by Henci Goer

Commentary on: Hannah ME et al. Induction of labor as compared with serial antenatal monitoring in post-term pregnancy. A randomized controlled trial. N Engl J Med 1992;326(24):1587-92. [Abstract]

Study design and results: multicenter randomized controlled trial (RCT) of routine induction versus expectant management. Healthy women reaching 41 weeks gestation at 22 hospitals were randomly assigned to induction within 4 days (1701 women) or expectant management until 44 weeks (1706 women). This is the largest RCT ever done on this issue, and is the largest individual trial included in the systematic review, making its results, along with its methodological problems, affect that review's outcomes disproportionately.

• perinatal mortality: 0 routine induction vs. 1.2 per 1000 (2/1706); not significant. One baby “was confirmed” as having died 3 days after assignment to the expectant management group, which means that the death might equally well have occurred during the lag period between assignment and induction had the woman been assigned to the induction group. The other baby developed fetal distress during spontaneous labor. An emergency cesarean was performed, but the baby died of meconium aspiration.
• neonatal morbidity: no differences for any of a long list of neonatal adverse outcomes, including low Apgar scores, need for various resuscitation measures, blood gas measurements, neurologic problems, infection, or admission to NICU. c/sec rates: 21.2% routine induction vs 24.5% expectant management; p = 0.03

Problems include but are not limited to the following:

• The trial had an extraordinarily high rate of crossover: One-third of women in the planned induction group actually began labor spontaneously, and one-third of the expectant management group were induced. This was because women assigned to induction went into labor before they could be induced in the induction group, while in the expectant management group, women or doctors requested elective induction or fetal surveillance tests were abnormal in the expectant management group. In the case of abnormal fetal test results, cesarean rates would likely be high because of concern that the baby was compromised. A follow-up study by Hannah herself permits calculation of cesarean rates according to whether labor was induced or spontaneous, and a very different picture emerges: 26% of induced women had cesareans compared with 18% of women with spontaneous labor onset. (1)
• Cesarean rates were high in both arms of the trial: The trial participants lacked every high-risk factor. They were healthy women with intact membranes carrying healthy, fullterm, singleton, head-down babies. Yet the cesarean rates were over 20% in both arms of the trial. The Hannah trial, like virtually all comparisons of two brands of obstetric management, is a classic example of comparing the “frying pan” with the “fire.” Results may seem somewhat better with the “frying pan” (planned routine induction) than the “fire” (planned expectant management), but both are extremely problematic. To carry the metaphor further, what is really needed is to get off the “stove” of conventional obstetric management. Other studies using a model of care more supportive of normal birth have achieved cesarean rates well under 10% in low-risk populations and under 15% in mixed-risk populations.
• The investigators combined data from first-time mothers and women with previous vaginal births: Investigators in the Hannah trial did not present cesarean section rates according to whether women had prior births, but the follow-up analysis did. Thirty-four percent—one-third—of primiparous women who were induced had c-sections versus 26% —one-quarter—of those beginning labor on their own. Eight percent of induced multiparous women had a cesarean versus 5% of women beginning labor spontaneously. The difference in first-time mothers is almost certainly statistically significant and comes as no surprise.
• Prior to this study, women were not considered “overdue” until 42 weeks. This effectively makes the study a trial of elective induction. Studies of elective induction consistently find roughly double the rate of c-section in primiparous women with induction compared with spontaneous labor onset.
• Women expectantly managed had fetal surveillance testing: The reason for inducing labor in 17% of the expectant management group was abnormal results on fetal testing. These tests have high false positive rates, which means that labor would have been induced in many women with nonreassuring test results who had perfectly healthy fetuses. This exposes more women to the risks of induction without improving outcomes.
• The investigators failed to report a severe adverse outcome that resulted from
  routine induction: Two researchers investigating another issue stumbled on a case in which one of the Hannah trial participants in the routine induction group had a precipitatous labor with fetal distress. The birth culminated in a forceps delivery that injured the spinal cord, causing quadraplegia. (2)
• Hannah has since publicly revealed herself to be biased in favor of universal c-section: Hannah chaired a 2002 conference entitled “Choosing Delivery by Caesarean: Has Its Time Come?” (3) Investigator bias has an obvious strong potential for compromising research.

Comment: After publication, the Hannah trial was hailed as proving the superiority of induction at 41 weeks, even though women were not entered into the trial until reaching 41 weeks or more and were then induced sometime within the next 4 days. Proponents of routine induction also claimed that the Hannah trial showed that inducing at 41 weeks improved infant outcomes, despite investigators finding no statistically significant (meaning unlikely to be due to chance) differences in perinatal death rates or a long list of adverse neonatal outcomes.

References:

1. Hannah ME et al. Postterm pregnancy: putting the merits of a policy of induction of labor into perspective. Birth 1996;23(1):13-9.
2. Menticoglou SM and Hall PF. Routine induction of labour at 41 weeks gestation:
   nonsensus consensus. BJOG 2002;109:485-91.
3. Maternal, Infant and Reproductive Health Research Unit (MIRU). Choosing Delivery by Caesarean: Has Its Time Come? Conference sponsored by MIRU, University of Toronto, Toronto, Ontario, Canada, Nov 7, 2002.