Prenatal Paternal Selective Serotonin Reuptake Inhibitors Use and Risk of ADHD in Offspring

Study Population

We identified all singletons born alive in Denmark from January 1, 1996, to December 31, 2008, in the Danish Medical Birth Registry (DMBR). The DMBR contains records of all deliveries in Denmark since 1973 and includes information on gestational age at birth from 1978.²⁸ A total of 817 842 singletons were identified during the study period. We excluded 20 349 children for whom we could not identify their fathers, 74 children without information on their mothers, 208 children without information on their mothers’ parity, 485 children who had missing or extreme values of gestational age (≤23 or ≥45 weeks), and 15 256 children who died or emigrated before 3 years of age. Finally, 781 470 children were included in the analysis.

Paternal SSRI Use

From the Danish National Prescription Registry, we obtained information on paternal SSRI use before conception. Since 1995, this registry has recorded all redeemed prescriptions in Denmark with the following information: the civil registration number of the patient, the dispensing date, the medication code (the World Health Organization’s Anatomic Therapeutic Chemical Classification System), the number of packages prescribed, and the number of dose units in each package.²⁹ SSRI use was identified on the basis of the Anatomic Therapeutic Chemical code N06AB. Because spermatogenesis is estimated to take ∼74 days,¹⁵ we defined the susceptible exposure period as the last 3 months before conception. Accordingly, a child was considered exposed if the dispensing date fell within the specified exposure window or the number of days for which the SSRIs were supplied overlapped with the exposure window. Children born to fathers who had no prescriptions for SSRIs and no supply overlap during the entire exposure window were considered unexposed. For further analyses, we extracted data on paternal SSRI use during the last 1 year before the conception of the index pregnancy and data on maternal SSRI use during the index pregnancy.

ADHD

Using the Danish Psychiatric Central Research Register (DPCRR) and the Danish National Patient Register, we identified children with hospital-based diagnoses of ADHD. The DPCRR contains diagnostic information on every admission from psychiatric hospitals and psychiatric wards in general hospitals in Denmark since 1969 and includes data on all outpatient visits and emergency department contacts since 1995.³⁰ The Danish National Patient Register has collected data on all inpatients from all somatic hospitals in Denmark since 1977 and outpatients since 1995.³¹ The coding classification used during the study period was the International Classification of Diseases, 10th Revision (ICD-10). Children with a diagnosis of ADHD (ICD-10 code F90) in the study cohort were defined as case patients. Children were followed up from 3 years of age until the first diagnosis of ADHD, death, emigration, or December 31, 2013, whichever came first.

Covariates

Using the Danish nationwide health registers, we retrieved data on characteristics that may be associated with ADHD in the children or paternal SSRI use. For each child, we obtained information on the calendar year of birth, sex, birth weight, Apgar score at 5 minutes, gestational age, maternal parity, maternal and paternal age at the child’s birth, maternal years of education, and maternal smoking status during pregnancy from the DMBR. Parents’ psychiatric history before the birth of the child was obtained from the DPCRR by International Classification of Diseases, Eighth Revision codes 290-315 from 1977 to 1993 and ICD-10 codes F00-F99 from 1994 and onward. Furthermore, we identified parents diagnosed with affective disorders before the birth of the child (specifically, International Classification of Diseases, Eighth Revision codes 296.09, 296.19, 296.29, 296.39, 296.99, 298.09, 298.19, 300.49, and 301.19 and ICD-10 codes F30–F34 and F38–F39).

Statistical Analysis

All analyses were performed by using SAS version 9.1 (SAS Institute Inc, Cary, NC). Cox regression models were used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) of ADHD in children following exposure to paternal SSRI use before conception.

We adjusted the analyses for the calendar year of birth (1996–1998, 1999–2001, 2002–2004, and 2005–2008), sex of the child (male or female), parity (1, 2, or ≥3), parental age at the child’s birth (≤25, 26–30, 31–35, and >35 years), maternal years of education (≤9, 10–14, and >14 years), maternal smoking status during pregnancy (yes or no), maternal history of psychiatric disorders before child’s birth (yes or no), and maternal antidepressant use during pregnancy (yes or no) in model 1. We additionally adjusted for paternal history of psychiatric disorders before child’s birth (yes or no) in model 2.

To distinguish the direct effect of SSRI use from that of the underlying indication for SSRI use, we extended the exposure window to 1 year before conception and examined the effect of SSRI use before, but not during, the defined 3-month susceptible period. This so-called negative-control strategy allowed us to identify a group of men who used SSRIs only during the period from 1 year to 3 months before conception (former users) and the other 2 groups of men who used SSRIs during the 3-month susceptible period: those who used SSRIs only during the 3-month susceptible period (current users) and those who used SSRIs both before and during the 3-month susceptible period (both former and current users). The reference group consisted of those children born to fathers who did not use SSRIs at any time in the last 1 year before conception. Considering that spermatogenesis happened during 3 months before conception, the ADHD risk observed among the former users could reflect the effects of underlying indication rather than the SSRI use itself.

Stratified analysis was performed to examine whether the association between paternal SSRI use and ADHD in children differed by sex. Considering the potential role of maternal antidepressant medication use as well as maternal mental health during pregnancy in offspring’s neurodevelopment, we restricted the analyses to the children whose mothers neither received antidepressants during pregnancy nor had affective disorders. To further distinguish the effect of SSRI medication from that of the main indication (ie, affective disorders), we performed stratified analyses according to paternal history of affective disorder before the birth of the child. As for those children born to fathers with affective disorders, the ADHD risk we examined could be solely attributable to paternal SSRI use.

To control for unmeasured family-related confounding factors, we conducted a sibling study by including only families with multiple children born in our study period and with at least 1 child with paternal SSRI preconception exposure. We then compared exposed siblings to unexposed siblings to estimate the association between paternal SSRI exposure and ADHD in children.

Considering that there were significant differences among the baseline characteristics of the exposed group and unexposed group, we further created a propensity score–matched (PSM) subcohort to reduce the potential selection bias. The propensity scores were estimated by using logistic regression as the probability of paternal SSRI exposure given some of the baseline factors. The independent variables included in the propensity score model were parental age at the child’s birth, maternal smoking and antidepressant use during pregnancy, parental history of affective disorder, and parental history of psychiatric disorder. We then used greedy match algorithms to match the exposed group to the unexposed group. After the matched pairs were identified, differences between matched pairs were evaluated. We then repeated the analysis among the PSM subcohort with additional adjustment for the calendar year of birth, sex of the child, and parity.

The 3-month cutoff point that we defined as the susceptible period was based on the fact that spermatogenesis takes ∼70 to 90 days, whereas it may also be possible that SSRIs induce adverse effects (including sperm damage) at the primitive stage. Hence, we defined another exposure period as 6 months before conception and repeated all the above analyses.