in the rate of cesarean section between labors monitored by STAN and those monitored only by CTG [relative risk (RR) 0.99, 95% CI 0.91–1.08], nor was there any significant difference in the rates of low Apgar scores at 5 min (RR 0.95, 95% CI 0.73–1.24). Although when STAN was used there were lower levels of severe metabolic acidosis at birth (RR 0.78), neonatal encephalopathy (RR 0.54) and babies requiring neonatal intubation (RR 0.79), in all three cases, the 95% CIs were wide and included 1.0, and were therefore not significant. He did find a significant reduction in the rate of fetal blood sampling (RR 0.61, 95% CI 0.41–0.91), and a modest reduction in the number of operative deliveries (RR 0.89, 95% CI 0.81–0.98) and admissions to the special care baby unit (RR 0.89, 95% CI 0.81–0.99). He concluded that “these findings provide some support for the use of fetal ST waveform analysis when a decision has been made to undertake continuous electronic fetal heart rate monitoring during labor. However, the advantages need to be considered along with the disadvantages of needing to use an internal scalp electrode, after membrane rupture, for ECG waveform recordings” [30]. A similar analysis by Potti and Berghella published in 2012 came to essentially the same conclusions [31]. The Swedish Health Technology Assessment report published in 2012 included only the four randomized trials using STAN in its computerized version [32]. They opined that “there is not enough scientific evidence to conclude that computerized ST analysis reduces the incidence of metabolic acidosis. Cesarean sections and instrumental vaginal deliveries due to fetal distress or other indications are the same, regardless of method… (although) STAN reduces the number of instances which require scalp blood sampling”.

ST-analysis – the clinical perspective

  1. Top of page
  2. Abstract
  3. Introduction
  4. ST-analysis – the clinical perspective
  5. To what extent does ST-analysis fulfil these demands?
  6. References

Clinicians aim to deliver babies in the best condition possible. To accomplish this goal, they need a method for fetal monitoring. Continuous electronic fetal heart rate and uterine contraction monitoring (by CTG) fulfils the requirements of a screening test. It has a high sensitivity, i.e. almost all hypoxic fetuses will have a nonreassuring CTG. Specificity, however, is low because many normal fetuses also have a nonreassuring CTG. Therefore, to identify the true hypoxic fetuses, a diagnostic test is needed as a supplement to the screening test.

The ideal method for fetal monitoring would be simple, noninvasive, and applicable for all women in labor, irrespective of gestational age. Furthermore, the method should satisfy (at least) three criteria/demands/needs/challenges. First, the method should give an early warning, that is when the baby is only pre-acidemic (i.e. with a pH/base deficit outside the normal range, but not abnormal enough to result in permanent damage) to give time for the clinician to deliver the baby without putting the mother at unnecessary risk. Second, system feedback should be clear (“do we have a problem? yes or no”) and easy to follow guidelines should be available. Third, the test should be able to identify the nonhypoxic fetuses so unnecessary intervention can be avoided.

To what extent does ST-analysis fulfil these demands?

  1. Top of page
  2. Abstract
  3. Introduction
  4. ST-analysis – the clinical perspective
  5. To what extent does ST-analysis fulfil these demands?
  6. References

As a consequence of the publication of cases with adverse fetal outcome from deliveries monitored with ST-analysis [23, 24], recommendations for current use of the automatic ST segment analysis system (“STAN”) [33] have been revised. The new guidelines include a “Checklist at start-up”, “CTG classification”, and “STAN clinical guidelines”. These guidelines recommend that intervention is based on both CTG abnormalities and ST-events. When intervention is indicated, according to guidelines, the baby should be delivered within 20 minutes.

Despite the reviewed guidelines, there are important limitations to the use and reliability of STAN. STAN should not be used when gestational age is ≤36 weeks or when maternal bearing down (“pushing”) has already started. STAN monitoring cannot be relied on during the first 20 min after it has been commenced, or if CTG is abnormal at the onset of STAN monitoring, or in cases where there is poor signal quality with gaps in T/QRS ratio of more than 4 min. Furthermore, STAN is thought to be unreliable if intrauterine infection is present. Finally, an abnormal CTG without ST-events does not rule out that the fetus is pre-acidemic.

In most labors, the CTG is normal and ST-events are absent. In the rest, three scenarios are possible. First, ST-events are present and CTG is abnormal: action is indicated. Second, ST-events are present and the CTG is normal: action is not indicated. Thirdly, ST-events are not present and CTG is abnormal: action is not usually indicated but there are exceptions – when the CTG is preterminal, the CTG is abnormal for 60 min, or the CTG is rapidly worsening – action is needed, and the fetus may be acidemic. This means that the combination of information from the CTG trace and STAN is not always simple or straight forward.

Data from a blinded case–control study among newborns with varying degrees of acidemia (cases) and newborns with normal cord artery acid–base status (controls) [25] shows that good signal quality – one of the basic requirements for reliable STAN monitoring – was absent in up to 50% of cases. The presence of ST-events increased the probability of acidosis, but only two-thirds of cases with metabolic acidosis had ST-event(s). Moreover, ST-events with abnormal CTG appeared late in the development of hypoxia. ST-events were also found among a large proportion of controls (50%). Finally, the CTG was frequently misclassified.

Analysis of cases with neonatal metabolic acidosis without significant ST-events has identified some of the pitfalls associated with clinical use of STAN [23, 24]. These pitfalls include poor signal quality, problems following disconnection of STAN registration, misclassification of preterminal CTG, no intervention in cases where the CTG deteriorates, and the unknown effect on STAN in cases of infection (chorioamnionitis). In addition, the narrow time window to deliver in cases of a significant ST-event was found to be a risk factor for the delivery of babies with a severe metabolic acidosis (the staff found it difficult to respond with the speed required).

One study concluded that “The most important limitation of ST analysis is deviation from STAN clinical guidelines by labor ward personnel rather than a fault in technology” [23]. However, the performance of any system depends on the combination of technology, the patient population, and the ability of the professionals to use the system. If a system is too complicated, the risk of nonadherence to guidelines increases. Standardization and elimination of unnecessary complexity are two very important aspects of improved patient safety [34, 35].

Recommendations for STAN use do not contain strict guidelines about when STAN should be initiated and probably as a result of this there is no consensus decision on which cases should be offered surveillance by STAN. In a Swedish and Dutch randomized controlled trial [14, 36] of STAN compared with CTG, women were eligible whenever a scalp electrode was applied regardless of the classification of the external CTG or presence of risk factors. In a district hospital in Sweden 69% of all women were monitored by STAN [37]. The hospital used routine CTG monitoring in the second stage of labor, and STAN was initiated whenever a scalp electrode was applied. The indications for STAN use in a Norwegian University hospital were high-risk patients and patients with an abnormal CTG and only 26% were monitored by STAN [38].

According to data on clinical use, departments that have implemented STAN end up having two labor ward routines for fetal surveillance. Some will be monitored by CTG alone – probably supplemented by fetal blood sampling in case of an abnormal CTG – and others will be monitored by STAN. This means that an abnormal CTG combined with ST-analysis without ST-events indicates continued monitoring, whereas an abnormal CTG without ST-analysis indicates some kind of action – such as a scalp pH measurement.

The introduction of STAN in an obstetrical department includes a “whole package” of education and training, with systematic CTG classification, and clinical STAN guidelines. If outcome improves, it is not obvious which part of the package was effective. Also, observational studies have shown different results. A study from Sweden showed a decreasing prevalence of metabolic acidosis parallel to an increased frequency of STAN use [38]. In contrast, five comparable obstetric units in Danish university hospitals had an equally low prevalence of birth asphyxia defined as umbilical artery pH <7.0 or Apgar score <7 at 5 min in the absence of umbilical artery pH: 0.4–0.6% [39]. Four units have used CTG/STAN for several years and the fifth uses CTG and fetal blood sampling. All hospitals have systematic teaching programs in CTG/STAN and use the same CTG classification. The rates of birth asphyxia were between 0.5% and 0.6% in the four obstetric units using STAN, and 0.4% in the unit using CTG and fetal blood sampling. Despite the limitations of such a comparison, these observational data indicate that a systematic approach to fetal monitoring, rather than STAN itself, may provide excellent fetal outcome.

In conclusion, STAN does not seem to be the ideal method for fetal monitoring. STAN is not simple and interpretations are not straightforward and it does not give a clear feedback, especially not early warning in case of pre-acidemia. Observational studies seem to indicate that most important is a systematic approach to fetal monitoring, rather than STAN itself.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. ST-analysis – the clinical perspective
  5. To what extent does ST-analysis fulfil these demands?
  6. References