In the latest media barrage on autism, fertility treatment has come into question as a possible cause for this increasingly common developmental disorder. The reason is two research abstracts recently presented at the International Meeting for Autism Research in Philadelphia.
One study assessed the history of IVF (in vitro fertilization) among 574 children evaluated at a special center for autism in Israel. The researchers found that 10% of the group diagnosed as autistic had had IVF, compared to a background rate in the overall population which they quote as 3.5%. Not surprisingly, maternal age was higher in the IVF group and the rate of prematurity was higher in the autistic children.
The second study was a look into a pre-existing database — the Nurse’s Health Study — which collects data from a cohort of nurses over time. The researchers compared the reproductive history reported by women who also reported having a child with autism and compared it to that of women who did not report having an autistic child. Of those with autistic children, 48% reported infertility with 34% having used ovulation inducing drugs, compared with 33% and 24%, respectively, in women without autistic children, a difference that was statistically significant when controlled for maternal age and self-reported pregnancy complications.
A Time article getting a lot of media play calls the results of the second study “some of the strongest evidence to date” linking autism to fertility treatment. Unfortunately, that’s just not true.
Studies certainly raise questions, but in reality do nothing to answer them. They are nothing more than preliminary forays that are fraught with problems when one tries to use the results for anything other than to inform further, better designed research. The problem is not necessarily with the studies themselves, which are clearly preliminary — it is with how the media is reporting the results. Which gets into the whole problem of PR and media reporting of research meetings, which I think is getting out of control, but that’s another post for another day…
For now, let’s go through the more obvious limitations of the data presented:
1. The Nurses Study Population — A 33% rate of infertility in the control group and close to 50% in the study group? The background rate of infertility in the US is about 10%. Is this study population representative? I doubt it.
2. Recall bias — This is when an individual who has experienced an adverse outcome does a better job of remembering exactly what drugs and treatments she took than someone who has not had the adverse outcome. It’s human nature, after all, to spend hours, even days looking back and asking “What did I do that caused this?” But it means you remember a lot more than folks with no reason to be so retrospective. The Nurses Study, being a prospective collection of data, does not have this bias, but the Israeli study may.
3. Controlling for the underlying problem: infertility — The Nurse’s Study found that women having more cycles of IVF or clomid had higher rates of autism. But if you need to undergo multiple cycles of infertility treatment, isn’t it just possible that there is something about your gametes — either egg or sperm — that are the problem, and not the drugs themselves? The fact that autism is a highly hereditable disorder supports that there may be an inherent association between the state of infertility and autism. Maybe infertility is nature’s way of controlling the gene pool, so to speak, and we’re messing with it by helping folks get pregnant who maybe weren’t supposed to. Don’t tell that, of course, to the millions of normal, healthy and intelligent individuals born as a result of IVF. But it might explain small increases in certain conditions among children born after fertility treatments compared to the general population, mightn’t it?
While the researchers did control for age, that’s just not good enough. We all know that while age is an excellent marker for IVF success, it is ultimately the quality of the oocyte and sperm that is the most important factor in determining success in fertility treatment.
4. Cause of infertility — One way to tease out the effect of the gametes from the fertility drugs is to focus only on those women with tubal infertility — their eggs are fine, it’s the passageway that’s the problem. If this group had higher autism rates with ovulation inducing drug use, then that’s a stronger association. Unfortunately, the data presented were not detailed enough to address this question.
5. Consequences of fertility treatments — Ovulation inducing agents all lead to higher rates of multiple pregnanies with their associated complications of prematurity and low birth weight, both factors associated with autism. How much of the reported associations were due to these factors and not the fertility treatments themselves?
6. Diagnostic bias — Could it be that parents who persist through multiple fertility cycles against the ever increasing odds that they may not have a child might also be more persistent in getting their child diagnosed with autism? It’s possible, I believe.
To date, well done research into developmental outcomes of children born as a result of assisted reproductive technology (ART) has not found significant risks, but most studies have been confined to infancy and early development, prior to when some children may be diagnosed with autism. Recent studies on longer term childhood outcomes among children born as a result of ICSI (intracytoplasmic sperm injection) have been quite reassuring in this regard. One small study did find a higher rate of autism in children born after ICSI, but not IVF, and the numbers were so small (3 children only had autism), that it’s difficult to make larger conclusions from the data. A recent review article found no reliable studies of autism and ART, and calls for more research.
I would agree.
*This blog post was originally published at The Blog that Ate Manhattan*