More Evidence to Avoid Hospital Birth: A Critique of the Results of the ARRIVE Study
Subscribe to Midwifery Today Magazine
A 2018 article in the New England Journal of Medicine by Grobman et al., regarding “A Randomized Trial of Induction Versus Expectant Management” (commonly referred to as the ARRIVE Trial), has gotten a lot of press, telling women that induction at 39 weeks has better mother and baby outcomes compared to going into labor naturally. However, it is not as straightforward as the authors would have you believe.
The protocol to induce for postdates at 41 weeks was accepted by medical professionals about 10 years ago. The randomized controlled trials (RCT) for this protocol are based on poor quality studies that included births at high risk of perinatal death, such as premature births and those of diabetics, and entail high rates of noncompliance with protocol (Cohain 2015). There is no good evidence that induction for postdates among low-risk women results in better outcomes for mother or baby.
After a decade of inducing at 41 weeks, the ARRIVE trial now supposedly provides RCT evidence to change the protocol again—to induction at 39 weeks. The excuse is not to improve neonatal outcomes—because it does not—but is supposedly to significantly reduce cesareans.
Most cesareans are done for non-medical reasons (Cohain 2009). Half of these surgical procedures are justified by the undefined term “dystocia,” in which the cesarean is performed after an arbitrarily decided time limit has elapsed. One of the main reasons not to perform a cesarean is because major abdominal surgery results in the death of 1 in 10,000 women in high income countries, due to hemorrhaging or anesthesia complications.
The reason not to induce is also to avoid maternal death. Induction results in the death of about 1 in 10,000 women, caused by amniotic fluid embolism (AFE) (although most research results fail to explicitly link induction with AFE). Research shows that 50%–70% of AFE is “associated” with induction (Knight 2012; McDonnell 2015; Stolk 2012; Kramer 2013; Roberts 2010). In 1970, the AFE rate was 1 in 120,000, but now that we induce 25% of births and augment another 25% of births, the rate of AFE is up to 1 in 15,000 births (Wagner 2005). If the women participating in the ARRIVE RCT trial, which promoted 39 week inductions, had been told before they signed up of their risk of dying of AFE if they underwent induction, there would have been no trial—because most women would not have agreed to join it. The danger of medical induction is comparable to driving to work five days a week for a year never wearing a seat belt or having airbags. Because the participating women were not informed of this risk, informed consent was not obtained. Yet the research received Institutional Review Board (IRB) approval. In addition, the study was approved despite the plan not to include enough women to pick up an increase in maternal deaths.
Low-risk births have better outcomes when they are allowed to proceed with minimal medical intervention. The study clearly demonstrated this. Both the induction and expectant management groups underwent high rates of interventions and the outcomes were horrendous. At 5 minutes after birth, 15% of the babies were either not breathing at all or were weakly breathing; 1% needed respiratory support for a day or more; 12% were admitted to NICU; 0.6% suffered hypoxic ischemic encephalopathy (HIE) and 0.2% had seizures, both of which cause long-term outcomes that are wide-ranging and may affect the babies’ motor, sensory, cognitive, and behavioral outcome; 0.3% had infections, which also often have long-term sequelae like brain damage; 0.7% had meconium aspiration syndrome; 0.3% had intracranial hemorrhage; and 5% had hyperbilirubinemia.
Two percent of these babies suffered from shoulder dystocia, despite the lack of a single birth weight over 3560 g (7 lb 14 oz). Because the biggest baby in the study weighed 3560 g, one would not expect any cases of shoulder dystocia. It has been suggested that, in theory, inductions may cause shoulder dystocia due to Pitocin forcing the shoulder to get stuck above the pubic bone.
Maternal outcomes were also terrible: Five percent had severe postpartum hemorrhages (PPH) of >1500 cc, requiring blood, hysterectomy, or blood products; 4% had third- or fourth-degree tears; and 2% had a postpartum infection.
The authors failed to report or address the effect of epidural analgesia on these labors.
Five percent, or 1 in every 20 births, did not comply with the assigned protocol. This means that for 5% of births, the doctor decided not to induce women in the induction group, while 5% of participants in the expectant management group were induced. However, their outcomes were reported according to the group in which the woman was originally randomized. As a result, we don’t know the actual cesarean rate for each protocol. There might have been no difference between the two groups or the induction group may have had a higher cesarean rate.
Three perinatal deaths that were not stillbirths were reported, one in the “induction group” and two in the “expectant management group.” But because of the crossover, it is possible that all three resulted from overzealous inductions or something entirely unrelated. There was no reason not to report the outcomes both by intention to treat and also according to what protocol the woman actually received. In any case, cesarean rates of 19% versus 22% are not different enough to change or direct a protocol, particularly in an unblinded study in which the doctors knew who was being induced.
Similar to the protocol to induce for postdates, Grobman et.al. reflects more evidence of research that uses bad methodology to make incorrect recommendations, once again medicalizing the birth process.
- Cohain, JS. 2009. “Is Action Bias One of the Numerous Causes of UnneCesareans?” MIDIRS Midwifery Digest 19(4): 495–99.
- Cohain, JS. 2015. “To what extent do English language RCT meta-analysis justify induction of low risk pregnancy for postdates?” J Gynecol Obstet Biol Reprod. 44(5): 393–97.
- Grobman, WA, et al. 2018. “Labor induction versus expectant management in low-risk nulliparous women.” N Engl J Med 379(6): 513–23.
- Knight, M. 2012. “Amniotic fluid embolism incidence, risk factors and outcomes: a review and recommendations.” BMC Pregnancy Childbirth 12: 7.
- Kramer, MS. 2013. “Incidence, risk factors, and consequences of amniotic fluid embolism.” Paediatr Perinat Epidemiol 27(5): 436–41.
- McDonnell, N. 2015. “Amniotic fluid embolism: an Australian-New Zealand population-based study.” BMC Pregnancy Childbirth 15: 352.
- Stolk, KH. 2012. “Severe maternal morbidity and mortality from amniotic fluid embolism in the Netherlands.”Acta Obstet Gynecol Scand 91(8): 991–95.
- Roberts, CL, et al. 2010 “Amniotic fluid embolism in an Australian population-based cohort.” BJOG 117(11): 1417–21.
- Wagner, M. 2005. “From caution to certainty: hazards in the formation of evidence-based practice—a case study on evidence for an association between the use of uterine stimulant drugs and amniotic fluid embolism.” Paediatr Perinat Epidemiol 19(2): 173–76.