Pregnancy intention in relation to maternal and neonatal outcomes in women with versus without psychiatric diagnoses

Studies have identified adverse maternal and neonatal outcomes for women with psychiatric disorders. Additionally, psychiatric disorders may pose an increased risk for unintended pregnancies (UPs) which in turn may also impact negatively on outcomes. The present study aims to compare the incidence of UPs in women with versus without current/past psychiatric diagnoses and investigates whether psychiatric history modifies the relation between delivery outcomes in women with and without UPs.


| INTRODUCTION
The perinatal period is a vulnerable period for women with psychiatric diagnoses and their offspring.2][3] Women with previous psychiatric disorders such as psychotic, bipolar, and depressive disorders have a considerable risk of relapse in the peripartum period. 1,2,4,5][8] Mothers with psychiatric disorders are at higher risk of admission to an intensive care unit, unplanned cesarean section, gestational diabetes, preeclampsia, induced labor, and have lower rates of breastfeeding. 6,7,9Neonates more often have low 5-min Apgar scores, low birthweight or are born preterm. 6,9,10Besides, psychiatric disorders may increase the risk for unintended pregnancies (UPs). 11Similarly, UPs are associated with various adverse outcomes such as increased risk for depression in mothers, 12,13 lower rates of breastfeeding 14,15 and preterm birth and low birthweight in neonates. 16,17In a systematic review and meta-analysis, we demonstrated an increased risk for UPs in women with psychiatric diagnosis compared to women without psychiatric diagnosis (n = 2650, 11 studies). 11The overall weighed prevalence of UPs was 65% in women with psychiatric diagnoses.There are various factors that may attribute to the UP risk in women with psychiatric disorders such as victimization by sexual violence 18 or disruption of menstrual cycles. 19,20Additionally, advanced planning capabilities, required for adequate use of contraceptive methods and

Significant Outcomes
• Current/past psychiatric diagnoses are associated with a higher odd of UPs and demand attention for pregnancy planning in psychiatric healthcare.

• Clinical maternal and neonatal outcomes after
UPs were comparable between women with and without current/past psychiatric diagnoses, except from a higher gestational age at delivery in women with current/past psychiatric diagnoses (mean difference 1.83 days, p = 0.006).

Limitations
• Pregnancy intention is notoriously difficult to assess and should be assessed prospectively with a validated instrument.Our retrospective study design did not allow prospective assessments, since we used reports by health care providers registered in health records.• We included women with any current/past psychiatric diagnosis, but did not adjust for severity of symptoms, timing, or current presence of the diagnosis.• As our study presents data from a birth cohort, women with abortions could not be included.The actual UP rates might thus be underestimated in the current study.
family planning, might be diminished in women with psychiatric disorders. 20,21Although possible adverse consequences of both UPs and psychiatric disorders have been studied separately, the impact of UPs on maternal and neonatal outcomes in women with psychiatric disorders has not been studied so far.The present study elaborates on previous literature by primarily investigating the incidence of UPs amongst women with various current/past psychiatric diagnoses versus women without psychiatric diagnoses.The secondary aim is to compare maternal and neonatal outcomes between women with UPs versus non-UPs who delivered in the hospital and to assess the modifying role of current/past psychiatric diagnoses in the association between pregnancy intention and maternal/neonatal outcomes.We hypothesize that women with both psychiatric diagnoses and UPs have the highest risk for adverse outcomes compared to women without psychiatric diagnoses or with intended pregnancies.

| Study design and eligibility criteria
A retrospective cohort was compiled of women who gave birth in OLVG hospital, Amsterdam, the Netherlands.Women ≥18 years old with birth registrations in the electronic patient file during January 1, 2015 to March 1, 2020 were included.Women with twin pregnancies were excluded.We excluded all women with missing data on psychiatric history and/or pregnancy intention.After selection of all eligible women in the cohort, we included the most recent birth of each individual woman in the cohort, hence, all women were included only once (Figure 1).

| Data collection
Data collection was conducted using CTcue, a software program that searches free text in pseudonymized electronic hospital charts.Multiple study investigators performed manual checks in hospital charts in case of missing data.

| Current/past psychiatric diagnosis
The presence of a current/past psychiatric diagnosis, screened at the pregnancy intake according to the standardized protocol, was defined as one or more of the following: presence and/or history of a psychiatric diagnosis as reported by the pregnant woman herself or by one of her caretakers (general practitioner, midwife, gynecologist, psychiatrist, or psychologist).For the current study, the current/past psychiatric diagnosis had to be present at the beginning of the pregnancy.Aside from diagnosis (yes/no), diagnoses were subsequently classified according to the DSM-5 (depressive disorders, anxiety disorders etc.). 22If women used psychopharmacological drugs, they were included in the "current/past psychiatric diagnosis" group if they were also diagnosed with a psychiatric illness.However, if their diagnosis was not clear, women were included in the control group.

| Unintended pregnancy
Unintended pregnancy was defined as a pregnancy that was not planned for, intended, and/or wanted at the time of conception as judged by the pregnant woman herself in retrospect.This information was written down in charts by a midwife or doctor.This variable was divided into four groups: "planned and wanted", "planned and unwanted", "unplanned and wanted", or "unplanned and unwanted".For the primary outcome we categorized all unplanned pregnancies as "unintended" pregnancies, both wanted and unwanted.All planned pregnancies were categorized as "intended" pregnancies (even if a pregnancy was planned but considered unwanted at the pregnancy intake, as these pregnancies were planned at conception).In case of an uncertain pregnancy intention after assessment by the researcher, a second researcher was asked to independently assess the pregnancy intention.If consensus could be reached, the pregnancy was included in the cohort.If uncertainty remained, the pregnancy was excluded from the study.

| Demographic characteristics
Maternal age at delivery was calculated through delivery date of the neonate and date of birth of the mother.Ethnical background was determined by country of birth of the mother's parents and grouped according to the Dutch obstetric registration system. 23Employment status of the mother was defined as having a paid job (yes/no) at time of conception.We gathered data on smoking and alcohol use during pregnancy, defined as any use of these substances from the moment of conception onward.Parity was defined as being primiparous (never delivered) or multiparous (delivered previously).We collected data on common somatic conditions and other reasons that require hospital pregnancy care: diabetes mellitus, hypertension, thyroid conditions, asthmatic conditions, epilepsy, history of cesarean section, bariatric surgery, obesity, gynecological conditions, and autoimmune diseases.Psychotropic medication use was defined as use of one or more psychotropic drugs during pregnancy.

| Maternal outcomes
Hypertensive disorders of pregnancy (HDP) and gestational diabetes (GDM) were defined in accordance with local guidelines.Deliveries were grouped by spontaneous vaginal delivery, assisted vaginal delivery and cesarean section.

| Neonatal outcomes
Prematurity was defined as gestational age (GA) at delivery <37 weeks.Birthweight percentile was calculated according to the Hoftiezer curve, which includes sex, GA at delivery and birthweight. 24Small for gestational age (SGA) was defined as a birthweight <10th percentile and being large for gestational age (LGA) as a birthweight >90th percentile.Apgar scores after 5 min were collected.Low Apgar score was defined as <7 after 5 min. 25mission to the neonatal ward was registered for those neonates in need of pediatric care.In standard care, neonates are admitted to the maternity ward with their mothers in case of absence of pediatric complications.We defined breastfeeding intention as all women who fully or partially breastfed their infants at the moment they left the hospital after delivery.

| Statistical methods
A power analysis indicated a minimum required sample size of 119 participants in both groups of the primary analyses (current/past psychiatric diagnoses and UPs) based on a UP rate of 43%-82% in women with psychiatric diagnosis. 11All analyses were conducted in R studio version 4.2.2. 26For descriptive analyses of normally distributed variables, means and standard deviations were reported.Continuous variables that were not normally distributed were described with median and interquartile range.Categorical variables were reported with numbers and percentages.Differences in demographics between women with intended pregnancies and UPs were described using T-tests or Chi-square tests.For the primary aim, the number of UPs was compared between women with and without current/past psychiatric diagnosis using logistic regression analysis.Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated.Subsequently, we performed logistic regression analyses with a dummy variable for having any current/past psychiatric diagnosis to assess the association of various psychiatric diagnosis with pregnancy intention.All models were adjusted for parity, employment status, psychotropic medication use, age, and ethnicity.
For the secondary aim, we compared maternal and neonatal outcomes between women with UPs versus non-UPs.Generalized linear models were fitted, depending on the nature of the outcome variable: linear regression models for continuous outcomes and logistic regression models for dichotomous outcomes.All models included a main effect for UP (yes/no) and were adjusted for parity, employment status, age, alcohol use, smoking, and psychotropic medication use.To evaluate a differential effect of psychiatric diagnosis on the association between UP and maternal and neonatal outcomes, each model was extended with a main effect current/past psychiatric diagnosis (yes/no), and an interaction term for UP with current/past psychiatric diagnosis.Main effects and interaction effects were presented as mean difference (MD) or OR with 95% CI.Wald p-values were estimated for all effect sizes, except the interaction term with psychiatric diagnosis, for which we used a Likelihood Ratio Test to assess the added value of the interaction term.P-values of <0.05 were considered statistically significant.

| Ethical considerations
The study protocol was approved by the Medical Research Ethics Committees United and the Advisory Committee Scientific Research at OLVG.The study was granted exempt from the Human Subjects Act, as declared by the Medical Research Ethics Committees United and exempt from informed consent procedures given the numbers of inclusion and mostly coded data collection.If patients indicated not to be willing to participate in any chart research (as noted in their medical chart), their data were not included in the study.

| General characteristics
In total, 1219 women with and 1093 women without current/past psychiatric diagnoses were included.Figure 1 presents the inclusion process.

| Association between unintended pregnancies and adverse maternal and neonatal outcomes
In Supplementary Table 1 we present demographic features of all four study groups in the secondary analyses (UP yes/no and current/past psychiatric diagnosis (yes/no).Table 3 displays the estimated associations of UPs on maternal and neonatal outcomes in two models: model 1 including main effects for UPs and current/past psychiatric diagnosis and model 2 including an interaction for both variables.There was no association between UPs and maternal outcomes (gestational diabetes mellitus, hypertensive disorders of pregnancy, assisted vaginal delivery and cesarean section).In addition, there was no interaction effect of UPs and current/past psychiatric diagnoses maternal outcomes.Neonatal outcomes were comparable between women with UPs and with intended pregnancies.Except from GA at delivery, current/past psychiatric diagnoses did not modify neonatal outcomes in women with UPs.After addition of the interaction term (model 2), women with UPs had a À 2.63 days lower mean GA at delivery than women in the reference group ( p = 0.096), women with current/ past psychiatric diagnoses had À1.67 days lower mean GA at delivery than women in the reference group (p = 0.204), but women with both UPs and current/past psychiatric diagnosis had a 2.21 days higher mean GA at delivery than women in the reference group (p-value interaction = 0.001).All other neonatal outcomes were comparable for UPs versus intended pregnancies, and there were no interaction effects.

| Post hoc sensitivity analyses
To comprehend the interaction effect of psychiatric diagnosis on the association between pregnancy intention and GA age at delivery, we performed a post hoc sensitivity analysis focusing on the two largest groups of DSM classified diagnoses; women with a current/past depressive disorder (N = 704) and women with a current/past anxiety disorder (n = 399).We compared GA at delivery in these two groups of women versus women without any current/past psychiatric diagnosis (n = 1093).Current/past depressive disorders modified the association between UPs and mean GA ( p-value interaction = 0.003).Current/past anxiety disorders did not modify the association between UPs and GA at delivery ( p-value interaction = 0.068).

| Key results
The results from our retrospective cohort study show that current/past psychiatric diagnoses are significantly associated with UPs after adjustment for relevant confounders (OR 1.56, CI 1.23-2.00,p < 0.001).In sub analyses, women with depressive (OR 1.67, CI 1.24-2.26,p = 0.001), personality (OR 2.64, CI 1.38-5.11,p = 0.004) and substance-related and addictive disorders (OR 4.29, CI 1.90-10.03,p = 0.001) showed a higher incidence of UPs compared to women without these disorders.In our population, maternal and neonatal outcomes were comparable between women with and without current/past psychiatric diagnoses in case of UPs, except from GA at delivery.Women with UPs and current/past psychiatric diagnoses had higher GA at delivery (mean difference 2.21 days, p = 0.001) compared to the reference group, whereas women with UPs without current/past psychiatric diagnoses and women with planned pregnancies and current/past psychiatric diagnoses showed a lower GA at delivery (resp.-2.63 and -1.67 days).This effect was mainly driven by the group of women with a depressive disorder.

| Strengths and limitations
We are the first to compare maternal and neonatal outcomes in a large group of women with and without UPs, also considering the effect of current/past psychiatric diagnosis.Our study is subject to several important limitations.Our dependability on uncontrolled naturalistic medical record information has serious implications on the availability of information on pregnancy intention and the reliability of our findings.Reporting bias probably played a role through various mechanisms.First, there could be an overreporting of UPs as UPs may draw more attention than intended pregnancies and subsequently are more likely to be reported in charts.We excluded files with missing data on pregnancy intention and/or psychiatric history.A comprehensive overview of incidences of UPs would require data on (elective abortions) amongst women with and without current/ past psychiatric diagnoses.Unfortunately, these data were not available as our study presents data from a retrospective cohort of women with ongoing pregnancies.Although missing data on abortions could lead to an overall underestimation of UPs in the current study, there are no implications on the difference between women with and without current/past psychiatric diagnoses.Additionally, it is important to consider the possible response bias in women when inquiring for pregnancy intention.Pregnancy intention is notoriously difficult to assess and should be assessed prospectively with a validated instrument, if possible. 27With our retrospective study design, we were unable to perform this assessment.Instead, we used registrations in medical charts by health care professionals.Moreover, pregnancy planning could be perceived as a sensitive topic that women do not wish to discuss during a pregnancy intake, resulting in underreporting.Another important source of uncertainty is the severity of psychiatric symptoms at conception.We included women with a current/past psychiatric diagnosis but could not adjust for severity, current presence, or duration of the diagnosis.In fact, as mental health symptoms are more common than psychiatric diagnoses amongst women in the fertile stage of life, UPs amongst all women with mental health symptoms might be even more prevalent.Our study was however limited in the possibility to assess mental health symptoms, and the presence of current/past psychiatric diagnoses was found to be a more reliable predictor variable in our retrospective design.Although the overall sample size of the study was adequate for the performed main analyses, caution must be applied for subgroup analyses of individual psychiatric disorders, specifically with small samples of patients with substance use disorders and personality disorders.As women with psychiatric disorders might underreport the presence thereof, our main predictor (current/past psychiatric diagnosis) may have created a significant bias as it is partially based on self-reported history of diagnosis.Nonetheless, our findings are in concordance with previous studies that also found that women with substance use disorders and personality disorders have an increased odd of UPs. 28,29Finally, we used the ethnicity variable according to the Dutch obstetric system as this is recorded in the patient file by all health professionals.However, since country of birth of the mother's parents does not always correspond to the ethnical background, this may be only partially correct and may have influenced our findings with regards of ethnicity.
Notwithstanding these limitations, our findings have value as they are based on a large dataset that was compiled independently by several researchers which decreases the possibility of researcher bias.Information bias was diminished by performing manual checks on chart data.The hospital provides with thousands of well documented patient charts including data from pregnant patients with psychiatric diagnoses, enabling us to collect detailed data on psychiatric history and pregnancy intention.Although prospective measurement of pregnancy intention is ideal, our work offers valuable insights in the incidence UPs amongst women with current/past psychiatric diagnoses.

| Psychiatric diagnoses and UPs
Our findings consolidate previous research that establishes psychiatric disorders as a predictor of UPs. 11There are several hypotheses that explain why.First, psychiatric symptoms can influence psychological mechanisms that are key for adequate use of contraceptive methods, such as planning, overview and impulse control regarding reproductive decision making. 30Moreover, reduced autonomy, lack of information, perceived stigma and worries about safety of contraceptive methods complicate pregnancy planning. 31Studies also showed that stress levels and depressive symptoms in young women with mental health problems 32 and longer disease duration in patients with severe mental illnesses 33 predict UPs.Alternatively, an overlap between psychiatric disorders and social and psychological predictors of ineffective contraceptive use could explain risk for UPs. 34Intimate partner violence, lack of social support 35,36 and low self-esteem 37 are related to both reproductive decision making and psychiatric disorders.In our study, presence of UPs was found in the overall group of women with a current/past psychiatric diagnosis (39.0%UPs, p < 0.001), and even higher in women with depressive (43.2% UPs, p = 0.001), substance use (66.7%UPs, p = 0.001) and personality disorders (49.1% UPs, p = 0.004).As depressive disorders are a common psychiatric disorder, especially for women of reproductive age, 38,39 the risk for UPs in this group is highly relevant.Previous studies show comparable findings amongst women with personality disorders and substance use disorders. 28,29Contrary to expectations, we did not report and increased odds of UPs in women with schizophrenia and eating disorders. 11,40Analyses in these relatively small samples of women could have been underpowered to find associations.This could explain relatively low UP rates in women with these severe mental illness compared to UP rates in literature. 41Also, our data did not show increased odds of UPs for women with anxiety disorders, which accords with previous data from a Japanese birth cohort. 42However, Tenkku et al. found a lower incidence of UPs in women with anxiety disorders, as opposed to a similar incidence between women with versus without anxiety disorders in our cohort.As qualitative studies on the psychological mechanisms behind UPs amongst women with anxiety disorders are currently lacking, it is challenging to hypothesize why women with anxiety disorders are less likely to become pregnant unintendedly.One of the possibilities is that individuals with anxiety disorders in general show harm avoidance and a drive to maintain control, which positively impact pregnancy planning behavior.

| UPs and maternal and neonatal outcomes
We found that on itself, UPs were not related to adverse maternal or neonatal outcomes.Our findings oppose most previous studies that showed robust associations between UPs and adverse outcomes in retrospective studies, prospective studies [43][44][45] and meta-analysis originating from various geographic regions. 16,17However, a recent large Swedish prospective cohort study found no association between UPs and severe pregnancy outcomes such as hypertensive disorders of pregnancy, gestational diabetes mellitus, assisted birth or cesarean section. 46Possibly, free health care settings with possibilities to choose abortion, like in Sweden and the Netherlands, can influence the consequences of UPs.Another reason is that more positive feelings towards the pregnancy could have mediated stress levels during pregnancy and thus positively impact birth outcomes after UPs such as premature birth and low birthweight. 36In our study, likewise, current/ past psychiatric diagnoses did not predict adverse maternal and neonatal outcomes.[49][50]

| Mediation by current/past psychiatric diagnoses
Our data showed a higher GA in women with UPs and current/past psychiatric diagnoses ( p = 0.001).This is probably a chance finding, as no other maternal or neonatal outcomes differed between the groups in our data, and no previous studies have shown more positive outcomes in women with both UPs and current/past psychiatric diagnoses.However, the lack of comparative literature illustrates the need for further evaluation of maternal and neonatal outcomes in these women.Additionally, there was no difference in the clinically relevant outcome preterm birth ( p = 0.618).A previous paper showed that adverse outcomes after UPs were similarly adverse in subsamples of women with prenatal depression. 17Although antenatal depression severity was not measured in our population, we also found that in our sensitivity analysis of women with current/past depression versus without, UPs were not related to adverse maternal or neonatal outcomes.Due to the limited available literature on this association, it is challenging to interpret our findings.

| Implications for future research
This research has thrown up several questions in need for further investigation.It would be interesting to study pregnancy planning amongst women with psychiatric diagnoses (other than severe mental illnesses) with qualitative research methods, to gain insight in the psychological mechanisms behind the wish for children and pregnancy planning.Both qualitative and quantitative studies could consider the role of intimate partner violence, social support, and autonomy in studying the impact of UPs in women with psychiatric diagnoses.To further evaluate neonatal outcomes, psychological adaptation to an UP during pregnancy and consequent presence or absence of psychological distress could be included in studies as a moderating variable.Last, our data provide insights in the short-term outcomes after UPs.Little is known about long-term outcomes after UPs in women with (severe) psychiatric disorders, such as mother child interaction and parenting stress. 51,52o conclude, our data confirm the hypothesis that current/past psychiatric diagnoses are associated with UPs.Women with depressive, personality and substance abuse disorders have increased odds for UPs compared to women without these diagnoses.These findings underscore the need for attention for pregnancy planning in psychiatric healthcare, especially as these psychiatric disorders have a high prevalence in society.We found no differences in maternal and neonatal outcomes after UPs versus intended pregnancies, neither did we find clinically relevant modification of psychiatric diagnosis on the relation between pregnancy intention and outcomes.Our data attenuate previous findings on adverse neonatal outcomes after UPs in addition to adverse pregnancy outcomes for women with psychiatric diagnoses.Although our study is limited by several factors, short-term maternal and neonatal outcomes for women with current/past psychiatric diagnoses in a hospital population might be better than expected.Increased efforts are needed to ensure that psychoeducation and conversations about pregnancy planning and UPs are available for women with current/past psychiatric diagnoses.
maternal health and pregnancy outcomes: a Swedish cohort study.PloS One.2023;18 (5) Demographic features of women with and without a current or past psychiatric diagnosis.
F I G U R E 1 Figure displays the inclusion process of the retrospective cohort.T A B L E 1

Table 1
Associations between current and/or past psychiatric diagnoses and pregnancy intention.Associations of UPs and current or past psychiatric diagnosis for all maternal and neonatal outcomes.
displays demographic features for both study groups.Women with current/past psychiatric diagnoses were on average 0.6 years younger (p = 0.015), more often unemployed (p < 0.001) and smoked more often during pregnancy (p < 0.001) than women without current/past psychiatric diagnoses.The incidence of UPs differed between women with (39.0%) versus without current/past psychiatric diagnosis (29.6%) in unadjusted (OR 1.56, 95% CI 1.24-1.97,p<0.001) and adjusted models (OR 1.56, 95% CI 1.23-T A B L E 2Note: Odds ratios for unintended pregnancies are presented for any psychiatric diagnosis or a type of psychiatric diagnosis (presented in the first column) versus the reference group.Some women were diagnosed with ≥1 disorder.In this table, only psychiatric diagnoses with n > 10 were included.Abbreviations: CI, confidence interval; UPs, unintended pregnancies.aAdjustedfor parity, employment status, psychotropic medication, and age.bAdjusted for parity, employment status, psychotropic medication use, age, and ethnicity.T A B L E 3Abbreviations: CI, confidence interval; LGA, large for gestational age; LRT, likelihood ratio test; MD, mean difference; OR, odds ratio; PD, current or past psychiatric diagnosis; SGA, small for gestational age; UP, unintended pregnancy.a Model included main effects of UP and PD, and was adjusted for psychoactive medication use, parity, employment status, age, alcohol use and smoking.b Model was adjusted for psychoactive medication use, parity, employment status, age, alcohol use, smoking, and included the interaction term psychiatric diagnosis :e0286052.47.Miller ES, Saade GR, Simhan HN, et al.Trajectories of antenatal depression and adverse pregnancy outcomes.Am J Obstet Gynecol.2021;226(1):108.e1-108.e9.48.Jablensky AV, Morgan V, Zubrick SR, Bower C, Yellachich LA.Pregnancy, delivery, and neonatal complications in a population cohort of women with schizophrenia and major affective disorders.Am J Psychiatry. 2005;162(1):79-91.49.Scrandis DA.Bipolar disorder in pregnancy: a review of pregnancy outcomes.J Midwifery Womens Health.2017;62(6):673-683.50.Jarde A, Morais M, Kingston D, et al.Neonatal outcomes in women with untreated antenatal depression compared with women without depression: a systematic review and meta-analysis.JAMA Psychiatry.2016;73(8):826-837.51.Shreffler KM, Spierling TN, Jespersen JE, Tiemeyer S. Pregnancy intendedness, maternal-fetal bonding, and postnatal maternalinfant bonding.Infant Ment Health J. 2021;42(3):362-373.52.McCrory C, McNally S. The effect of pregnancy intention on maternal prenatal behaviours and parent and child health: results of an irish cohort study.Paediatr Perinat Epidemiol.2013;27(2):208-215.