By: Emily Willingham, Forbes
The headlines linking labor induction and autism risk are splashy–aren’t they always with “autism linked to” stories? My favorite misstatement of the research is in this headline from Bloomberg: “Autism risk for children may be raised when labor induced,” as though the cause-and-effect association is established and inducing labor is The Factor that leads to the risk.
It could just as easily have read, “Labor induction risk may be raised when child is autistic.” Why? Because the study in question did not show a cause and effect between induced (initiated) or augmented (hastened) labor and autism. It found an increased odds that a child born following a labor induction and augmentation would later be labeled as autistic by special education services. Yet there are problems with reaching even that conclusion.
In their analysis, first author Simon Gregory of Duke University and colleagues relied on a North Carolina birth records database covering 1990 to 1998 and on child education records from 1997 to 2007 that identified children in special education as having an autism diagnosis. They excluded any children from the analysis if the records included a diagnosis in addition to autism.
What they did not include in the study (likely because it was unavailable) are the following possible influential factors: mother’s BMI pre-pregnancy, father’s age, child head circumference, specific child birth weight, mother’s insurance status, family socioeconomic status (there appears to have been a stand-in factor for that), the presence of any sibling births in the cohort, and whether or not the child had autistic siblings. Lack of availability of relevant data can sometimes make a study untenable, at least, and at best should warrant considerable caution in interpretation and speculation.
The authors report an increased odds that a child will be diagnosed with autism if the child’s birth involved a labor induction. You can’t see the data behind the article paywall, so I’ve reproduced some here. The authors performed up to five analyses that included and excluded various factors. What you see below is the results as a series of numbers in a row, each followed by a range in parentheses. The bolded values are the odds that a child born with the factor present would later be labeled as autistic in the education system. Any number higher than 1 represents an increased odds. For example, if the value given for the “induced and augmented birth” factor is 1.23, a child born under these conditions would have an odds ratio of 1.23 for a 23% increase in the odds of a later autism tag in special ed.
Odds of later identification of autism in special education records, as related to birth interventions (5 calculations)
Induced and augmented: 1.23 (1.02-1.47); 1.21 (1.01-1.46); 1.21 (1.01-1.46); 1.2 (1-1.44); 1.27 (1.05-1.52)
Induced only: 1.1 (1.01-1.19); 1.11 (1.02-1.2); 1.11 (1.02-1.2); 1.1 (1.02-1.19); 1.13 (1.04-1.22)
Augmented only: 1.15 (1.07-1.24); 1.14 (1.06-1.23); 1.14 (1.06-1.23); 1.14 (1.05-1.22); 1.16 (1.07-1.25)
As you can see from these numbers, induction alone results in a limited–at most, 13%–increase in the odds that a child will later be ID’d in special education records as autistic. The odds increase with augmentation isn’t much more impressive. For the two combined, the maximum increase in odds that the authors found was 27%.
But wait! There’s more. What the headlines don’t report is that there is an even greater odds a child will have this label in special ed if the child’s mother has a college education (I think this is the stand-in for socioeconomic status; again, the authors don’t specify). A high school education was set as 1.
Maternal educational attainment level and odds of a later ID of autism in special ed (4 calculations)
Middle school: 0.99 (0.8-1.22); 1.01 (0.82-1.25); 1.01 (0.82-1.25); 1.01 (0.82-1.24)
Some high school: 0.96 (0.88-1.05); 0.98 (0.9-1.07); 0.98 (0.9-1.06); 0.98 (0.9-1.06)
High school: 1
Some college: 1.1 (1.03-1.19); 1.09 (1.02-1.17); 1.09 (1.02-1.17); 1.11 (1.03-1.19)
College: 1.33 (1.24-1.44); 1.3 (1.21-1.41); 1.31 (1.21-1.41); 1.33 (1.23-1.44)
Note that the odds that a child will have a special ed autism label if mom has a college education are pretty consistently 30 to 33%. That is two or three times the odds linked to induction or augmentation and greater than the odds of labor and induction together. But no headlines.
And yet, there is more. Smoking during pregnancy, it seems, reduces the odds of a later autism label:
Maternal smoking and odds of a child’s later autism diagnosis in special education (3 calculations)
Smoking: 0.87 (0.81-0.94); 0.86 (0.79-0.93); 0.85 (0.79-0.92)
Those numbers indicate decreased odds of around 14% on average if mom smoked during pregnancy. No big headlines on that one, either. There’s a likely explanation for this result, and it’s not because smoking prevents autism.
The authors offer up a wildly speculative explanation for why induction might have a role in later autism diagnosis, even though induction alone didn’t do anything very spectacular in terms of those odds:
One possible explanation for the induction/augmentation-ASD association is through exposure to exogenous oxytocin. An estimated 50% to 70% of women who undergo labor induction receive exogenous oxytocin, a suggested contributor to the development of autism. Furthermore, 2 large European studies demonstrated that 30% to 50% of women in spontaneous labor required oxytocin augmentation during the course of labor.
Of course, the going story on oxytocin is that autistic people need to be treated with it.
A strange little aside: The authors also looked at the link between birth year and being labeled autistic in special ed. The odds were up by more than 50% for children born in 1994 but at -11% for children born in 1998. What ho! Is that a decrease in odds over the years?
And more. We have limited information about events during the birth. No one knows what the child’s head circumference was, and head circumference can influence the progress of birth and has been linked–although not at all conclusively–to autism. The authors also found an increased odds for the autism label if the mother had gestational diabetes (which has been found before), but no one knows what the mothers’ pre-pregnancy BMI values were, and the authors didn’t include birthweight–high birthweight is an outcome of gestational diabetes–in their analyses. [ETA: They had a minimum cutoff for birthweight of 400 g, as science writer Tara Haelle reports here. She also points out how limited the application of these findings would be, even if they did show cause-effect with induction.] Low and high birthweight have both been linked (and not linked) to autism. Indeed, the authors did find a 25% increase in autism label odds for children born at 34 weeks of gestation or less, suggesting an influence of lower birth weight, at least.
We also don’t know the family’s socioeconomic status or insurance status; maternal educational status appears to have been the stand-in for socioeconomic status, although the authors don’t tell us that. Any one of these could be a factor that muddies up a link between induction/augmentation and the presence of an autism label. For insurance, research has found a higher rate of birth-related interventions among women who are privately insured compared to women on Medicaid. Certainly, these factors could blur any odds involving maternal education status (linked with higher income) and smoking (linked with lower income). Income levels (and by association, insurance quality) have been tied to whether or not children receive an autism diagnosis or receive services. In other words, that question of socioeconomic status and insurance is a huge one affecting both birth interventions and diagnosis/services. Given that the autism “diagnosis” here was based on a special education label and thus receipt of services, that information is probably very, very relevant, but missing.
Finally, there’s the meconium. The authors found an increased odds (about 22%) of an autism label in special ed if the child’s birth involved the appearance of meconium–a.k.a., fetal poop–which can be an indication of fetal distress. That’s an odds almost as high as the induction and augmentation odds combined. A previous large analysis of 40 studies found that
Factors associated with autism risk in the meta-analysis were abnormal presentation, umbilical-cord complications, fetal distress, birth injury or trauma, multiple birth, maternal hemorrhage, summer birth, low birth weight, small for gestational age, congenital malformation, low 5-minute Apgar score, feeding difficulties, meconium aspiration, neonatal anemia, ABO or Rh incompatibility, and hyperbilirubinemia. Factors not associated with autism risk included anesthesia, assisted vaginal delivery, postterm birth, high birth weight, and head circumference.
And that takes us back to our beginning: This study didn’t show that induction or augmentation during childbirth substantially increases the risk for autism, although it hints at a greater influence of socioeconomic status and by implication, healthcare access. If anything, based on earlier literature, it adds a slight if only mathematical confirmation of the perception that births involving autistic children can be associated with more complications, such as the presence of meconium, gestational diabetes, and fetal distress, than births involving non-autistic children. And that points to induction and augmentation as useful in these situations, not as problematic, and certainly does not affirm them as a risk.