by Henci Goer

Commentary on: Nicholson, J. M., Yeager, D. L., & Macones, G. (2007). A preventive approach to obstetric care in a rural hospital: Association between higher rates of preventive labor induction and lower rates of cesarean delivery. Ann Fam Med, 5(4), 310-319. [Abstract]

Study design and results: retrospective comparison of cesarean surgery rates and neonatal outcomes in 794 women attended by practitioners who were high users of preventive induction with prostaglandin E2 (PGE2) (defined as induction rate overall = 21% and preventive induction rate = 15%) versus 1075 women attended by low users (defined as induction rate overall < 21% or preventive induction rate < 15%). The theory behind preventive induction was that need for cesarean surgery for “cephalopelvic disproportion” and “uteroplacental insufficiency” increases with pregnancy duration past 37 completed weeks. Practitioners can minimize the cesarean rate, the researchers hypothesize, by inducing labor according to an individualized risk profile so as to ensure “optimal” gestational length. The pilot study for this project was paid for by a PGE2 manufacturer who continues to supply PGE2 to the investigators gratis.

• Indications for preventive induction: impending macrosomia, previous cesarean delivery, impending preeclampsia, gestational diabetes, other. Appendices downloadable from the journal’s website further clarify indications for preventive induction. Examples of these are cigarette use, “advanced age” (= 35 y) at delivery, “excess weight gain” (= 30 lb) [30 lb is the recommended pregnancy weight gain for an average-weight woman], “short stature” (= 62 in), 1st trimester anemia (Hgb = 11.0 g/dL), borderline oligohydramnios (AFI 5-7), previous assisted vaginal delivery, actual or high risk of domestic abuse, and increasing or severe depression.

• Induction rates, high vs. low user: Note: Mean gestational age at admission differed between groups by one day (39 w 5 d high user vs. 39 w 6 d low user), although this difference achieved statistical significance, meaning “difference not likely to be due to chance.”

  • induction rate overall: 31% high user vs. 20% low user
  • induction rate using PGE2: 23% high user vs. 16% low user
  • indicated induction: 10% high user vs. 12% low user, difference not statistically significant (NS), meaning difference likely to be due to chance
  • preventive induction: 21% high user vs. 8% low user
  • elective induction, defined as induction in a woman who did not meet eligibility criteria for either indicated or preventive induction [calculated from study data]: none in either high or low users
• Cesarean rates overall, high vs. low user: 5% high user vs. 12% low user

• Cesarean rates with spontaneous labor onset, high vs. low user:

  • nulliparous women (those giving birth for the first time): 7% high user vs. 7% low user, NS
  • multiparous women (those who had given birth previously) with no prior cesarean: 1% high user vs. 2% low user, NS
  • women laboring after a prior cesarean: 5% high user vs. 33% low user

• Cesarean rates with induced labor, high vs. low user:

  • nulliparous women: 8% high user vs. 26% low user
  • multiparous women no prior cesarean: 0% high user vs. 10% low user
  • women laboring after a prior cesarean: 7% high user vs. 26% low user

• Cesarean rates, induced labor vs. spontaneous labor onset:

  • nulliparous women: 17% induced vs. 7% spontaneous
  • multiparous women: 4% induced vs. 2% spontaneous Note: It is unclear whether this subgroup included women with and without prior cesareans.
• VBAC rate, defined as the vaginal birth rate in women planning VBACs [calculated from study data], high vs. low user: 93% high user vs. 69% low user. Note: Similar proportions of women with prior cesarean planned VBAC (87% high user vs. 86% low user).
• Instrumental vaginal delivery rates were similar (18% vs. 16%) between high vs. low users. [This means that the reduction in cesarean rates was not offset by a rise in vaginal instrumental deliveries.]

• Factors independently associated with cesarean surgery (multiple logistic regression performed to account for correlating factors)

  • high-user practitioner: odds ratio 0.6
  • family practitioner (FP) as caregiver: odds ratio 0.6 Note: High-use group consisted of 4 FPs and 1 obstetrician (OB); low-use group consisted of 3 FPs, 6 OBs, and 2 nurse-midwives, both employed by an obstetrician group.
  • labor epidural: odds ratio 3.2 Note: The epidural rate during the study period was low (7% high user vs. 16% low user).
  • Other factors: stature = 62 in (odds ratio 1.9), BMI > 30 kg/m2 (odds ratio 2.4), nulliparity (odds ratio 4.8), prior cesarean (odds ratio 20.6), nonvertex presentation (odds ratio 146.9)
• Neonatal outcomes:
  • Women attended by high users had reduced risk of thick meconium (risk ratio: 0.45, absolute difference 2.6 babies per 100), repetitive late decelerations (risk ratio: 0.23, absolute difference 1.3 babies per 100), birth weight > 4500 g (risk ratio: 0.48, absolute difference 1.4 babies per 100), and neonatal intensive care unit (NICU) admission (risk ratio: 0.66, absolute difference 1.9 babies per 100). [All absolute differences calculated from study data.]
  • No differences were found for mean birth weight, birth weight > 4000 g, head circumference = 37 cm, 5-min Apgar scores < 7, or cord blood pH < 7.2.

Problems include but are not limited to the following:

• Investigators did not measure what they thought they were measuring: The overarching problem with this study is that investigators assumed they were measuring the effect of prophylactic treatment on factors intrinsic to the woman that increased her risk of cesarean. However, they actually measured differences in practice style and the placebo effect of thinking that preventive induction is efficacious. In support of this argument:

  • Having a family practitioner as a provider cut the odds of having a cesarean by 40%: This was an independent factor, which means having an FP had this effect regardless of whether the FP was a high or low user or the presence of other factors that might increase a woman’s risk of cesarean. In fact, having a high-user care provider did no better at reducing the odds of cesarean than just having an FP.
  • Despite similar percentages of women planning VBAC, the likelihood of vaginal birth in a VBAC labor was much higher in high versus low users: This commitment to VBAC indicates a major difference in practice style and philosophy that reduces cesarean rates but is unrelated to induction policies.
  • Inducing labor increased cesarean rates in all subgroups but to a much greater degree in low users. Cesarean rates were similar between groups for spontaneous labors. This indicates differing approaches to induced labors. Previous research has demonstrated the attending physician is an independent risk factor for cesarean surgery in induced labors (Luthy, 2004).
  • Inducing labor greatly increased cesarean rates compared with spontaneous labor: Investigators explained this away by saying that one of this study’s strengths was that it looked at outcomes according to whether the practitioner was a high or low user of preventive induction. This approach, investigators argue, unmasked that high users had lower cesarean rates, which they attribute to high preventive induction rates. A more plausible explanation is that high preventive induction rate serves as a marker for practitioners who prefer to avoid cesarean surgeries and who believe preventive induction can help achieve that goal. This explanation is supported by the investigators’ statement that high users were “openly enthusiastic (p. 313)” about preventive induction while low users were “somewhat wary (p. 313).” In other words, the paradox actually provides clear evidence that a “provider effect,” not a “clinical effect,” was at work.
  • Preventive induction had no clinically significant effect on reducing macrosomia, one of the two theoretical reasons for doing it: Mean birth weights, percentages of babies weighing > 4000 g, and percentages with head circumference = 37 cm were similar between high and low users. Birth weights > 4500 g differed significantly, but only 1.4 more babies in every hundred were born in this weight range in the low-user group, a trivial difference. The real issue is that high users believed that induction would prevent “cephalopelvic disproportion” due to macrosomia, which could affect their decision to proceed to surgery. Other studies have shown that the belief that women will not be able to birth their babies can become a self-fulfilling prophecy. For example, two studies of ultrasound birth weight estimation reported that cesarean rates were much higher when the obstetrician believed the baby would weigh > 4000 g than if the baby actually weighed that much but was estimated to weigh less (Levine, 1992; Weeks, 1995). Another study found that women whose prior cesareans were for labor dystocia were given less time during the VBAC labor compared with women who had prior cesareans for other indications (Shipp, 2000).
  • Mean gestational length differed only by 1 day, and in both groups, fell short of 40 completed weeks, which means “uteroplacental insufficiency” associated with post maturity, the other theoretical reason for preventive induction was not an issue. The increase in meconium, nonreassuring fetal heart rate patterns, and NICU admissions among low users, which the study authors see as evidence for increasing risk of placental deterioration, remains unexplained, but might be explained by differences in labor management—using high-dose oxytocin protocols, for example—or practice style variations in criteria for admission to the NICU. In any case, the first two are surrogate, not clinical outcomes, and only 2 more babies per 100 were admitted to the NICU in the low-use group.
• Evidence suggests investigator bias: The pilot study was funded by a company that makes PGE2 inserts and that continues to supply it gratis. All three investigators are proponents of preventive induction: two of them were high users in this study and the third is the co-author of another study reporting positive effects. The choice of terminology betrays bias: “Active Management of Risk in Pregnancy at Term (AMOR-IPAT)” for elective induction of labor, which, indeed, study authors are at pains to distinguish from what they deem elective induction (see next bullet), and “cephalopelvic disproportion,” which implies a concrete diagnosis that has long since been discarded because of its false implications in favor of “labor dystocia.” Bias also explains their failure to address the body of research that has consistently found no benefits for inducing labor for suspected macrosomia, and certainly not for what the authors call “impending macrosomia.” Bias is revealed as well in the assumption that normal pregnancy results in “uteroplacental insufficiency” even before reaching median duration. Finally, as the above analysis makes clear, investigators did not perceive what was in plain sight to anyone with a different perspective. Logic gaps are an inevitable effect of bias.
• No elective inductions were performed: The investigators define “elective induction” as inductions not supported by ACOG as “indicated” inductions or by preventive rationales. Not one induction in this series, whether among low or high users of preventive induction, was elective. By creating an extensive list of categories, many of which are merely deviations from average converted to pathology by biased language, such as “advanced age,” and some of which, such as “domestic abuse” or “depression,” would not be rectified by early delivery, investigators have ensured that virtually all women will fall into at least one of them. The real payoff of this study is it enables clinicians to induce whomever they please behind the façade that it is for the woman’s own good.

Comment: This study was chosen because it exemplifies a problematic genre of obstetric research. It, like episiotomy, elective primary cesarean surgery, Active Management of Labor, Active Management of Third Stage, and now Active Management of Risk in Pregnancy at Term (notice a pattern?), proposes routine or frequent intervention into normal, healthy pregnancies or labors in the belief that this can improve them. The recurrence of this theme derives from the underlying tenet that pregnancy and labor are inherently pathological, which is self-contradictory. Disease by definition is a disruption of normal physiologic processes. If the premise is wrong, everything that follows from it must be wrong too. And so it proves here. Every medical intervention carries risks; if you expose women to them who aren’t having a problem, you expose them to their potential harms with no counterbalancing benefits. This truth gives you a touchstone for evaluating studies: Reject out of hand any study that purports to show that intervening in the absence of a problem improves outcomes. In fact, distrust any study where investigators display a bias against normal birth. That bias, as can be seen in this study, can distort their research on every level from study design to data interpretation.

This study was also chosen because it raises concern that it is a harbinger of things to come. It replicates the pattern of the early studies of Active Management of Labor, that is, uncontrolled studies by believers resulting in enthusiastic advocacy for a highly interventive protocol that supposedly reduces cesarean rates. In that case, as in this one and all other studies of this ilk, what investigators truly measure is the effect of belief on provider judgment and practice style. Imposition of these protocols, however well-intentioned, though, do much harm and no good while misdirecting attention away from policies and practices that truly promote optimal outcomes. Consider yourself forewarned.

References:
Levine, A. B., Lockwood, C. J., Brown, B., Lapinski, R., & Berkowitz, R. L. (1992). Sonographic diagnosis of the large for gestational age fetus at term: Does it make a difference? Obstet Gynecol, 79(1), 55-58.

Luthy, D. A., Malmgren, J. A., & Zingheim, R. W. (2004). Cesarean delivery after elective induction in nulliparous women: The physician effect. Am J Obstet and Gynecol, 191(5), 1511-1515.

Shipp, T. D., Zelop, C. M., Repke, J. T., Cohen, A., Caughey, A. B., & Lieberman, E. (2000). Labor after previous cesarean: Influence of prior indication and parity. Obstet Gynecol, 95(6 Pt 1), 913-916.

Weeks, J. W., Pitman, T., & Spinnato, J. A., 2nd. (1995). Fetal macrosomia: Does antenatal prediction affect delivery route and birth outcome? Am J Obstet Gynecol, 173(4), 1215-1219.