Maternal Schizophrenia and the Risk of a Childhood Chronic Condition

Abstract

Background and Hypothesis

Maternal schizophrenia heightens the risk for certain perinatal complications, yet it is not known to what degree future childhood chronic health conditions (Childhood-CC) might arise.

Study Design

This population-based cohort study using health administrative data from Ontario, Canada (1995–2018) compared 5066 children of mothers with schizophrenia to 25 324 children of mothers without schizophrenia, propensity-matched on birth-year, maternal age, parity, immigrant status, income, region of residence, and maternal medical and psychiatric conditions other than schizophrenia. Cox proportional hazard models generated hazard ratios (HR) and 95% confidence intervals (CI) for incident Childhood-CCs, and all-cause mortality, up to age 19 years.

Study Results

Six hundred and fifty-six children exposed to maternal schizophrenia developed a Childhood-CC (20.5/1000 person-years) vs. 2872 unexposed children (17.1/1000 person-years)—an HR of 1.18, 95% CI 1.08–1.28. Corresponding rates were 3.3 vs. 1.9/1000 person-years (1.77, 1.44–2.18) for mental health Childhood-CC, and 18.0 vs. 15.7/1000 person-years (1.13, 1.04–1.24) for non-mental health Childhood-CC. All-cause mortality rates were 1.2 vs. 0.8/1000 person-years (1.34, 0.96–1.89). Risk for children exposed to maternal schizophrenia was similar whether or not children were discharged to social service care. From age 1 year, risk was greater for children whose mothers were diagnosed with schizophrenia prior to pregnancy than for children whose mothers were diagnosed with schizophrenia postnatally.

Conclusions

A child exposed to maternal schizophrenia is at elevated risk of chronic health conditions including mental and physical subtypes. Future research should examine what explains the increased risk particularly for physical health conditions, and what preventive and treatment efforts are needed for these children.

Introduction

Schizophrenia is a chronic and severe mental illness that affects close to 75 million people worldwide.¹ For many affected individuals and their families, schizophrenia results in high rates of psychiatric and medical morbidity, with important impacts on quality of life.¹ Schizophrenia affects men and women in equal numbers,² yet women are more likely to engage in longer-term relationships, and to have children.^3–5 In the latter half of the 20th century, shifts from institutionalization to community-based psychiatric care, followed by increased use of newer, fertility-sparing antipsychotics have led to increased opportunities for pregnancy in this population.^6,7

Large epidemiological studies show pregnancies in women with schizophrenia to be at elevated risk for perinatal complications including maternal preeclampsia, preterm birth, fetal growth disturbances, and other neonatal morbidities that may indicate a risk for continued chronic health problems during childhood and beyond.^8–10 Prior research on the offspring of parents with schizophrenia has focused on the heritability of schizophrenia and other mental disorders.^11–13 There has been little systematic exploration of other chronic health conditions these children may experience past the neonatal stage, including specifically the risk for chronic medical conditions that could result from various in utero and/or early postnatal exposures.

Developmental origins influence the risk for many chronic diseases, including cardiac, metabolic, lung, and kidney diseases, mental illness, and even some cancers.^9,14,15 Increased sensitivity in early life development is thought to be related to extensive plasticity during that time, where a wide variety of stressors (e.g. psychosocial, infectious, nutritional, chemical) can alter developmental trajectories.¹⁶ Children born to women with schizophrenia may be at risk for a variety of such adverse early developmental exposures. Prenatal exposures might be social (e.g. poverty, poor housing, and unemployment),³ behavioral (e.g. suboptimal eating, smoking, or substance use)⁷ and/or medical (e.g. maternal diabetes, hypertension).¹⁷ As one example, poor maternal nutrition is linked to congenital malformations due to folate deficiency,¹⁸ and to a higher risk for asthma from vitamin D deficiency.¹⁹ A child exposed to poorly controlled diabetes in utero may be at higher risk for multiple medical problems, including malformations, and later metabolic disease.^20–22 Postnatal factors, such as compromised parenting capacity, maternal separation, poverty, second hand smoke, and poor diet, could further increase a child’s vulnerability to various chronic diseases.²³

This study used population-based health administrative data collected in a universal healthcare system to evaluate an array of serious chronic health conditions in children (Childhood-CC), including both mental health and non-mental health conditions from birth to age 19 years.

Methods

Study Design and Data Sources

This population-based cohort study in Ontario, Canada, used valid, comprehensive province-wide administrative health data housed at ICES, a nonprofit research organization that evaluates health care services and their effectiveness in Ontario (www.ices.on.ca) where patient-level records in the data sources are de-identified and linked to each other by a unique identifier number. These datasets permit accurate linkage of mothers characteristics to the outcomes of their offspring, and description of a mother’s psychiatric diagnoses and her child’s risk for Childhood-CC. All obstetrical deliveries in hospital are captured in ICES’ MOMBABY datafile, which has 98% successful linkage of maternal and newborn health records. The MOMBABY datafile also provides date of birth, child sex, and gestational age at birth (largely based on first-trimester dating ultrasonography²⁴), allowing accurate reverse estimation of conception date. Linked ICES databases were also used to measure exposure, covariates, and outcomes: Registered Persons Database (RPDB): date of birth (death), sex, and postal code, using Canadian census data to allow neighborhood-level income estimates; Immigration, Refugee, and Citizenship Canada (IRCC) Permanent Resident Database: immigration application records for people who initially applied to land in Ontario since 1985, allowing for classification of immigrant and refugee status; Canadian Institutes of Health Information Discharge Abstract Database (CIHI-DAD): hospital admissions, discharges, transfers using standard diagnosis (ICD-9/ICD-10-CA) and procedure codes (CCP/CCI); Ontario Mental Health Reporting System (OMHRS): mental health hospital data from all facilities with designated adult inpatient mental health beds; and Ontario Health Insurance Plan (OHIP) database: inpatient and outpatient, primary care, and specialist physician billing and diagnostic data submitted on all Ontario residents since 1992. These datasets are complete and reliable for billing, procedural, and primary diagnostic data.²⁵

Participants

All children born between April 1, 1995 and March 31, 2018 in Ontario were considered for inclusion, and were followed from their birth until March 31, 2019, age 19 years, date at migration out of the province, or death. Maternal schizophrenia was defined as one or more maternal hospitalizations or 3 or more outpatient visits for schizophrenia within 3 years of each other between database inception and the index birth (ICD-9 codes: 295, 298: ICD-10 codes: F20, 25, 29). This algorithm has a sensitivity of 90.1%, specificity of 68.0%, positive predictive value of 77.1%, and negative predictive value of 88.7% versus clinical charts in our datasets.²⁶

Children exposed to maternal schizophrenia were first matched to unexposed children on child sex to allow for stratified subanalyses. Then, they were matched in up to a 1:5 ratio based on a propensity-score comprising birth-year and a set of variables measured at delivery in the mother that were considered key potential confounders of the relation between maternal schizophrenia and the outcomes. These variables were as follows: sociodemographics (age, parity, immigrant status, neighborhood income quintile, region of residence), maternal medical conditions (pre-pregnancy diagnoses of asthma, diabetes, cancer, chronic obstructive pulmonary disease, congestive heart failure, HIV, myocardial infarction, hypertension, and rheumatoid arthritis, all ascertainable using validated disease registries at ICES and collapsed into 1 variable of maternal medical chronic condition), and other maternal mental health conditions (predelivery mood disorders and predelivery substance use disorders).^27–29 Pregnancy, delivery, and neonatal complications were not entered into the propensity-score as they were conceptualized as variables that might lie along the causal pathway between exposure and outcome.

Outcomes

The framework used to identify child chronic health conditions (Childhood-CCs) in the current study was the U.S Agency for Healthcare Research and Quality’s (AHRQ) Chronic Condition Indicator (CCI)—pediatric adaptation.³⁰ In the AHRQ’s CCI, a chronic health condition is defined as a condition lasting 12 or more months that (1) places limitations on self-care, independent living, and social interactions, and/or (2) results in need for ongoing intervention with medical products, services or special equipment.³¹ The CCI indicator is set up such that children with one or more CCI codes on hospital admission are identified as having a Childhood-CC. Childhood-CC identified using this coding framework therefore represent significant morbidity and medical complexity, so this forms an appropriate outcome framework from which to systematically evaluate meaningful long-term child morbidity past the neonatal stage associated with maternal schizophrenia.

The algorithm underlying the CCI indicator originated with work by Hwang et al.³² where selected International Classification of Disease (ICD-9) codes were classified as Childhood-CC based on a physician panel who reviewed each diagnosis code and determined whether it represented an acute or chronic condition. For the pediatric adaptation, an expert pediatric group revised the categories, including redesignating some diagnoses from chronic to nonchronic (e.g. cystitis) as more appropriate for children.³⁰ In the pediatric adaptation, Childhood-CCs include conditions such as asthma, congenital anomalies, diabetes, obesity, and mental illness. A CCI macro obtained from the authors of the pediatric adaptation was used to collapse ICD codes classified as Childhood-CCs into 531 clinically meaningful, mutually exclusive diagnosis categories, mapped to 21 overarching groups, largely organized by organ system (supplementary table S1).

The primary outcome for the current paper was any Childhood-CC, up to age 19 years, measured using the AHRQ-CCI pediatric adaptation ICD codes in the Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD). We also examined as secondary outcomes from within Childhood-CCs: (1) mental health Childhood-CC (category #8); (2) non-mental health Childhood-CC (all categories except #8); and (3) non-mental health Childhood-CC subcategories. Finally, we measured all-cause mortality, using Ontario’s Registered Persons Database (RPDB).

Statistical Analysis

Exposed and unexposed groups were described at the time of the index birth in terms of all covariates, before and after propensity-score matching. Groups were compared on covariates using standardized differences, a measurement of the magnitude of the difference between groups that is independent of sample size. The standardized difference is an advantage in studies using large datasets such as this where very small differences can be “statistically significant”; a standardized difference > 0.10 suggests a clinically meaningful difference.³³ We reported the risk for the outcomes in each exposure group overall, and we reported the top 3 most frequent Childhood-CC ICD codes by child age group (0–1, 2–5, 6–9, 10–14, >15 years). When identifying outcomes, all of a child’s Childhood-CCs were considered, not only the first diagnosis. So, for clarity, a child could have one or more mental health Childhood-CCs, one or more non-mental health Childhood-CCs, and could be represented in more than one subcategory of non-mental health Childhood-CC. Risks for each outcome were compared between the exposed and unexposed using a Cox proportional hazards model, generating hazard ratios, and 95% confidence intervals (CI) overall, and stratified by child sex. We used a robust sandwich matrix to account for multiple deliveries per mother.

In subsequent analyses, we added to the Cox models an adjustment for cesearean delivery, preterm birth, and birth weight small-for gestational age, one at a time, and then together—as each are potential causal pathway variables to some unknown degree. Within a subgroup of children born from 2002 onward, data were available to identify children removed from their mother’s care during the delivery hospitalization, and discharged into the care of social services. We created 4 exposure groups of children: maternal schizophrenia/discharged to social services, no maternal schizophrenia/discharged to social services, maternal schizophrenia/discharged home with their mother, and no maternal schizophrenia/discharged home with their mother (the referent). Children were followed from birth to March 31, 2019, date at migration out of province or death. For this analysis, due to the complexity of propensity-score matching multiple groups, Cox proportional hazards models were adjusted for maternal age, parity, birth-year, child sex, neighborhood income of mother at birth, region of residence, pre-delivery maternal chronic medical condition, mood disorder, and alcohol or substance use disorder.

In a final additional analysis designed to attempt to isolate the in utero period from early postnatal effects, children alive at the time of their first birthday were further classified into 4 groups: maternal schizophrenia diagnosed prior to conception, maternal schizophrenia diagnosed in utero, maternal schizophrenia diagnosed between birth and child age 365 days, and no maternal schizophrenia. Children in each of the schizophrenia exposure groups were matched by birth-year in a 1:5 ratio to children with no exposure to maternal schizophrenia, and followed for outcomes from age 366 days until March 31, 2019, age 19 years, date at migration out of the province or death. Cox proportional hazards models compared each exposure group to the referent, adjusted for maternal age, parity, and birth-year. Additional covariates were not entered into the model as the number of participants in the in utero and postnatally diagnosed groups was expected to be too small for adjustment with more than 3 variables.

The use of data was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require a Research Ethics Board review. The study was approved by the ICES privacy office (ICES logged study 2019 0990 113 000).

Results

From 1995 to 2018, there were 2 943 844 live-born children with valid maternal and infant health card numbers, including 5066 whose mothers were diagnosed with schizophrenia prior to their birth. All exposed children were matched, almost all in a 1:5 ratio, to unexposed children on the propensity-score; fewer than 10 children had a 1:4 match. Accordingly, there were 25 324 matched unexposed. Among mothers diagnosed with schizophrenia, the mean age at the time of the index birth was 30.7 years (SD 6.0) (table 1). A high proportion (36.4%) resided in the lowest neighborhood income quintile, and about 9.6% lived in a rural region. The prevalence of chronic medical conditions and mental health conditions other than schizophrenia among the mothers was high (e.g. 12.1% with asthma, 4.3% with diabetes and 3.7% with chronic hypertension, 67.9% with mood disorder, 8.5% with substance or alcohol use disorder). After propensity-score-matching, these potential confounders were well-balanced between groups (table 1). Matched groups were also similar on pathway variables (maternal pregnancy complications, labor, and delivery outcomes and neonatal complications), except for an elevated risk of neonatal intensive care unit admission in the exposed (supplementary table S2).

In the propensity-matched cohort, children were followed for a median of 5.2 years (interquartile range [IQR] 2.1–9.9 years), with a median follow-up of 4.9 years (IQR 1.8–9.7) in the exposed group and 5.3 years (IQR 2.1–10.0 years) in the matched group of children unexposed to maternal schizophrenia. Overall, there were 656 (13.0%) exposed children diagnosed with a Childhood-CC, compared to 2872 (11.2%) unexposed, corresponding to incidence rates of 20.5 vs. 17.2/1000 person-years, an HR of 1.18 (95% CI 1.08–1.28) (figure 1, table 2). For mental health Childhood-CC, the corresponding incidence rates were 3.3 vs. 1.9 per 1000 person-years (HR 1.77, 95% CI 1.44–2.18). For non-mental health Childhood-CC, they were 18.0 vs. 15.7 per 1000 person-years (HR 1.13, 95% CI 1.04–1.24). The HRs were above 1 in 9 of the 20 non-medical Childhood-CC subcategories, with statistically significant elevations in the cardiovascular, neurological, and nutritional subcategories (figure 2). All-cause mortality occurred at rates of 1.2 vs. 0.8 per 1000 person-years, respectively (HR 1.34, 95% CI 0.96–1.89). Non-mental health Childhood-CCs predominated in early childhood years and mental health Childhood-CCs predominated in later childhood for both groups (supplementary table S3). However, in the exposed, mental health Childhood-CCs comprised the top 3 most frequent conditions in both the 10–14 years and 15–19 years age groups, while they were the top 3 most frequent conditions for the unexposed only among those 15–19 years. In results stratified by sex, estimates were similar to those of main analysis, although the HR for mental health Childhood-CCs appeared to be greater among girls than among boys (table 2; supplementary table S3). Adjusting for preterm birth, small-for gestational age, and cesarean section individually and together, did not change results (supplementary table S4).

In the additional analysis of children for whom data were available on discharge to social service care at the time of the delivery hospitalization, all exposure groups had elevated aHRs relative to the referent group (table 3). Children with maternal schizophrenia—whether discharged to social services or discharged home with their mothers—had the highest aHRs for mental health Childhood-CCs. Children without maternal schizophrenia who were discharged to social services had the highest aHRs for non-mental health Childhood-CCs. Among children with maternal schizophrenia there were some differences in point estimates between who were and were not discharged home with their mothers for both mental health and non-mental health Childhood-CCs, but 95% CI overlapped considerably.

In the additional analysis of children from age 366 days onward, there were elevated aHRs for any Childhood-CC and for any non-mental health Childhood-CC among children whose mothers were diagnosed with schizophrenia diagnosed prior to pregnancy and among those whose mothers were diagnosed during pregnancy, but not for children whose mothers were diagnosed postnatally (supplementary table S5). For mental health Childhood-CC, there was increased risk in all groups that had maternal schizophrenia exposure vs. non-exposure, with the greatest magnitude of effect among children whose mothers were diagnosed prior to pregnancy. The number of deaths was too small for stratified analysis.

Discussion

In this population-based study, children born to women with schizophrenia were at higher risk than their unexposed peers for being diagnosed with a serious chronic health condition up to age 19 years. The novel finding of this study was an elevation in risk for non-mental health Childhood-CCs. This risk was small in relative terms, but has impact in absolute terms given the high prevalence of these non-mental health Childhood-CCs. The risk elevation was seen broadly across medical subcategories with an HR above 1.00 in 9 of the medical Childhood-CC subcategories. Statistically significant elevations were seen in 3 subcategories: (1) cardiac, which includes chronic conditions such as rheumatic heart disease, hypertension, cardiomyopathy, arryhythmias, ischemic disease; (2) neurological, which includes CCs such as cerebral palsy, epilepsy, spinal cord injury, migraine, tics, muscular dystrophy; and (3) nutritional, which includes problems such as malnutrition, and Vitamin D deficiency (rickets). Finally, there was also a non-significantly elevated risk for all-cause mortality, which is a potential cause for concern clinically. Taken together, these results support the idea that health risks for children born to women with schizophrenia extend beyond intergenerational risk for psychosis and other mental disorders. Monitoring, prevention and early identification are warranted to address a broad range of chronic conditions in this population.

Epidemiological studies have mostly evaluated the adult children of one or both parents with schizophrenia, and showed elevated risk for schizophrenia.^11–13,34 Large population-based studies report a significantly increased risk of other mental disorders associated with parental schizophrenia, and genetic links with ADHD have also been reported.³⁵ The results of these studies, and ours, align with those of genetic studies showing shared heritability across a wide range of mental and developmental disorders.^36,37 The higher HR for mental health Childhood-CCs observed in girls was novel, but this may reflect the age groups studied, as mood disorders in youth are more common in girls than boys.³⁷

This study is most novel in that it is among the first to explore the impact of maternal schizophrenia on medical conditions in children beyond the neonatal stage. These children were previously known to be at increased risk for preterm birth, small-for-gestational age birth weight, and other neonatal morbidities.⁸ One large register-based study in Denmark found an elevated risk for a broad range chronic and non-chronic medical conditions in children of parents with schizophrenia, bipolar disorder, and depressive disorders. Therein, children of mothers with schizophrenia were at elevated risk of a magnitude very similar to that seen in the current study (overall incidence 22.9/100 person-years, adjusted incidence rate ratio 1.15, 95% CI 1.12–1.17).³⁸ The same group found elevated risk for gastrointestinal and undefined somatic disorders in children of mothers with psychotic disorders up to age 6.³⁹ Similar to our study, a register-based study in Sweden did not find an increased risk for autoimmune disorders specifically among children exposed to maternal psychotic disorders.⁴⁰ However, overall, these data are consistent with our current report of a significantly increased risk for chronic medical conditions across childhood. Interestingly, we did not show an elevated risk for these conditions among children whose mothers were diagnosed with schizophrenia postnatally. This suggests that, from a developmental origins of health and disease perspective, in utero exposures may have been more important in the risk for non-mental health Childhood-CC than postnatal exposures. We also did not show differential risk for non-mental health Childhood-CC based on whether children were discharged home with their mothers. Yet, it is difficult to make inferences about postnatal exposures based on this analysis as we did not have further postnatal data about child placement, reunification, or apprehension.

Health administrative data have the advantage of large sample sizes, and so allow for accurate and efficient detection of health disparities in high-risk populations such as this one where it would otherwise take years to accrue sufficient sample size. The AHRQ-CCI framework is set up to capture outcomes in hospital-based data so the findings pertain to serious chronic disease. However, they may not be generalizable to conditions managed only in the ambulatory setting. Further, there were covariates we were not able to directly measure (or measure with reasonable confidence), including maternal (e.g. maternal smoking, nutrition, obesity, and maternal medication use), and child factors (e.g. the child’s physical and social environment, single vs. dual parent familial setting, exposure to adverse life events, nutrition, and exercise). We also did not conduct analyses to determine whether all children of the same mother were at increased risk. Finally, we did not attempt to elucidate mechanisms for increased risk of specific chronic conditions. It will be important to explore potential pathways to specific health outcomes in future research.

Over the last number of years, the need to attend to perinatal health of women with schizophrenia and their neonates has been increasingly recognized. The results of the current study around the risk for chronic conditions in older children add to the literature around the complexity of schizophrenia as an illness, and its potential implications for health into the next generation. Further understanding of the mechanisms underlying the risk for longer-term child chronic health conditions in this population, as well as whether there are specific preventive or treatment strategies that might be adapted for them, is warranted. Research suggests that discriminatory attitudes and behaviors of healthcare professionals toward individuals, and mothers with schizophrenia, may affect the quality of health care that these women and their children receive, thus impacting their health outcomes.^41–43 It will be essential to ensure that families where a mother has been diagnosed with schizophrenia are appropriately supported in managing child and adolescent chronic medical and psychiatric conditions.

Funding

This study was funded by an operating grant from the Canadian Institutes for Health Research (PJT156021). The funder/sponsor did not participate in the work.

Acknowledgments

We thank the Ontario Ministry of Health (MOH) and the Ontario Ministry of Long-Term Care (MLTC) for their data. Parts of this material are based on data and/or information compiled and provided by the Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions and statements expressed herein are those of the authors and not necessarily those of the CIHI. Parts or whole of this material are based on data and/or information compiled and provided by Immigration, Refugees and Citizenship Canada (IRCC) current to 2018. However, the analyses, conclusions, opinions and statements expressed in the material are those of the author(s), and not necessarily those of IRCC. This work was supported by the ICES, which is funded by an annual grant from the Ontario MOH and MLTC. The opinions, results and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario MOH or MLTC is intended or should be inferred.

Conflict of Interest

Dr. Vigod reports royalties from UpToDate Inc for authorship of materials on depression and pregnancy, outside of the submitted work. Dr. Morrison outside of the submitted work—Advisory board member for NovoNordisk. Dr. Holloway outside of the submitted work—Advisory board member for Taylored interventions. Dr. Tu receives a Research Scholar award from the Department of Family and Community Medicine at the University of Toronto. Dr. Saunders receives an honorarium from the BMJ Group, Archives of Diseases in childhood, outside of the submitted work. The other authors have no relevant conflicts of interest to disclose.

Author Contribution

Dr. Vigod conceptualized and designed the study and its analyses, supervised data analysis, interpreted the data, drafted the initial manuscript, and reviewed, and revised the manuscript. Drs. Ray, Cohen, Saunders, Barker, Berard, Dennis, Holloway, Morrison, Oberlander, Hanley, Tu, and Brown substantially contributed to the conceptualization and design of the study, its analyses, and interpretation, and critically reviewed, and revised the manuscript for important intellectual content. Mr. Wilton substantially contributed to the acquisition of data, conducted the analyses, aided in data interpretation, and critically reviewed, and revised the manuscript for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

References