Offspring outcomes when a parent experiences one or more major psychiatric disorder(s): a clinical review

Abstract

We sought evidence on quantifiable offspring outcomes, including problems, needs and strengths, associated with their experience of major parental psychiatric disorder(s), focusing on schizophrenia, affective illnesses and personality disorder(s). We were motivated by the absence of any systematic exploration of the needs of offspring of parents in secure hospitals. Seven electronic databases were searched to identify systematic reviews of studies quantifying offspring outcomes when a parent, or parent surrogate, has major psychiatric disorder(s). Our search (updated in February 2018) identified seven high-quality reviews, which incorporated 291 unique papers, published in 1974–2017. The weight of evidence is of increased risk of poor offspring outcomes, including psychiatric disorder and/or behavioural, emotional, cognitive or social difficulties. No review explored child strengths. Potential moderators and mediators examined included aspects of parental disorder (eg, severity), parent and child gender and age, parenting behaviours, and family functioning. This clinical review is the first review of systematic reviews to focus on quantifiable offspring problems, needs or strengths when a parent has major psychiatric disorder(s). It narratively synthesises findings, emphasising the increased risk of offspring problems, while highlighting limits to what is known, especially the extent to which any increased risk of childhood problems endures and the extent to which aspects of parental disorder moderate offspring outcomes. The absence of the reviews’ consideration of child strengths and protective factors limits opportunity to enhance offspring resilience.

Introduction

Interest in the outcomes among offspring who have experienced major parental psychiatric disorder is long-standing. However, offspring who have a parent in secure psychiatric care have received little examination in the research literature, although in England and Wales alone 5000–7000 people are likely to currently have a parent detained in a secure psychiatric hospital.¹ We sought to examine one facet of how these offspring may present: measures of offspring problems, needs or strengths when a parent has one or more of the psychiatric disorder(s) commonly suffered by secure hospital patients—schizophrenia, major affective illness and/or personality disorder(s) (eg, Coid²). A preliminary search of the PsycINFO database alone yielded over 30 000 references when entering terms for psychiatric disorder and parent–child relationship (see online supplementary text 1). Due to this volume of literature, we performed a review of systematic reviews in order to bring together high-quality research into one narrative synthesis and determine implications for future research. Due to the diversity of research within the reviews anddue to concern that not all findings were applicable to offspring of secure hospital patients, we kept our question broad, while retaining focus on the parental disorders most commonly suffered by secure hospital patients.

Objective

Our objective was to answer, in a clinical review, the question ‘what quantifiable offspring outcomes, including problems, needs and/or strengths, are associated with their experience of major parental psychiatric disorder(s)?’ We focused on schizophrenia, affective illnesses and personality disorder(s).

Methods

Seven web-based search engines were searched (OVID MEDLINE, EMBASE, PsycINFO, AMED (Allied and Complementary Medicine), Joanna Briggs Institute Evidence-Based Practice Database, Scopus and Web of Science) from inception to February 2018. Keywords were used as search terms to capture major psychiatric disorder(s), focusing on schizophrenia, severe affective illnesses and personality disorders, and to capture the parent–child relationship (online supplementary text 1). Search terms for offspring outcomes were not applied to avoid our expectations limiting the search. Major subject heading (MeSH) terms were used where available.

The search identified 3637 unique records. The first 100 titles and, where necessary, abstracts were rated independently by two researchers (SEA, NK) against the inclusion/exclusion criteria. The inclusion criteria required papers to be systematic reviews of studies of quantifiable outcome measures, concerning sons/daughters of a parent with a major psychiatric disorder. The exclusion criteria eliminated reviews with unreported/non-replicable methods, solely subdiagnostic threshold parental problems or reviews without quantitative child measures (for the full criteria, see online supplementary table 1). Inter-rater reliability was 99%. Discrepancies were resolved through discussion. The remaining title and abstract screening was completed by one researcher (SEA), leaving 86 papers. Full-text examination by two researchers (SEA, NK) left 20 reviews (figure 1).

Figure 1.

Search and selection process.

Quality assessment of these 20 reviews was undertaken by one researcher (SEA) according to the criteria across the five areas that Maniglio³ describes as being consistent with the Centre for Reviews and Dissemination guidelines⁴; these are that there should be sufficient information for replication, regarding evidence identification, study selection, data extraction, formal study quality assessment, and objective data synthesis and analysis. Reviews which were rated ‘good’ (meeting all criteria) and ‘fair’ (meeting all but one of the criteria) were included; seven reviews were retained (see online supplementary table 2). References from the included reviews were scrutinised for further reviews, but none met our criteria. Two researchers (SEA, NK) extracted data from all seven reviews, according to review questions and methods (tables 1 and 2). An a priori decision was made to synthesise the findings into a clinical review. The review was conducted according to a protocol (available on request), but this was not registered. Studies reported in this review, from the seven systematic reviews, are referenced in online supplementary text 2, with reference numbers preceded by ‘S’.

Table 1.

Offspring outcomes when a parent experiences major psychiatric disorder: research questions, review sources, design patterns of included studies and their country of origin

Authors and publication year	Main research question(s)/aim(s)	Databases, final search date and other sources	Studies (n), study design, sample origins	Countries of origin of the studies included in the review
Rasic et al, 20148	What is the risk of mental disorder(s) in offspring of parents with severe mental illness (schizophrenia, bipolar disorder or major depressive disorder)? Does risk extend beyond the disorder present in the parent?	MEDLINE/PubMed, EMBASE and PsycINFO, to end of 2012. Reference lists were also searched.	Total: 33 An unspecified mix of cross-sectional and longitudinal studies. 21 (64%) studies recruited unwell parents through clinics/hospital admissions±other methods.	No geographical restrictions: studies were from the USA (11, 33%), 3 from Canada, 2 each from UK, Australia, Romania and Israel, and 1 each from Turkey, India, Spain, Holland, Palau, Switzerland, Finland, Sweden, Brazil, Denmark and Japan.
Goodday et al, 201910	To test for associations between exposure to parental psychopathology in childhood and subsequent suicide-related behaviours in the offspring. To determine if any such associations differ by the type(s) and timing of parental psychopathology, gender of the parent and of the child, type of child psychiatric symptoms, and family functioning.	MEDLINE, CINAHL (Cumulative Index to Nursing and Allied Health Sciences), EMBASE, PsycINFO and Web of Science, to March 2017. Grey literature, including dissertations/theses, Child Welfare Information Gateway and Google Scholar. Reference lists were also searched.	Total: 54 19 cross-sectional, 24 longitudinal, 11 case–control studies. No further details given.	No geographical restrictions: included USA (n=26, 48%), Europe (n=18, 33%—Belgium, Denmark, Finland, Germany, The Netherlands, Norway, Poland, Sweden, UK), ‘Africa’, Brazil, Korea, New Zealand, Puerto Rico, South Korea and Taiwan.
Connell and Goodman, 20027	To examine the relative strength of the association between psychopathology in mothers vs fathers and internalising or externalising disorders in their children.	PsycINFO and ERIC (Education Resources Information Center), 1888–2001. Authors were contacted. Notices were posted on internet-based discussion lists for psychologists. Reference lists were also searched.	Total: 134 28 studies used parent clinical samples, 39 used child clinical samples and 67 used community samples.	Not stated.
Mendes et al, 201211	To test the association between maternal depression and depression in school-age children. To test for moderating effects of environmental and social covariates.	MEDLINE, LILACS (Latin American and Caribbean Health Sciences Literature), SciELO, Index Psi and PsycINFO, from 2004 to 2010. Portuguese, Spanish and English search terms were used*.	Total: 30 16 longitudinal (9 community, 4 clinical and 3 mixed samples). 14 cross-sectional (6 community, 6 clinical and 6 mixed samples).	24 (80%) from the USA; 1 each from Brazil, Chile, England, Germany, Hungary and Malaysia.
Lau et al, 20189	To examine the relative risk of a range of affective and non-affective psychopathologies among offspring of at least one parent with bipolar disorder compared with offspring with no parental major psychiatric history.	MEDLINE, PsycINFO, EMBASE and Scopus, to July 2015.	Total: 17 7 longitudinal, 7 cross-sectional, 3 cross-sectional from the longitudinal BIOS study. Samples recruited from inpatient (2), outpatient (2), both inpatient and outpatient (4), outpatient plus other methods (6), other methods (2) and ‘psychiatric clinic’ (1).	Australia, Canada, Romania, Spain, Switzerland and USA (details of numbers not available). (Papers from other countries may have been included but only these were listed.)
Eyden et al, 20165	What are the psychopathological and psychosocial outcomes for offspring of mothers with borderline personality pathology? What are the mechanisms (parenting or mother/offspring characteristics) underpinning associations between maternal borderline personality pathology and offspring outcomes? (Two other questions were not relevant to our review.)	PsycINFO, PubMed, EMBASE, Web of Science, Scopus and ASSIA (Applied Social Sciences Index and Abstracts), from 1980 to July 2015. Manual searching of journals: Journal of Personality Disorders and Personality Disorders: Theory, Research and Treatment, from January 2010 to July 2015. Reference lists were also searched.	Total: 33‡ 8 cross-sectional, 4 longitudinal, 21 case–control. Samples: 15 community, 10 clinical, 4 community and clinical, and 4 not specified beyond ‘high risk’ (overall 12 of 33 studies involved ‘high risk’ samples NOS).	Studies were from the USA (n=15, 45%), Australia (n=6, 18%), UK (n=4, 12%), Germany (n=3, 9%), Canada (n=3, 9%), France (n=1, 3%) and China (n=1, 3%). NB: does not equal 100% due to rounding.
Petfield et al, 20156	What difficulties are experienced by children of mothers with borderline personality disorder? (Another question was not relevant to our review.)	PsycINFO and MEDLINE, up to July 2014. Reference lists were searched. 8 authors were contacted.	Total: 17 All described as cross-sectional in Petfield et al§. Samples’ sources are not detailed in Petfield et al.	No geographical restrictions imposed; country of each included sample not stated.

*Search terms only in English unless otherwise specified.

‡Eyden et al ⁵ stated that 33 papers met their inclusion criteria, and they have given details of the 33 papers in their tables; however, in their narrative synthesis, a 34th paper appears (Macfie and Swan^S8), and a number of important findings are attributed to this paper, which is in their reference list. We have therefore also included the findings of the Macfie and Swan^S8 paper, as reported by Eyden et al, in our review.

§All 17 included studies are also in Eyden et al, but only 16 are described in Eyden et al’s tables; only 12 are considered by Petfield et al ⁶ to report child outcomes, although 13 appear to include child outcomes. All 17 studies are described as cross-sectional in Petfield et al, but only 1 of 17 (Herr et al ^S12) was described as cross-sectional in Eyden et al, which labelled all the remaining studies (except for one, Macfie and Swan,^S8 which does not appear in their table describing the studies) as case–control.

BIOS, Pittsburgh Bipolar Offspring Study; NOS, not otherwise specified.

Table 2.

Offspring outcomes when a parent experiences major psychiatric disorder: sample descriptions and offspring outcome descriptions

Authors and publication year	Parental gender	Parental diagnosis	Offspring gender	Offspring age	Offspring outcomes measured	Offspring outcome measures and data sources
Rasic et al, 20148	5 of 33 studies, mothers only. 28 of 33 studies, not specified.	Schizophrenia or similar psychoses. Bipolar affective disorder. Major depressive disorder.	1 study, daughters only. 32 studies, sons and daughters.	Target: mean age of 10 years or more. Actual: mean age of 10 years or more.	Diagnosed psychiatric illness or disorder, excluding: Minor depression. Bipolar spectrum disorder.	All 33 included data directly from offspring. All 33 studies used standardised interviews.
Goodday et al, 201910	10 of 54 studies, mothers only. 44 of 54 studies, not specified.	Any type of parental psychopathology according to the ICD/DSM criteria or parental SRB*.	1 study stratified for gender. 1 study, sons only. 45 studies, sons and daughters. 7 studies, gender unspecified.	Target: 0–25 years. Actual: mean age <25 years in all studies, but in 14 (26%) studies upper age limit >25 years.	SRB* included: Severity of ideation. Planned/unplanned attempts. Lethality of attempts.	38 studies included information directly from the offspring (±other sources). 12 studies only used offspring medical records/ICD codes or cause of death registries. 2 studies only included information about the child from parent/caregiver. 2 studies used the SSAGA, which is for adults so it is likely it was parent-rated, given the offspring age ranges (7–14 years in 1 study, 12–26 in the other). 50 studies used at least one structured measure. It was unclear how structured data collection was in 2 studies with data collected from the child and in 2 studies which relied solely on medical records.
Connell and Goodman, 20027	53 of 127† studies, mothers only. 19 of 127† studies, fathers only. 55 of 127† studies, mothers and fathers.	‘Parent mental health problems’: (1) alcohol/other substance abuse or dependence, (2) depression, (3) anxiety, (4) schizophrenia, (5) antisocial personality disorder, (6) bipolar disorder, (7) mixed, including symptoms or ratings of mental distress.	22 of 127† studies, sons only. 105 of 127† studies, both sons and daughters‡.	Target: age 2–18 years. Actual: mean age ranged from 1.70 to 17.50 years. Overall mean of 9.37 years (SD 3.97 years).	‘Childhood internalising behaviour problems’: Symptom ratings of: Depressed mood. Anxiety. Social withdrawal. Diagnoses of: Childhood depression. Anxiety disorders. ‘Childhood externalising behaviour problems’: Symptom ratings of: Aggression. Conduct problems. Delinquency. Diagnoses of: Conduct disorder. Oppositional defiant disorder. ADHD.	Child symptoms of internalising behaviour problems: 35 studies collected data from the child, 17 used ‘combined sources’§. Other studies collected data from parent(s) or teacher(s). Child diagnoses of internalising behaviour problems: 38 studies collected data from the child, 23 studies used ‘combined sources’§. Other studies collected data from parent(s) or ‘chart review(s)’. Child symptoms of externalising behaviour problems: 11 studies collected data from the child, 21 studies used ‘combined sources’§. Other studies collected data from parent(s) or teacher(s). Child diagnoses of externalising behaviour problems: 7 studies collected data from the child, 47 studies used ‘combined sources’§. Other studies collected data from parent(s) or ‘chart review(s)’. No information on how offspring outcomes were measured—structured/unstructured.
Mendes et al, 201211	Mothers only.	Depression.	3 of 30 studies, daughters only. 27 of 30 studies, sons and daughters.	Target: 6–12 years. Actual: 1–17 years.	Childhood depression.	21 studies collected data directly from the child using questionnaires/structured self-report assessment/semistructured interview, using a schedule, with the child (±separate interview with the parent). 1 CBCL (parent report) and SDQ (responder not specified). 3 various structured tools (responder unspecified). 5 various structured tools (parent only responder). All 30 studies used at least one structured tool.
Lau et al, 20189	4 of 17, mothers only. 4 of 17, mothers and fathers. 9 of 17, not stated.	Bipolar I. Bipolar II. Schizoaffective disorder.	1 study, sons only. 16 studies, sons and daughters.	Target: 2–30 years. Actual: 2–30 years.	Psychiatric disorder diagnoses. Other emotional and behavioural outcomes.	All 17 studies used a standardised diagnostic interview with the child±others. CBCL was used to rate offspring internalising/externalising behaviours.
Eyden et al, 20165	Mothers, 33 studies.	Emotionally unstable personality disorder.	1 study, daughters only. 1 study, sons only. 31 studies, gender not specified.	Target: unrestricted. Actual: 2 months to adulthood (defined here as 19+ years). No mean child age given.	Psychiatric disorder: Diagnoses. Symptoms. Psychosocial outcomes: Self-esteem. Interpersonal. Home environment.	15 studies collected data directly from child—interview/questionnaire or direct observation/experiment. 5 studies child data only from other sources. 2 studies unclear from whom data were collected. 11 studies no offspring outcomes specified. 32 studies used structured tools. 1 retrospective case record review, tool unspecified.
Petfield et al, 20156	Mothers, 17 studies.	Emotionally unstable personality disorder.	Never specified.	Target: 0–18 years. Actual: 0–18 years.	Child ‘difficulties’, including: Cognitive. Behavioural. Parent–child relationships. Mental health.	13 studies included child outcomes, either/both: Data collected from the child. Data collected through direct observation of the child, for example, video of mother–child interactions.   All 13 studies employed at least one structured measure.

*Suicide-related behaviours (SRBs), which here include suicidal thoughts, suicide plans, self-inflicted potentially or actually harmful behaviour, with or without clear intent to end one’s life, and completed suicide.

†Connell and Goodman reported having identified 134 separate samples; however, to ascertain this information on offspring gender, one had to manually count the studies listed in Connell and Goodman’s⁷, online supplementary appendix 4, in which the current authors could only identify 127 separate studies.

‡No studies in this review only included daughters. The table in the review refers to the child gender being either ‘mix’, ‘both’ or ‘boy’, but it was not clear what the difference between ‘mix’ and ‘both’ was, and we have considered these to both represent samples which included sons and daughters.

§It is unclear if one of these sources was the child.

ADHD, attention deficit hyperactivity disorder; CBCL, Child Behaviour Checklist (parent report); DSM, Diagnostic and Statistical Manual of Mental Disorders; ICD, International Classification of Mental and Behavioural Disorders; SDQ, Strengths and Difficulties Questionnaire; SSAGA, Semi-Structured Assessment for the Genetics of Alcoholism.

Presentation

Methodological comparability of included reviews

The review questions and methods and thus the nature of the included studies varied (tables 1 and 2). Of the 291 unique papers in the reviews, published between 1974 and 2017, only 17 were common to the two reviews of offspring of mothers with emotionally unstable personality disorder (EUPD)^{5 6} and only 5 were shared across the remaining reviews. Cross-sectional, longitudinal epidemiological studies, general population-based studies, clinical cohorts, case-controlled studies and convenience samples were all represented. Although all reviews included data collected directly from offspring, the extent to which this was true varied (table 2). Three reviews employed meta-analysis^7–9 and four narrative synthesis alone.5 6 10 11

Reviews varied in the extent to which they considered methodological differences between included papers. Although all reviews considered moderators, mediators or confounders in the narrative, only two^{7 8} of the three meta-analyses statistically tested for them; they found none with significant effect, except where findings lacked confidence due to small sample sizes and subgroup heterogeneity. Limitations reported narratively included a lack of clarity on recruitment methods or on diagnostic comorbidities,⁹ risk of bias, for example, studies measuring offspring exposure to parental psychopathology retrospectively,¹⁰ reliance on a single data source,¹¹ predominance of cross-sectional study design, and overdependence on the parent for information about the child.^{5 6}

We aimed to cover all forms of parent–offspring relationship, but reviews rarely specified this. Only Rasic et al ⁸ required a genetic parent–offspring relationship, and only Connell and Goodman⁷ explicitly included biological or social parent–offspring relationships. Neither parent nor offspring gender was consistently reported (table 2). Although all reviews included sons and daughters, three reviews confined parental gender to mothers.^{5 6 11} All reviews specified offspring age, which varied from no restrictions⁵ to a narrow focus (6–12 years¹¹).

Outcomes

All reviews focused on difficulties experienced by offspring, including psychiatric disorder,5 6 8 9 11 subdiagnostic threshold emotional difficulties and self-directed aggression,5–7 9–11 behavioural difficulties,^{5–7 9} cognitive dysfunctions,^{5 6 11} and social difficulties,^{5 6} each of which is detailed in the following sections. The only advantage, thus reported indirectly by the reviews, was an absence of disorder.

Offspring diagnosis of psychiatric disorder

Only one⁸ of the five reviews examining offspring diagnosis focused on offspring of parents with schizophrenia. These offspring were seven times more likely to develop schizophrenia than offspring of healthy parents, but were not at increased risk of any of the other six disorders examined (depression, bipolar disorder, anxiety disorders, attention deficit hyperactivity disorder (ADHD), behavioural disorders and substance misuse disorders) (table 3). Two reviews involving offspring of parents with bipolar disorder^{8 9} found an elevated risk of all disorders tested, except schizophrenia, and one⁸ found this also for parental depression, apart from with respect to offspring bipolar disorder (table 3). One review¹¹ reported that most of their included studies (17 of 22) found an increased risk of offspring depression at ages 6–12 years when mothers were depressed. The two reviews of offspring of mothers with EUPD^{5 6} made little reference to offspring diagnosis, but both included one study^S1 of 140 children, aged 6–14 years, which found that they had increased susceptibility to depression compared with offspring of mothers with depression. One other paper in these reviews^S2 identified an increased risk of offspring ADHD at a mean age of 11 years.

Table 3.

Absolute rates and relative risks of psychiatric disorder(s) in offspring of a parent diagnosed with schizophrenia, bipolar affective disorder or major depressive disorder, compared with offspring of psychiatrically healthy parents, according to the two reviews which tested these (Rasic et al ⁸, Lau et al)⁹

Outcome of diagnosed psychiatric disorder in offspring	Parental disorder	Offspring (n)*	Absolute rate†	95% CI	Relative risk	95% CI	P value	Review authors
Schizophrenia	Bipolar	581	0.04	0.02 to 0.10	2.76	0.67 to 11.27	0.158	R‡
		NA	NA	NA	NA	NA	NA	L§
Bipolar affective disorder	Bipolar	1415	0.06	0.04 to 0.09	4.06	1.91 to 8.62	0.000	R
		1290	NA	NA	8.97	3.85 to 20.91	<0.0001	L
Depression	Bipolar	1466	0.14	0.11 to 0.18	2.07	1.27 to 3.35	0.003	R
		1494	NA	NA	2.43	1.64 to 3.60	<0.0001	L
Anxiety	Bipolar	1288	0.27	0.22 to 0.33	1.92	1.56 to 2.36	0.000	R
		1572	NA	NA	2.14	1.63 to 2.81	<0.0001	L
Behavioural disorder¶	Bipolar	1027	0.14	0.10 to 0.19	1.84	1.24 to 2.72	0.002	R
		1410	NA	NA	2.48	1.64 to 3.74	<0.0001	L
ADHD	Bipolar	1234	0.14	0.09 to 0.21	1.62	1.23 to 2.13	0.001	R
		1181	NA	NA	2.59	1.87 to 3.60	<0.0001	L
Substance-related disorder	Bipolar	1137	0.15	0.09 to 0.24	1.45	1.07 to 1.97	0.016	R
		1033	NA	NA	1.70	1.17 to 2.45	<0.05	L
Any disorder	Bipolar	1285	0.60	0.53 to 0.67	1.66	1.50 to 1.83	0.000	R
		1214	NA	NA	1.98	1.70 to 2.32	<0.0001	L
Schizophrenia	Schizophrenia	816	0.12	0.08 to 0.18	7.54	4.02 to 14.13	0.000	R
	Depression	266	0.04	0.01 to 0.11	1.52	0.63 to 3.64	0.349	R
Bipolar affective disorder	Schizophrenia	481	0.03	0.02 to 0.05	1.84	0.73 to 4.66	0.197	R
	Depression	553	0.03	0.01 to 0.13	5.03	0.90 to 28.18	0.066	R
Depression	Schizophrenia	740	0.15	0.09 to 0.25	1.31	0.78 to 2.20	0.312	R
	Depression	1339	0.26	0.15 to 0.41	2.38	1.94 to 2.91	0.000	R
Anxiety	Schizophrenia	511	0.15	0.07 to 0.29	0.97	0.68 to 1.39	0.87	R
	Depression	1298	0.29	0.19 to 0.43	1.78	1.41 to 2.25	0.000	R
Behavioural disorder¶	Schizophrenia	69	0.29	0.20 to 0.41	1.90	0.81 to 4.49	0.142	R
	Depression	1380	0.16	0.08 to 0.30	1.80	1.56 to 2.09	0.000	R
ADHD	Schizophrenia	69	0.10	0.05 to 0.20	1.76	0.34 to 9.03	0.500	R
	Depression	1053	0.11	0.08 to 0.15	2.40	1.66 to 3.47	0.000	R
Substance-related disorder	Schizophrenia	528	0.20	0.11 to 0.34	1.72	0.88 to 3.37	0.112	R
	Depression	884	0.11	0.06 to 0.20	1.72	1.30 to 2.27	0.000	R
Any disorder	Schizophrenia	729	0.47	0.34 to 0.60	1.45	1.17 to 1.79	0.001	R
	Depression	1273	0.57	0.46 to 0.67	1.64	1.40 to 1.92	0.000	R

NA (not available) means data are not presented in the review(s).

*‘n’ is the number of offspring of a parent with the specified psychiatric disorder on whom the particular calculation is based.

†Absolute rate, so 0.12 means 12% of all offspring of a parent with the specified psychiatric illness.

‡R, data from Rasic et al.⁸

§L, data from Lau et al.⁹

¶Behavioural disorder in Rasic et al includes oppositional defiant disorder (ODD), conduct disorder (CD) and antisocial personality disorder, and in Lau et al it includes ODD and CD.

ADHD, attention deficit hyperactivity disorder.

Offspring emotional difficulties and self-directed aggression

All reviews except one⁸ explored offspring ‘emotional difficulties’, which included subdiagnostic threshold depression, anxiety, and emotional instability, unspecified emotional problems and/or suicide-related behaviours (SRBs). Five reviews5–7 9 11 concluded that offspring are at a higher risk of these when a parent has a major psychiatric disorder. One review⁵ specified a significant association between offspring problems and maternal ‘symptoms’ of EUPD (four studies^{S3, S4, S5, S6}). The review cited one study^S2 as describing higher rates of personality disorder, but as offspring had a mean age of 12 years these problems are arguably personality traits. Other studies in their review focused on, and found, higher rates of offspring ‘emotional dysregulation’,^{S7, S8, S9, S10} insecure attachment,^{S1, S11, S12, S13} unstable self-image^{S8, S14} and suicidal ideation.^S15 The overlapping Petfield et al ⁶ review similarly reported higher offspring insecure attachment^{S1, S8, S12, S13} and suicidal ideation.^S15 Of the 34 studies from these two reviews, only one was clearly reported to have found no higher risk of ‘child emotional problems’.^S16 Emotional dysregulation in the offspring also emerged in two studies^{S17, S18} in the Mendes et al ¹¹ review of children of mothers with depression. Connell and Goodman,⁷ in their meta-analysis, found higher prevalence of combined anxiety and depression diagnoses and subdiagnostic symptoms. Lau et al ⁹ found significant differences in the Child Behaviour Checklist scores for internalising disorders between offspring of a parent with bipolar/schizoaffective disorder (n=145) and offspring of psychiatrically well parents (n=148), drawing on three studies.^{S19, S20, S21} Relevant review findings with only one supporting study are detailed in online supplementary table 3a,b.

One review¹⁰ focused on offspring suicide and SRBs. The likelihood of offspring suicide was raised if a parent had schizophrenia^S22 or had died by suicide,^{S23, S24, S25} and also in one study^S23 which examined a wide range of parental psychopathology. Offspring SRBs were consistently associated with parental schizophrenia, maternal suicide and maternal suicide-related actions, and inconsistently with parental affective disorder, parental personality disorder, maternal suicide-related thoughts and paternal suicide (for a full breakdown of offspring types of SRB, see online supplementary table 3a,b).

Offspring behavioural difficulties

Four reviews^{5–7 9} included reference to offspring subdiagnostic threshold ADHD, oppositional defiant disorder and conduct disorder, antisocial personality traits, and unspecified behavioural problems. Connell and Goodman’s meta-analysis⁷ found a small offspring population mean elevation of risk of ‘externalising’ subdiagnostic threshold states and diagnoses with a wide range of parental psychiatric disorder(s). Lau et al ⁹ found significantly higher Child Behaviour Checklist externalising disorder scores among offspring of a parent with bipolar/schizoaffective disorder (n=145) compared with offspring of psychiatrically well parents (n=148), drawing on three studies.^{S19, S20, S21} Both reviews focusing on offspring of mothers with EUPD^{5 6} concluded that offspring were more likely to have behavioural difficulties.^{S2, S15, S16, S26}

Offspring cognitive dysfunctions

Three reviews^{5 6 11} included offspring cognitive outcomes, encompassing thoughts, perceptions and social cognitions, and all reported higher rates of difficulties for offspring of a parent with psychiatric disorder.

Mendes et al ¹¹ reported that offspring of mothers with depression have difficulty recognising ‘positive bases’, focus more on negative stimuli and show decreased flexibility in changing their focus of attention.^{S17, S18} Both reviews of offspring of emotionally unstable mothers^{5 6} reported poorer offspring self-representations,^S8 and self-perception of friendship-forming ability and of their own social acceptability.^S12 In preschool children, both reviews^{5 6} reported less developed offspring theory of mind and poorer offspring recognition of emotional expressions.^S27

Offspring social outcomes

Only the two reviews of mothers with EUPD^{5 6} explored offspring experience of home and/or school and/or work, and between them they identified only two relevant studies.^{S2, S28} Both found higher problem rates at home, school and social life compared with children of mothers with other personality disorders, although changes in household composition, witnessing of suicide attempts or parental abusive behaviour may have been the key problems here.

Offspring outcomes not reported in the reviews

By inference, the reviews collectively demonstrate that around half of offspring whose parent has a major psychiatric disorder remain psychiatrically well at the time of assessment. However, positive outcomes, such as possible higher offspring self-esteem from taking caring roles, were not encompassed in the reviews’ questions, although review discussions sometimes indicated offspring protective factors, such as secure attachment (Gratz et al ^S11 in Eyden et al ⁵).

Moderators and mediators

Five reviews5–7 10 11 explicitly considered the extent to which variables partly explain or ‘mediate’ any relationship between parental disorder and child outcomes, or the extent to which they ‘moderate’ any effect size found. Three reviews^{5 6 10} quantified the included studies’ examination of this and suggest that less than a fifth did so (Eyden et al ⁵ (6 of 33, 18%); Goodday et al ¹⁰ (9 of 54, 17%); Petfield et al ⁶ (2 of 17, 12%)). Mendes et al ¹¹ discussed moderators and mediators without labelling or quantifying the studies which examined them, while Connell and Goodman⁷ tested for potential moderators and mediators but found sample sizes to be generally too small and heterogeneous for confidence in findings.

Additional parental factors

Only one review considered severity, complexity and/or chronicity of parental disorder.¹¹ According to two included papers,^{S29, S30} child depression was more severe when mothers experienced depression with comorbid disorder(s); another study suggested inter-relationships between maternal depression, comorbidity and maladaptive mother–child interactions.^S31 This review¹¹ further reported that three (all from the longitudinal STAR*D (The Sequenced Treatment Alternatives to Relieve Depression) study) of their five papers assessing change in maternal depression found that as mother improved so did the child.^{S32, S33, S34}

The impact of chronicity of exposure was considered—by definition in the two personality disorder reviews,^{5 6} but also in Mendes et al.¹¹ The latter concluded that, overall, there was no relationship between offspring depression and chronicity of exposure to maternal depression, although they cited one longitudinal study as reporting that child depressive symptoms worsened the longer the parent had been depressed.^S35

Two reviews^{7 10} examined parental gender effects. Maternal psychiatric disorder was more strongly associated with negative offspring outcomes than paternal disorder. Goodday et al ¹⁰ found this in all three studies which tested for it.^{S36, S37, S38} Connell and Goodman⁷ found that child ‘internalising behaviours’ were more strongly associated with maternal than paternal depression (g=0.02, p<0.05), and with maternal than paternal psychiatric disorder overall (g=0.04, p<0.05), although child ‘externalising behaviours’ were not differentially associated with parental gender. They also found evidence of interaction between child age and parental gender. Younger children were more likely to have emotional and/or behavioural difficulties when the mother was affected rather than the father, but older children when the father was affected, this applying especially to paternal depression.

Child age at the time of parental illness was specifically addressed by Goodday et al,¹⁰ who noted that just a third (17 of 54) of their studies considered timing at all, most of them retrospectively. Among the seven prospective studies, two examined child age at exposure, one^S25 of which found a linear trend for timing of parental psychiatric inpatient admission and subsequent offspring suicide, with a slightly stronger association when the child was younger (under 3 years OR=2.5, 95% CI 2.0 to 3.0; 3–10 years OR=1.9, 95% CI 1.5 to 2.4; or >10 years OR=2.1, 95% CI 1.8 to 2.6). The control group was randomly selected from the general population; however, these analyses did not control for type, chronicity or severity of parental illness.

Offspring characteristics

Two^{7 11} of the three reviews^{7 10 11} considering offspring gender as a moderator indicated that girls may be more vulnerable than boys, although there was also an interaction between offspring gender and age, and between offspring gender and parental gender; the third review¹⁰ was equivocal. Mendes et al ¹¹ cited six studies^{S18, S39, S40, S41, S42, S43} as finding that older girls were more vulnerable to depression when their mother was depressed. Connell and Goodman⁷ found that girls seemed to have more externalising problems when fathers experienced psychiatric disorder. Goodday et al ¹⁰ reported that of the three studies testing for moderation by offspring gender, one reported a significant interaction between offspring gender and maternal depression in offspring suicide-related thoughts (OR 5.99, p<0.01; girls OR 5.18, p<0.01; boys OR 0.78, p=0.78),^S44 while two reported non-significant findings.^{S45, S46}

Possible mediating effects of child cognitive, temperamental and interpersonal vulnerabilities were considered in three reviews,^{5 6 11} yielding few relevant studies. Both maternal EUPD reviews^{5 6} identified a single paper (n=140 children; n=102 parents) which found that offspring cognitive problems (rumination, negative attributions, dysfunctional attitudes, self-criticism, excessive reassurance seeking and insecure attachment) appeared to mediate between maternal EUPD and child depression.^S1 Mendes et al ¹¹ found one study incorporating children’s ‘verbal competence’ (n=164 dyads)^S47 and one children’s temperament (n=337 dyads),^S41 which provided modest evidence that these were associated with more depressive symptoms in the children of mothers with depression.

The complexity of relationships was highlighted by two reviews,^{10 11} which suggested that if offspring developed psychiatric disorder, this could precipitate more maladaptive mother–child interactions, which in turn may worsen offspring outcomes. The first of these reviews¹⁰ evidenced such a pathway towards SRBs in adolescents and young adults from four supporting studies, two of which focused on our parental diagnoses of interest.^{S48, S49} Although the second review¹¹ included two relevant longitudinal studies,^{S50, S51} pathway analyses were not done.

Parenting and parent support

Three reviews^{5 6 11} examined parenting style, finding that it is an important mediator. Mendes et al ¹¹ found that five studies showed that ‘negative parenting’ mediates between maternal depression and the severity and duration of offspring depression.^{S31, S43, S52, S53, S54} Eyden et al ⁵ used the term ‘maladaptive parenting’ and found two studies evidencing its mediation of the association between maternal personality disorder and negative child outcomes.^{S7, S55} Petfield et al ⁶ also concluded that the ‘few’ (unspecified) studies examining this all found evidence of a mediating role for parenting style.

Two reviews^{10 11} considered family dynamics. One reported that ‘family functioning’ partially mediated the associations between maternal depression and offspring SRBs.^{S56, S57} The other found that marital conflict could be related directly to maternal depression and indirectly to child depression,^{S41, S54, S58} but could also follow a pathway from maternal depression through family dynamics, affecting child emotional security and raising the risk of child depressive symptoms.^{S17, S18}

Three reviews^{7 10 11} considered the other parent’s presence and health, with two finding offspring problems associated with single parenthood^{10 11} and one finding no enhanced risks.⁷ Mendes et al ¹¹ observed, from three studies, that being a single mother with a ‘low level of education’ was associated with both maternal depression and child emotional problems,^{S33, S35, S53} and from one study that the father’s presence could reduce the risk of child depression.^S59 Goodday et al ¹⁰ cited one study showing that the risk of SRBs among offspring of a psychiatrically unwell parent was further elevated when that parent was single.^S25 Connell and Goodman,⁷ however, found no such enhanced risk and speculated that a higher risk of intrafamilial conflict in intact families might balance the risk of higher exposure to the troubled parent in single-parent families.

Genetic factors

Six reviews^{5 7–11} acknowledged a genetic component to offspring susceptibility to adverse outcomes, although none offered evidence or discussed heritability or genetic mechanisms at length.

Conclusions and clinical implications

This is the first review of systematic reviews examining quantified offspring outcomes when a parent has major mental disorder. Seven high-quality reviews including almost 300 studies were identified. Three general conclusions follow. First, the weight of evidence appears to indicate an increased risk of offspring psychiatric disorder or behavioural, emotional, cognitive or social difficulties. Second, such conclusions, with the exception of an increased risk of offspring psychiatric disorder, tend to be based on a small number of original studies. Third, potential child strengths, such as enhanced coping skills, were not explored. The materials predominantly involved dependent children, suggesting a paucity of quantitative studies involving adult offspring, although qualitative research makes no such omission.¹²

The reviews provided useful pointers for future research, highlighting individual studies which suggest the relevance of the severity of parental illness and of comorbidities. The nearest any review came to considering parental violence (other than self-harm), which may accompany these serious psychiatric disorders, was in Petfield et al,⁶ which reported that children of mothers with EUPD were more likely to witness violence than children of parents with other personality disorder(s), although it is unclear who perpetrated this violence.

None of the reviews sought studies evaluating potential positive correlates of experiencing parental psychiatric disorder, although three referred to this in their discussion.^{5 9 11} There are studies of children with psychiatrically unwell parents in which the children desire recognition and acknowledgement of any positives their situation presents (eg, Cooklin¹³), and those demonstrating offspring protective factors (eg, Collishaw et al ¹⁴), but this requires further research.

Our review identifies the potential value of intervention. One review¹¹ found evidence of improvements in child psychiatric disorder when the parent’s disorder remitted, as demonstrated in their included papers from STAR*D,^{S32, S33, S34} which suggests that effective parental treatment may itself have offspring benefits. Indications that parenting may mediate between parental and child disorders highlight an opportunity for intervention, and parenting programmes are effective.^{15 16} Family processes also mediate the relationship between parental psychiatric disorder and child psychiatric disorder¹⁷ or other variations in child functioning.¹⁸

Our review has several limitations. Only 22 of the 291 original studies featured in more than one of the included reviews, which presents difficulties in making generalisable findings; findings from studies focusing on offspring of a parent with first-episode depression will not necessarily apply to offspring of a parent with chronic psychosis. Generalisable findings were also challenged by terms such as ‘externalising’ being variously defined and alternative terms being used without definition, for example ‘disruptive behaviour disorder’ in Eyden et al.⁵ However, the large number of unique original studies extends the range of findings. We included only systematic reviews which survived quality appraisal, but excluding materials, even justifiably, risks loss of pertinent findings. Furthermore, the time lag between original papers being published and being included in a review means potential loss of recent relevant evidence.

In conclusion, this review of reviews consolidates evidence indicating that offspring of a parent with psychiatric disorder(s) are at increased risk of diagnosed psychiatric disorder and/or subthreshold symptoms. It highlights that even when explicitly sought by reviews, studies rarely examined the potential moderating effects of parental disorder timing, severity, chronicity and/or comorbidity,^{10 11} nor offspring experience in adulthood.⁵ It also highlights that research questions concerning offspring social outcomes emerge readily when a parent has EUPD but less readily when a parent has mental illness. Associated features, such as violence, and mediating effects, such as parenting support and offspring protective factors, were minimally considered.

Recommendations encompass exploration of why some offspring experience greater adversity, including examination of the moderating roles of aspects of parental disorder and examination of offspring protective factors and resilience. Examination of offspring outcomes in adulthood and interest in a broader range of outcomes irrespective of parental disorder would enable a more complete understanding of the impact of parental psychiatric disorder on offspring.