The associations between maternal illnesses during pregnancy and autism in children may have been overestimated in past research, according to a study that analyzed data from more than a million children in Denmark. Other factors shared between a mother and her child, such as genetics, could explain nearly all of the connections suggested in past work.
“We do not claim that maternal health definitely makes no difference to autism,” says principal investigator Magdalena Janecka, associate professor of child and adolescent psychiatry and population health at New York University. But “it should be more of a standard practice to examine familial confounding in this type of research.”
Many women who have children with autism say they feel that something they did during pregnancy caused the condition, Janecka adds. “There should be a really high bar for how the associations between pregnancy factors and child neurodevelopment are communicated to the general public. As a field we should be very careful.”
Multiple studies in the past two decades have linked maternal illness during pregnancy to autism in children. But whether the former causes the latter has remained uncertain. For instance, autism is linked to gene variants that influence the immune system, so the mothers of autistic children may just have heightened vulnerability to serious infections.
O
f the 1.1 million children born in Denmark between 1998 and 2015, 18,374 are diagnosed with autism. The researchers examined potential associations between these autistic children and 236 diagnoses their mothers experienced in the one to four years before the children were born. Just 30 diagnoses had a potentially significant association with autism, after accounting for a broad range of demographic and socioeconomic factors.These diagnoses included obstetric, cardiometabolic and psychiatric conditions, such as diabetes during pregnancy and depression. But only one maternal diagnosis—pregnancy complications related to the fetus—remained strongly associated with autism when the researchers analyzed whether any of the 30 diagnoses occurred among the autistic children’s fathers or in the mothers a few years prior to or during pregnancies involving non-autistic siblings. These fetal complications likely did not cause autism but were instead early signs of it, the researchers suggest.
This triangulation method reduces bias and “is often the best way to assess the evidence that an observed association represents a causal relationship,” says Ezra Susser, professor of epidemiology and psychiatry at Columbia University, who was not involved in this study. “This paper shows that for the association between maternal conditions and autism, triangulation is especially important.”
“With their elegant design, [the investigators] show that many of the associations seem to be non-causally related,” says Michael Benros, professor and head of research on biological and precision psychiatry at the University of Copenhagen, who was not involved in the work.
Future research can investigate what genetic and non-genetic factors shared by family members confound the link between maternal health and autism, Janecka says.
The scientists detailed their findings in January in Nature Medicine.
