Glantz, J. Christopher MD, MPH; Bertoia, Natalie MD
The nonstress test (NST) is an indirect measure of fetal oxygenation. Traditional criteria for a reassuring (reactive) NST are the presence of 2 fetal heart rate accelerations in 20 minutes, with each acceleration 15 beats per minute or more above baseline and 15 seconds in duration or more (15×15). Reactive NSTs are associated with normal fetal pH level and survival.1
Many studies have assessed heart rate accelerations in the term fetus, but few have focused on preterm pregnancies. This is important because, although the total number of accelerations per unit time does not differ between term and preterm fetuses, the number of 15-beat accelerations per unit time in preterm pregnancies is less than the number of 15-beat accelerations in term gestations.2–4 Using the same reactivity criteria in a preterm fetus as in a term fetus often necessitates follow-up fetal testing to determine whether a “nonreactive” preterm NST has a true positive result. The number of preterm 10-beat accelerations is similar to the number of 15-beat accelerations at term, such that a modified 10×10 preterm reactivity criterion (10 beats or more for 10 seconds or more) has been proposed.5,6 No studies have examined use of the 10-beat modification specifically for preterm pregnancies; however, some older studies in term or near-term pregnancies performed before adoption of the 15×15-beat criterion used varying thresholds, including 10-beat, without differences in perinatal outcomes.7–9 Despite a lack of comparative study, the 10-beat criterion has been accepted as reactive for gestational age less than 32 weeks.10 The primary purposes of this study were to determine whether preterm NSTs are associated with perinatal outcome and whether interpretation using the 10×10 criterion is equivalent to the 15×15 criterion. Secondary purposes were to determine the effect of decelerations and the pattern of sequential NSTs on outcome.
MATERIALS AND METHODS
This study is a retrospective review and was approved by the University of Rochester Research Subjects Review Board. Participants were consecutive women 18–45 years of age with singleton pregnancies who had undergone NST between 23 and 32 weeks of gestational age and delivered before 34 weeks at Strong Memorial Hospital between July 2006 and December 2008. Candidates were identified from a computerized hospital perinatal database, and the electronic medical records were reviewed for maternal age, gravidity, parity, indications for NST, formal interpretation of NST, gestational ages at NST and at delivery, birth weight, Apgar scores, and fetal demises. Fetuses with central nervous system anomalies or cardiac arrhythmias were excluded because of potential interference with interpretation of NSTs. Nearly all NSTs in this study were performed on inpatients. The practice at Strong Memorial Hospital when gestational age was less than 32 weeks was that a 20-minute NST demonstrating at least two accelerations 10 or more beats above baseline and lasting 10 or more seconds was considered reactive (ie, neither further testing nor intervention was performed if these criteria were met). When gestational age was 32 weeks or more, the criteria for reactivity were at least two accelerations 15 or more beats above baseline and lasting 15 or more seconds over the course of 20 minutes. When an NST initially was nonreactive, the time was extended to 40 minutes and vibro-acoustic stimulation was used. Nonreassuring NSTs were followed-up by biophysical profile tests.
We retrospectively reviewed all NSTs performed within 1 week of delivery. Blinded to neonatal outcome, a single reader (N.B.) classified each NST into one of three categories: 1) nonreactive according to the 10-beat criterion (ie, fewer than two 10×10-beat accelerations in 20 minutes); 2) reactive according to the 10-beat criterion but not the 15-beat criterion (reactive 10-beat); or reactive according to the 15-beat criterion (reactive 15-beat). The presence or absence of decelerations was recorded, including the type of deceleration (variable, late, prolonged). For women with three NSTs within the week before delivery, patterns were characterized as decreasing (a consistent decline in acceleration height, beginning at 15- or 10-beat reactivity and ending at a “lower” degree, whether 10-beat or less, or nonreactive, respectively), increasing (a consistent increase in acceleration height, beginning at nonreactive or 10-beat and ending at 10-beat or more, or 15-beat, respectively), irregular (varying inconsistently over the three NSTs, regardless of the final reading), or consistent (eg, always the same result each time). There are multiple possible permutations of NST results within each category.
Once the list of participants was compiled, it was linked to the hospital neonatal intensive care unit (NICU) database that provided detailed information on each neonate. Neonatal fields included intensive delivery room resuscitative measures (endotracheal tube, epinephrine, or chest compressions), 5-minute Apgar score less than 7, respiratory distress or mechanical ventilation more than 6 hours in NICU, necrotizing enterocolitis, intraventricular hemorrhage, renal failure, hypoxic-ischemic encephalopathy, seizures, periventricular leukomalacia (PVL), pneumonia, sepsis, hypotension, and neonatal death.
For analysis, six outcomes were chosen for comparison between NST groups: intensive delivery room resuscitation, 5-minute Apgar score less than 7, prolonged mechanical ventilation, necrotizing enterocolitis, intraventricular hemorrhage, and perinatal death. The other various outcomes were too infrequent for statistical comparison (see Results). The χ2 test was used to compare the frequency of each outcome by the result of the proximate NST (the NST closest to delivery), the presence or absence and type of deceleration on the proximate NST, and the pattern of NST results when there were three within 1 week of delivery. To determine independent odds ratios between criteria, nonreactivity, and the presence of decelerations, we performed multiple logistic regression entering reactive 10-beat, nonreactive, and decelerations for each outcome and using reactive 15-beat as the reference. Because birth weight and gestational age are strongly associated with perinatal outcomes, we then added these two variables to the regression to adjust for them, again using reactive 15-beat as the reference. The analyses were performed using SPSS 18.0 for Mac.
Seven hundred fifty-one NSTs from 488 women were assessed; 97 women had three sequential NSTs. The most common NST indications were preterm labor (124 of 488 [25.4%]), hypertension or preeclampsia (97 of 488 [19.9%]), and premature rupture of membranes (71 of 488 [14.5%]). Other less common indications included diabetes, placenta previa, trauma or abruption, cerclage, fetal growth restriction, or fever (all less than 5%). There were multiple NST indications in 31 women, and in 116 women (23.8%) the indication was unclear.
Patient characteristics, NST gestational age or intervals, and delivery information are shown in Table 1. Two thirds of the women were white, and one third were African American (1% were “other”). Fifty percent of women received tocolytics, and 84% received at least one dose of betamethasone before delivery; 234 (48%) deliveries were cesarean. Delivery occurred before 26 weeks in 68 (13.9%), between 26 and 27 weeks in 84 (17.2%), between 28 and 29 weeks in 104 (21.3%), between 30 and 31 weeks in 152 (31.1%), and at 32 week or later in 80 (16.3%) women. Umbilical cord pH was performed selectively and was performed in only 13% of neonates; therefore, it was not included in our analysis. There were differences between proximate NST results and unadjusted gestational age at delivery, gestational age at proximate NST, 5-minute Apgar score, birth weight, and NICU length of stay, with increasing reactivity associated with higher gestational age, birth weight, and shorter NICU stay (data not shown; P<.001). There were no differences between proximate NST result and maternal age, gravidity, parity, interval between NST and delivery, gestational age at nonproximate NSTs, or intervals between NSTs (data not shown; P>.08).
For all proximate NSTs at any gestational age, 105 were characterized as nonreactive, 151 were characterized as reactive 10-beat, and 232 were characterized as reactive 15-beat. Our NST reading agreed with the formal reading 85% of the time (95% confidence interval 82–88%, κ=0.75, P<.001). When there was a difference, our reading was a higher degree of reactivity 81% (95% confidence interval 72–90%) of the time. Of proximate NSTs, rates of nonreactivity (by either criterion) declined with increasing gestational age, and rates of 15-beat reactivity increased significantly (P<.001, Table 2). The proportions of NSTs reactive only by the 10-beat criterion were stable through 29 weeks and then declined. Decelerations were present in 274 (56.1%) proximate NSTs: 60–79% in gestational age intervals before 30 weeks and 44–48% (less frequent but still common) at or after 30 weeks (P<.001). When decelerations were present, they were variables 93.8% of the time (257 of 274).
A neonatal team was present at all deliveries, and all neontates in the study were admitted to the NICU. Numbers of each neonatal outcome are shown in Table 3. A few fields in the NICU database were incomplete, so denominators in Table 3 are not all identical. Outcomes too infrequent to analyze included hypoxic-ischemic encephalopathy (one 590-g fetus at 25 weeks with Apgar scores 1, 1, and 7 after a reactive 10-beat NST without decelerations on the day of delivery), seizures (one 1,010-g fetus at 25 weeks with Apgar scores 8 and 9 after a reactive 15-beat NST without decelerations 6 days before delivery), and PVL (n=3; gestational age 27–29 weeks, birth weights 1,060–1,210 g, one nonreactive NST the day of delivery with subsequent Apgar scores 3, 4, and 9; one reactive 10-beat NST 5 days before delivery with subsequent Apgar scores 6 and 8; one reactive 15-beat NST on the day of delivery with subsequent Apgar scores 7 and 8; all three fetal NSTs had decelerations), sepsis (n=6), and renal failure (n=5). Fetal demise occurred in seven cases (1.4%, also too infrequent for independent subgroup analysis); these occurred between 23 and 28 weeks of gestation and were combined with the 40 neonatal deaths to comprise perinatal mortality. Five of seven NSTs proximate to death were nonreactive and two were reactive 10-beat. Decelerations were present in six of the seven NSTs preceding fetal demise.
The unadjusted associations between NST result and outcomes are shown in Table 3, and the unadjusted associations between the presence or absence of decelerations and outcome are shown in Table 4. With the exception of necrotizing enterocolitis, there was an unadjusted association between degree of NST reactivity and perinatal outcome (reactive 15-beat=best outcomes; reactive 10-beat=mid-range outcomes; nonreactive=worst outcomes; P<.001). There were statistically significant unadjusted associations between presence of decelerations and higher rates of resuscitation, 5-minute Apgar score less than 7, mechanical ventilation, intraventricular hemorrhage, and hypotension (all P<.03), but not for perinatal death or necrotizing enterocolitis (P>.08).
Using logistic regression to calculate independent associations with outcome (relative to 15-beat reactivity) of reactive 10-beat or nonreactive and decelerations, 10-beat reactivity initially appeared to be associated with worse outcome than was 15-beat reactivity for all outcomes except low 5-minute Apgar score (Table 5). With the exception of necrotizing enterocolitis, a nonreactive NST was associated with worse outcome than a reactive NST by either criterion. Presence of decelerations was associated with higher rates of resuscitation and 5-minute Apgar score less than 7 than was absence of decelerations. When gestational age and birth weight were included in the regression to adjust for possible confounding, however, there no longer were any statistically significant differences between 10-beat and 15-beat criteria with any perinatal outcome, although there was a trend that did not quite meet statistical significance toward more perinatal deaths when the final NST was reactive by the 10-beat rather than 15-beat criterion (P=.05). After adjustment for gestational age and birth weight, the only outcome associated with nonreactive NSTs was perinatal death (relative to 15-beat reactivity, P<.001). Decelerations were not associated with adverse outcome after adjustment for gestational age and birth weight. When indication for NST, NST-to-delivery time, and route of delivery were added to the regression, the only change in results was that cesarean delivery was associated with higher frequency of resuscitation, low Apgar score, and intraventricular hemorrhage. The significance of the NST results did not change (data not shown).
Post hoc power calculations were performed using as baseline the frequencies of adverse outcomes after a proximate 15-beat NST. For a doubling in frequency of adverse outcome in the 10-beat as compared with the 15-beat group at α of 0.05, the study had 89–99% power to show differences in all outcomes except hypotension (74% power) and necrotizing enterocolitis (64% power).
For the 97 women with three NSTs in the week preceding delivery, “consistent” was the most frequent pattern (n=36; 37.1%), followed by “increasing” (n=23; 23.7%) and “decreasing” (n=22; 22.7%), with “irregular” being least frequent (n=15; 16.5%). There was no significant association between pattern and gestational age (Table 2) or outcome (Table 6), or pattern and indication for NST (P=.32, data not shown). Comparing women who had three NSTs in 1 week with whose with less than three NSTs, there were no differences in indications for NST (P=.15, data not shown). Because of small numbers in each group, however, power to detect differences in most outcomes was low, although there were no consistent trends between the increasing and decreasing groups and outcomes.
Although the appropriate criterion for interpretation of NST in the severely preterm pregnancy has not been well-established, the use of the 10-beat criterion has gained popularity as a means of minimizing unnecessary follow-up antepartum testing. In our study, there first appeared to be a graded worsening in most perinatal outcomes associated with NST criteria: reactive by 15-beat had the best outcomes, followed by reactive by 10-beat having intermediate outcomes, followed by nonreactive portending the least favorable outcomes. Because of the association between reactivity and gestational age (in this study and others) and because of the graded associations of increased risk of adverse outcomes with decreasing gestational age,11 when degree of reactivity was adjusted for gestational age and birth weight, there no longer were any statistically significant differences between the 10-beat and 15-beat criteria in predicting perinatal outcome. After adjustment, although nonreactivity still was associated with increased risk of perinatal death, it was not associated with any other perinatal outcome. With the exception of nonreactivity and perinatal death, gestational age and birth weight—regardless of the NST results—were the primary predictors for most perinatal outcomes in neonates born at less than 33 weeks. In addition, variable fetal heart rate decelerations were common and had no demonstrable predictive value regarding fetal outcome in these preterm fetuses. These findings are consistent with those of other investigators.12,13
There has been uncertainty as to whether the maturational effect is such that a fetus that has met 15-beat criteria once should always thereafter be expected to meet 15-beat criteria,14,15 or whether meeting it once may reflect catching an epoch by chance during which there were two 15-beat accelerations. The maturational theory states “when (reactivity by 15-beats) is exhibited at an early age ... it is expected that the fetus will continue to achieve a reactive (by 15-beats) result during all subsequent monitoring sessions.”14 “Once a 15-beat, always a 15-beat,” although theoretically plausible based on the neurologic maturational theory of NST, was not borne out by our data. Outcomes did not differ depending on the consistency or variation in NST results (although the numbers were small for some pattern subgroups, which may have limited statistical power). Our results are more consistent with those of other investigators' reports that there will be some 20-minute epochs even in preterm gestations in which by chance there will be two 15-beat accelerations.16 In this construct, randomly catching such an epoch does not mean that the next epoch also will have two 15-beat accelerations or that subsequent reactivity by 10-beats is less reassuring if a previous NST had met 15-beat criteria any more than a term NST with 15-beat accelerations is less reassuring if the previous NST had 20-beat accelerations.
Strengths of this study include a novel blinded approach with a large sample size and consistent reading to minimize interpretive variability (although variability was not large in this study). Although some fetal outcomes were too infrequent for analysis, we achieved adequate power for multiple perinatal outcomes. Using Ovid and Medline, we used the terms “preterm,” “premature,” “nonstress testing,” “antepartum testing,” “criteria,” and “criterion,” with no restriction on language or time. Various publications acknowledge the use of modified criteria for interpreting preterm NSTs,6,17 but this is the first study we have been able to find that examines the relation of different criteria to outcome.
Limitations of our study include the inherent subjectivity of NST interpretation and the retrospective nature of the analysis. Although the hospital practice is that preterm NSTs meeting the 10-beat criterion were considered reassuring and that nonreactive NSTs required some form of follow-up testing (compared with delivery), management was not restricted to an experimental protocol. Another limitation is that there is no group that did not undergo NST, ie, there was no untested control group. It is possible that if there were such a group (however inconsistent with current standards of care), their outcomes may have been worst of all—or perhaps no different! Many outcomes analyzed in this study have associations both with gestational age and with hypoxemia,11,18 whereas neurologic outcomes more strongly associated with fetal hypoxia (hypoxic-ischemic encephalopathy, PVL, seizures) were rare in our study and could not be analyzed separately. There was no obvious trend, however, between NST result and these outcomes. Information on long-term outcomes (eg, late PVL, cerebral palsy) was not available in our databases.
In our population, reactivity by 10 beats appears to be as predictive of perinatal outcome in pregnancies lasting less than 33 weeks as is reactivity by 15 beats. Variable decelerations are common and usually have little relation to outcome in markedly preterm gestations. Gestational age and birth weight are much more strongly associated with preterm perinatal outcome than are NST results. Because of the high morbidity and mortality associated with gestational age less than 33 weeks, the threshold for delivery generally should be weighted in favor of gaining time in utero than in the NST results alone.
1. Phelan JP. The nonstress test: a review of 3,000 tests. Am J Obstet Gynecol 1981;139:7–10.
2. Druzin ML, Fox A, Kogut E, Carlson C. The relationship of the nonstress test to gestational age. Am J Obstet Gynecol 1985;153:386–9.
3. Gagnon R, Campbell K, Hunse C, Patrick J. Patterns of human fetal heart rate accelerations from 26 weeks to term. Am J Obstet Gynecol 1987;157:743–8.
4. Natale R, Nasello C, Turliuk R. The relationship between movements and accelerations in fetal heart rate at twenty-four to thirty-two weeks' gestation. Am J Obstet Gynecol 1984;148:591–5.
5. Harman CR. Assessment of Fetal Health. In: Creasy RK, Resnik R, Iams JD, Lockwood CJ, Moore TM, editors. Maternal-fetal medicine: principle and practice. 6th ed. Philadelphia (PA): Saunders; 2009. p. 361–95.
6. Management of intrapartum fetal heart rate tracings. ACOG Practice Bulletin No. 116. American College of Obstetricians and Gynecologists. Obstet Gynecol 2010;116:1–9.
7. Amankwah KS, Kaufmann RC, Roller RW, Prentice RL. A new definition of the nonstress test. Obstet Gynecol 1980;56:48–51.
8. Koller WS Jr, Curet LB. Fetal activity determinations and oxytoxin challenge tests for assessment of fetal well-being. Obstet Gynecol 1978;52:176–8.
9. Mendenhall HW, O'Leary JA, Phillips KO. The nonstress test: the value of a single acceleration in evaluating the fetus at risk. Am J Obstet Gynecol 1980;136:87–91.
10. Macones GA, Hankins GD, Spong CY, Hauth J, Moore T. The 2008 National Institute of Child Health and Human Development workshop report on electronic fetal monitoring. Obstet Gynecol 2008;112:661–6.
11. Hack M, Fanaroff AA. Outcomes of children of extremely low birthweight and gestational age in the 1990s. Semin Neonatol 2000;5:89–106.
12. Meis PJ, Ureda JR, Swain M, Kelly RT, Penry M, Sharp P. Variable decelerations during nonstress tests are not a sign of fetal compromise. Am J Obstet Gynecol 1986;154:586–90.
13. Sorokin Y, Dierker LJ, Pillay SK, Zador IE, Schreiner ML, Rosen MG. The association between fetal heart rate patterns and fetal movements in pregnancies between 20 and 30 weeks' gestation. Am J Obstet Gynecol 1982;143:243–9.
14. Menihan CA, Zottoli EK. Electronic fetal monitoring: concepts and applications. Philadelphia (PA): Lippincott; 2001.
15. Premature rupture of membranes. ACOG Practice Bulletin No. 80. American College of Obstetricians and Gynecologists. Obstet Gynecol 2007;80:1–13.
16. Baskett TF. Gestational age and fetal biophysical assessment. Am J Obstet Gynecol 1988;158:332–4.
17. Druzin ML, Smith JF, Gabbe SG, Reed KL. Antepartum Fetal Evaluation. In: Gabbe SG, Niebyl JR, Simpson JL, editors. Obstetrics: normal and problem pregnancies. 5th ed. Philadelphia (PA): Churchill Livingstone; 2007. p. 267–300.
18. Freeman RK, Garite TJ, Nageotte MP. Fetal heart rate monitoring. 3rd ed. Philadelphia (PA): Lippincott Williams & Wilkins; 2003.