McWhorter, Jeannie MD*; Carlan, S J. MD*; O’Leary, Timothy D. MD*; Richichi, Kris MSN, ARNP*; O’Brien, W F. MD†
Preterm birth is the most common cause of perinatal morbidity and mortality in nonanomalous neonates,1 and spontaneous preterm labor is responsible for more than half of all cases of preterm birth.2 Prostaglandins play a crucial role in preterm as well as term labor, and agents that inhibit prostaglandin synthesis are effective tocolytics.3–5
Recently, several inhibitors of cyclooxygenase-2 (COX-2), the enzyme required to produce the prostaglandins (PGs) most likely involved in preterm labor, have been developed.6–8 These agents may be more effective tocolytics or exhibit a better side effect profile than the nonspecific nonsteroidal antiinflammatory (NSAIDs) drugs currently used as tocolytics.9 The purpose of this study was to compare the ability of oral rofecoxib, a preferential COX-2 inhibitor, with magnesium sulfate to arrest preterm labor for 48 hours.
MATERIALS AND METHODS
Pregnant women who were admitted to the high-risk obstetric service at Arnold Palmer Hospital for Children and Women between December 1999 and December 2002 were considered eligible for the study if they had a gestational age of 22 to 34 weeks, intact amniotic membranes, a diagnosis of preterm labor, and a cervical dilatation of 4 cm or less. Preterm labor was defined as progressive cervical dilatation or effacement associated with regular uterine contractions.2 Exclusion criteria included medical complications contraindicating tocolysis, nonreassuring fetal surveillance, evidence of fetal growth restriction, allergy to nonsteroidal antiinflammatory drugs, and sonographic evidence of lethal congenital anomalies. At randomization, participating subjects had genital cultures taken for Neisseria gonorrhoeae, group B Streptococcus, and Chlamydia trachomatis if bacteriostatic lubricant had not been used in the vagina preceding the culture collection. All were treated intravenously with 2 g of ampicillin every 6 hours, until urogenital culture results returned. If cervical culture results were negative, antibiotics were stopped. If cervical culture results were positive, the subjects were treated with the appropriate oral antibiotic for a total of 7 days. Each woman underwent an ultrasound examination (Ultramark IV ultrasound unit; Advanced Technology Laboratories, Bothell, WA) and had the option of amniocentesis for an infection assessment and a fetal lung maturity test.
This study was approved by the Orlando Regional Healthcare Institutional Review Board. After preterm labor was diagnosed, the patient was counseled about the study by the resident and offered an institutional review board–approved informed consent document. Subjects were fully informed that they would be randomly assigned to 1 of 2 tocolytic techniques. All women who met the selection criteria and were randomly assigned were given an overview of the study and a copy of their informed consent document. Each participant in the study was randomly assigned by the pharmacy with a random number table to receive either intravenous magnesium sulfate or oral rofecoxib. The hospital pharmacy supplied the patient with the medications, and the investigators and patients were blinded as to which preparation the patient was taking. At no time before data analysis did any clinical investigator have access to or knowledge of the identity of assigned drug. Those randomly assigned to the magnesium sulfate group received our hospital’s traditional protocol consisting of 4–6 g in a 20% solution as an intravenous loading dose followed by a continuous infusion at a rate of 2–4 g/h. The rofecoxib group received 50 mg orally once a day in an opaque gel capsule. The magnesium sulfate group was given an oral rofecoxib look-alike placebo in an identical opaque gel capsule with lactose filler dispensed at prescribed intervals for the active drug. The rofecoxib group received intravenous physiologic saline solution at a rate of 80 mL/h for the duration of the study. The medication was given for a maximum of 48 hours. The medication was discontinued before 48 hours if the preterm labor stopped, if the patient delivered, if the preterm labor persisted and the medication was switched, or if significant side effects developed and the patient requested the medication be stopped. Persistent preterm labor was defined as continued contractions of at least 6 per hour with associated cervical change. The medication was also stopped before 48 hours if any clinical condition occurred requiring discontinuation of tocolysis. The patient’s participation in the study was completed at 48 hours or if the medication was stopped for any reason within 48 hours. All women enrolled in the study had continuous electronic fetal monitoring for the duration of the study. The subjects were offered pharmacologic acceleration of fetal pulmonary maturation, which consisted of 12 mg of intramuscular betamethasone every 24 hours for 2 doses. If stable and undelivered, they were discharged to be followed up in the high-risk obstetric clinic or their assigned physician’s office.
Success was defined as arrest of labor and no delivery within 48 hours in women who received only their randomized medication. Therapy was considered unsuccessful if the medication was stopped, switched, or another agent added before 48 hours for any reason other than arrest of labor; if the initial dose of medication was not given for any reason; or if tocolysis was continued after the initial 48-hour study period. In those women who required continued treatment for persistent preterm labor after the 48-hour study period, the selection of tocolytic was determined by the attending physician. After discharge from the hospital, those women who resumed preterm labor and required readmission for repeat tocolysis were not assigned to the failed group.
During the study period, no patient received aspirin or other tocolytics other than the study agent. An ultrasound examination was scheduled before the start of treatment, when the medication was stopped, and at 48 hours. At each examination we attempted to evaluate the deepest pocket of amniotic fluid and amniotic fluid index by using transabdominal ultrasonography and cervical length by using vaginal probe ultrasound. The method of obtaining cervical length included scanning in the dorsal lithotomy position after micturition. The entire cervical length was measured along the closed endocervical canal between the triangular area of echodensity at the external os, and the V-shaped notch at the internal os. In cases of cervical funneling, the apex of the funnel was considered the beginning of the endocervical canal.
Throughout the treatment, the investigators reviewed nursing notes and conducted patient interviews to assess the incidence of maternal side effects, such as headache, nausea, emesis, indigestion, lethargy (including visual disturbance), shortness of breath, palpitations, local pain at the intravenous site, or flushing.
All neonates not lost to follow-up were examined within 24 hours of birth by a pediatrician. The diagnosis of respiratory distress syndrome, intraventricular hemorrhage, sepsis, pulmonary hypertension, or necrotizing enterocolitis was based on clinical, imaging, and laboratory criteria when necessary. We attempted to contact by telephone those patients lost to follow-up.
The primary outcome variable was delay in delivery for 48 hours. Secondary outcome variables included maternal and neonatal side effects. The research hypothesis was that rofecoxib was at least as effective as magnesium sulfate when used to arrest labor for 48 hours. Statistical analysis was conducted by using the Mann–Whitney U test for comparing difference of medians, Student t test for comparing differences of means, and Pearson χ2 analysis or Fisher exact test for categoric data (SPSS, Chicago, IL). Comparisons between the groups with respect to amniotic fluid index, frequency of uterine contractions, cervical dilatation, and cervical length at randomization when the medication was stopped and at 48 hours were analyzed by using repeated-measures analysis of variance. Comparison between the groups for the number of twin gestations, positive fetal fibronectin tests, successful amniocenteses, amniotic fluid samples with glucose less than 15 mg/dL, genital culture results, patients not completing the 48 hours, reason(s) for not completing the 48-hour study period, incidence of maternal side effects, incidence of success, incidence of delayed delivery past 48 hours, number of women receiving betamethasone, number of patients not initially dilated on admission to our triage unit, number of patients dilated more than 2 cm on admission to our triage unit, and the number of patients with delivery information available were analyzed by using the χ2 or Fisher exact test, where appropriate. The length of days in the hospital on the original admission was analyzed with the Mann–Whitney U test. To determine sample size, we assumed rofecoxib would be at least as effective as magnesium sulfate in arresting labor for 48 hours. Previous studies reported a failure rate of 20% with magnesium sulfate.10,11 To decrease the failure rate to 6%, which has been reported with the NSAID indomethacin,12 103 patients in each group would be required, which would yield a power of 80% at a 95% confidence level (CI). Statistical significance was defined as P < .05. We analyzed data based on intention to treat. Two-tailed P values are reported throughout.
Two hundred fourteen patients were randomly assigned, 105 to the rofecoxib group and 109 to the magnesium sulfate group. Four (4%) and 3 (3%) of the patients in the rofecoxib and magnesium sulfate groups, respectively, did not receive their assigned drug after randomization because of ruptured membranes before medication (n = 2), misdiagnosis of preterm labor (n = 2), chorioamnionitis (n = 1), patient refusal (n = 1), and physician error (n = 1). The groups were similar in demographic characteristics (Table 1), including racial distribution. The only multiple gestations enrolled were twins, and 8 patients in each group had a twin gestation (P = .94). Forty-two (19.6%) of 214 patients had vaginal fetal fibronectin performed at admission, and 9 (8.5%) and 7 (6.4%) of the rofecoxib and magnesium sulfate groups, respectively, had a positive test, (P = .55). One hundred fifty (70%) of 214 women had urogenital cultures performed. There was no difference between the groups in the incidence of positive cultures for Neisseria gonorrhoeae, group B Streptococcus, and Chlamydia trachomatis. Of the 73 attempted amniocenteses 72 were successful, 39 (37.1%) and 33 (30.2%) in the rofecoxib and magnesium sulfate groups, respectively (P = .29), and the number of patients with amniotic fluid glucose less than 15 mg/dL was 2 (1.9%) and 6 (5.5%) from the rofecoxib and magnesium sulfate groups, respectively (P = .17).
The overall response was comparable in the 2 groups (Table 2). The flow of patients through the study is shown in Figure 1. Discontinuation of the study before 48 hours occurred in 80 (76.1%) and 92 (84.4%) of the rofecoxib and magnesium sulfate groups who received medication, respectively (relative risk [RR] 1.10; 95% CI 0.97, 1.24; P = .15). In those women who did not complete the 48-hour study period, the only significant difference in reason was a maternal side effect with 0 versus 6 (6%), rofecoxib versus magnesium sulfate, respectively (P =. 03). There was a higher reported incidence of maternal side effects in the magnesium sulfate group (RR 1.81; 95% CI 1.07, 3.06). There were 5 patients who did not complete the 48 hours because of 2 physician errors, 1 amniocentesis documenting lung maturity, and 2 preterm premature rupture of the membranes before the completion of the study. The incidence of side effects was higher in the magnesium sulfate group (Table 3). No patient from either group developed a cardiac dysrhythmia or was admitted to the intensive care unit.
The number of patients receiving at least 1 injection of betamethasone was similar between the groups at 103 of 105 (98%) and 105 of 109 (96.3%), rofecoxib and magnesium sulfate, respectively (P = .68). The number of patients not initially dilated on admission to our triage unit was similar between the groups at 6 of 105 (6%) and 8 of 109 (7%), rofecoxib and magnesium sulfate, respectively (P = .63). The number of patients dilated more than 2 cm on admission to our triage unit was similar between the groups at 28 of 105 (27%) and 33 of 109 (30%), rofecoxib and magnesium sulfate, respectively (P = .56).
The incidence of success was similar between the groups at 78 (74%) and 80 (73%), rofecoxib and magnesium sulfate, respectively (RR 0.99; 95% CI 0.84, 1.16). Delivery was delayed for 48 hours in 95 (90.4%) and 96 (88%) patients in the rofecoxib and magnesium sulfate groups, respectively (RR 0.97; 95% CI 0.89, 1.06). We would need to treat 2,686 patients with rofecoxib to see a significant improvement in delay in delivery past 48 hours assuming equal numbers of patients in the magnesium sulfate and rofecoxib groups. The median length of days in the hospital on the original admission was also similar at 2 versus 2, rofecoxib versus magnesium sulfate, respectively (P = .10). Ninety-two (87.6%) and 102 (93.5%) patients in the rofecoxib and magnesium sulfate groups, respectively, had delivery information available (RR 1.07; 95% CI 0.98, 1.17). Of those with delivery data available, there was no difference between the groups in selected obstetric or neonatal outcomes (Table 4). There were 4 neonatal deaths in the magnesium sulfate group, and all received at least a portion of their study drug (Table 5).
Prostaglandin production in the cervix and gestational tissues is a central component to the process of labor.13,14 The 2 recognized isoforms of the enzyme that catalyze the transformation of arachidonic acid to PG are cyclooxygenase-1 (COX-1) and COX-2. Cyclooxygenase-1 is expressed in most tissues and regulates normal cellular processes. Thus inhibition of this enzyme is thought to be responsible for many of the adverse side effects observed with the use of the nonselective NSAIDs, such as indomethacin. Cyclooxygenase-2 is an inducible enzyme stimulated by mechanical stretch15 and inflammatory cytokines13,16 and is up-regulated during labor.17 Although both COX-1 and COX-2 can be found in fetal membranes and may play a role in myometrial contractility, it is felt that the synthesis of PGs that stimulate labor is a result of COX-2 exclusively.7,8,14,18
In 1997 Sawdy et al19 reported the first case of a COX-2 inhibitor (nimesulide) used to prevent preterm birth. Daily exposure to 200 mg from 16 to 34 weeks did not result in detectable fetal ductus arteriosus constriction or decreased amniotic fluid volume and the neonate had an uncompromised course.
In 2001, Slattery et al,20 using in vitro isometric tension recordings, concluded that COX-2 inhibitors could exert significant relaxation in the human myometrial tissue. Other investigators also suggested that the COX-2 inhibitors may be useful tocolytics.21,22 Animal studies confirmed the potential efficacy as a tocolytic, but there has been evidence that prolonged use of the agents may result in ductal constriction and oliguria, 2 of the prominent fetal side effects known to occur with the use of nonselective cyclooxygenase inhibitors.23 Further animal research indicated that in utero ductal patentcy may be mediated, in fact, by both the COX-124 and the COX-2 pathway.25 Two subsequent nonrandomized reports in humans indicated that prolonged use of a COX-2 inhibitor could delay delivery but was also associated with fetal renal impairment.26,27 Peruzzi et al28 reported a case of permanent neonatal end-stage renal failure after 4 weeks of in utero exposure to a COX-2 inhibitor (nimesulide) from 26 to 30 weeks of gestation. Short courses of COX-2 inhibitors, however, have not been associated with detectable adverse fetal or neonatal side effects when given after 22 weeks of gestation, but the data in humans are limited.
We chose magnesium sulfate as the control drug because it is our first-line tocolytic. Although the overall efficacy of magnesium sulfate as a tocolytic is arguable,29 it has been used for more than 3 decades.10,30 The frequency of maternal side effects has been a consistent concern, and it appears that the margin between maternal safety and efficacy of magnesium sulfate infusion for tocolysis is quite narrow.31 We likewise noted a high incidence of maternal side effects (29%) and discontinuation rate (6%) in the magnesium sulfate group. This is consistent with an earlier study32 showing a 10% discontinuation rate in patients treated with magnesium sulfate infusion for tocolysis with an overall 31% incidence of reported side effects.
Despite the encouraging results of recent studies33,34 demonstrating improved methods to predict prematurity and prevent preterm labor, the incidence of preterm delivery has remained stable during the last 20 years.35 Clearly, new methods to arrest acute preterm labor are needed.21 The coxibs are a new group of antiinflammatory drugs that selectively inhibit COX-2.36 Thus, they should be at least as effective as nonselective agents for the inhibition of fetal membrane PG synthesis with fewer side effects and may represent a new strategy for tocolysis. In the present study, a new COX-2 inhibitor, rofecoxib, demonstrated comparable clinical efficacy in delaying delivery for 48 hours in direct comparison with magnesium sulfate. This is consistent with nonselective PG inhibitors that have been used for acute as well as maintenance tocolysis.37 The potential for fetal and neonatal side effects, however, have limited the widespread acceptance of this class of drug as first-line tocolytic agents.
Our data suggest that short courses of oral rofecoxib may have acceptable maternal and fetal side effect profiles when used for no more than 48 hours. However, based on the case reports26–28 implicating long-term COX-2 inhibitors in disrupting normal fetal urinary dynamics, it appears that these agents may indeed temporarily eliminate or significantly reduce concentrations of an essential PG involved in normal cellular processes, and caution should also be recommended when using the current generation of COX-2 inhibitors, especially if the objective is lengthy maintenance tocolysis rather than acute tocolysis.
Our study may have several limitations and could be improved in several ways. First, we may have overdiagnosed preterm labor and treated some women in both groups unnecessarily. When preterm labor is diagnosed on the basis of contractions alone, between 30% and 70% of women will resolve without treatment.38,39 In this circumstance, many more patients would be required to determine a benefit of rofecoxib. It is likely that, despite the entry criteria, many of our patients were not in true preterm labor and the high success rate of both groups is the result of our definition of preterm labor.
Second, we did not exclude twins in our study. The dynamics of premature uterine contractions in twins may be different from those in singleton pregnancies because of uterine overdistension,40 but strict entry criteria and the blinded randomization process should have eliminated bias. In addition, there was no difference in the number of twins in each randomized group.
Finally, the most serious limitation of our study is the loss to follow-up of 13 fetuses in the rofecoxib group. Serious unrecognized complications could have occurred in this group, which would have significantly changed our conclusions regarding safety of the medication.
Accumulating data suggest that prolonged tocolytic treatment is not effective in improving neonatal outcome and remains controversial. There is a strong consensus, however, that in those patients who present with active labor before 34 weeks of gestation that a 48-hour delay in delivery until a full course of betamethasone is completed will have improved neonatal outcomes. Our data indicate that short-term use of the COX-2 inhibitor, rofecoxib, is at least as effective as magnesium sulfate as a tocolytic and appears to have no greater risk of side effects than magnesium sulfate. In addition, rofecoxib is discontinued less frequently for maternal side effects. Further research on both the potential fetal side effects and proposed second generation COX-2 inhibitors is necessary.
1. da Fonseca EB, Bittar RE, Carvalho MH, Zugaib M. Prophylactic administration of progesterone by vaginal suppository to reduce the incidence of spontaneous preterm birth in women at increased risk: a randomized placebo-controlled double-blind study. Am J Obstet Gynecol 2003;188:419–24.
2. Goldenberg RL. The management of preterm labor [review]. Obstet Gynecol 2002;100:1020–37.
3. O’Brien WF. The role of prostaglandins in labor and delivery [review]. Clin Perinatol 1995;22:973–84.
4. Olson DM, Mijovic JE, Sadowsky DW. Control of human parturition [review]. Semin Perinatol 1995;19:52–63.
5. Zuckerman H, Reiss U, Rubinstein I. Inhibition of human premature labor by indomethacin. Obstet Gynecol 1974;44:787–92.
6. FitzGerald GA, Patrono C. The coxibs, selective inhibitors of cyclooxygenase-2 [review]. N Engl J Med 2001;345:433–42.
7. Slater D, Allport V, Bennett P. Changes in the expression of the type-2 but not the type-1 cyclo-oxygenase enzyme in chorion-decidua with the onset of labour. Br J Obstet Gynaecol 1998;105:745–8.
8. Fuentes A, Spaziani EP, O’Brien WF. The expression of cyclooxygenase-2 (COX-2) in amnion and decidua following spontaneous labor. Prostaglandins 1996;52:261–7.
9. Hayes EC, Rock JA. COX-2 inhibitors and their role in gynecology [review]. Obstet Gynecol Surv 2002;57:768–80.
10. Petrie RH. Preterm parturition: tocolysis using magnesium sulfate [review]. Semin Perinatol 1981;5:266–73.
11. Nochimson D, Petrie R, Shah B, Pampati N, Brunelle D. Comparison of conservative and dynamic management of premature rupture of the membranes/premature labor syndrome: new approaches to delivery of which may minimize the need for intensive care. Clin Perinatol 1980;7:17–31.
12. Zuckerman H, Shalev E, Gilad G, Katzuni E. Further study of the inhibition of premature labor by indomethacin. Part II. Double-blind study. J Perinat Med 1984;12:25–9.
13. Perkins D, Kniss D. Tumor necrosis factor-alpha promotes sustained cyclooxygenase-2 expression: attenuation by dexamethasone and NSAIDs. Prostaglandins 1997;54:727–43.
14. Sawdy RJ, Slater DM, Dennes WJ, Sullivan MH, Bennett PR. The roles of the cyclo-oxygenases types one and two in prostaglandin synthesis in human fetal membranes at term. Placenta 2000;21:54–7.
15. Kanayama N, Fukamizu H. Mechanical stretching increases prostaglandin E2
in cultured human amnion cells. Gynecol Obstet Invest 1989;28:123–6.
16. Rauk PN, Chiao JP. Interleukin-1 stimulates human uterine prostaglandin production through induction of cyclooxygenase-2 expression. Am J Reprod Immunol 2000;43:152–9.
17. Slater DM, Berger LC, Newton R, Moore GE, Bennett PR. Expression of cyclooxygenase types 1 and 2 in human fetal membranes at term. Am J Obstet Gynecol 1995;172:77–82.
18. Habermehl DA, Janowiak MA, Vagnoni KE, Bird IM, Magness RR. Endothelial vasodilator production by uterine and systemic arteries. IV. Cyclooxygenase isoform expression during the ovarian cycle and pregnancy in sheep. Biol Reprod 2000;62:781–8.
19. Sawdy R, Slater D, Fisk N, Edmonds DK, Bennett P. Use of a cyclo-oxygenase type-2-selective non-steroidal anti-inflammatory agent to prevent preterm delivery. Lancet 1997;350:265–66.
20. Slattery MM, Friel AM, Healy DG, Morrison JJ. Uterine relaxant effects of cyclooxygenase-2 inhibitors in vitro. Obstet Gynecol 2001;98:563–9.
21. Bukowski R, Saade GR. New developments in the management of preterm labor [review]. Semin Perinatol. 2001;25:272–94.
22. Shellhaas CS, Iams JD. The diagnosis and management of preterm labor [review]. J Obstet Gynaecol Res 2001;6:305–11.
23. Lee PR, Kim SR, Jung BK, Kim KR, Chung JY, Won HS, Kim A. Therapeutic effect of cyclo-oxygenase inhibitors with different isoform selectivity in lipopolysaccharide-induced preterm birth in mice. Am J Obstet Gynecol 2003;189:261–6.
24. Guerguerian AM, Hardy P, Bhattacharya M, Olley P, Clyman RI, Fouron JC, et al. Expression of cyclooxygenase in ductus arteriosis of fetal and newborn pigs. Am J Obstet Gynecol 1998;179:1618–26.
25. Loftin CD, Trivedi DB, Langenbach R. Cyclooxygenase-1-selective inhibition prolongs gestation in mice without adverse effects on the ductus arteriosus. Clin Invest 2002;110:549–57.
26. Holmes RP, Stone PR. Severe oligohydramnios induced by cyclooxygenase-2 inhibitor nimesulide. Obstet Gynecol 2000;96:810–1.
27. Locatelli A, Vergani P, Bellini P, Strobelt N, Ghidini A. Can a cyclo-oxygenase type-2 selective tocolytic agent avoid the fetal side effects of indomethacin? BJOG 2001;108:325–6.
28. Peruzzi L, Gianoglio B, Porcellini MG, Coppo R. Neonatal end-stage renal failure associated with maternal ingestion of cyclo-oxygenase-type-1 selective inhibitor nimesulide as tocolytic. Lancet 1999;354:1615.
29. Crowther CA, Moore V. Magnesium for preventing preterm birth after threatened preterm labour [review]. Cochrane Database Syst Rev 2000;2:CD000940.
30. Steer CM, Petrie RH. A comparison of magnesium sulfate and alcohol for the prevention of premature labor. Am J Obstet Gynecol 1977;129:1–4.
31. Hollander DI, Nagey DA, Pupkin MJ. Magnesium sulfate and ritodrine hydrochloride: a randomized comparison. Am J Obstet Gynecol 1987;156:631–7.
32. Glock JL, Morales WJ. Efficacy and safety of nifedipine versus magnesium sulfate in the management of preterm labor: a randomized study. Am J Obstet Gynecol 1993;169:960–4.
33. Goldenberg RL, Iams JD, Mercer BM, Meis PJ, Moawad A, Das A, et al. The Preterm Prediction Study: toward a multiple-marker test for spontaneous preterm birth. Am J Obstet Gynecol 2001;185:643–51.
34. Iams JD, Casal D, McGregor JA, Goodwin TM, Kreaden US, Lowensohn R, et al. Fetal fibronectin improves the accuracy of diagnosis of preterm labor. Am J Obstet Gynecol 1995;173:141–5.
35. Goldenberg RL, Rouse DJ. The prevention of premature birth. N Engl J Med 1998;339:313–20.
36. Kurumbail RG, Stevens AM, Gierse JK, McDonald JJ, Stegeman RA, Pak JY, et al. Structural basis for selective inhibition of cyclooxygenase-2 by anti-inflammatory agents [published erratum appears in Nature 1997;385:555]. Nature 1996;384:644–8.
37. Higby K, Xenakis MJ, Pauerstein CJ. Do tocolytic agents stop preterm labor? A critical and comprehensive review of efficacy and safety [review]. Am J Obstet Gynecol 1993;168:1247–56; discussion 1256–9.
38. Kragt H, Keirse MJ. How accurate is a woman’s diagnosis of threatened preterm delivery? Br J Obstet Gynecol 1990;77:317–23.
39. Castren O, Gummerus M, Saarikoski S. Treatment of imminent preterm labor: a comparison between the effects of nylidrin chloride and isoxsuprine chloride as well as of ethanol. Acta Obstet Gynecol Scand 1975;54:95–100.
40. Goldenberg RL, Iams JD, Miodovnik M, Van Dorsten JP, Thurnau G, Bottoms S, et al. The preterm prediction study: risk factors in twin gestations. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 1996;175:1047–53.