MONITORING OR AUSCULTATION?
During the last few years, perinatologists have asked themselves many questions regarding the real usefulness of intrapartum fetal monitoring. In fact, do we impose it on patients because we are really convinced of its diagnostic efficacy or has it become a mere product of routine?
Tran-Ngoc Tran and Millet (16) published a paper in 1980 analyzing data from our Department for 1075 delivering women who had had normal pregnancies and had no risk factors. The aim of the study was to evaluate if a routine cardiotocographic registration at the beginning of delivery was a method for precocious screening of fetal distress. In spite of the total absence of risk factors in the studied populations, the authors found a pathologic pattern in 71 (6.6%) of patients and 9 (12.6%) of these women underwent an intervention (cesarean section, forceps or vacuum extractor). Even when the pattern was judged physiological (1004 cases) an intervention for fetal distress was necessary in 63 cases (6.3%). The pH of the umbilical artery was pathological (pH <7.10) in 5.4% of cases in the group with normal registrations and in 6.9% of the cases in the group of the pathological registrations. This experience and the specialized literature had convinced us, at the time, to propose continuous fetal monitoring to all women in labour with or without risk factors. Were we right ?
Already in 1974, Paul and Hon (13) compared the outcome for monitored and unmonitored deliveries in Los Angeles (USC). Of the 28,621 births that had occurred in 1970 - 1972, 24.3% underwent monitoring, and in most cases they were high risks births. The intrapartum mortality was of 1.3/1000 in the group monitored and 4.4/1000 in the control group. Neonatal mortality in the same groups was 10/1000 and 17/1000 respectively. This retrospective study seemed very promising but presented biases that could have altered its statistical value.
In a study published in 1979, a group of authors, whose global results are to be found in Freeman and Garite (4), a book dedicated to monitoring, reviewed 11 retrospective non-randomized studies. One of the studies reviewed was that of Paul and Hon concerning intrapartum fetal death during continuous electronic fetal monitoring and during simple auscultation.
The total number of deliveries controlled by simple auscultation was 99,482, with 176 (1.76/1000) intrapartum deaths. The total nimber of monitored deliveries was 38,785 with only 21 (0.54/1000) intrapartum deaths (p <0.001). These studies have been heavily criticized, because the periods of monitoring and auscultation were frequently different, and in sequence.
Even if not perfect, these studies were nonetheless convincing, as the monitorized cases concerned pathological or high risk pregnancies much more frequently than the cases for which simple auscultation was applied.
During the seventies, several randomized prospective studies were begun, but including a smaller number of patients. Freeman and Garite also reviewed 4 very well-conducted studies. Overall, the number of deliveries controlled by simple auscultation was 901, with 1 (1.1/1000) intrapartum death, and the number of monitored patients was 1,125, with no reported fetal death. Larger numbers (at least 18,000 deliveries) were clearly necessary to show any statistically significant difference. It is also necessary to underscore that these studies required the continuous presence at the patients’ bed side of a well-trained nurse. As noted in the last edition of Williams Obstetrics (2): " It cannot be overemphasized that the techniques for continuous recording of fetal heart rates and uterine pressures do not by themselves provide continuous surveillance of the fetus. Appropriately trained personnel must be immediately available to activate the electronic techniques, to inspect and analyze almost continuously the data that are being recorded, and to act promptly on the findings. "
The first of the randomized prospective studies had been that of Haverkamp et al. (6) in Denver, Colorado, on 483 high risk patients. No intrapartum death was reported. In the " auscultation " group 16 out of 241 mothers (6.6%) underwent a cesarean section, 3 of which were performed because of fetal distress, whereas in the monitored group 40 of the 242 mothers (16.5%) underwent cesarean section, 18 because of fetal distress. At the beginning of monitoring increases in cesarean section rates were criticized by many observers. This negative effect associated with the difficulty of demonstrating the actual well-being of the fetus or of the newborn was at the origin of the banning of this diagnostic method in some cases. After a few years, however, a more accurate reading of the diagrams allowed the majority of clinical research teams to obtain a decrease or even the disappearance of this negative effect, as demonstrated in 1985 by Greenland et al. (5).
With such small numbers, it is difficult, especially on the basis of randomized studies, to demonstrate any effect by using the criterion of intrapartum mortality. Neonatal morbidity and mortality are, however, more useful criteria.
In the review by Freeman and Garite (4), 12 non-randomized retrospective studies were also analyzed. These studies had compared the rate of neonatal deaths after intrapartum monitoring and after auscultation. In the group followed by auscultation (90,713 deliveries) 1382 neonatal deaths occurred (15.2/1000) against 179 (4.4/1000) registered in the monitored group (40,013) (p <0.001). It must be noted however that children whose birthweight was less than 1500 g or who presented congenital malformations were not taken into account in these statistics. Again, the same remarks and doubts advanced concerning the intra-partum deaths in the non-randomized studies with respect to sample size apply.
In the randomized studies, however, the numbers are so small that the neonatal criterion is not significant. In fact, contrary to expectation, one death for the 901 deliveries of the auscultation group (1.11/1000) was registered as against 5 neonatal deaths for the 1,125 deliveries of the monitored group (4.44/1000). The study of neonatal morbidity in the prospective and randomized studies is, however, more enlightening.
In an article published in 1976, Renou et al. (14), reported study of a population at high obstetrical risk and analyzed the following criteria of neonatal morbidity in 222 control patients and in 218 monitored patients: 1) neonatal intensive care, 2) neurologic pathology, and 3) cerebral lesions. The number of newborns that fit these criteria were respectively 11, 2 and none in the monitored group, and 30, 13 and 4 in the control group.
Wood et al. (17) and Kelso et al. (8) carried out randomized studies on low risk populations and found that monitoring did not influence the rate of perinatal deaths, maternal infections, Apgar scores, pH in the cord blood, or in the newborns, intensive care admissions, pathological neurological signs, infections, or cerebral lesions. In both studies, the rate of assisted deliveries was increased in the monitored group.
In 1978, Neutra et al. (11) had shown a positive influence of monitoring for all classes of obstetrical risk, except the lowest which represented in fact 80% of the studied population (see Table 1). Differences were not statistically significant.
Ballas et al. (1) compared the perinatal results of 7604 deliveries over two successive years. In 1976, 15.4% of births were monitored. In 1977, the team decided to switch to routine monitoring. Only mothers who entered the department at a stage of delivery that was too advanced, or those who were programmed for an elective cesarean section were not subjected to the routine. The percentage of monitoring increased during the second year to 56%. The results did not show any difference in perinatal, precocious neonatal, antepartum and intrapartum deaths. The only significant advantage concerned the percentage of Apgar scores lower than 7 at 1 minute: 1.66% in 1976 and 0.96% in 1977.
Ingemarsson and Herbst (7), called once more into question the necessity of monitoring low risk deliveries, and they re-evaluated the admission test as Tran-Ngoc Tran and Millet (16) had done earlier. Their results (see Table 2) show that 1% of patterns was pathologic at admission and 1.4% of fetal distress occurred in cases with normal patterns. These rates are very different from ours. These differences could probably be explained by a different selection of cases and by a possible increase in the capacity to read the patterns during the last 10 years. Ingemarsson and Herbst conclude that only periodic monitoring by auscultation is necessary in the case of low risk deliveries with a normal pattern at admission.
A study by Leveno et al. (9) is important because of the high number of studied pregnancies, its prospective character and its methodology. This study was carried out on 34,995 births over a period of 36 months. The center had in alternating periods only 7 monitoring machines available (selective monitoring) or 19 machines (routine monitoring).
During the " selective " months, 37% of delivering women were followed by monitoring against 79% during the " routine " months. The results for the fetuses and for the newborns in this study are shown in Table 3. The perinatal results were similar in the 2 groups. The rate of cesarean sections, however, was slightly higher in the " routine " group (11%) than in the selective group (10.2%), p <0.05.
The authors concluded that continuous monitoring during labor was not necessary for all deliveries. In particular, the monitoring did not seem useful if the risk of perinatal complications could be defined as low.
The so-called Dublin study, caused most perinatologists to question the usefulness of intrapartum monitoring. MacDonald et al. (10) conducted a randomized study of 12,964 mothers. One group underwent continuous intrapartum fetal monitoring and the other periodic fetal auscultation. The rate of cesarean sections was similar in the two groups (see Table 4), but the study revealed more forceps extractions in the monitored group, as a consequence of so-called pathologic patterns.
There were 14 intrapartum or neonatal deaths in each group and no differences in pathological Apgar scores, neonatal resuscitations or admissions to the intensive care unit. Neonatal convulsions were more frequent in the auscultation group (Table 5) and this difference was related to a longer labour. When checked a year after birth, however, only three of the twenty-one surviving babies presented neurological abnormalities.
These studies came under heavy criticism, and some of the arguments appear quite convincing, in particular regarding the interpretation of monitoring patterns: as clearly demonstrated by Nielsen et al. (12), a great variability in the evaluation of monitoring patterns exists from one observer to the other, and even in the same observer at different times. This variability overshadows the predictive value of fetal monitoring, and should be taken into account when analyzing the cost/benefit ratio of this method.
In a recent article, Ellison et al. (3) re-examined the data from the Dublin study. In particular, they studied a group of newborns who had undergone a neurological examination, because of admission to an intensive care unit, presenting neurological symptoms, or because the treating team was preoccupied with outcome. They demonstrated (Table 6) that the Apgar scores at 1 and 5 minutes were lower and the standard deviation wider in the group that underwent neurological assessment. The length of labour had been longer in the group monitored that had later undergone a neurological examination.
The authors studied the results of the neonatal neurological assessments conducted within 0 to 48 hours of birth and 72 hours to 1 week of life, and tried to find the best correlations between the results of these tests and the data from auscultation or monitoring. In the auscultation group, abnormalities of auscultation, the presence of meconium and the number of scalp pH performed showed significant correlations with neonatal neurological state. A better correlation was found, however, in the monitored group between late decelerations and severe bradycardia, and neurological abnormalities.
Shy et al. (15) compared the neurological development of 93 babies born prematurely and followed-up by monitoring during delivery to 98 premature babies whose follow-up had been realized by periodic auscultation. All the babies weighed fewer than 1750 grams at birth. The incidence of neurological sequelae was of 20% in the monitored group and 8% in the auscultation group (p <0.03). The mental or psychomotor development scores were slightly better in the auscultation group.
Fetal monitoring is a useful diagnostic tool but we have probably attached more importance to it than warranted in the last 20 years.
The use of continuous biophysical monitoring in case of high risk deliveries is no longer a matter of discussion today, but we should not forget the value of parallel use of clinical evaluation and biochemical monitoring.
On the other hand, the advantage of biophysical monitoring during low risk deliveries which represent the majority of deliveries is much more difficult to demonstrate. All authors whose data we discussed here agree that their studies are criticizable, and their conclusions must be carefully evaluated because of numerous bias that cannot always be eliminated, and can influence results. It is well to remember that the character of data on monitoring patterns are fundamentally relative and this is not the smallest problem for scientific evaluation of the method. The cost/benefit ratio of cardiotocographic monitoring of birth does not appear to favour its use in low risk cases on medical, human, or financial grounds. It is probably correct today to propose a decrease in the surveillance of certain deliveries, by utilizing periodic auscultation and associating with it all we have learnt from our experience with monitoring.
A lighter monitoring protocol for delivery
During the last months, we have introduced in our delivery unit in Geneva the following monitoring protocol.
We maintain cardiotocographic monitoring in the presence of evident risk factors such as:
In the absence of risk factors, the patient undergoes at admission an external 20-30 minutes of monitoring. If the pattern is normal we go on following the patient by periodic auscultation either by stethoscope or by Doppler detector. The fetal heart rate (FHR) is checked at least every 15 minutes during dilatation and after each contraction during expulsion. The FHR is controlled at some distance from a contraction to determine the baseline and then again after the next contraction for at least 30 seconds.
In the case of abnormality discovered at auscultation, such as bradycardia of fewer than 100 beats per minute (bpm), decrease of the fetal heart rate under the level of the baseline during the 30 seconds after a contraction or tachycardia over 160 bpm during more than 10-15 minutes, monitoring is re-established.
If the patient has not yet delivered 6 hours after admission monitoring, a new cardiotocographic recording is made.
All detected abnormalities during delivery trigger re-establishment of monitoring. The request of a patient who prefers to be continuously monitored by cardiotocography even in the absence of risk factors must also be accorded.
Edited by Aldo Campana,