CCBYNC Open access
Research

Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum disorders: population based case-control study

BMJ 2013; 346 doi: http://dx.doi.org/10.1136/bmj.f2059 (Published 19 April 2013) Cite this as: BMJ 2013;346:f2059

Re: Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum disorders: population based case-control study

The article unfortunately adds virtually nothing to our understanding of causes of autism.

Approximately two third of cases which can be diagnosed as autism remain undiagnosed and untreated1. It is established that both autism and depression are often under- and misdiagnosed. Probabilities of an under- or misdiagnosis are not known with confidence. Therefore, statistical inference regarding correlations between depression and autism ultimately depends on this uncertainty.

Authors do not make any effort to exclude or at least evaluate the likelihood of diagnosed maternal depression and diagnosed autism being co-related via a third factor. An example of such factor is the following. Mother who is interested in or concerned about her own mental state and is likely to trust doctors to be able to fix it, is more likely to pay close attention to her child's mental development and trust doctors to evaluate and fix it.

Science is prediction. Using the results presented in the paper it is virtually impossible to answer an elementary scientific question: what is the likelihood of the child to be autistic if the mother has mental depression.

In the absence of the causation model, for example a mechanism by which SSRIs or depression in general are affecting fetal development, statistical results alone are useless, however small are the reported p-values.

Also, there seems to be either a typo or an error in the table 1. Paternal depression in the case of autism with ID. 0.5 (presumably %) versus 0.3 for controls. The probability that the proportion of diagnosed paternal depression is greater for the autism cases than for controls is 0.93, which indicates that the p-value in the table (0.293) is likely to be wrong by almost an order of magnitude.

1. Prevalence of autism spectrum disorders in a total population sample. Kim YS, Leventhal BL et al, Am J Psychiatry. 2011 Sep;168(9):904-12. doi: 10.1176/appi.ajp.2011.10101532. Epub 2011 May 9.

Competing interests: No competing interests

20 April 2013
Vacslav Glukhov
Researcher
Independent
Abingdon Villas, London W8 6BS
Click to like:
97
Vote down!