194

Randomised trial of cardiotocography alone or with ST waveform analysis for intrapartum monitoring

It

is

record the fetal possible to electrocardiographic waveform (ECG) from the scalp electrode used in labour for detection of fetal

heart rate. Animal and observational studies of changes in the ST waveform of the ECG during hypoxia suggest that a combination of heart rate and ST waveform analysis might improve the predictive value of intrapartum monitoring. In a randomised trial, we have studied intervention rates and neonatal outcome for high-risk labours monitored either by conventional cardiotocography (CTG) or by ST waveform analysis plus CTG. 1200 women with pregnancy of at least 34 weeks’ gestation were assigned to the groups when the decision to apply a fetal scalp electrode was made. Neonatal outcome was assessed by umbilical-cord blood gas analysis, Apgar scores, resuscitation needed, and postnatal course. All recordings were retrospectively viewed by an observer unaware of clinical details to check adherence to the trial protocol. The addition of ST waveform monitoring to CTG substantially reduced the proportion of deliveries for fetal distress (ST + CTG 27/615 vs CTG 58/606; p < 0·001). The groups did not differ in rate of operative delivery for other reasons, incidence of asphyxia at birth, or neonatal outcome. Metabolic acidosis and low 5 min Apgar scores were less common in the ST + CTG than the CTG group, but not significantly so. The only case of birth asphyxia in the ST + CTG group was identified by both heart rate and ST changes. The review of recordings showed that the reduction in intervention rate was among cases with CTG patterns classified as normal or intermediate, whereas there was no difference in intervention rates among cases with abnormal

recordings. Our findings confirm that ST waveform analysis discriminates CTG changes in labour and that our protocol for interpretation is safe. Further randomised studies

are

warranted. Introduction

Electronic fetal monitoring is widely used but its value remains uncertain. Fetal heart rate and uterine contraction pattern are recorded on the cardiotocograph (CTG). Difficulty in interpretation of CTG patterns during labour can result in unnecessary operative intervention,’ whereas some important changes go unrecognised.2,3 Fetal blood sampling (FBS) can be used along with CTG monitoring but it requires additional expertise, is time consuming, and gives only intermittent information, and is not therefore widely used.4,5

By contrast, the fetal electrocardiographic (ECG) waveform can be recorded continuously from the fetal scalp electrode used for measurement of fetal heart rate, and it can provide additional information on fetal responses to labour. The filters used in existing CTG monitors to remove baseline noise and improve heart rate recording prevent the detection of all but gross ECG waveform changes. However, the waveform can be studied in detail with appropriate filter characteristics (0-05-100 Hz) and signal processing. An increase in the ST segment and T wave of the fetal ECG in response to hypoxia has been identified in fetal animal studies.6 This change can be expressed as a ratio of T wave height to QRS height (T/QRS ratio). Testing of a microprocessor-based system’ in observational studies in human beings showed that a combination of fetal heart rate and ST waveform change may be clinically useful.8 The aim of our randomised trial was to find out whether monitoring of ST waveforms usefully adds to CTG monitoring, by comparing intervention rates and neonatal outcome in labours monitored by CTG alone with those in labours monitored by ST waveform analysis and CTG.

Subjects and methods The trial took place at Plymouth General Hospital, where there than 5200 deliveries per year. In 1989, before the trial, the total caesarean-section rate was 9-5%, the rate of forceps delivery 8-5%, and the perinatal mortality rate 6-1 per 1000 livebirths. The overall rate of delivery for fetal distress was 45%; the rate for women who had continuous monitoring with a fetal scalp electrode was 11 -0%. We calculated that a sample size of 450 in each group would give us a 90% chance of detecting a 50% change from this fetal-distress delivery rate at p < 0.05 with a two-sided test to compare proportions.9 The trial size required to assess an effect on neonatal morbidity was more difficult to calculate owing to the lack of definite endpoints. Objective assessment of immediate outcome was possible with umbilical-cord acid-base measurement. We assumed a 2% incidence of metabolic acidosis;lO thus a sample size of 1300 in each group would give us a 50% chance of detecting a in a two-sided test. Based on these 50% difference at p > 60 dropped beats, > 60 s duration TABLE II-PROTOCOL ACTION REQUIRED

’Atter jU mm tor

one

abnormality, immediately tor two

or more

We used the HP 8040A CTG monitor (Hewlett-Packard, Boblingen, Germany) for CTG monitoring alone and the STAN 8801 monitor (Cinventa, Molndal, Sweden) for the ST + CTG group. We independently tested the filter characteristics of the STAN monitoring system before the study and found it to be adequate for ECG analysis (report available from Cinventa on request). We used a single spiral electrode, previously found to be the best for ECG recording," in both study groups. In the ST + CTG group the lead collection system also required use of a maternal thigh electrode to standardise the ECG vector.12 Labour ward staff were given extensive training in the use of the STAN monitor and the concepts of CTG and ST waveform before and throughout the trial. In a pilot phase of 100 cases, the trial protocol was used but data for these cases were not included in the analysis. In the CTG group, the interpretation of CTG traces (table I) and management (table II) followed accepted clinical guidelines, including an FBS option*. In the ST + CTG group, the interpretation of first-stage traces was based on the same CTG criteria, modified by the T/QRS ratio (table I). In the second stage, acute changes in the ST waveform over 5 min were regarded as important. An FBS pH of 7-25 or more was considered normal but if the CTG pattern that prompted FBS persisted, a further sample had to be taken 60 min later; if the pattern deteriorated, FBS had to be repeated sooner. A pH of7-20-7’25 was suspicious; FBS had to be repeated at 45 min or sooner if the trace deteriorated. A pH of 7 20 or lower was abnormal and indicated immediate delivery. We recorded the frequency of FBS (both total numbers done and numbers of cases with an FBS) and the reasons for FBS given in the hospital notes. Similarly, operative deliveries were classified according to the reasons given. Deliveries for fetal distress included both those done when the FBS pH was abnormal and those done because of concern about an abnormal trace. Deliveries for failure to progress included those for inadequate progress in first-stage or second-stage labour and elective forceps deliveries done on account of maternal condition (eg, hypertension). Neonatal outcome was assessed in four ways. Umbilical-cord arterial and venous blood gas analysis (Corning 178 blood gas analyser, CIBA Corning Diagnostics Ltd, Halstead, Essex, UK) was used throughout. In the first 400 cases, cord-blood gas analysis

analysis

*The protocol documents used in the trial

are

available from The Lancet.

restricted to patients undergoing operative delivery or FBS. When a second blood-gas analyser became available, we carried out cord-blood analysis for all deliveries. The cord was double-clamped immediately after delivery and 1-5 ml blood was withdrawn from an artery and the vein into separate heparinised syringes. If samples could not be analysed within 10 min of delivery they were placed on ice and were analysed within the next 30 min. The base deficit calculated from the blood compartment was corrected to that in the extracellular fluid compartment by means of the SiggaardAndersen acid-base chart,13 since this is more correct in the perinatal period.14 We also recorded Apgar scores at 1 min and 5 min and details of any resuscitation needed. Finally, babies were followed up daily to discharge. Any abnormalities were noted, together with details of admissions to neonatal intensive care, treatment given, and diagnosis at discharge. A diagnosis of birth asphyxia required that four criteria were satisfied: (1) cord artery pH below 7-05 and extracellular fluid base deficit greater than 12 mmol/1; (2) Apgar score at 5 min 7 or less; (3) active resuscitation for at least 4 min; and (4) the presence of one of the following features-hypoglycaemia (< 1.6 mmol/1) within the was

TABLE III-BASELINE CHARACTERISTICS cT

*Hypertension, diabetes, reduced fetal

movements, urinary-tract

n

rT

infection, asthma.

196

TABLE IV-INDICATIONS FOR OPERATIVE DELIVERIES

first 12 h (mild asphyxia), neurological abnormalities or need for assisted ventilation in the first 48 h (moderate asphyxia), or death

(severe asphyxia). It is inevitable that clinical staff do not adhere strictly to protocols occasion. All traces were therefore reviewed retrospectively by one of us (J. W.); she started at the beginning of each trace and worked to the end in 30 min segments without preview. Classification was strictly according to the trial protocol without awareness of neonatal outcome. The CTG for each trace was classified as normal, intermediate, or abnormal (table i) if the CTG pattern persisted for at least 30 min in the first stage or 15 min in the second stage. ST + CTG traces were also classified as normal raised (>0-24), or negative. Any T/QRS ratio (

12 mmol/I

deliveries for failure to progress, 2 had FBS followed by normal deliveries, and 1 had a forceps delivery for fetal distress. All had normal neonatal outcome with no asphyxia. On the retrospective review of traces, a further 6 women in the ST + CTG group had ECG signals of such poor quality that the ST waveform could not be assessed though the CTG was satisfactory. All these women had normal CTGs and outcome. Of necessity, the 27 cases described here were excluded from the blinded retrospective review. There was no significant difference between the CTG and ST + CTG groups in the frequency of FBS (80 samples from 58 cases, 9-6% vs 70 samples from 52 cases, 8-5%; p = 056, odds ratio 1-15 [0’77-1’70]). The addition of ST to CTG monitoring significantly reduced the rate of operative deliveries for fetal distress (table IV). This category includes deliveries by caesarean section and rotational forceps and mid-cavity vacuum extractions (CTG 19 vs ST + CTG 5; p=0007, odds ratio 395 [1-46-10-6]). The numbers of operative deliveries for failure to progress did not differ

significantly. There were no significant differences between the groups in neonatal outcome (table v). The CTG group baby with mild asphyxia was bom at term, weighing 3200 g, by spontaneous vaginal delivery. The CTG showed a loss of baseline variability followed by sudden bradycardia in the second stage, which lasted just 5 min. The cord artery pH was 6 87 and the base deficit in extracellular fluid 16 mmol/1. 30 min later the capillary glucose concentration was unrecordable but the baby was fed nasogastrically every 2 h and had no further difficulties. The baby with moderate asphyxia was delivered at term by lower segment caesarean section because of an abnormal trace (tachycardia, loss of variability, and late decelerations). He weighed 2380 g. The cord artery pH was 7-03, base deficit in extracellular fluid 13 mmol/1. The baby was ventilated for 48 h and treated with prophylactic chloral hydrate. The 1 case of birth asphyxia in the ST + CTG group was in the baby of a multigravida induced at term because of suspected growth retardation. The heart rate pattern became very abnormal (see figure) shortly after an epidural was administeredthere was ST-segment depression, the T wave inverted, and the T/QRS ratio became negative (T/QRS - 0. 1 to - 0-2). The baby was delivered spontaneously, weighing 2880 g, 40 min later. The Apgar scores were 4 at 1 min and 6 at 5 min; the baby needed 15 min of active resuscitation. Cord blood gas analysis was not done since the criteria for the first 400 trial entries were not satisfied. The baby was jittery and was treated for hypoglycaemia. TABLE VI-RETROSPECTIVE CLASSIFICATION OF INTERVENTIONS FOR FETAL DISTRESS

197

of information according to a protocol decreased the intervention rate. Other sources of bias should have been controlled by the randomisation process. The well-known improvement in general management that accompanies any randomised trial would affect both groups equally and, if anything, would decrease the chance that we would find any difference. Two important findings support our contention that the reduction was primarily due to the addition of ST waveform information. First, the trial did not bias the indications given in each group for an operative delivery. When a clinician is concerned about a heart rate recording he or she is likely to deliver the baby and distort the indications given for delivery. Since the two groups had similar proportions of deliveries for failure to progress, it seems unlikely that clinicians were misreporting indications for intervention in either group. Second, retrospective review of recordings showed that the reduction in operative deliveries for fetal distress was in cases classified as normal or intermediate with CTG alone. Thus, ST plus CTG specifically prevented unnecessary intervention. It has been assumed that similar levels of nonintervention can be achieved by means of better in-service training in CTG interpretation. In practice, such improvements are difficult to achieve and indeed have not been achieved despite 30 years of electronic heart rate monitoring in labour. Both before and during the trial we put much effort into teaching CTG interpretation as well as the concepts of ST waveform analysis. We explained the physiology of fetal responses to labour and the resulting heart rate changes. The difference in operative intervention is therefore even more surprising. It seems that the addition of ST waveform analysis to CTG improved clinical interpretation of CTG traces. We were not surprised that the FBS rate was not significantly changed. With the introduction of a new monitoring method into our clinical practice, we had to maintain the safeguard of an FBS option. Our findings and those of Johansen et al 16 lead us now to recommend a reduction in the FBS frequency in the CTG-abnormal group when the ST waveform is normal. The suggestion of beneficial effects on Apgar scores and metabolic acidosis is interesting and supports the safety of ST + CTG monitoring. The full trial of 2400 cases will provide further information. There is no universally accepted definition of birth asphyxia. We believe a combination of factors is most useful and our definition (metabolic acidosis, need for resuscitation, and an effect on neonatal behaviour) is the same as that of Gilstrap et al,17 but it may underestimate the number of mild cases. It is difficult to distinguish fetal distress from fetal stress. The very high circulating catecholamine concentrations in a healthy infant after a normal delivery exceed even those of an adult with a phaeochromocytoma18 and anaerobic metabolism is an important part of the fetal response to hypoxaemia. In fetal lambs, an increase in the ST waveform is directly correlated with the rate of myocardial glycogenolysis,19 which is increased by hypoxaemia or beta-receptor stimulation .20 A persistently high, stable T/QRS ratio accompanied by a normal CTG was seen in 5% of ST + CTG cases. This observation may reflect increased sympathoadrenal stimulation from the general arousal of labour or may be in response to mild but compensated hypoxaemia. In all these cases, neonatal outcome was good. The only case of asphyxia in the ST + CTG group was clearly identified by both an abnormal CTG and negative T waves but no clinical action was taken. ST depression with a negative T wave has been

piece

ST+CTG trace from

a baby with mild birth asphyxia. A= recording during epidural insertion; ECG waveform complexes

are

normal. No contractions shown since belt had been removed for

epidural insertion B = recording 20 min later showing abnormal fetal heart rate pattern with tachycardia, loss of heart rate variability, and prolonged late decelerations accompanied by ST-segment depression, negative T waves, and a negative T/QRS ratio.

retrospective review, the CTG and ST + CTG had similar numbers of normal (478 vs 474), groups intermediate (59 vs 57), and abnormal (66 vs 60) CTGs (table VI). Operative intervention among patients with normal and intermediate traces was significantly lower in the ST + CTG group than in the CTG group (8/531 vs 31/537; p=0 008, odds ratio 0 25 [0’11-0’55]). ST monitoring did not significantly affect the rate of operative deliveries among subjects with abnormal traces (ST + CTG 18/60 vs CTG 27/66; p = 0-28, odds ratio 0 62 [0’30-1.30]). Both babies with normal CTGs and negative ST waveforms had falsely negative waveforms; 1 was a monitor error and 1 was a breech fetus with very unusual complexes. In all the CTG-normal, ST-raised cases the T/QRS ratio was persistently high (0-25-0-32) throughout the labour. All cases of acute and progressive ST changes with increasing T/QRS occurred during the second stage of labour, parallel with progressive CTG abnormalities. In the

Discussion We believe this study is the first in which a new means of fetal assessment, also involving a new monitor, has been tested in a randomised trial before adoption into obstetric practice. Evaluation of the ST waveform of the fetal ECG closely followed standard pharmaceutical procedures for the introduction of a new drug-animal studies to investigate the basic physiology, observational studies in man so that a clinical protocol can be devised, then a large randomised trial to establish the clinical value and practicability. The trial showed that the addition of ST waveform analysis to CTG monitoring reduced the rate of operative deliveries for fetal distress by 53% and the use of caesarean section and rotational forceps by 74% with no change in neonatal outcome. Furthermore, intervention rates at this hospital were already low. This study could not be blinded. We specifically examined whether subjective interpretation of an additional

198

observed in growth-retarded guineapig fetuses ;21 it may result from depletion of myocardial glycogen reserves, which would prevent anaerobic metabolism. Previous observational studies22-24 on the value of ST waveform analysis based on the T/QRS ratio alone did not take account of complexity of ST waveform change and the underlying physiology. So far, the combination of an abnormal CTG plus acute ST changes has accurately identified settings in which action was required to deliver the fetus; thus the potential to improve outcome exists. A possible criticism of the trial design is that little is known of ST waveform changes in fetuses of less than 36 weeks gestational age. Only 1 of the 12 fetuses of less than 36 weeks in the ST + CTG group showed ST changes (with intermittent negative T waves) and an abnormal CTG; the baby was delivered with forceps. He was not asphyxiated (cord artery pH 7-15, base deficit in extracellular fluid 6 mmol/l, 5 min Apgar score 9). All other fetuses of less than 36 weeks gestational age (CTG group 11) had normal outcomes with no asphyxia. It is technically easier to obtain a continuous fetal-heartrate record than to record the whole ECG essential for waveform analysis. High-quality ECG signals are required since filtering cannot be used. For 15 (2-4%) of 615 randomised subjects we could not obtain ECG signals of sufficient quality, whereas adequate signals for standard heart rate detection were obtained for all but 2 (04%). Correct choice of scalp electrode" and proper electrode application are essential. Further improvements in signal processing and R wave detection are being investigated. We believe this trial confirms both the potential of ST waveform change to discriminate fetal heart rate change in labour and the safety of the model of interpretation. We emphasise that this model depends on the use of both variables and adequate education of staff in their interpretation. Randomised studies in other units and further development of ST + CTG monitors seems

justified. We thank the Swedish Board of Technical Development for support, Dr J. Smith for help with the initial trial organisation, teaching, and data collection, Dr D. Wright (Department of Mathematics and Statistics, Polytechnic South West) for statistical advice, and medical and midwifery staff at Plymouth General Hospital for cooperation and support. The study was supported by grants from the South Western Regional Health Authority Locally Organised Research Scheme, the Northcott Devon Medical Foundation, and TSW Telethon Appeal Fund.

REFERENCES Haverkamp AD, Orleans M, Langerdoerfer S, McFee J, Murphy J, Thompson H. A controlled trial of the differential effects of intrapartum fetal monitoring. Am J Obstet Gynecol 1979; 134: 399-408. 2. Ennis M, Vincent CA. Obstetric accidents: a review of 64 cases. BMJ 1990; 300: 1365-67. 3. Murphy KW, Johnson P, Moorcraft J, Pattinson R, Russell V, Turnbull A. Birth asphyxia and the intrapartum cardiotocograph. Br J Obstet Gynaecol 1990; 97: 470-79. 4. Wheble AM, Gillmer MDG, Spencer JAD, Sykes GS. Changes in fetal monitoring practice in the UK: 1977-1984. Br J Obstet Gynaecol 1989; 1.

96: 1140-47. 5. Clark SL, Paul RH. Intrapartum surveillance: the role of fetal scalp blood sampling. Am J Obstet Gynecol 1985; 153: 717-20. 6. Greene KR. The ECG waveform. In: Whittle M, ed. Ballière’s clinical obstetrics and gynaecology, vol 1, no 1. London: Baillière Tindall, 1987: 131-55. 7. Rosén KG, Lindcrantz K. STAN—the Gothenburg model for fetal surveillance during labour by ST analysis of the fetal ECG. Clin Phys Physiol Meas 1989; 10 (suppl B): 51-56. 8. Rosén KG, Arulkumaran S, Greene KR, et al. Clinical validity of fetal ECG waveform analysis. In: Saling E, ed. Perinatology. Nestlé Nutrition Workshop Series. New York: Raven Press, 1991: 95-110.

9. Low JA. The role of blood gas and acid-base assessment in the diagnosis of intrapartum fetal asphyxia. Am J Obstet Gynecol 1988; 159: 1235-40. 10. Casagrande JT, Pike MC, Smith PG. An improved formula for calculating sample sizes for comparing binomial distributions. Biometrics 1978; 34: 483-86. 11. Westgate J, Keith RDF, Curnow JSH, Ifeachor EC, Greene KR. Suitability of fetal scalp electrodes for monitoring the fetal electrocardiogram during labour. Clin Phys Physiol Meas 1990; 11: 297-306. 12. Lindecrantz K, Lilja H, Widmark C, Rosén KG. Fetal ECG during labour: a suggested standard. J Biomed Eng 1988; 10: 351-53. 13. Siggaard-Andersen O. An acid-base chart for arterial blood with normal and pathophysiological reference areas. Scand J Clin Lab Invest 1971; 27: 239-42. 14. Rosén KG, Murphy KW. How to assess fetal metabolic acidosis from cord samples. J Perinat Med 1991; 19: 221-26. 15. Gardner MJ, Gardner SB, Winter PD. Confidence interval analysis (CIA). Microcomputer program version 1.0. London: British Medical Journal, 1989. 16. Johansen RB, Rice C, Shokr A, Doyle M, Chenoy R, O’Brien PMS. ST-waveform analysis of the fetal electrocardiogram could reduce fetal blood sampling. Br J Obstet Gynaecol 1992; 99: 167-68. 17. Gilstrap LC, Leveno KJ, Burris J, Williams MC, Little BB. Diagnosis of birth asphyxia on the basis of fetal pH, Apgar score and newborn cerebral dysfunction. Am J Obstet Gynecol 1989; 161: 825-30. 18. Lagercrantz H, Bistoletti P. Catecholamine release in the newborn infant at birth. Pediatr Res 1977; 11: 889-93. 19. Greene KR, Dawes GS, Lilja H, Rosén KG. Changes in the ST waveform of the fetal lamb electrocardiogram with hypoxemia. Am J Obstet Gynecol 1982; 144: 950-57. 20. Hökegård KH, Karlsson K, Kjellmer I, Rosén KG. ECG changes in the fetal lamb during asphyxia in relation to beta-adrenoreceptor stimulation and blockade. Acta Physiol Scand 1979; 105: 195-203. 21. Widmark C, Jansson T, Lindecrantz K, Rosén KG. ECG waveform, short term heart rate variability and plasma catecholamine concentrations in response to hypoxia in intrauterine growth retarded guinea-pig fetuses. J Develop Physiol 1991; 15: 161-68. 22. Newbold S, Clewlow F, Wheeler T. The effect of uterine contractions on the T/QRS ratio of the fetal electrocardiogram and the fetal heart rate during labour and the relation of these variables to condition at delivery. Br J Obstet Gynaecol 1991; 98: 173-78. 23. Maclachlan NA, Spencer JAD, Harding K, Arulkumaran S. Fetal acidemia, the cardiotocogram and the T/QRS ratio of the fetal ECG in labour. Br J Obstet Gynaecol 1992; 99: 26-31. 24. Murphy KW, Russell P, Johnson P, Valente J. Clinical assessment of fetal electrocardiogram monitoring in labour. Br J Obstet Gynaecol 1992; 99: 32-37.

From The Lancet African fever We have recently received reports from various quarters throwing grave doubts on the received etiology and on the nature of African fever. Is the disease due to malaria? If so, is the malaria in Africa produced by the same, or approximately the same, causes by which it is produced elsewhere? This naturally involves the whole subject of the nature, cause, prophylaxis, and cure of malaria. Recent bacteriological research seems certainly to point to a micro-organism, a polymorphic haematozoon, being concerned in the production of malaria in most parts of the world. At any rate, the researches of Laveran in France, ofMarchiafava, Celli, and Camillo Golgi in Italy, of Vandyke Carter and Hehir in India, and Councilman and Osler in America, all result in a common agreement that such is the case, and the examintion of the blood of a few patients who have returned from various parts of Africa also point to the same haematozoon being at work in that continent. On the other hand, it will not do to ignore the fact that there are observers, and observers of no mean order, who in India, in America, and in Africa deny even that malaria exists as a disease sui generis, and who maintain that the phenomenon witnessed in, for instance, intermittent fever are due to either errors of personal hygiene or to the action of "chill". It is certainly difficult to reconcile these last opinions with the widespread ravages of malaria, which... possess the unenviable position of killing more human beings than any other malady enumerated in the nomenclature of disease.

(July 2, 1892)