Althaus, Janyne E. MD; Petersen, Scott M. MD; Fox, Harold E. MD, MSc; Holcroft, Cynthia J. MD; Graham, Ernest M. MD
Electronic fetal monitoring entered clinical practice in the late 1960s to serve as a screening test for asphyxia severe enough to cause neurologic damage or fetal death. It was hoped that electronic fetal monitoring would allow the recognition of asphyxia at a sufficiently early stage so that timely obstetric intervention would avoid hypoxic–ischemic brain injury or fetal death.1 Randomized controlled trials performed after its adoption as standard of care in the United States have failed to confirm its efficacy in decreasing the incidence of long-term neurologic morbidity.2 Despite this and the American College of Obstetricians and Gynecologists (ACOG) statement that electronic fetal monitoring does not offer an advantage over intermittent auscultation,3 it continues to be used in 80% of labors in North America, with almost all pregnant patients having at least some electronic fetal monitoring either before or during labor.4
Although preterm deliveries less than 37 weeks occur in only 10% of all pregnancies, they account for about a third of all children with cerebral palsy. Cerebral palsy is a heterogeneous upper motor neuron disorder whose diagnosis is not made until 1–2 years of life. Although there is a general perception that all forms of neurologic injury in children are related to birth events, the amount of time between birth and the diagnosis of cerebral palsy may introduce many confounding factors. Cerebral white matter injury is the most common form of brain injury in preterm infants, and can be diagnosed in the neonatal period shortly after birth. This lesion occurs almost exclusively in neonates younger than 34 weeks, and 60–100% of survivors develop cerebral palsy.5
The cerebral white matter is one of the most vulnerable regions of the brain in preterm infants because of its decreased blood flow (it receives only 25% of the blood flow of cortical gray matter), its maturational inability to autoregulate blood flow, and a decreased number of anastomoses between the short and long penetrating arteries that provide its blood supply.6 The decreased number of blood vessels produces end zones in the white matter that are the sites for the focal necroses of periventricular leukomalacia and white matter atrophy producing ventriculomegaly. Late decelerations on electronic fetal heart rate monitoring have been associated with uteroplacental insufficiency. Because white matter lesions in the preterm brain should be very sensitive to changes in uteroplacental perfusion, our objective in this study was to attempt to identify electronic fetal heart rate abnormalities that will identify preterm neonates diagnosed with cerebral white matter injury.
MATERIALS AND METHODS
This institutional review board–approved case–control study of all births between 23 and 34 weeks gestation at a single university hospital between May 1994 and September 2001 identified 150 infants with cerebral white matter injury characterized by ventricular dilatation due to white matter atrophy or periventricular leukomalacia. All these infants were identified through a database that includes admission, hospital course, and discharge information for all infants admitted to the neonatal intensive care unit. A control group of 150 neonates was formed by matching each case by the subsequent delivery at that gestational age within 7 days. Pregnancy dating was by best clinical estimate using last menstrual period confirmed by ultrasonography. It is our practice to perform an ultrasound examination on all our patients at around 20 weeks gestation, and patients whose dating is uncertain get an ultrasonogram as soon as possible. Ultrasound dating was used if the last period was uncertain or ultrasound dating differed by more than 7 days in the first trimester or 10 days in the second trimester. Patients presenting in the third trimester were dated by ultrasonography if they were unsure of their last period or if there was a discrepancy between the period and ultrasonogram of more than 2 weeks. Neonates with chromosomal abnormalities and congenital anomalies were excluded.
Preeclampsia was defined as proteinuria, edema, and the presence of new-onset hypertension. Intraventricular hemorrhage was defined in the standard fashion, with grade 1 indicating hemorrhage limited to the germinal matrix, grade 2 intraventricular hemorrhage, grade 3 hemorrhage with ventricular dilatation, and grade 4 ventricular dilatation with parenchymal extension of hemorrhage. To qualify as receiving a course of antenatal steroids, the patient must have received two 12-mg doses of betamethasone given 24 hours apart. The clinical diagnosis of chorioamnionitis was made in the presence of maternal fever, with the presence of at least one other finding of fetal tachycardia, uterine tenderness, or purulent vaginal discharge. Histologic chorioamnionitis was diagnosed when any polymorphonuclear leukocytes were seen in either the chorion or amnion, or in significant amounts in the subchorionic space. Histologic funisitis was diagnosed when polymorphonuclear leukocytes were seen in the umbilical cord.
Electronic fetal heart rate tracings were obtained for 125 (83%) of the cases and 121 (81%) of the controls. For those delivered by cesarean, the last hour of electronic fetal monitoring before delivery was reviewed. For cases and controls delivering vaginally, the very terminal portion of the fetal heart rate tracing was often unreadable as the fetal head moved through the low pelvis; therefore, the last hour of interpretable fetal heart rate tracing before delivery was reviewed, and the end of reading-to-delivery time was recorded for each vaginal delivery. The tracings were evaluated according to the National Institute of Child Health and Human Development guidelines by 3 independent maternal–fetal medicine specialists blinded to neonatal outcome.1 Each reviewer recorded the baseline fetal heart rate, time with fetal heart rate more than 160 beats per minute (bpm) (tachycardia) or less than 110 bpm (bradycardia), number of accelerations, reactivity, total number of decelerations, and number of late, variable, or early decelerations. Short-term variability was classified according to the National Institutes of Health guidelines, with grade 1 indicating undetectable variability, grade 2 minimal variability with amplitude range less than or equal to 5 bpm, grade 3 moderate variability with amplitude range from 6 to 25 bpm, and grade 4 marked variability with amplitude range more than 25 bpm. Severe variable decelerations were those with a decrease to less than 70 bpm or lasting more than 60 seconds. The number of bradycardic episodes lasting more than 2 minutes was recorded, as well as the nadir and length of the most severe bradycardic episode. In addition to determining if electronic monitoring could identify fetuses with cerebral white matter injury, we also wished to determine if electronic monitoring could identify neonates with an umbilical arterial pH less than 7.0 or base excess less than −12 mmol/L, values that have been linked to an increased risk of long-term neurologic morbidity. Umbilical arterial blood gases were obtained for 231 (76%) of the patients in this study, 100 of 121 (82.6%) for cases and 93 of 125 (74.4%) for controls. Although it is our policy to obtain a cord gas sample on all deliveries, difficulty in obtaining samples in small preterm cords prevented all deliveries from having a cord gas result.
The diagnosis of cerebral white matter injury was made by neonatal head ultrasonogram. All neonates born between 23 and 32 weeks had at least 3 head ultrasonograms: the first at 24–72 hours after birth, the second at 10–14 days of life, and the third at 6 weeks to specifically look for periventricular leukomalacia. Infants born between 32 and 34 weeks underwent head ultrasonography only if it was felt warranted by the attending neonatologist.
Although the initial cases and controls were matched pairs, we were able to obtain the fetal heart rate tracings for 83% of the cases and 81% of the controls. Because the patients for whom we were unable to locate the fetal heart rate tracings were randomly distributed throughout each group, the groups were compared as independent samples rather than as matched pairs. Assuming that the control group would have 1 ± 1 late decelerations per hour, to show an increase to 2 ± 1 late decelerations per hour in the case group (a 100% increase), with an alpha of 0.05, a sample size of 23 patients in each group would have a 90% power to detect this difference. Continuous data were analyzed using the t test, and categorical data with χ2 or Fisher exact test using Stata 7.0 (Stata Corporation, College Station, TX) and SPSS 12.0 (SPSS Inc, Chicago, IL) software. Linear regression with determination of a Pearson correlation coefficient was performed to examine the relationship between the number of late decelerations per hour and umbilical arterial pH and base excess. A P < .05 was considered significant. Kappa correlation for interobserver reliability was calculated to measure the agreement among the 3 reviewers. Kappa values can vary from −1 to +1, with −1 indicating perfect inverse correlation, 0 no correlation, and +1 perfect positive correlation. For this study, a kappa value less than 0.2 indicated poor agreement; 0.2–0.6, fair/moderate agreement; and more than 0.6, substantial agreement.
Of the 125 cases with cerebral white matter injury whose electronic fetal monitoring tracings could be obtained for this study, 64 (51.2%) delivered vaginally and 61 (48.8%) delivered by cesarean. Of the 121 control patients without brain injury, 72 (59.5%) delivered vaginally and 49 (40.5%) delivered by cesarean, which was not significantly different. There was 1 forceps delivery of an infant in the case group and no vacuum-assisted deliveries. Tracings for vaginal deliveries and cesarean deliveries were analyzed separately.
There were no differences in maternal demographics between cases and controls for those delivering vaginally (Table 1). There was a statistically significant increase in preeclampsia in the case group. Antenatal betamethasone exposure was the same in both groups. Both cases and controls had a similar frequency of exposure to magnesium antenatally. Clinical chorioamnionitis and histologic chorioamnionitis and funisitis were similar in both groups. There was a significant increase in intraventricular hemorrhage and decrease in neonatal death in the case group. There was no difference in 1 or 5 minute Apgar scores less than 7 or umbilical arterial pH or base excess between the cases and controls. No distinguishing features were identified on electronic fetal monitoring to identify neonates with cerebral white matter injury (Table 2). The end of reading-to-delivery time was the same for both groups, indicating that the same late portion of the fetal heart rate tracing was reviewed for both cases and controls. The fetal heart rate baseline, incidence and duration of tachycardia or bradycardia, incidence of decreased short-term variability, number of accelerations, percent reactive, number and type of decelerations, and bradycardic episodes were not significantly different between the two groups.
Of the 110 patients who delivered by cesarean, there were 61 cases with neonatal cerebral white matter injury and 49 controls without brain injury (Table 3). There was no difference in maternal demographics, gestational age, or birth weight. Although there was no increase in multiple gestations among the cases delivering vaginally, there was a significant increase in multiple gestations among the cases delivering by cesarean. There was a significant decrease in preeclampsia among the cases that delivered by cesarean; because there was a significant increase in preeclampsia among cases delivering vaginally, overall for the study there was no difference in preeclampsia for the cases and controls (21/125, 16.8%; 20/121, 16.5%; P = .95). There were no significant differences in antenatal steroids, magnesium exposure, clinical chorioamnionitis, or histologic chorioamnionitis or funisitis between the cases and controls. As in the vaginal delivery group, there was a significant increase in intraventricular hemorrhage among the cases. Review of the last hour of the electronic fetal heart rate tracing before delivery did not identify any distinguishing characteristics for the cases later diagnosed with cerebral white matter injury (Table 4). There was no difference in the incidence of bradycardia between the cases and controls, but the 2 controls with bradycardia had a significantly longer bradycardic episode than the 4 cases with bradycardia.
The kappa correlation for the 3 reviewers in determining reactivity of the fetal heart rate tracing using the National Institutes of Health guidelines was 0.53 indicating fair/moderate agreement. A nonreactive tracing during the hour before delivery did not have any effect on umbilical arterial pH (7.29 ± 0.10 nonreactive, 7.31 ± 0.08 reactive, P = .41) or base excess (−2.7 ± 3.8 mmol/L nonreactive, −2.9 ± 3.4 mmol/L reactive, P = .85). Although there was no difference in the number of late decelerations between neonates with cerebral white matter injury and controls, linear regression of the number of late decelerations during the hour before delivery versus umbilical arterial pH (r = −0.12, P = .08, Fig. 1) and base excess (r = −0.18, P = .01, Fig. 2) did show that as the number of late decelerations increased there was a decrease in both umbilical arterial pH and base excess that reached the level of statistical significance for base excess only, which may reflect the fact that base buffer must be consumed before the pH decreases.
Umbilical arterial gas results linked with intrapartum injury severe enough to increase the risk of long-term neurologic problems are a pH less than 7.0 and a base excess less than −12.0 mmol/L.7 Of the 246 neonates in this study, there were 6 with either an umbilical arterial pH less than 7.0 or a base excess less than −12.0 mmol/L. Only two of the patients who delivered vaginally had umbilical arterial gases below these cutoffs, one in the case group and one in the control group, and from the cesarean delivery group there were 4 neonates, 2 cases and 2 controls (Tables 1 and 3). The tracings of these 6 patients were compared with a control group of 118 patients without brain injury and umbilical arterial gas results above these cutoffs (Table 5). Among the neonates with severe acidosis, there was a statistically significant increase in baseline variability less than 5 bpm. However, in this group of high-risk patients, the positive predictive value of baseline variability less than 5 bpm to predict severe metabolic acidosis was only 7.7%. There was no significant difference in the number of late decelerations during the last hour before delivery between the acidotic and control groups.
A number of antenatal and intrapartum factors have been shown to affect the incidence of neonatal neurologic morbidity. Preeclampsia has been found to have a protective effect in the development of neonatal neurologic morbidity,8 and although there was a statistically significant increase in preeclampsia in the vaginal delivery case group and in the cesarean delivery control group, overall there was no difference in the incidence of preeclampsia between cases and controls. Antenatal betamethasone is associated with a decreased risk of cystic periventricular leukomalacia among very preterm infants,9 and the incidence of its use was the same in both groups regardless of the mode of delivery. Exposure to magnesium sulfate before preterm delivery may decrease neurologic morbidity and was not significantly different between the groups.10 Clinical chorioamnionitis and histologic chorioamnionitis and funisitis, which have been associated with an increased risk of neonatal brain injury,11 were similar in both groups. Intraventricular hemorrhage increases the risk of neonatal cerebral white matter injury because the hemorrhage provides a rich source of iron for the generation of free radicals, which can injure oligodendrocyte progenitor cells.12 Multiple gestation is an independent risk factor for cerebral palsy and long-term neurologic impairment13 and was significantly increased among cases delivering by cesarean. The incidence of neonatal death was the same in the cesarean delivery groups, but was significantly higher in the vaginal delivery controls, reflecting the fact that the neonate will have to survive around 6 weeks for white matter brain injury to be diagnosed. We did not want to exclude neonatal deaths because we were afraid that this would artificially remove neonates with severe metabolic acidosis severe enough to increase the risk of long-term neurologic morbidity (umbilical arterial pH < 7.0 or base excess < −12 mmol/L). Even when all neonatal deaths are included, the incidence of severe metabolic acidosis is still very low: 3 out of 93 cases (3.2%) and 3 out of 100 controls (3.0%). Even when neonatal deaths were removed, we were unable to identify any fetal heart rate characteristics that identified fetuses in whom cerebral white matter injury was later diagnosed.
The basis of electronic fetal monitoring is that changes in the fetal heart rate are related to fetal brain function. The introduction of electronic fetal monitoring has created an expectation of perfect outcome. However, despite a 5-fold increase in the rate of cesarean delivery from 5% to almost 25% of all deliveries in this country, based in part on the electronically derived diagnosis of “fetal distress,” cerebral palsy prevalence has remained unchanged.14 Some researchers have concluded that because of the lack of standard definitions or management algorithms for fetal heart rate interpretation there is wide variability in clinical decision making, and that this problem has confounded the randomized trial method, suggesting that randomized trials should not be considered as grade I evidence.15 They felt that new research should be done using the National Institutes of Health workshop standard definitions to create practice guidelines. This case–control study used these standard definitions to determine if electronic monitoring could identify preterm fetuses with neonatal cerebral white matter injury, a major precursor of cerebral palsy. Although computerized analysis of fetal heart rate tracings has been used, it has not been successful in identifying abnormal patterns that correlate with neonatal neurologic morbidity, and because computerized analysis is not used in routine clinical practice, we chose to have the tracings reviewed by clinicians.
Studies on monkey fetuses have shown that during the course of progressive hypoxia that led to death over 2–13 days that variability of the fetal heart rate disappeared, and late decelerations invariably developed before acidemia occurred.16 It is generally believed that decreased short-term variability is the single most reliable sign of fetal compromise. Short-term variability is produced by the competing effect of sympathetic and parasympathetic nerve input, and when the fetus is alert and active, short-term fetal heart rate variability is 6–25 bpm. Normal short-term variability is a good indicator of the absence of severe hypoxia and acidosis; however, there are many things other than hypoxia and acidosis that produce decreased short-term variability. A study of 488 term fetuses found that the most significant intrapartum fetal heart rate parameter in predicting the development of significant acidemia was the presence of decreased variability less than 5 bpm for at least 1 hour.17 A study of 78 children with cerebral palsy and 300 controls with birth weights more than 2,500 g who had electronic fetal monitoring and survived to 3 years of age found a higher incidence of decreased short-term variability and multiple late decelerations in those with cerebral palsy; however, cerebral palsy occurred in only 0.2% of all the children with these electronic monitoring findings, for a false-positive rate of 99.8%.18
Attempts to correlate fetal heart rate patterns associated with brain damage in humans have been based primarily on studies that identified cases as a result of medicolegal action. A review of the intrapartum fetal heart rate tracings for 55 brain injured neonates and a literature review of 10 additional studies failed to find fetal heart rate patterns associated with neurologic injury or protocols for intervention to avoid neurologic injury.19 A retrospective descriptive study of 44 children, of whom 9 were preterm, identified through litigation, described a unique fetal heart rate pattern during labor that they thought may be associated with preexisting fetal brain damage.20 This pattern consisted of a normal baseline rate with absent variability and small, variable decelerations with overshoot. However, acidosis, as defined as a cord pH less than 7.0 or base excess less than −15 mmol/L, occurred in only a third of the infants, and they could not find any discrete hypoxic–ischemic episode for any fetus either before or during labor. They emphasized that theirs was neither a controlled nor prospective study, and this population was not easily reproducible. We were unable to find any such pattern in the preterm infants with cerebral white matter injury in our study. This study looked only at the electronic fetal heart rate tracings of preterm infants born at less than 34 weeks, and metabolic acidosis severe enough to be linked with intrapartum injury and increase the risk for long-term neurologic morbidity was rare and not significantly different between the cases and controls.
In conclusion, this study found that although there are electronic fetal monitoring findings associated with decreasing umbilical arterial pH and base excess, severe metabolic acidosis that may reflect an intrapartum event occurs only rarely in preterm infants with cerebral white matter injury. The correlation between reviewers in determining something as basic as reactivity was only fair/moderate, illustrating the nonspecificity of this monitoring technique. With adequate power to detect a small increase in the number of late decelerations occurring during the last hour of monitoring, we were unable to detect any difference, and the number of late decelerations during the last hour before delivery for neonates with cerebral white matter injury was very small. Although fetuses with severe metabolic acidosis had a significant increase in decreased short-term variability, this finding had a poor predictive value. These findings show that electronic fetal monitoring is not capable of identifying preterm infants with neonatal cerebral white matter brain injury, a major precursor of cerebral palsy.
1. National Institute of Child Health and Human Development Research Planning Workshop. Electronic fetal heart rate monitoring: research guidelines for interpretation. Am J Obstet Gynecol 1997;177:1385–90.
2. Thacker SB, Stroup DF, Peterson HB. Efficacy and safety of intrapartum electronic fetal monitoring: an update. Obstet Gynecol 1995;86:613–20.
3. American College of Obstetricians and Gynecologists. Fetal heart rate patterns: monitoring, interpretation, and management. ACOG Technical Bulletin. Washington DC: ACOG; 1995.
4. Parer JT. Electronic fetal heart rate monitoring: a story of survival. Obstet Gynecol Surv 2003;58:561–63.
5. Leviton A, Paneth N. White matter damage in preterm newborns: an epidemiologic perspective. Early Hum Dev 1990;24:1–22.
6. Volpe JJ. Neurology of the newborn. 4th ed. Philadelphia (PA): W.B. Saunders; 2000.
7. American College of Obstetricians and Gynecologists, American Academy of Pediatricians. Neoncephalopathy and cerebral palsy: defining the pathogenesis and pathophysiology. Washington, DC: ACOG; 2003.
8. Murphy DJ, Sellers S, MacKenzie IZ, Yudkin PL, Johnson AM. Case-control study of antenatal and intrapartum risk factors for cerebral palsy in very preterm singleton babies. Lancet 1995;346:1449–54.
9. Badawi N, Kurinczuk JJ, Keogh JM, Alessandri LM, O'Sullivan F, Burton PR, et al. Antepartum risk factors for newborn encephalopathy: the Western Australian case-control study. BMJ 1998;317:1549–53.
10. Crowther CA, Hiller JE, Doyle LW, Haslam RR. Effect of magnesium sulfate given for neuroprotection before preterm birth: a randomized controlled trial. JAMA 2003;290:2669–76.
11. Wu YW, Escobar GJ, Grether JK, Croen LA, Greene JD, Newman TB. Chorioamnionitis and cerebral palsy in term and near-term infants. JAMA 2003;290:2677–84.
12. Volpe JJ. Neurobiology of periventricular leukomalacia in the premature infant. Pediatr Res 2001;50:553–62.
13. Williams K, Hennessy E, Alberman E. Cerebral palsy: effects of twinning, birthweight, and gestational age. Arch Dis Child Fetal Neonatal Ed 1996;75:F178–82.
14. Clark SL, Hankins GD. Temporal and demographic trends in cerebral palsy–fact and fiction. Am J Obstet Gynecol 2003;188:628–33.
15. Parer JT, King T. Fetal heart rate monitoring: is it salvageable? Am J Obstet Gynecol 2000;182:982–7.
16. Murata Y, Martin CB Jr, Ikenoue T, Hashimoto T, Taira S, Sagawa T, et al. Fetal heart rate accelerations and late decelerations during the course of intrauterine death in chronically catheterized rhesus monkeys. Am J Obstet Gynecol 1982;144:218–23.
17. Williams KP, Galerneau F. Intrapartum fetal heart rate patterns in the prediction of neonatal acidemia. Am J Obstet Gynecol 2003;188:820–3.
18. Nelson KB, Dambrosia JM, Ting TY, Grether JK. Uncertain value of electronic fetal monitoring in predicting cerebral palsy. N Engl J Med 1996;334:613–8.
19. Rosen MG, Dickinson JC. The paradox of electronic fetal monitoring: more data may not enable us to predict or prevent infant neurologic morbidity. Am J Obstet Gynecol 1993;168:745–51.
20. Schifrin BS, Hamilton-Rubinstein T, Shields JR. Fetal heart rate patterns and the timing of fetal injury. J Perinatol 1994;14:174–81.