- Email: [email protected]
A randomized clinical trial of daily nonstress testing versus biophysical profile in the management of preterm premature rupture of membranes
A randomized clinical trial of daily nonstress testing versus biophysical profile in the management of preterm premature rupture of membranes David F. Lewis, MD,a C. David Adair, MD,a Jonathan W. Weeks, MD,a P. Scott Barrilleaux, MD,a Michael S. Edwards, MD,a and Thomas J. Garite, MDb Shreveport, Louisiana, and Orange, California OBJECTIVE: Our purpose was to evaluate the ability of 2 different antepartum testing modalities to predict infectious morbidity in patients with preterm premature rupture of membranes. STUDY DESIGN: During a 36-month period, patients with preterm premature rupture of membranes (at 23 to 34 weeks of gestation) were randomly assigned to either a daily nonstress test or a biophysical profile, after a 24-hour observational period. We used the original scoring system of Manning et al for the biophysical profile, with a score of ≤6 considered abnormal. Nonstress test results were considered abnormal if the test was nonreactive or if the patient had late decelerations or significant variable decelerations; abnormal results led to further evaluation with a biophysical profile. Results of the last test before delivery were evaluated to determine whether infectious complications had been predicted. RESULTS: One hundred thirty-five patients were enrolled in the study. Demographics, pregnancy characteristics, and neonatal outcomes were similar. Neither the daily nonstress test nor the daily biophysical profile had good sensitivity for predicting infectious complications (39.1% and 25.0%, respectively). However, both had good specificity (84.6% and 92.6%, respectively). Positive and negative predictive values were 52.9% and 75.9%, respectively, for the daily nonstress test and 66.7% and 68.4%, respectively, for the daily biophysical profile. Cost was significantly higher in the daily biophysical profile group. Nonstress testing of patients at <28 weeks’ gestation generally required a backup biophysical profile. CONCLUSION: Neither the daily nonstress test nor the daily biophysical profile had good sensitivity for predicting infectious complications after preterm premature rupture of membranes. (Am J Obstet Gynecol 1999;181:1495-9.)
Key words: Preterm premature rupture of membranes, antenatal testing, intra-amniotic infections, neonatal sepsis
Preterm premature rupture of membranes complicates >1% of all pregnancies.1 Currently, management that is both expectant and conservative is recommended unless fetal lung maturity is documented or infectious complications are diagnosed.2, 3 The rationale of this management scheme is that the risk of prematurity and its attendant morbidities outweighs the risk of perinatal or neonatal infections. Pregnancies complicated by preterm premature rupture of membranes result in an increased risk of perinatal morbidity and death. In an effort to decrease neonatal From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Louisiana State University Medical Center,a and the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of California-Irvine.b Received for publication November 30, 1998; revised January 18, 1999; accepted June 9, 1999. Reprint requests: David F. Lewis, MD, Woman’s Hospital, MaternalFetal Medicine Center, 9050 Airline Highway, Baton Rouge, LA 70815. Copyright © 1999 by Mosby, Inc. 0002-9378/99 $8.00 + 0 6/1/100663
morbidity, antepartum fetal testing has been advocated.4 Some authors recommend a daily nonstress test,5, 6 whereas others contend that a full biophysical profile is necessary.7, 8 The purpose of this investigation was to compare the efficacy of both recommended testing modalities in pregnancies complicated by preterm premature rupture of membranes. Specifically, we sought to evaluate the ability of each test to predict infectious morbidity of the mother and neonate. Material and methods All patients with preterm premature rupture of membranes (at ≤34 weeks’ gestation) admitted to Louisiana State University School of Medicine in Shreveport were asked to participate in this randomized clinical trial. Preterm premature rupture of membranes was diagnosed by a history of fluid leakage with confirmation by either sterile speculum examination documenting ferning or positive results on Nitrazine evaluation or both. At the time of diagnosis, the cervix was visually assessed for cervical dilatation and effacement. We obtained appro1495
1496 Lewis et al
December 1999 Am J Obstet Gynecol
Table I. Maternal demographics Nonstress test (n = 69) Age (y, mean ± SD) Race (No.) White African American Primigravid (No.) History of preterm delivery (No.)
Biophysical profile (n = 66)
24.2 ± 7.0
25.7 ± 6.6
19 (27.5%) 50 (72.5%) 21 (30.4%) 14 (20.3%)
14 (21.2%) 52 (78.8%) 13 (19.7%) 16 (24.3%)
Table II. Obstetric pregnancy characteristics
Estimated gestational age at admittance (wk, mean ± SD) Latency period (d, mean ± SD) Antibiotic use (d, mean ± SD) Estimated fetal weight at admittance (g, mean ± SD) Amniocentesis (No.) Endometritis (No.) Apgar score ≤7 at 5 min (No.) Delivery for maturity (No.) Cesarean delivery (No.) Corticosteroid use (No.)
Nonstress test (n = 69)
Biophysical profile (n = 66)
29.7 ± 3.0 13.6 ± 11.3 6.17 ± 1.5 1492.1 ± 549 32 (46.4%) 5 (7.3%) 9 (13.1%) 23 (33.3%) 14 (20.3%) 34 (49.3%)
29.0 ± 3.4 12.6 ± 11.4 5.85 ± 1.7 1347.3 ± 522 28 (42.4%) 4 (6.1%) 11 (16.7%) 14 (21.2%) 21 (31.8%) 37 (56.1%)
priate culture specimens to ascertain whether group B Streptococcus, Neisseria gonorrhoeae, Chlamydia trachomatis, or bacterial vaginosis was present. Bacterial vaginosis was diagnosed by microscope evaluation and with the “whiff test.” All patients received ampicillin-sulbactam (Unasyn) or, if allergic to penicillin, clindamycin (Cleocin), administered intravenously until delivery or for a total of 7 days. If specific pathogens were identified, appropriate antibiotic selection was used for treatment. Some patients (55 patients) received corticosteroids and were included in a previous report.9 Tocolytic agents were not used. All patients underwent complete ultrasonography and a biophysical profile on admission. Some patients underwent amniocentesis to rule out infection and to assess for fetal lung maturity. These patients were usually at >32 weeks’ gestation, or the clinicians suspected intra-amniotic infection. All patients remained in the labor and delivery unit for at least 24 hours with continuous electronic fetal monitoring and hourly monitoring of maternal vital signs. All patients without obvious clinical infection or without a condition requiring immediate delivery, who had remained stable for 24 hours before transfer to our antenatal ward, were asked to participate in this randomized trial. All patients with preterm premature rupture of membranes, including patients who had undergone cerclage or digital vaginal examination before tertiary transfer, were included. No digital examinations were performed in our institution. Patients were randomly assigned to undergo either daily nonstress testing or daily biophysical profiling after all volunteers provided informed consent. Participants underwent fre-
quent maternal temperature assessments and were asked to perform fetal-kick counts 4 times daily. A nonstress test was considered reactive if it resulted in 2 accelerations with 15 beats/min above the baseline that lasted for ≥15 seconds during a 20-minute period. The test result was considered abnormal if these criteria were not met, a late deceleration occurred, or a significant variable deceleration (30 beats for 30 seconds) occurred. All tests were reviewed by the maternal-fetal medicine faculty. Patients with abnormal results on nonstress testing had a complete biophysical profile as a backup confirmatory test. The biophysical profile, as described by Manning et al,10 was used with the modification of fluid assessment described by Chamberlain et al11 (total score of 10, with 2 points each for tone, amniotic fluid, movement, breathing, and nonstress test). With an initial score of ≤6, the test was repeated within 12 hours, with continuous fetal monitoring on its first occurrence. A score of ≤6 was considered abnormal. A score of ≤4 prompted delivery if it persisted when the examination was extended to a 1hour interval. For the purpose of this study, the last test score obtained before delivery was used for data analysis. All tests were performed by a third- or fourth-year resident in obstetrics and gynecology or by the attending maternal-fetal medicine faculty. Delivery was prompted by spontaneous labor, clinical evidence of intra-amniotic infection, a mature fetal lung profile, or abnormal antenatal fetal test results with confirmation by a faculty member. All patients received antibiotics during the intrapartal period for prophylaxis against group B Streptococcus.The
Lewis et al 1497
Volume 181, Number 6 Am J Obstet Gynecol
Table III. Neonatal characteristics Nonstress test (n = 69) Hospital stay (d, mean ± SD) Admittance weight (g, mean ± SD) Antibiotic use (d, mean ± SD) Hyaline membrane disease (No.) Intraventricular hemorrhage (No.) Necrotizing enterocolitis (No.) Bronchopulmonary dysplasia (No.) Retinopathy of prematurity (No.)
28.2 ± 25.4 1740.8 ± 585 5.31 ± 1.8 14 (20.3%) 3 (4.4%) 2 (2.9%) 4 (5.8%) 7 (10.2%)
Biophysical profile (n = 66) 33.8 ± 39.6 1614.6 ± 495 4.95 ± 1.9 23 (34.9%) 4 (6.1%) 3 (4.6%) 5 (7.6%) 5 (7.6%)
Table IV. Incidence of infectious morbidity
Intra-amniotic infections (No.) Diagnosis by amniocentesis (No.) Sepsis (No.) Presumed sepsis (No.) ≥2 infections (No.) Total infections*
Nonstress test (n = 69)
Biophysical profile (n = 66)
11 (15.9%) 1 3 (4.4%) 11 (15.9%) 2 (2.9) 23 (33.3%)
19 (28.8%) 2 1 (1.5%) 10 (15.2%) 6 (9.0) 24 (36.4%)
*Total infections equal number of patients with ≥1 infections.
main outcome variables were intra-amniotic infection, neonatal pneumonia, sepsis, and presumed sepsis. Intraamniotic infection was diagnosed clinically, by analysis of amniotic fluid obtained from an amniocentesis (positive Gram stain or culture), or by the findings of a maternal temperature of 100.4°F, foul-smelling fluid, and uterine tenderness. The diagnosis of neonatal sepsis required positive results on culture of blood or spinal fluid or the presence of congenital pneumonia. Neonatal or congenital pneumonia was diagnosed by the neonatal staff and required a positive radiographic finding plus evidence of sepsis. The diagnosis of presumed sepsis was made by the attending neonatologist and included clinical signs of infection with negative culture results and an abnormal leukocyte count (leukopenia, ≤5000 cells/mm3; neutropenia, ≤1500 cells/mm3; or leukocytosis, ≥28,000 cells/mm3 with a left shift). The clinical signs of presumed sepsis included shock, poor perfusion, temperature instability, respiratory distress, hypotonia, lethargy, and feeding intolerance. On the basis of a 30% incidence of infectious morbidity7 (all the measured infections, including intra-amniotic infection, sepsis, and presumed sepsis) in patients with preterm premature rupture of the membranes, 90% sensitivity of the biophysical profile,7 and 45% sensitivity of the nonstress test12, 13 for predicting infections, we calculated that 133 patients would be needed for 80% power (β = .2). Statistical analysis included the Fisher exact test, χ2 analysis, 2-tailed Student t test, and the Mann-Whitney U test. Randomization was accomplished by using random number tables with a 1:1 match. Cards
were kept in opaque envelopes and not opened until after informed consent was obtained. For cost analysis, $100 was used for a nonstress test and $250 for a biophysical profile. Results Within a 36-month period, 135 patients were enrolled and completed the study. Sixty-six patients were randomly selected for daily biophysical profiles and 69 patients for daily nonstress tests. Thirty-seven of the patients randomly selected for the daily nonstress tests required 147 backup biophysical profiles. Tables I, II, and III contain demographic data, pregnancy characteristics, and neonatal characteristics, respectively, for patients in the daily nonstress test and the daily biophysical profile groups. No significant differences (P ≥ .05) in baseline characteristics were found between these groups. The incidence of infectious morbidity is presented in Table IV. No differences were found between the 2 groups. Of the patients randomly selected to undergo daily nonstress testing, 23 (33%) had a total of 25 infections, and of the patients randomly selected to have daily biophysical profiles, 24 (36.4%) had a total of 30 infections. To evaluate the effectiveness of these antepartum tests for predicting infection, we used data from the last test before delivery. Seventeen patients had abnormal nonstress test results; 9 of these tests correctly predicted infections in the total of 23 patients in whom infection developed. Nine patients in the group monitored with biophysical profiles had abnormal scores, with 6 profiles
1498 Lewis et al
December 1999 Am J Obstet Gynecol
Table V. Clinical relevance of nonstress tests and biophysical profile for predicting infectious morbidity
Daily nonstress test Daily biophysical profile Backup biophysical profile Neonatal infection* Nonstress test Biophysical profile Backup biophysical profile
Sensitivity (%)
Specificity (%)
Positive predictive value (%)
Negative predictive value (%)
39.1 25.0 33.3
84.6 92.9 87.5
52.9 66.7 75.0
75.9 68.4 53.9
42.9 27.3 33.3
83.3 89.1 81.8
35.3 33.3 50.0
87.3 86.0 69.2
*Neonatal infection included presumed sepsis, sepsis, and congenital pneumonia.
correctly predicting infections in the 24 infected patients. We also evaluated the ability of the 2 tests to predict neonatal infections (presumed sepsis and sepsis). There were 14 neonatal infections in the group of patients having daily nonstress tests, with 6 infections being correctly predicted by antenatal testing; 11 neonatal infections occurred in the group having daily biophysical profiles, with 3 infections being correctly predicted. Table V presents the clinical utility of the 2 tests at predicting infectious morbidity. Neither test had acceptable sensitivity or positive predictive values. The cost per patient of a daily biophysical profile was considerably higher than the cost of a daily nonstress test with a backup biophysical profile when necessary ($2917 vs $1731; P < .001). Of the patients in the nonstress test group, 27 (40%) were at ≤28 weeks’ gestation, and those patients required 103 (71%) of the 147 backup tests. Comment Antepartum fetal testing was developed to detect fetal hypoxia and acidosis and to prevent fetal deaths by timely intervention.4 Use of these tests in patients with preterm premature rupture of membranes has been recommended because of the high rate of perinatal morbidity and intrapartum fetal distress.5, 6 The original test was the nonstress test. Several authors demonstrated good sensitivity of a nonreactive nonstress test and its ability to detect infectious morbidity.12, 13 The biophysical profile, since introduced by Manning et al,10 has been shown to detect poor fetal condition and has a high direct correlation with fetal and neonatal acidbase status. Vintzileos et al7 first suggested its use in the management of preterm premature rupture of membranes and noted its excellent sensitivity for predicting infections. These authors modified the scoring system to obtain these results, whereas others have found dissimilar results14-16 using traditional scoring systems. In a prospective study, Carroll et al17 also found normal biophysical profile scores in the majority of infected pregnancies complicated by preterm premature rupture of
membranes. Several possibilities could explain the differences between our data and those of Vintzileos et al.7 First, we did not use their scoring modifications because they are not widely used, and we wanted our data to be readily extrapolated. Second, management practices have changed since the original data of Vintzileos et al7 were presented. These changes include the use of antibiotics and corticosteroids, plus a more aggressive use of amniocentesis for the detection of early infections. Many infected patients would not be in the study because of delivery. Third, we included only patients who remained stable for 24 hours, and many infected patients would have been excluded because of spontaneous delivery. It must also be emphasized that in the original work of Vintzileos et al7 the biophysical profile was not used to time delivery. Therefore, if the test is an “early” predictor of infections, some patients in our study would have been delivered, improving their outcome. We also used fetalkick counts, which have been shown to indicate fetal compromise.18 The ability of the nonstress test to predict infectious morbidity has also been called into question. One study reported that preterm premature rupture of membranes promotes a more reactive fetal heart rate tracing.8 Several retrospective studies evaluating intrapartum fetal heart rate analysis in cases of known sepsis noted that the majority of the tests showed normal results.12, 19 Gonen et al13 refuted the ability of a nonstress test to predict congenital sepsis, noting that its sensitivity was only 15.9%. Our study supports the findings of Carroll et al,17 who noted a lower rate of nonreactive nonstress tests in pregnancies complicated by infections with preterm premature rupture of membranes than in those without infections. Although we found that the nonstress test and the biophysical profile have poor sensitivities, we are not suggesting that these tests be disregarded. In fact, the specificities and negative predictive values are acceptable, and a normal test result is reassuring. Reassurance with this major complication is important. We believe that a daily nonstress test is sufficient in most cases; however, daily
Lewis et al 1499
Volume 181, Number 6 Am J Obstet Gynecol
biophysical profiles may be more reasonable in extreme prematurity (≤28 weeks’ gestation). Previous work has shown that decreased amniotic fluid volume,19, 20 a nonreactive nonstress test,20, 21 and infections are more likely when premature rupture of membranes occurs at ≤28 weeks than later in gestation.22 Patients who were at <28 weeks of gestation and who were monitored with a daily nonstress test accounted for the majority of the backup biophysical profiles in our series. In our current health care system, cost is of paramount concern. A significant increase in health care expenditures was found in the group receiving a daily biophysical profile. The total daily testing expenditure was $78,000 greater than that for the daily nonstress test with a backup biophysical profile test. This study represents the first prospective clinical trial comparing these 2 tests in clinical practice. Both seem to be reasonable alternatives; however, because of significant cost differences, we currently use a daily nonstress test in gestations >28 weeks, with a backup biophysical profile for an abnormal test result and a daily biophysical profile for all gestations ≤28 weeks. Our data support this clinical approach in terms of efficacy and cost savings.
8.
9.
10.
11.
12.
13. 14.
15.
16.
REFERENCES
1. Wolf EJ, Mallozzi A, Rodis JF, Campbell WA, Vintzileos AM. The principal pregnancy complications resulting in preterm birth in singleton and twin gestations. J Matern Fetal Med 1992;1:20612. 2. American College of Obstetricians and Gynecologists. Premature rupture of membranes. Washington: The College; 1988. ACOG Technical Bulletin No.: 115 . 3. Garite TJ. Premature rupture of the membranes: the enigma of the obstetrician. Am J Obstet Gynecol 1985;151:1001-5. 4. Vintzileos AM, Campbell WA. Antepartum surveillance in patients with preterm premature rupture of the membranes. Clin Obstet Gynecol 1991;34:779-93. 5. Moberg LJ, Garite TJ, Freeman RK. Fetal heart rate patterns and fetal distress in patients with preterm premature rupture of the membranes. Obstet Gynecol 1984;64;60-4. 6. Vintzileos AM, Campbell WA, Nochimson DJ, Weinbaum PJ. The use of the non-stress test in patients with premature rupture of the membranes. Am J Obstet Gynecol 1986;155:149-53. 7. Vintzileos AM, Campbell WA, Nochimson DJ, Connolly ME, Fuenfer MM, Hoehn KJ. The fetal biophysical profile in patients
17.
18.
19. 20.
21.
22.
with premature rupture of membranes: an early predictor of fetal infection. Am J Obstet Gynecol 1985;152:510-6. Roberts AB, Goldstein I, Romero R, Hobbins JC. Comparison of total fetal activity measurement with the biophysical profile in predicting intra-amniotic infection in preterm premature rupture of membranes. Ultrasound Obstet Gynecol 1991;1:36-9. Lewis DF, Brody K, Edwards MS, Brouillette RM, Burlison S, London SN. Preterm premature ruptured membranes: a randomized trial of steroids after treatment with antibiotics. Obstet Gynecol 1996;88:801-5. Manning FA, Baskeh TF, Morrison I, Lange I. Fetal biophysical profile scoring: a prospective study in 1184 high-risk patients. Am J Obstet Gynecol 1981;140:284-93. Chamberlain PF, Manning FA, Morrison I, Harman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume: the relationship of marginal and decreased amniotic fluid volume to perinatal outcome. Am J Obstet Gynecol 1984;150:245-9. Day D, Ugol JH, French JI, Haverkemp A, Will RE, McGregor JA. Fetal monitoring in perinatal sepsis. Am J Perinatol 1992;9:28-33. Gonen R, Ohlsson A, Farine D, Mulligan JE. Can the non-stress test predict congenital sepsis? Am J Perinatol 1991;8:91-3. Gauthier DW, Meyer WJ, Bieniarz A. Biophysical profile as a predictor of amniotic fluid culture results. Obstet Gynecol 1992;80:102-5. DelValle GO, Joffe GM, Izquierdo LA, Smith JF, Gilson FT, Curet LB. The biophysical profile and the non-stress test: poor predictors of chorioamnionitis and fetal infection in prolonged preterm premature rupture of membranes. Obstet Gynecol 1992;80:106-10. Miller J, Kho MS, Brown HL, Gubert HA. Clinical chorioamnionitis is not predicted by an ultrasound biophysical profile in patients with premature rupture of membranes. Obstet Gynecol 1990;76:1051-4. Carroll SG, Papaioannou S, Nicolaides KH. Assessment of fetal activity and amniotic fluid volume in the prediction of intrauterine infection in preterm prelabor amniorrhexis. Am J Obstet Gynecol 1995;172:1427-35. Moore T, Piacquadio K. A prospective assessment of fetal movement screening to reduce fetal mortality. Am J Obstet Gynecol 1989;160:1075-9. Youchah J, Chazotte C, Cohen WR. Heart rate patterns and fetal sepsis. Am J Perinatol 1989;6:356-9. Hardin JA, Jackson DN, Lewis DF, Major CA, Nageotte MP, Asrat T. Correlation of the amniotic fluid index and the non-stress test in the patient with preterm premature rupture of membranes. Am J Obstet Gynecol 1991;165:1088-94. Zeevi D, Sodovsky E, Youris J, Granat M, Ohel G. Antenatal fetal heart rate characteristics in cases of premature rupture of membrane. Am J Perinatol 1988;5:260-3. Gonik B, Bottoms SSF, Cotton DB. Amniotic fluid volume as a risk factor in preterm premature rupture of the membranes. Obstet Gynecol 1985;65:456-9.
Key words: Preterm premature rupture of membranes, antenatal testing, intra-amniotic infections, neonatal sepsis
Preterm premature rupture of membranes complicates >1% of all pregnancies.1 Currently, management that is both expectant and conservative is recommended unless fetal lung maturity is documented or infectious complications are diagnosed.2, 3 The rationale of this management scheme is that the risk of prematurity and its attendant morbidities outweighs the risk of perinatal or neonatal infections. Pregnancies complicated by preterm premature rupture of membranes result in an increased risk of perinatal morbidity and death. In an effort to decrease neonatal From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Louisiana State University Medical Center,a and the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of California-Irvine.b Received for publication November 30, 1998; revised January 18, 1999; accepted June 9, 1999. Reprint requests: David F. Lewis, MD, Woman’s Hospital, MaternalFetal Medicine Center, 9050 Airline Highway, Baton Rouge, LA 70815. Copyright © 1999 by Mosby, Inc. 0002-9378/99 $8.00 + 0 6/1/100663
morbidity, antepartum fetal testing has been advocated.4 Some authors recommend a daily nonstress test,5, 6 whereas others contend that a full biophysical profile is necessary.7, 8 The purpose of this investigation was to compare the efficacy of both recommended testing modalities in pregnancies complicated by preterm premature rupture of membranes. Specifically, we sought to evaluate the ability of each test to predict infectious morbidity of the mother and neonate. Material and methods All patients with preterm premature rupture of membranes (at ≤34 weeks’ gestation) admitted to Louisiana State University School of Medicine in Shreveport were asked to participate in this randomized clinical trial. Preterm premature rupture of membranes was diagnosed by a history of fluid leakage with confirmation by either sterile speculum examination documenting ferning or positive results on Nitrazine evaluation or both. At the time of diagnosis, the cervix was visually assessed for cervical dilatation and effacement. We obtained appro1495
1496 Lewis et al
December 1999 Am J Obstet Gynecol
Table I. Maternal demographics Nonstress test (n = 69) Age (y, mean ± SD) Race (No.) White African American Primigravid (No.) History of preterm delivery (No.)
Biophysical profile (n = 66)
24.2 ± 7.0
25.7 ± 6.6
19 (27.5%) 50 (72.5%) 21 (30.4%) 14 (20.3%)
14 (21.2%) 52 (78.8%) 13 (19.7%) 16 (24.3%)
Table II. Obstetric pregnancy characteristics
Estimated gestational age at admittance (wk, mean ± SD) Latency period (d, mean ± SD) Antibiotic use (d, mean ± SD) Estimated fetal weight at admittance (g, mean ± SD) Amniocentesis (No.) Endometritis (No.) Apgar score ≤7 at 5 min (No.) Delivery for maturity (No.) Cesarean delivery (No.) Corticosteroid use (No.)
Nonstress test (n = 69)
Biophysical profile (n = 66)
29.7 ± 3.0 13.6 ± 11.3 6.17 ± 1.5 1492.1 ± 549 32 (46.4%) 5 (7.3%) 9 (13.1%) 23 (33.3%) 14 (20.3%) 34 (49.3%)
29.0 ± 3.4 12.6 ± 11.4 5.85 ± 1.7 1347.3 ± 522 28 (42.4%) 4 (6.1%) 11 (16.7%) 14 (21.2%) 21 (31.8%) 37 (56.1%)
priate culture specimens to ascertain whether group B Streptococcus, Neisseria gonorrhoeae, Chlamydia trachomatis, or bacterial vaginosis was present. Bacterial vaginosis was diagnosed by microscope evaluation and with the “whiff test.” All patients received ampicillin-sulbactam (Unasyn) or, if allergic to penicillin, clindamycin (Cleocin), administered intravenously until delivery or for a total of 7 days. If specific pathogens were identified, appropriate antibiotic selection was used for treatment. Some patients (55 patients) received corticosteroids and were included in a previous report.9 Tocolytic agents were not used. All patients underwent complete ultrasonography and a biophysical profile on admission. Some patients underwent amniocentesis to rule out infection and to assess for fetal lung maturity. These patients were usually at >32 weeks’ gestation, or the clinicians suspected intra-amniotic infection. All patients remained in the labor and delivery unit for at least 24 hours with continuous electronic fetal monitoring and hourly monitoring of maternal vital signs. All patients without obvious clinical infection or without a condition requiring immediate delivery, who had remained stable for 24 hours before transfer to our antenatal ward, were asked to participate in this randomized trial. All patients with preterm premature rupture of membranes, including patients who had undergone cerclage or digital vaginal examination before tertiary transfer, were included. No digital examinations were performed in our institution. Patients were randomly assigned to undergo either daily nonstress testing or daily biophysical profiling after all volunteers provided informed consent. Participants underwent fre-
quent maternal temperature assessments and were asked to perform fetal-kick counts 4 times daily. A nonstress test was considered reactive if it resulted in 2 accelerations with 15 beats/min above the baseline that lasted for ≥15 seconds during a 20-minute period. The test result was considered abnormal if these criteria were not met, a late deceleration occurred, or a significant variable deceleration (30 beats for 30 seconds) occurred. All tests were reviewed by the maternal-fetal medicine faculty. Patients with abnormal results on nonstress testing had a complete biophysical profile as a backup confirmatory test. The biophysical profile, as described by Manning et al,10 was used with the modification of fluid assessment described by Chamberlain et al11 (total score of 10, with 2 points each for tone, amniotic fluid, movement, breathing, and nonstress test). With an initial score of ≤6, the test was repeated within 12 hours, with continuous fetal monitoring on its first occurrence. A score of ≤6 was considered abnormal. A score of ≤4 prompted delivery if it persisted when the examination was extended to a 1hour interval. For the purpose of this study, the last test score obtained before delivery was used for data analysis. All tests were performed by a third- or fourth-year resident in obstetrics and gynecology or by the attending maternal-fetal medicine faculty. Delivery was prompted by spontaneous labor, clinical evidence of intra-amniotic infection, a mature fetal lung profile, or abnormal antenatal fetal test results with confirmation by a faculty member. All patients received antibiotics during the intrapartal period for prophylaxis against group B Streptococcus.The
Lewis et al 1497
Volume 181, Number 6 Am J Obstet Gynecol
Table III. Neonatal characteristics Nonstress test (n = 69) Hospital stay (d, mean ± SD) Admittance weight (g, mean ± SD) Antibiotic use (d, mean ± SD) Hyaline membrane disease (No.) Intraventricular hemorrhage (No.) Necrotizing enterocolitis (No.) Bronchopulmonary dysplasia (No.) Retinopathy of prematurity (No.)
28.2 ± 25.4 1740.8 ± 585 5.31 ± 1.8 14 (20.3%) 3 (4.4%) 2 (2.9%) 4 (5.8%) 7 (10.2%)
Biophysical profile (n = 66) 33.8 ± 39.6 1614.6 ± 495 4.95 ± 1.9 23 (34.9%) 4 (6.1%) 3 (4.6%) 5 (7.6%) 5 (7.6%)
Table IV. Incidence of infectious morbidity
Intra-amniotic infections (No.) Diagnosis by amniocentesis (No.) Sepsis (No.) Presumed sepsis (No.) ≥2 infections (No.) Total infections*
Nonstress test (n = 69)
Biophysical profile (n = 66)
11 (15.9%) 1 3 (4.4%) 11 (15.9%) 2 (2.9) 23 (33.3%)
19 (28.8%) 2 1 (1.5%) 10 (15.2%) 6 (9.0) 24 (36.4%)
*Total infections equal number of patients with ≥1 infections.
main outcome variables were intra-amniotic infection, neonatal pneumonia, sepsis, and presumed sepsis. Intraamniotic infection was diagnosed clinically, by analysis of amniotic fluid obtained from an amniocentesis (positive Gram stain or culture), or by the findings of a maternal temperature of 100.4°F, foul-smelling fluid, and uterine tenderness. The diagnosis of neonatal sepsis required positive results on culture of blood or spinal fluid or the presence of congenital pneumonia. Neonatal or congenital pneumonia was diagnosed by the neonatal staff and required a positive radiographic finding plus evidence of sepsis. The diagnosis of presumed sepsis was made by the attending neonatologist and included clinical signs of infection with negative culture results and an abnormal leukocyte count (leukopenia, ≤5000 cells/mm3; neutropenia, ≤1500 cells/mm3; or leukocytosis, ≥28,000 cells/mm3 with a left shift). The clinical signs of presumed sepsis included shock, poor perfusion, temperature instability, respiratory distress, hypotonia, lethargy, and feeding intolerance. On the basis of a 30% incidence of infectious morbidity7 (all the measured infections, including intra-amniotic infection, sepsis, and presumed sepsis) in patients with preterm premature rupture of the membranes, 90% sensitivity of the biophysical profile,7 and 45% sensitivity of the nonstress test12, 13 for predicting infections, we calculated that 133 patients would be needed for 80% power (β = .2). Statistical analysis included the Fisher exact test, χ2 analysis, 2-tailed Student t test, and the Mann-Whitney U test. Randomization was accomplished by using random number tables with a 1:1 match. Cards
were kept in opaque envelopes and not opened until after informed consent was obtained. For cost analysis, $100 was used for a nonstress test and $250 for a biophysical profile. Results Within a 36-month period, 135 patients were enrolled and completed the study. Sixty-six patients were randomly selected for daily biophysical profiles and 69 patients for daily nonstress tests. Thirty-seven of the patients randomly selected for the daily nonstress tests required 147 backup biophysical profiles. Tables I, II, and III contain demographic data, pregnancy characteristics, and neonatal characteristics, respectively, for patients in the daily nonstress test and the daily biophysical profile groups. No significant differences (P ≥ .05) in baseline characteristics were found between these groups. The incidence of infectious morbidity is presented in Table IV. No differences were found between the 2 groups. Of the patients randomly selected to undergo daily nonstress testing, 23 (33%) had a total of 25 infections, and of the patients randomly selected to have daily biophysical profiles, 24 (36.4%) had a total of 30 infections. To evaluate the effectiveness of these antepartum tests for predicting infection, we used data from the last test before delivery. Seventeen patients had abnormal nonstress test results; 9 of these tests correctly predicted infections in the total of 23 patients in whom infection developed. Nine patients in the group monitored with biophysical profiles had abnormal scores, with 6 profiles
1498 Lewis et al
December 1999 Am J Obstet Gynecol
Table V. Clinical relevance of nonstress tests and biophysical profile for predicting infectious morbidity
Daily nonstress test Daily biophysical profile Backup biophysical profile Neonatal infection* Nonstress test Biophysical profile Backup biophysical profile
Sensitivity (%)
Specificity (%)
Positive predictive value (%)
Negative predictive value (%)
39.1 25.0 33.3
84.6 92.9 87.5
52.9 66.7 75.0
75.9 68.4 53.9
42.9 27.3 33.3
83.3 89.1 81.8
35.3 33.3 50.0
87.3 86.0 69.2
*Neonatal infection included presumed sepsis, sepsis, and congenital pneumonia.
correctly predicting infections in the 24 infected patients. We also evaluated the ability of the 2 tests to predict neonatal infections (presumed sepsis and sepsis). There were 14 neonatal infections in the group of patients having daily nonstress tests, with 6 infections being correctly predicted by antenatal testing; 11 neonatal infections occurred in the group having daily biophysical profiles, with 3 infections being correctly predicted. Table V presents the clinical utility of the 2 tests at predicting infectious morbidity. Neither test had acceptable sensitivity or positive predictive values. The cost per patient of a daily biophysical profile was considerably higher than the cost of a daily nonstress test with a backup biophysical profile when necessary ($2917 vs $1731; P < .001). Of the patients in the nonstress test group, 27 (40%) were at ≤28 weeks’ gestation, and those patients required 103 (71%) of the 147 backup tests. Comment Antepartum fetal testing was developed to detect fetal hypoxia and acidosis and to prevent fetal deaths by timely intervention.4 Use of these tests in patients with preterm premature rupture of membranes has been recommended because of the high rate of perinatal morbidity and intrapartum fetal distress.5, 6 The original test was the nonstress test. Several authors demonstrated good sensitivity of a nonreactive nonstress test and its ability to detect infectious morbidity.12, 13 The biophysical profile, since introduced by Manning et al,10 has been shown to detect poor fetal condition and has a high direct correlation with fetal and neonatal acidbase status. Vintzileos et al7 first suggested its use in the management of preterm premature rupture of membranes and noted its excellent sensitivity for predicting infections. These authors modified the scoring system to obtain these results, whereas others have found dissimilar results14-16 using traditional scoring systems. In a prospective study, Carroll et al17 also found normal biophysical profile scores in the majority of infected pregnancies complicated by preterm premature rupture of
membranes. Several possibilities could explain the differences between our data and those of Vintzileos et al.7 First, we did not use their scoring modifications because they are not widely used, and we wanted our data to be readily extrapolated. Second, management practices have changed since the original data of Vintzileos et al7 were presented. These changes include the use of antibiotics and corticosteroids, plus a more aggressive use of amniocentesis for the detection of early infections. Many infected patients would not be in the study because of delivery. Third, we included only patients who remained stable for 24 hours, and many infected patients would have been excluded because of spontaneous delivery. It must also be emphasized that in the original work of Vintzileos et al7 the biophysical profile was not used to time delivery. Therefore, if the test is an “early” predictor of infections, some patients in our study would have been delivered, improving their outcome. We also used fetalkick counts, which have been shown to indicate fetal compromise.18 The ability of the nonstress test to predict infectious morbidity has also been called into question. One study reported that preterm premature rupture of membranes promotes a more reactive fetal heart rate tracing.8 Several retrospective studies evaluating intrapartum fetal heart rate analysis in cases of known sepsis noted that the majority of the tests showed normal results.12, 19 Gonen et al13 refuted the ability of a nonstress test to predict congenital sepsis, noting that its sensitivity was only 15.9%. Our study supports the findings of Carroll et al,17 who noted a lower rate of nonreactive nonstress tests in pregnancies complicated by infections with preterm premature rupture of membranes than in those without infections. Although we found that the nonstress test and the biophysical profile have poor sensitivities, we are not suggesting that these tests be disregarded. In fact, the specificities and negative predictive values are acceptable, and a normal test result is reassuring. Reassurance with this major complication is important. We believe that a daily nonstress test is sufficient in most cases; however, daily
Lewis et al 1499
Volume 181, Number 6 Am J Obstet Gynecol
biophysical profiles may be more reasonable in extreme prematurity (≤28 weeks’ gestation). Previous work has shown that decreased amniotic fluid volume,19, 20 a nonreactive nonstress test,20, 21 and infections are more likely when premature rupture of membranes occurs at ≤28 weeks than later in gestation.22 Patients who were at <28 weeks of gestation and who were monitored with a daily nonstress test accounted for the majority of the backup biophysical profiles in our series. In our current health care system, cost is of paramount concern. A significant increase in health care expenditures was found in the group receiving a daily biophysical profile. The total daily testing expenditure was $78,000 greater than that for the daily nonstress test with a backup biophysical profile test. This study represents the first prospective clinical trial comparing these 2 tests in clinical practice. Both seem to be reasonable alternatives; however, because of significant cost differences, we currently use a daily nonstress test in gestations >28 weeks, with a backup biophysical profile for an abnormal test result and a daily biophysical profile for all gestations ≤28 weeks. Our data support this clinical approach in terms of efficacy and cost savings.
8.
9.
10.
11.
12.
13. 14.
15.
16.
REFERENCES
1. Wolf EJ, Mallozzi A, Rodis JF, Campbell WA, Vintzileos AM. The principal pregnancy complications resulting in preterm birth in singleton and twin gestations. J Matern Fetal Med 1992;1:20612. 2. American College of Obstetricians and Gynecologists. Premature rupture of membranes. Washington: The College; 1988. ACOG Technical Bulletin No.: 115 . 3. Garite TJ. Premature rupture of the membranes: the enigma of the obstetrician. Am J Obstet Gynecol 1985;151:1001-5. 4. Vintzileos AM, Campbell WA. Antepartum surveillance in patients with preterm premature rupture of the membranes. Clin Obstet Gynecol 1991;34:779-93. 5. Moberg LJ, Garite TJ, Freeman RK. Fetal heart rate patterns and fetal distress in patients with preterm premature rupture of the membranes. Obstet Gynecol 1984;64;60-4. 6. Vintzileos AM, Campbell WA, Nochimson DJ, Weinbaum PJ. The use of the non-stress test in patients with premature rupture of the membranes. Am J Obstet Gynecol 1986;155:149-53. 7. Vintzileos AM, Campbell WA, Nochimson DJ, Connolly ME, Fuenfer MM, Hoehn KJ. The fetal biophysical profile in patients
17.
18.
19. 20.
21.
22.
with premature rupture of membranes: an early predictor of fetal infection. Am J Obstet Gynecol 1985;152:510-6. Roberts AB, Goldstein I, Romero R, Hobbins JC. Comparison of total fetal activity measurement with the biophysical profile in predicting intra-amniotic infection in preterm premature rupture of membranes. Ultrasound Obstet Gynecol 1991;1:36-9. Lewis DF, Brody K, Edwards MS, Brouillette RM, Burlison S, London SN. Preterm premature ruptured membranes: a randomized trial of steroids after treatment with antibiotics. Obstet Gynecol 1996;88:801-5. Manning FA, Baskeh TF, Morrison I, Lange I. Fetal biophysical profile scoring: a prospective study in 1184 high-risk patients. Am J Obstet Gynecol 1981;140:284-93. Chamberlain PF, Manning FA, Morrison I, Harman CR, Lange IR. Ultrasound evaluation of amniotic fluid volume: the relationship of marginal and decreased amniotic fluid volume to perinatal outcome. Am J Obstet Gynecol 1984;150:245-9. Day D, Ugol JH, French JI, Haverkemp A, Will RE, McGregor JA. Fetal monitoring in perinatal sepsis. Am J Perinatol 1992;9:28-33. Gonen R, Ohlsson A, Farine D, Mulligan JE. Can the non-stress test predict congenital sepsis? Am J Perinatol 1991;8:91-3. Gauthier DW, Meyer WJ, Bieniarz A. Biophysical profile as a predictor of amniotic fluid culture results. Obstet Gynecol 1992;80:102-5. DelValle GO, Joffe GM, Izquierdo LA, Smith JF, Gilson FT, Curet LB. The biophysical profile and the non-stress test: poor predictors of chorioamnionitis and fetal infection in prolonged preterm premature rupture of membranes. Obstet Gynecol 1992;80:106-10. Miller J, Kho MS, Brown HL, Gubert HA. Clinical chorioamnionitis is not predicted by an ultrasound biophysical profile in patients with premature rupture of membranes. Obstet Gynecol 1990;76:1051-4. Carroll SG, Papaioannou S, Nicolaides KH. Assessment of fetal activity and amniotic fluid volume in the prediction of intrauterine infection in preterm prelabor amniorrhexis. Am J Obstet Gynecol 1995;172:1427-35. Moore T, Piacquadio K. A prospective assessment of fetal movement screening to reduce fetal mortality. Am J Obstet Gynecol 1989;160:1075-9. Youchah J, Chazotte C, Cohen WR. Heart rate patterns and fetal sepsis. Am J Perinatol 1989;6:356-9. Hardin JA, Jackson DN, Lewis DF, Major CA, Nageotte MP, Asrat T. Correlation of the amniotic fluid index and the non-stress test in the patient with preterm premature rupture of membranes. Am J Obstet Gynecol 1991;165:1088-94. Zeevi D, Sodovsky E, Youris J, Granat M, Ohel G. Antenatal fetal heart rate characteristics in cases of premature rupture of membrane. Am J Perinatol 1988;5:260-3. Gonik B, Bottoms SSF, Cotton DB. Amniotic fluid volume as a risk factor in preterm premature rupture of the membranes. Obstet Gynecol 1985;65:456-9.