Article Text

Original research
Nervous system drugs taken by future fathers and birth defects in offspring: a prospective registry-based cohort study
  1. Maarten Wensink1,2,
  2. Ying Lu3,
  3. Lu Tian3,
  4. Tina Kold Jensen4,
  5. Niels Erik Skakkebæk5,
  6. Rune Lindahl-Jacobsen1,2,
  7. Michael Eisenberg6
  1. 1Department of Epidemiology, Biostatistics and Biodemography, University of Southern Denmark, Odense, Denmark
  2. 2Interdisciplinary Center on Population Dynamics, University of Southern Denmark, Odense C, Denmark
  3. 3Department of Biomedical Data Science, Stanford University School of Medicine, Stanford, California, USA
  4. 4Department of Environmental Medicine, University of Southern Denmark, Odense, Denmark
  5. 5Department of Growth and Reproduction, Juliane Marie Centre, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark
  6. 6Male Reproductive Medicine and Surgery, Department of Urology, Stanford University School of Medicine, Stanford, California, USA
  1. Correspondence to Dr Maarten Wensink; mwensink{at}


Objectives To evaluate the association of paternal intake of antipsychotics, anxiolytics, hypnotics and sedatives, antidepressants, selective serotonin reuptake inhibitors (SSRIs) and (benzo)diazepines during the development of fertilising sperm with birth defects in offspring.

Design Prospective registry-based cohort study.

Setting Total Danish birth cohort 1997–2016 using Danish national registries.

Participants All 1 201 119 Danish liveborn singletons born 1997–2016 were eligible, 39 803 (3.3%) of whom had at least one major birth defect.

Exposure Offspring were considered exposed if their father had filled at least one prescription in the relevant drug category during development of fertilising sperm (the 3 months prior to conception).

Primary and secondary outcome measures Primary outcome was the diagnosis, in the first year of life, of at least one major birth defect as categorised in the EUROCAT guidelines. Secondary outcome was the diagnosis, in the first year of life, of at least one major birth defect in any of the EUROCAT subcategories. Adjusted ORs (AORs) were calculated, along with their 95% CIs, adjusted for year, education, smoking status and age of the mother, and education, disposable income and age of the father.

Results This study found weak or null associations between birth defects and selected drugs. Specifically, antidepressants (17 827 exposed births) gave 3.5% birth defects (AOR 0.97 (0.89 to 1.05)). Diazepines, oxazepines, thiazepines and oxepines (as antipsychotics, 1633 offspring) gave 4.7% birth defects (AOR 1.22 (0.97 to 1.54)), attenuated to 1.13 when excluding by mothers’ prescriptions. The study was well powered assuming 100% therapy adherence, while assuming 50% therapy adherence, the study remained well powered for the largest groups (SSRIs and antidepressants overall).

Conclusions Antipsychotics, anxiolytics, hypnotics and sedatives, antidepressants, SSRIs and benzodiazepine-derived anxiolytics, when taken by the father during development of fertilising sperm, are generally safe with regard to birth defects.

  • epidemiology
  • neurology
  • preventive medicine
  • public health
  • reproductive medicine

Data availability statement

Data may be obtained from a third party and are not publicly available. Data from Statistics Denmark cannot be made publicly available but can be applied for through the usual ways at The grant proposal is summarised here:

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • High-quality registry data give full coverage of population.

  • Highly powered study for most of the investigated drugs.

  • Unable to assess actual drug intake.

  • Unable to assess associations between drugs and fertility.


Certain neurological drugs have been associated with adverse changes in semen quality. Beyond common reproductive outcomes like sperm motility, selective serotonin reuptake inhibitors (SSRIs) have been associated with increased frequencies of DNA fragmentation and abnormal sperm morphology.1–4 Anxiolytics, in particular benzodiazepines, have been associated with chromosomal abnormalities in sperm5 6. Of concern, many of these drugs are commonly prescribed to prospective fathers with increasing use over time.7 In Denmark, the proportion of births where the father had been prescribed neurological drugs in the 6 months preceding conception more than doubled between 1997 and 2017, from approximately 4% to almost 9%. Importantly, prescriptions of antidepressants, mostly SSRIs, increased threefold to 2.5%.7

It is known that paternal factors are associated with birth outcomes such as preterm birth, low birth weight and neonatal intensive care unit stays.8 9 Given the association of sperm DNA damage in certain neurological drugs, the safety of neurological drugs regarding offspring health needs to be evaluated. In particular, it is unknown whether paternal use of these drugs during sperm development is associated with the risk of birth defects.

Hence, we performed a cohort study on all singleton live births in Denmark 1997–2016 (1 201 119 births), linking national registries: the birth registry, the prescription registry and the patient registry. We then assessed for any association between specific neurological drugs prescribed to the father to be in the 3 months just prior to conception (sperm development) and birth defects diagnosed in the first year of life.


Data and inclusion criteria

We obtained the Danish Medical Birth Registry (MFR10) 1997–2016, which contains all births in Denmark from 20 weeks of gestation onwards. In addition to characteristics of the newborn and pregnancy, such as gestational age and Apgar score, this registry contains the CPR (Centrale Personregister) number,11 a unique identifier that all Danish citizens and residents have been given since 1968, for newborn, mother and father (if known). We used this CPR number to link registries, meaning that entries with unusable or missing CPR number of either parent or offspring were deleted. Stillbirths were also deleted due to dissimilar ascertainment of birth defects (see Outcome). Approximate conception date is contained in the MFR as birth date minus estimated gestational age.

We linked this registry to the Danish National Prescription Registry (LMDB12), which we obtained from 1995 to mid-2018. This registry gives complete coverage of all prescriptions filled in Denmark by persons with a CPR number. In Denmark, over-the-counter drug prescriptions are limited; common pain medication like paracetamol is not freely available in large packages. From this registry, we created indicator variables for exposure (see Exposure). We also used this registry to identify those births where the mother had taken any of the investigated drugs up to giving birth (see Statistical Analyses).

We further linked with the Danish National Patient Registry13 from 1995 through mid 2018, which contains diagnoses for all inpatient and outpatient contacts, although not for diagnoses in the family doctor setting. This registry includes birth defects, which we classified according to the EUROCAT guidelines,14 allowing 1 year of follow-up on birth. Birth defects that EUROCAT classified as minor were excluded.

We incorporated information from Statistics Denmark, the central authority on Danish statistics. These variables were paternal disposable income, the amount of money that a person or household has available for spending and saving after income taxes and interest expenses have been accounted for, and highest achieved education (both by year). We further linked with the Population Registry to obtain birth date and sex of the parents. Births with fathers of unknown or female sex were removed, as were births to mothers of male sex.


The primary outcome was the diagnosis of at least one major birth defect in the first year of life (binary variable), categorised as per the EUROCAT guidelines,14 which provide International Classification of Diseases (ICD) codes of birth defects that they classify as major. The secondary outcome was being diagnosed with at least one major birth defect (binary variable) in any of the EUROCAT subcategories (by organ or tract).


As one spermatogenic cycle takes approximately 3 months,15 we considered offspring whose father filled a prescription in the relevant category during the 3 months preconception as exposed. We examined the following medication categories: antipsychotics (Anatomical Therapeutic Chemical classification code N05A), among which diazepines, oxazepines, thiazepines and oxepines (N05AH); anxiolytics (N05B), among which benzodiazepine-derived anxiolytics (N05BA); hypnotics and sedatives (N05C), among which benzodiazepines (N05CD); and antidepressants (N06A), among which SSRIs (N06AB).

Missing data

Approximately 15% of the merged records had at least one entry missing, in particular maternal smoking status (online supplemental table 1). We imputed 10 datasets in a procedure described in detail in the Statistical appendix under the assumption of missingness at random. Reported results are estimates and SEs pooled under Rubin’s rule. Imputation and pooling was handled with the R package mice16 (V.3.8.0).

Statistical analyses

We employed flexible logistic regressions using generalised additive models with R package mgcv17 V.1.8–33, which allow non-linear smooth associations between the exposure variable and the birth defect risk. Categorical variables were modelled by simple indicator variables for each level. From these models, we obtained ORs and their 95% CIs for being diagnosed with at least one major birth defect in the first year of life after adjusting for birth year, maternal factors (smoking status during pregnancy, highest achieved education, maternal age) and paternal factors (disposable income, highest achieved education and paternal age). These potential confounders were selected prior to the analysis for their potential relatedness to both the predictor and outcome18–21 and were not selected based on their significance.

We compared exposed versus unexposed groups for each drug group separately, first for all liveborn singletons. As a sensitivity analysis, we then repeated this analysis excluding births where the mother had taken any of the investigated drugs at any time prior to delivery. We then compared, by conditional logistic regression, exposed versus unexposed offspring of the same father, adjusting for birth year, maternal age, and nulliparity. We then analysed the distribution across EUROCAT organ subgroups without excluding births based on maternal drug use.

All data analyses were carried out on the secure server of Statistics Denmark and run in R22 V.3.6.3.

Minimum detectable risk and OR calculations

We calculated minimum detectable ORs at 80% and 90% power using the software PS Power and Sample Size, V.3.1.623 both for the actual exposure numbers and under the assumption that 50% of the fathers actually took their prescriptions. Because some drugs induced highly selective groups (see Results), we conservatively assumed an exposed:unexposed ratio of 1:10 for these calculations (the larger groups tended to be less selective, see Results).

Patient and public involvement statement

Patients or the public were not involved in the planning, executing and communication of this study.


The cohort

The Birth Register had 1 276 229 records for 1997–2016. After exclusion of records with unusable CPR of the offspring (2888) or father (1150), 1 272 750 records could be linked to the patient register, the prescription register, (socioeconomic) variables held at Statistics Denmark and the population register. Excluding births to fathers with registered unknown or female sex (19 163), mothers of male sex (7), and stillbirths (1927) left 1 251 653 records for multiple imputation. After imputation, excluding records of non-singleton births (50 534) and records with missing gestational age (27 080) left 1 174 727 offspring. Exclusion of births with mothers who filled a prescription of any of the investigated drugs at any time up to delivery left 936 706 offspring.

Among the 1 174 727 births available for the main analysis, that is, liveborn singletons without missing gestational age, 17 827 offspring were exposed to antidepressants, including 11 902 to SSRIs; 4301 to antipsychotics, including 1633 to diazepines, oxazepines, thiazepines and oxepines; 4918 to anxiolytics (primarily benzodiazepines, n=4742); and 5797 to hypnotics and sedatives, of which 1153 to benzodiazepines (tables 1 and 2). Grouping (benzo)diazepines resulted in 7057 exposed births. Exclusion of births where the mother had taken any of the investigated drugs prior to delivery reduced the exposure numbers (by approximately 1/3), representative of the correlation between parents for these drugs (table 2).

Table 1

Cohort characteristics by drug use for all liveborn singletons in Denmark 1997–2016

Table 2

Specific neurological drugs associated with sperm damage and their adjusted ORs (AORs) for having at least one major birth defect

Fathers who were prescribed any neurological medications before conception were older, as were their partners (table 1). Differences in education, income, maternal smoking and parity were also noted. Preterm percentages were slightly higher in the drug exposed groups (>6%) versus the non-exposed group (5%). The sex ratio was similar for all exposure groups relative to the non-exposed group.

Multiple imputation results suggested that missing data were unlikely to have influenced the results from the complete case analysis. The regression results with or without multiple imputation showed only very modest associations for potential confounders, mostly maternal education with an adjusted OR (AOR) just below 1.1 for low education.

Birth defects analysis

Birth defects in children of fathers exposed to neurological drugs just before conception were generally similar to those in the unexposed population (3.3%–3.9% exposed vs 3.3% unexposed, table 1). After multivariable adjustment, all 95% CIs included one (table 2). Results were similar in the siblings analysis (table 2). For antidepressants and SSRIs, the ORs were 0.97 (0.89 to 1.05) and 0.94 (0.85 to 1.04), respectively (all liveborn singletons), and 0.95 (0.85 to 1.05) and 0.96 (0.85 to 1.09) after exclusion. There was a moderate but not statistically significant tendency towards higher birth defect risk among children whose fathers were prescribed diazepines, oxazepines, thiazepines and oxepines (N05AH), which showed an AOR of 1.22 (95% CI 0.97 to 1.54) for all liveborn singletons, and 1.13 (95% CI 0.81 to 1.57) after exclusion of births to mothers ever prescribed any drug in the groups investigated here. In this group, birth defects appeared especially elevated in the urinary tract (0.73% vs 0.26%, p<0.001 (p=0.04 after Šidàk correction for multiple testing), table 3).

Table 3

EUROCAT subgroups (binary: ≥1) by drug class, all liveborn singletons, Denmark 1997–2016

Power and detectable odds

At 80% or 90% power, the minimum detectable OR was between 1.1 and 1.3 for the larger groups but approximately 1.5 for the smaller groups (N05AH and N05CD, table 4). Assuming a therapy adherence of 50%, minimum detectable ORs were approximately 1.3 for antidepressants or SSRIs, approximately 1.5 for antipsychotics, anxiolytics, hypnotics and sedatives, as well as for benzodiazepine-derived anxiolytics. For benzodiazepines as hypnotics and sedatives (N05CD), minimum detectable ORs could be as high as 2.1 (table 4).

Table 4

Minimum risks detectable as aberrant


Summary of findings

The current study found weak or null associations between offspring birth defects and prescriptions of common neurological drugs filled by the father during the 3 months preconception. The only medication group that suggested a possible association was diazepines, oxazepines, thiazepines and oxepines (as antipsychotics, N05AH), which showed a moderately elevated AOR of 1.22 (0.97 to 1.54)) for all liveborn singletons. The point estimate was similar in the sibling analysis, but reduced to 1.13 after excluding offspring whose mother had filled a prescription of any of the investigated drugs at any time prior to delivery. For SSRIs, a large group with the strongest prior evidence of associated sperm damage, the AOR was 0.94 (0.85 to 1.04) before exclusion and 0.96 (0.85 to 1.09) after exclusion. Results were similar when comparing exposed to unexposed siblings. The number of births with paternal exposure to each of the drugs was generally large enough to detect a what might be a clinically significant elevation in risk, for the larger groups even when assuming that only half the fathers took the medication that they had been prescribed.

Strengths and limitations

The design of a nationwide, registry-based cohort study allowed the inclusion of large numbers of fathers who were prescribed the investigated drugs just before conception and to ascertain whether their offspring had birth defects. The registries used are generally complete and of high quality, with (hospital) reimbursement generally depending on reporting and with cross-checks between registries in place. Further information can be found in references.10–13 Although our measure of paternal exposure was indirect—filling a prescription does not equate with taking the drugs—the study had power to overcome exposure misclassification. Dosage and exact timing of exposure were not considered, which could have biased our results towards the null.

We did not have information on paternal lifestyle factors, such as exercise or smoking, and there may have been maternal factors (eg, genetic predisposition, lifestyle factors like exercise) for which we could not control. We saw significant differences in demographics between fathers prescribed drugs and those who were not. However, these factors are unlikely to have biased the results towards the null because that would require paternal drug prescriptions to correlate with protective maternal or paternal factors.

Even using registry data, there remains a possibility that offspring of fathers prescribed neurological drugs are less visible to the healthcare system because of the fathers’ psychological or psychiatrical ailments. This could result in reduced birth defect ascertainment for these offspring and hence bias the results towards (or even below) the null. Nevertheless, Denmark has universal healthcare with scheduled check-ups for newborns, both at birth and in the first year of life, and we restricted to birth defects classified by EUROCAT as major. Thus, it seems reasonable to suspect that the majority of birth defects would be diagnosed. However, if there was an association between a paternal medication and an earlier reproductive outcome (e.g., failure to fertilize, miscarriage), the effect on birth defects could be interpreted as bias towards the null.

Interpretation, possible mechanism, comparison with the literature

Although sperm DNA damage suggests a risk to offspring, this risk may not materialise if sperm with damaged DNA fail to fertilise an egg cell, if the oocyte corrects any DNA damage, if the conceptus fails to develop into a viable fetus or if the fetus is aborted. Hence, sperm damage could lead to subfertility or infertility, but not birth defects. As the Danish Medical Bi

rth Registry covers only pregnancies from week 20 onwards, further studies are necessary to explore this hypothesis.

Literature on paternal effects on offspring is limited. Certainly, it is reasonable to expect that the 9 months a fetus spends developing in utero gives more scope for teratogenic effects from maternal exposure than preconception spermatogenic paternal contribution. Yet there is increasing evidence that sperm contributes more than DNA alone,24 and the early stages of pregnancy are also the most vulnerable stages with regard to birth defects.

The observation of a tendency towards increased risk in diazepines, oxazepines, thiazepines and oxepines (as antipsychotics, N05AH) may be due to the disease rather than drug, although antipsychotics as a whole only very mildly tended towards an increased OR, while neither birth defects of the nervous system nor chromosomal birth defects were elevated in this group. However, a prior study did suggest a possible association between paternal diazepam and perinatal mortality and growth retardation.25 The attenuation of the point estimate seen when excluding births where mothers had been on any of these drugs may indicate confounding by maternal associations. However, if a significant share of the N05AH-exposed offspring were actually unexposed, because the father may not have taken the filled prescription, 1.22 may be an underestimate of the true association.


The current study found weak or null associations between prescriptions of neurological drugs (antipsychotics, anxiolytics, hypnotics and sedatives, antidepressants, and SSRIs, benzodiazepine-derived anxiolytics) filled by the father during the development of fertilising sperm (3 months before conception) and birth defects in the offspring. Paternal use of diazepines, oxazepines, thiazepines and oxepines (as antipsychotics, N05AH) during the development of fertilising sperm may be associated with mildly elevated birth defect frequencies, although a maternal pathway is not excluded here and although this observation could be due to chance. As such, men can be counselled that these medications likely do not increase the risk of birth defects. Further studies are necessary to investigate whether these drugs lead to higher rates of stillbirths, early abortions or failure to fertilise, as well as the group N05AH.

Data availability statement

Data may be obtained from a third party and are not publicly available. Data from Statistics Denmark cannot be made publicly available but can be applied for through the usual ways at The grant proposal is summarised here:

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants, but registry studies are exempted in Denmark from IRB review. They are subject to strict regulations. They are executed on a secure server at Statistics Denmark, and only aggregate data are allowed to leave the server (tables, regression coefficients) after being controlled by someone else than the primary analyst.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors ME, RL-J, NES, YL and TKJ designed the study. MW, YL and LT handled data and statistical analysis. MW wrote the first draft. All authors interpreted the results, revised the manuscript and approved the final version. MW is the guarantor of the article.

  • Funding This study was funded by NIH grant HD096468 (to ME).

  • Disclaimer The funder had no role in the study design, collection, analysis and interpretation of the data and in the writing of the report.

  • Competing interests YL reports grants from Merck, and personal fees from United Health Care, Nektar, and Gilead, outside the submitted work. ME reports advisorships for Sandstone Diagnostics, Dadi, Hannah and Underdog, which are fertility-related companies.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.