[Editor’s Note: Lamaze International is in the process of moving the archives of our When Research is Flawed series to Science & Sensibility. When Research is Flawed is a series of brief critiques of influential studies that have shaped policy and practice, despite having serious flaws, significant limitations, or both. – AMR]
Commentary on: Hannah ME, Ohlsson A, Farine D, Hewson SA, Hodnett ED, Myhr TL, Wang EE, Weston JA, & Willan AR (1996). Induction of labor compared with expectant management for prelabor rupture of the membranes at term. TERMPROM Study Group. The New England journal of medicine, 334 (16), 1005-10 PMID: 8598837
Study design and results: multicenter, multinational randomized controlled trial in developed countries of 5041 women with confirmed PROM at ≥ 37 completed weeks of gestation. Women were not in active labor, had a singleton fetus in cephalic presentation, and had no contraindication to trial participation.
Investigators randomly allocated trial participants to one of four groups: (1) immediate induction with oxytocin, (2) expectant management for 4 days before oxytocin induction or until an indication for induction developed, (3) immediate induction with prostaglandin E2 (PGE2) followed by oxytocin if necessary, or (4) expectant management for 4 days before PGE2 induction or until an indication for induction developed.
Selected background information [Note: These represent ranges in rates reported among the 4 study groups. No significant differences across groups were detected for any of the following:]
- vaginal exam at trial admission: 35-39% digital, 64-67% speculum
- number of digital vaginal exams: 49-63% had ≥ 4
- Group B strep (GBS) status: 9-12% tested positive for GBS
- median time to active labor in expectant groups: 16-17 h
Selected maternal outcomes:
- cesarean rate: rates ranged among the 4 groups from 10-11% overall, 14-15% nulliparous women, 4-5% multiparous women
- any sign of chorioamnionitis:
– 4.0% induction/oxytocin vs. 8.6 % expectant/oxytocin, p < 0.001 [Absolute difference: 4.6%. Absolute difference for diagnosis based on criteria other than intrapartum fever (fever before labor, elevated white blood cell count, or foul-smelling amniotic fluid): 2.3%.]
– 6.2% induction/prostaglandin vs. 7.8% expectant/prostaglandin. Difference did not achieve statistical significance, meaning it was likely to be due to chance.
- neonatal infection: rates ranged from 2-3% and were not significantly different across the 4 groups
- stay in neonatal intensive care unit > 24 h:
– 7% induction/oxytocin vs. 12% expectant/oxytocin, p < 0.001.
– 9% induction/prostaglandin vs. 10% expectant/prostaglandin. Difference did not achieve statistical significance
– 8% induction/oxytocin vs. 14% expectant/oxytocin, p < 0.001.
– 11% induction/prostaglandin vs. 12% expectant/prostaglandin, p = 0.003.
- All other neonatal outcomes were similar, including, fetal distress, meconium-stained amniotic fluid, Apgar score < 7 at 1 or 5 min, cord blood pH < 7.1, need for oxygen resuscitation, jitteriness or irritability, seizures, hypotonia, abnormal level of consciousness, apnea, abnormal feeding at 48 h or more, and ventilation after resuscitation.
Problems include but are not limited to the following:
- Failure to consider the effect of epidural analgesia on intrapartum fever confounds chorioamnionitis results. Most diagnoses of chorioamnionitis were made on the basis of intrapartum fever. At the time of the trial, the association between epidural analgesia and intrapartum fever was not widely known, and no adjustment was made for this factor. Had this been done, an excess probably would remain in the expectant group, but infection rates might have been lower in all groups.
- Women who were colonized with GBS were not treated in labor. A secondary analysis looked at the effect of GBS status, based on vaginal swabs obtained at trial entry, on outcomes (Hannah, 1997). Calculations using that study’s data reveal that one-third of neonatal infections were in women testing positive for GBS. GBS also caused one of the four deaths in the expectant group in babies without lethal anomalies. Current standard practice—screening for GBS at the end of pregnancy and providing antibiotics in labor to those who are colonized—would have reduced, and might have eliminated neonatal infections in GBS + women, thus reducing infection rates overall, and it might have prevented the death. It is also possible that GBS status would not have been a factor or would have been less of a factor in neonatal infections were it not for women having vaginal exams at trial entry and multiple exams before delivery. (See next bullet points.)
- Chorioamnionitis rates were confounded by multiple digital vaginal exams. Leaving aside epidural analgesia as a confounding factor in diagnosing chorioamnionitis, yet another secondary analysis reported that chorioamnionitis increased steadily with number of digital vaginal exams independent of other factors (Seaward, 1997). Compared with less than three, the odds ratio climbed from a 2-fold increase for 3 to 4 exams to a 5-fold increase with more than 8. Seaward (1998) reported in their evaluation of risk factors for neonatal infection that chorioamnionitis had the strongest independent association. The rate among infants of women with chorioamnionitis was 16%, a six-fold increase over those not experiencing chorioamnionitis.
- Neonatal infection rates were confounded by vaginal exams at trial entry. A secondary analysis of trial data found that having a vaginal exam at trial entry increased the risk of neonatal infection by 250%, even after taking into account GBS status (Hannah, 1997). This difference is likely to be greater than appears because the analysis authors chose to combine digital and speculum exams, although only digital exams are believed to increase the risk of infection.
- Neonatal infection rates were confounded by multiple digital vaginal exams during labor. According to another secondary analysis, the percentage of infections trended upward with the number of vaginal exams independent of other factors, including time from rupture of membranes to labor onset and length of active labor (Seaward, 1998). It rose from 2% in women with 3 to 4 exams to 5% in women with more than 8. The odds roughly doubled compared with women having fewer than 3 vaginal exams, although the difference only achieved statistical significance when 7 to 8 exams were compared with fewer than 3.
Comment: Based solely on the TermPROM trial, the American College of Obstetricians and Gynecologists (ACOG) recommends immediate induction, generally with oxytocin, for women with term PROM on the grounds that inducing labor will reduce chorioamnionitis, febrile morbidity, and neonatal antibiotic treatments without increasing cesarean rates (ACOG, 2007). The primary argument for immediate induction has always been reducing neonatal infections, which ACOG acknowledges it does not do, and, as can be seen in this deconstruction, with optimal care other benefits are likely to be smaller than currently appear.
By contrast, a Cochrane systematic review published in 2006 also evaluates term PROM management (Dare, 2006). Despite being heavily dependent on the TermPROM trial—three-quarters of the 6800 participants among the 12 trials in total come from the TermPROM trial—the reviewers reach a more tempered conclusion: “Since differences in outcomes between planned and expectant management may not be substantial, women need to be able to access the appropriate information to make an informed choice (p. 12).”
In summary, in the absence of signs of infection, expectant management remains a viable option. Nonetheless, the secondary analyses have given us a more nuanced picture. While the original trial report found no difference in neonatal infection rates between immediate induction and expectant management overall, the secondary analyses make clear that length of time between rupture and delivery matters. They also found that modifiable factors affected infection rates, which means we do not know what they would have been with optimal care.
For those choosing expectant management, the question arises of how long to wait before inducing labor if one prefers to set a limit. Consider the following: Seaward (1998) reported that time from membrane rupture to labor onset of 24-48 hours versus less than 12 hours was an independent predictor of neonatal infection. Infection rates with 24 hours or more to onset of labor were 4% versus the background 2% rate. Hannah (1996) reported that the median time to active labor, not labor onset, after membrane rupture was 16-17 hours. It therefore seems reasonable to wait about 18 hours before inducing labor. Half the group of women will have achieved active labor by this time, and, if induced, the remaining half are likely to have started labor by the 24-hour cut point.
Women with PROM at term who are GBS + constitute a special subset. The Centers for Disease Control (2002) guidelines for management of GBS + women say nothing about inducing women with ruptured membranes at term, which suggests that awaiting spontaneous labor is acceptable provided that antibiotic therapy is initiated. And given that it takes time to instill the recommended dose of antibiotics, common sense dictates that women who prefer not to wait for labor should delay induction until they have an adequate dose of antibiotics on board.
In any case, regardless of GBS status or decisions around whether or when to induce, to minimize the risk of infection, women should avoid digital vaginal exams until established in labor, and their use should be minimized during labor. Data also suggest that oxytocin is the induction agent of choice. It appears to reduce infection rates compared with PGE2 without any offsetting disadvantages.
ACOG Committee on Practice Bulletins-Obstetrics (2007). ACOG Practice Bulletin No. 80: premature rupture of membranes. Clinical management guidelines for obstetrician-gynecologists. Obstetrics and gynecology, 109 (4), 1007-19 PMID: 17400872
Schrag S, Gorwitz R, Fultz-Butts K, & Schuchat A (2002). Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC. MMWR. Recommendations and reports : Morbidity and mortality weekly report. Recommendations and reports / Centers for Disease Control, 51 (RR-11), 1-22 PMID: 12211284
Dare MR, Middleton P, Crowther CA, Flenady VJ, & Varatharaju B (2006). Planned early birth versus expectant management (waiting) for prelabour rupture of membranes at term (37 weeks or more). Cochrane database of systematic reviews (Online) (1) PMID: 16437525
Hannah ME, Ohlsson A, Wang EE, Matlow A, Foster GA, Willan AR, Hodnett ED, Weston JA, Farine D, & Seaward PG (1997). Maternal colonization with group B Streptococcus and prelabor rupture of membranes at term: the role of induction of labor. TermPROM Study Group. American journal of obstetrics and gynecology, 177 (4), 780-5 PMID: 9369819
Seaward PG, Hannah ME, Myhr TL, Farine D, Ohlsson A, Wang EE, Haque K, Weston JA, Hewson SA, Ohel G, & Hodnett ED (1997). International Multicentre Term Prelabor Rupture of Membranes Study: evaluation of predictors of clinical chorioamnionitis and postpartum fever in patients with prelabor rupture of membranes at term. American journal of obstetrics and gynecology, 177 (5), 1024-9 PMID: 9396886
Seaward PG, Hannah ME, Myhr TL, Farine D, Ohlsson A, Wang EE, Hodnett E, Haque K, Weston JA, & Ohel G (1998). International multicenter term PROM study: evaluation of predictors of neonatal infection in infants born to patients with premature rupture of membranes at term. Premature Rupture of the Membranes. American journal of obstetrics and gynecology, 179 (3 Pt 1), 635-9 PMID: 9757963