Paternal use of antidepressants and offspring outcomes in Sweden: nationwide prospective cohort studyBMJ 2018; 361 doi: https://doi.org/10.1136/bmj.k2233 (Published 08 June 2018) Cite this as: BMJ 2018;361:k2233
All rapid responses
Dr. Saripanidis correctly comments that paternal use of antidepressants around conception was associated with an increased risk of autism in the offspring, in models without adjustment for confounding factors. It is however wrong to say that the effect disappears after ‘arbitrary covariate adjustment’. Instead the covariates in our statistical models were all carefully selected (prior to our analyses) and they all fulfill the standard requirement for being a confounder in a regression model.
The standard requirement is being associated with both the exposure and the outcome. These covariates are available through the Swedish national registers controlled by the Swedish government and constantly monitored to ensure high quality. The registers are managed by full time professionals with special skills in IT, public health work and register administration. Our analytic team have long experience in statistical and epidemiological methods, psychiatry and autism ascertainment and working with the Swedish registers. The variables we included are used in our previous studies in maternal exposures and offspring outcomes where confounding may be similar (1, 2). The un-adjusted results may serve as an example of a risk estimate severely affected by confounding which would indeed have been the main results if we would not have been able to perform our very detailed analysis correctly adjusting for confounding factors.
Dr Degner kindly commented on our work. He correctly acknowledged that antidepressants have been increasingly prescribed in the past few decades. He omitted that it is more accurate to state that SSRIs increased in use while others have comparatively dropped in use. He rightly noted that investigations on risks due to paternal antidepressant usage are scarce. Moreover, he noted our results are consistent with prior observational findings.
In his comments Dr Degner cited research by Pedersen et al that SSRIs and SNRIs are heterogeneous with regard to the degree of serotonin selectivity. Pedersen et al. examined septal heart defects among children whose mothers were treated with SSRIs in early pregnancy, particularly sertraline and citalopram . This outcome was not specifically examined in our study. Hence we cannot make claim to that outcome. However, we conducted sensitivity analysis by drug so Dr Dregner may have misinterpreted the scope of our work.
Dr. Degner's letter needs modifications owing to inaccuracy. Dr Degner erroneously stated that we had “limited access to psychiatric diagnoses”. That is inaccurate. Our target population are the 170 508 Swedish live born children, conceived from July 29, 2005 and born up to December 31, 2007. For the parent, mother as well as the father, we extracted data from one of the largest and most complete national databases of clinically ascertained psychiatric diagnoses. The diagnoses are assigned by specialist clinicians and has been used extensively in epidemiological research. Furthermore, these diagnoses were utilized in a detailed way in the analyses in order to adjust for confounding where we adjust for 10 different diagnostic groups of psychiatric conditions. They were selected a-priori and have been used in our earlier publications (1, 2).
In our opinion, Dr Degner did not convey the basic epidemiological rationale for the choice of a confounder. A confounder is associated with both the exposure and the outcome (3). A point in case is that Dr Dregner cited research that maternal use of mood stabilizers, like valproate, during pregnancy was associated with a significantly increased risk of ASD. It is unclear to us why maternal valproate use should be accounted for in our study. We chose our confounders as classically done in epidemiology (3). By definition, a confounder is related to both the outcome and exposure. Unless Dr Dregner argued that paternal antidepressant use is causally associated with maternal valproate use. Adjusting for genetic risk not associated with both the outcome and the antidepressant medication may instead introduce bias by opening for unmeasured confounding.
Dr Degner also writes, “A serious interpretation of paternal contribution must include the risk of adverse offspring outcomes from inherited genes and epigenetic changes directly or indirectly and consider spermicidal and antitrichomonas activities of antidepressants”. First, as a clarification to readers who have not read our manuscript, our analysis does indeed consider inherited genetic risk. Our approach is to adjust for psychiatric diagnosis at the time of birth of the offspring. Despite our study being large, there was not sufficient statistical power to perform an analysis adjusting for familial confounding (1). Second, in our manuscript we do mention several underlying potential causes/mechanisms which may have an adverse effect on the offspring through paternal sperm, including a spermicidical effect of (at least some of) SSRIs. However, spermicidal and antitrichomonas activities of antidepressants are factors on the causal path between use of antidepressant medication and potential offspring adverse outcomes and including such covariates in the statistical model may introduce unwanted biases. If we had found large effects of paternal use of antidepressants at conception a more detailed analysis of these factor would have been warranted. Given our results, we do acknowledge that residual confounding may still exist and we cannot rule out that there are unwanted effects in smaller sub-groups.
We hope we have clarified the misunderstandings and/or misinterpretations in Dr Degner's response. Paternal intake of antidepressants during the period around conception is safe with respect to the risk of the four major adverse outcomes in offspring: preterm birth, malformation, autism, and intellectual disability.
We welcome future constructive well conceived critiques of our work to promote academic debate.
1. Viktorin A, Uher R, Kolevzon A, et al. Association of Antidepressant Medication Use During Pregnancy With Intellectual Disability in Offspring. JAMA Psychiatry Published Online First: 12 July 2017. doi:10.1001/jamapsychiatry.2017.1727
2. Viktorin A, Uher R, Reichenberg A, et al. Autism risk following antidepressant medication during pregnancy. Psychological Medicine 2017;:1–10. doi:10.1017/S0033291717001301
3. Clayton D, Hills M. Statistical Models in Epidemiology. Oxford Scientific Publications
Competing interests: Support from the National Institutes of Health, Beatrice and Samuel A Seaver Foundation, Fredrik and Ingrid Thuring Foundation, and Swedish Society of Medicine for the submitted paper; no financial relationships with any organisations that might have an interest in the submitted paper in the previous three years; no other relationships or activities that could appear to have influenced the submitted paper.
Paternal use of antidepressants and offspring outcomes in Sweden: nationwide prospective cohort study
Antidepressants have been increasingly prescribed in the past few decades, but investigations on risks due to paternal antidepressant usage are limited. This large observational prospective cohort study  supports previous negative research findings.
A subgroup analysis of antidepressant drugs must consider that the group of SSRIs and SNRIs is heterogeneous, with varying degrees of serotonin selectivity. There was an increased prevalence of septal heart defects among children whose mothers were treated with SSRIs in early pregnancy, particularly sertraline and citalopram .
The authors had limited access to psychiatric diagnoses; there was only a low number of psychiatric diagnoses observed in fathers, paternal use of non-antidepressant drugs, somatic diseases, smoking or alcohol consumption. These factors have a high relevance for adverse effects on offspring, for example, the use of mood stabilizers as co medication. Maternal use of valproate during pregnancy was associated with a significantly increased risk of autism spectrum disorder and childhood autism in the offspring .
The setting of this study was Sweden nationwide. Genetic and epigenetic effects are very complex. A serious interpretation of paternal contribution must include the risk of adverse offspring outcomes from inherited genes and epigenetic changes directly or indirectly and consider spermicidal and antitrichomonas activities of antidepressants.
The strong limitations of this study encourages no accurate, straight message and confuses doctors, patients and parents.
1) Viktorin A et al. Paternal use of antidepressants and offspring outcomes in Sweden: nationwide prospective cohort study.BMJ 2018;361:k2233
2) Pedersen LH et al. Selective serotonin reuptake inhibitors in pregnancy and congenital malformations: population based cohort study.BMJ 2009;339:b3569
3) Christensen J et al. Prenatal valproate exposure and risk of autism spectrum disorders and childhood autism.. JAMA 2013;309(16):1696-703. doi: 10.1001/jama.2013.2270
4) Kumar VS, et al. The spermicidal and antitrichomonas activities of SSRI antidepressants. Bioorg Med Chem Lett2006;16:2509-12
Competing interests: No competing interests
Re: Paternal use of antidepressants and offspring outcomes in Sweden: nationwide prospective cohort study
In figure 2, we read that paternal use of antidepressants around conception was also associated to autism in their children.
Only after Authors' arbitrary successive covariate statistical adjustment the observed major adverse outcome vanishes.
Competing interests: No competing interests