Introduction

Higher rates of preterm births (<37 weeks gestation) resulting in increased mortality and adverse offspring outcomes [1–3] have been associated with maternal psychiatric history [4]. Although frequently attributed to environmental (antenatal distress, elevated neuro-endocrine activity, and homeostasis disruptions) [5] and genetic causes [1,6,7], critical knowledge gaps remain. First, the association between maternal psychiatric history and preterm delivery has not been extensively studied across a wide range of psychiatric disorders. One recent systematic review correlated mental illness before delivery with pregnancy complications including preterm delivery, with most evidence from depression and anxiety disorders [8]. Rather, most previous studies have generally focused on mood and anxiety disorders [8–14], with other psychiatric disorders receiving less attention [4,15–22]. Because these studies have typically relied on modest sample sizes and yielded mixed results, a comprehensive examination across different psychiatric disorders in a large population is required. Second, the association between paternal psychiatric history and preterm delivery has rarely been studied [23–25] and a rigorous study addressing the combined effect of psychiatric disorders in couples is notably absent. To understand the underlying mechanisms, it is essential to disentangle psychiatric history contributions in mothers from fathers. Third, evidence suggests that even moderately shorter gestation, as 1 or 2 weeks before full term, is associated with neonatal morbidity [26] and future health outcomes [2]. How the risk of preterm delivery in offspring of parents with psychiatric diagnoses extends to the entire range of gestational age is largely unknown, particularly at early term (37 to 38 weeks). We are only aware of one study that has examined the risk of birth before full term for several psychiatric disorders in mothers [4]. No studies have examined changes in distribution of gestational age for paternal psychiatric disorders and shift of gestational age in preterm and early term births. Support for such an association will identify a large group at risk with unmet needs.

Using Swedish national registers, we examined the association between a wide range of clinically ascertained psychiatric diagnoses in fathers, mothers, and both parents and gestational age in offspring, compared to offspring born to parents without psychiatric diagnoses.

Methods

Results

Of 1,499,701 infants, we excluded 32 with implausible death/emigration date. Only 10,749 (0.7%) had missing data on covariates (gestational age, sex, parental income and education) and were excluded from the analyses. Thus, our analytic cohort comprised 1,488,920 infants (Fig A in S1 Appendix). A total of 1,268,507 infants were born to parents without a psychiatric diagnosis, of whom 73,094 (5.8%) were born preterm. A total of 4,597 of 73,500 (6.3%) infants were born preterm to fathers with a psychiatric diagnosis, 8,917 of 122,611 (7.3%) infants were born preterm to mothers with a psychiatric diagnosis, and 2,026 of 24,302 (8.3%) infants were born preterm to both parents with a psychiatric diagnosis. The median lag years between the first observed diagnosis and the date of delivery was 7 years (interquartile range (IQR): 4 to 13 years) in fathers and 6 years (IQR: 3 to 11 years) in mothers; the corresponding years for the last observed diagnosis was 2 years (IQR: 0.6 to 6 years) in fathers and 2 years (IQR: 0.6 to 5 years) in mothers. The cumulative distribution of years between birth of the child and last previous diagnosis for mothers and fathers separately is shown in Fig C in S1 Appendix. Based on summary statistics, offspring of parents with psychiatric history tended to be born in recent years and to younger parents with lower level of education and income. Their mothers were more likely to be underweight or obese, to have smoked during pregnancy, and to have spontaneous labor (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Cohort characteristics.

https://doi.org/10.1371/journal.pmed.1004256.t001

Parental psychiatric history before delivery, gestational age, and preterm birth

Overall, compared to offspring of neither parent with a psychiatric diagnosis, the gestational age distribution was shifted towards earlier gestational age for offspring of parents with psychiatric history, particularly for a psychiatric history in mothers and in both parents and for the lower quantiles of gestational age, e.g., quantiles lower than 10% tile, approximately 38 weeks of gestational age (Fig 1 and Fig B in S1 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Differences in gestational age (days) for births to mothers, fathers, and both parents with psychiatric diagnosis compared to that of neither parent with psychiatric diagnosis (reference group) at each different percentile.

Percentile: A division of a data distribution into 100 parts where each part contains 1% of the data. For example, the lowest 5% of the data are found below the fifth percentile. For each psychiatric diagnostic category, the quantile regression models examined the effects of exposures at each different percentile of GA, comparing GA for births of mothers, fathers, and both parents with psychiatric diagnosis, to that of neither parent with psychiatric diagnosis, adjusted for offspring’s birth year in 5-year interval. The quantile regression estimates the shift of GA at the lower tails of the GA distribution, i.e., the most sensitive periods of preterm and early term births. To facilitate interpretation, the GA (approximately weeks) for the reference group is printed underneath each percentile. GA, gestational age.

https://doi.org/10.1371/journal.pmed.1004256.g001

Adjusted for birth year, for any psychiatric diagnosis, the RR for preterm birth was estimated at RR = 1.12 (95% CI [1.08, 1.15] p < 0.001) for fathers, at RR = 1.31 (95% CI [1.28, 1.34] p < 0.001) for mothers, and at RR = 1.52 (95% CI [1.46, 1.59] p < 0.001) for psychiatric history in both parents, compared to when neither parent had a psychiatric diagnosis. Additionally, adjusting for maternal and paternal age, income, and education, the RR was similar for maternal associations 1.27 (95% CI [1.24, 1.29] p < 0.001), but noticeably reduced for paternal associations RR = 1.04 (95% CI [1.01, 1.07] p = 0.006) (Table 2). Adjusting further for maternal BMI and maternal smoking did not further alter the results (Table C in S1 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Parental psychiatric diagnosis before delivery and RR of preterm birth.

https://doi.org/10.1371/journal.pmed.1004256.t002

The risk patterns remained consistent across different psychiatric disorders (Fig 2). Stress-related disorders were associated with the highest risks of preterm birth with corresponding RRs estimated at 1.23 (95% CI [1.16, 1.31] p < 0.001) for a psychiatry history in fathers, at 1.47 (95% CI [1.42, 1.53] p < 0.001) for mothers, and at 1.90 (95% CI [1.64, 2.20] p < 0.001) for both parents (Table 2, Model 1). The patterns of risk for preterm birth were also seen in infants born very preterm, moderate/late preterm, and early term (Fig 3A and Table D in S1 Appendix). Co-occurring diagnoses from different diagnostic categories increased risk; for fathers: RR = 1.10 (95% CI [1.07, 1.13] p < 0.001), 1.15 (95% CI [1.09, 1.21] p < 0.001), and 1.33 (95% CI [1.23, 1.43] p < 0.001), for diagnoses in 1, 2, and ≥3 categories; for mothers: RR = 1.25 (95% CI [1.22, 1.28] p < 0.001), 1.39 (95% CI [1.34, 1.44] p < 0.001), and 1.65 (95% CI [1.56, 1.74] p < 0.001) (Fig 3B and Table E in S1 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Parental psychiatric history before delivery and RR of preterm birth.

RRs with 95% CIs were calculated using log-binomial regression models, adjusted for birth year by cubic natural splines with 5 knots. The dots in the figure represent RRs and lines represent 95% CIs. For psychiatric category “other,” upper 95% CI = 5.48. CI, confidence interval; P.hist, psychiatric history; RR, relative risk.

https://doi.org/10.1371/journal.pmed.1004256.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Parental psychiatric history and risk of preterm birth, by gestational age categories and by co-occurring different psychiatric disorders.

Part A: Parental psychiatric history and risk of very preterm birth, moderate/late preterm birth, and early term birth versus full term birth. ORs of GA <32 weeks (very preterm birth), GA 32 to 36 weeks (moderate/late preterm birth), and GA 37–38 weeks (early term birth) with 95% CIs, versus GA ≥39 weeks (full term birth), were calculated using multi-nominal logistic regression models adjusted for birth year by cubic natural splines with 5 knots. The dots in the figure represent ORs and lines represent 95% CIs. The reference group was offspring of neither parent with a psychiatric diagnosis. Part B: Number of co-occurring psychiatric disorders from different diagnostic categories in fathers and mothers and risk of preterm birth. RRs of preterm birth with 95% CIs were calculated using log-binomial regression models, adjusted for birth year by cubic natural splines with 5 knots and any psychiatric diagnosis of the other parent. The dots in the figure represent RRs and lines represent 95% CIs. Diagnostic categories of psychiatric diagnoses: psychoactive substance use, schizophrenia and other non-mood psychotic disorder, mood disorders, neurotic/behavioral disorders, neurodevelopmental disorders, emotional and behavioral disorders of childhood origin and intellectual disability, and other psychiatric disorder (not otherwise specified). CI, confidence interval; GA, gestational age; OR, odds ratio; RR, relative risk.

https://doi.org/10.1371/journal.pmed.1004256.g003

Discussion

Gestational age is an established predictor for morbidity and mortality across life span. In this nationwide birth cohort study spanning over 20 years, including 1.5 million births, we observed a shift towards earlier gestational age in offspring of parents with a history of psychiatric diagnosis, particularly at lower quantiles of gestational age corresponding to preterm and early term births. Consequently, the risks of preterm and early term birth were increased, with increased risk when only fathers had a psychiatric history, even higher risks when only mothers had a psychiatric history and highest for psychiatric history in both parents, compared to neither parent with psychiatric history. One dose-dependent pattern was seen with increasing number of co-occurring psychiatric disorders from different diagnostic categories in fathers and mothers. After adjustment for parental age and socioeconomic status and maternal BMI and smoking, the risk was largely similar for maternal associations but was noticeably reduced for paternal associations, although it remained statistically significant. The preterm risk was higher among spontaneous deliveries, compared to non-spontaneous deliveries. There were no sex-specific differences in risk.

Strengths of our study include the large population-based cohort design, with prospective ascertainment of psychiatric diagnoses assigned by specialists and gestational age determined using ultrasound. Selection bias was minimized by utilizing data from a publicly financed health care system with universal access and close to complete follow-up. Compared to earlier studies, our sample size and detailed database enabled us to examine risks associated with a wide range of psychiatric disorders in one population and to examine, to the best of our knowledge, the additive effect of maternal and paternal exposure for the first time.

Our study also have several limitation. Despite the large sample size, statistical precision was limited in subgroups, mainly where both parents had specific psychiatric subtypes. Although we examined the risk by diagnostic categories and subtypes of psychiatric disorders, pathophysiology and genetics underlying different categories can still be very heterogeneous. We do not have data on psychiatric diagnoses set in primary care, which excludes subclinical conditions. Possibly these individuals can introduce measurement errors that could result in, if anything, underestimating the actual associations. We do not have information on disease severity and activity during pregnancy. However, we examined the co-occurrence of several psychiatric diagnoses as an indicator of disease load. Future studies will have to address the potential impact of psychiatric medications or other types of treatment. Finally, we cannot rule out residual confounding, e.g., due to maternal infection or other unknown factors, as well as bias due to fetal death, miscarriage, or abortion at extremely early gestational age range.

This is, as far as we know, the first large population-based study to comprehensively examine the full gestational range by psychiatric diagnoses for both the parents. Importantly, the increased risk was present not only for the 5% of the offspring born preterm, but also for the larger group of offspring born at early term—in Sweden approximately 20% of all births, and remained consistent across different psychiatric disorders. There is a growing literature supporting the importance of delivery in the optimal biological window around week 40. For example, morbidity and mortality are elevated not only for infants born preterm, but also for infants born “early term” [2,3], which opens a larger pathway of risk associated with gestational age, not only for a small subset of the population. Only 1 study examined the risk of birth before full term. It included only mothers with psychiatric disorders but reported an increased risk of births before 39 weeks, consistent with our finding of an increased risk of births at early term between 37 and 38 weeks [4]. Two earlier studies on prenatal paternal depression and schizophrenia suggested paternal depression as a risk factor for preterm birth [23], but not for paternal schizophrenia [24]. In our study, we observed increased risk of preterm birth in cases of any paternal psychiatric diagnosis and by a wide range of psychiatric diagnostic categories. Increased risk of preterm delivery in mothers with a psychiatric history has been reported in several studies [4,20,35–39]. The present study supported earlier findings and extended previous data by examining risks in 13 diagnostic categories/subtypes. Our results are in line with studies reporting higher preterm risks with maternal mood and anxiety disorders [8–11,14], stress-related disorders [12,13], eating disorders [15,16], neurodevelopmental disorders [18,19], substance use [20], schizophrenia [21], and OCD [22].

Our results show a consistent risk pattern across different psychiatric diagnoses and gestational weeks. The underlying pathophysiology behind the observed association is complex. The risk of preterm birth is more pronounced in spontaneous than in non-spontaneous delivery. Stress and inflammation/infection are main causes of spontaneous preterm and interact with genetic susceptibility [40]. Maternal stress is also regarded as an important link between maternal psychiatric disorders and preterm birth through endocrine, behavioral, and/or inflammation/immune-mediated mechanisms [5]. Suffering from a psychiatric condition, or having a partner with a psychiatric condition during pregnancy, may be challenging to the mother. Increased psychological stress has been linked to higher levels of stress hormones such as cortisol, which may induce premature uterus contractions [40,41]. Accordingly, the highest point estimates for risk of preterm birth is observed for paternal and maternal stress-related disorders. In addition, 2 meta-analyses show that paternal depression is associated with maternal depression [42,43]. Studies have identified factors such as violence, substance abuse, and smoking as important mediators for poor obstetric outcomes, including preterm delivery, in women with severe mental illness [44–46]. It is possible that the same social and environmental disadvantages that predispose to preterm birth cluster in couples in which both have mental illness [47]. A recent meta-analysis demonstrates that unemployment and financial strain are important risk factors for paternal depression [43]. In line with this, we found that adjustment for parental education, income, and age at delivery decreased risks associated with paternal psychiatric history but risk estimates remained essentially unchanged for maternal psychiatric history. This may indicate the involvement of environmental/social factors in the pathway between parental psychiatric history and preterm births, particularly for paternal associations. Still, genetic causes are the main drivers of risk for psychiatric disorders, and psychiatric disorders are among the most heritable diseases [6]. There is evidence for genetic components and gene–environment interactions in spontaneous preterm birth [1,7], and it is not clear if these are also associated with psychiatric conditions. It is possible that, through shared genetic or environmental factors, parents who carry genetic risk for psychiatric disorders may also carry a higher liability for preterm delivery. In our study, the risk of preterm birth increased with number of different psychiatric disorders in a parent. This dose–response relationship may also indicate that higher disease load, either genetic load or as a measure of disease severity, is linked to magnitude of the risk of preterm delivery. Consequently, if genetic, the pattern with lower association in fathers than in mothers suggests a risk driven by interactions between genetic susceptibility and environmental exposures; the genetic risk may in turn be modified by socioeconomic risks and other gene–environmental interactions [1,7].

At least 15% of couples giving birth in Sweden had a history of psychiatric diagnosis(es), which implies a substantial potential for public health initiatives. Paternal involvement during pregnancy has been shown to moderate maternal stress, and the lack of such paternal support is associated with adverse perinatal outcomes [48]. However, such psychological support may be lacking in which both members of the couple have psychiatric conditions. Linking psychological stressors with increased risk of preterm birth, factors such as social and environmental disadvantages are likely to cluster and psychological support to counteract such stress is more likely to be adversely affected in families were both parents have mental conditions. This raises the concern about preventive efforts in antenatal care and opens possibilities for clinical intervention, e.g., whether additional social support to these families may be needed. Causally, if we speculate that inherited genetic risk as well as “biological” or “psychological” stress can carry over to other populations, the results are generalizable beyond Sweden. However, risk from downstream factors, e.g., obesity and smoking, as well as public health interventions may differ between countries and populations.

Supporting information