Parental Mental Illness in Early Childhood: From Familial Liability to Everyday Caregiving, Adversity, and Prevention

Quick Takeaway

A parent’s mental illness can affect a child, but not in one simple or inevitable way. Children may inherit some vulnerability to mental health problems, yet genetics are only part of the story. What also matters is what happens in everyday life: whether a parent can provide comfort, supervision, routines, emotional availability, and dependable care during difficult periods.

This article argues that parental diagnosis alone should never be treated as evidence of harmful parenting. Two parents with the same condition may function very differently. The important question is what, if anything, changes in the child’s actual care when the parent is unwell.

Research suggests that genetics, parental symptoms, family stress, financial or social difficulties, caregiving patterns, and the presence of other supportive adults can all contribute to a child’s outcome. These factors often operate together rather than competing with one another.

The practical message is straightforward: treat the parent’s mental illness, identify specific caregiving tasks that become difficult, make sure another dependable adult can help when needed, address practical problems such as housing or childcare, and pay attention to the child’s own emotional and developmental needs.

Some approaches, including programs designed to improve parenting sensitivity, parent–child relationships, or a parent’s ability to understand a child’s feelings and behavior, are promising. However, the evidence does not yet show that any single psychological mechanism explains most of the risk or that one particular program prevents childhood trauma.

The goal, therefore, should not be to blame parents or assume that children of parents with mental illness are destined to develop problems. The goal is to identify where a family actually needs support and provide it early. Genetic risk may not be changeable, but many parts of a child’s environment, caregiving system, and access to support are.

Abstract

Parental mental illness is associated with elevated risk for offspring psychopathology, yet the route from a parental diagnosis to a young child’s developmental experience is neither singular nor deterministic. This critical conceptual review integrates population-register studies, genetically informed research, evidence on parenting and parental mentalization, qualitative studies, intervention trials, and professional guidance to clarify what is known, what remains uncertain, and what is actionable in early childhood. The central model separates four levels that are often conflated: familial liability, parental symptoms and functioning, caregiving tasks and child exposures, and developmental outcomes. Large cohorts document concentrated child-protection and out-of-home-care involvement among some families affected by parental mental illness, but administrative contact is not equivalent to maltreatment. Genetically informed studies show that familial confounding matters while also remaining compatible with environmentally mediated harm from maltreatment. Newer evidence further cautions against assuming that indirect parental genetic effects are large for depression and anxiety. Parenting sensitivity and parental mentalization are plausible relational mechanisms, but meta-analytic evidence suggests that their mediating effects are generally modest and that mentalization measures are not validated tools for predicting maltreatment. Intervention evidence is similarly mixed: whole-family and preventive programs show small and heterogeneous benefits, early-childhood multicomponent interventions can improve child developmental outcomes, and recent trials underscore the need for adequately powered comparisons and implementation measurement. The practical implication is a function-first family response: treat parental illness; identify which caregiving tasks are disrupted; maintain continuity of safe, dependable care during symptomatic periods; address material constraints; and assess the child directly. The field now needs longitudinal, genetically informed, and factorial intervention studies that test specific mechanisms rather than inferring them from diagnosis, association, or improvement in a proposed mediator.

Keywords: parental mental illness; early childhood adversity; intergenerational transmission; genetic liability; caregiving; parental reflective functioning; prevention

1. Introduction

For a young child, parental mental illness is not encountered as a diagnostic category. It is encountered through ordinary developmental transactions: whether distress is noticed, whether comfort is available, whether supervision is reliable, whether routines remain predictable, and whether another trusted adult can step in when the parent is acutely unwell. These moments are where an epidemiologic association becomes a developmental question. The relevant problem is therefore not simply whether a parent has a disorder, but how symptoms, treatment, resources, relationships, and the child’s own characteristics combine to shape the care the child actually receives.

The offspring of parents with mental disorders have elevated risks of mental disorder across diagnostic boundaries [1]. This transdiagnostic aggregation is important, but it does not identify a single mechanism. Familial resemblance can reflect direct genetic transmission, prenatal influences, indirect genetic effects, shared social conditions, caregiving processes, assortative mating, exposure outside the household, and child-to-parent effects. The strongest conceptual error is to treat these pathways as mutually exclusive. In many families they coexist, interact, and change over time.

This review argues that caregiving deserves explicit attention because it is developmentally proximal, observable, and often modifiable. That emphasis is pragmatic rather than causal: current evidence does not establish caregiving as the dominant source of intergenerational psychiatric risk. Nor does a parental diagnosis establish harmful parenting. Mental illness may affect care through withdrawal, irritability, disorganization, impaired concentration, altered threat perception, sleep disruption, or reduced capacity to coordinate daily tasks, but the developmental consequences depend on severity, timing, treatment, material resources, the availability of other caregivers, and what the child experiences repeatedly rather than momentarily.

Parents’ own accounts reinforce the need for this distinction. A systematic review and metasynthesis of 29 studies involving 562 parents with serious mental illness described caregiving strain, fear of professional judgment or child removal, and broad support needs [2]. Those findings illuminate barriers to help-seeking and service engagement; they do not provide prevalence estimates or demonstrate that a particular psychological mechanism causes adversity.

The review addresses three linked questions. First, what can population, genetically informed, and relational studies establish about pathways from parental mental illness to childhood adversity and psychopathology? Second, which caregiving and family processes are plausible intervention targets, and how strong is the evidence for them? Third, what should families, practitioners, services, and researchers do while the mechanism evidence remains incomplete? The resulting position is deliberately asymmetric: uncertainty about causal decomposition should make claims more precise, but it should not delay support for clearly identified parental, caregiving, material, or child needs.

2. Scope, approach, and definitions

This article is a critical conceptual review rather than a systematic review or meta-analysis. Targeted searches of scholarly databases, publisher pages, repositories, and professional guidance through 19 October 2026 focused on parental mental illness and offspring risk; child protection and out-of-home care; gene–environment correlation and indirect genetic effects; parenting sensitivity and parental mentalization; qualitative experiences of parenting with mental illness; family and dyadic interventions; epigenetics; and clinical guidance. Priority was given to large cohorts, genetically informative designs, randomized trials, systematic reviews and meta-analyses, and evidence that challenged or qualified the proposed framework. Bibliographic details for recent studies were checked against indexed journal records; where online-first and issue dates differed, the final issue year is used in the reference list. The search was not exhaustive, was not preregistered, and did not include duplicate screening or formal study-level risk-of-bias grading. Accordingly, the review should be read as an evidence-informed conceptual synthesis, not as a complete estimate of the literature or of certainty.

Early childhood is defined here as birth through age five. Evidence from older children, adolescents, or adult retrospective reports is used when it tests a mechanism that is relevant to early development, but its indirectness is identified. Adversity denotes harmful or substantially stressful conditions; maltreatment denotes abuse or neglect; and trauma-related symptoms require separate assessment. Attachment insecurity, transient misattunement, parental distress, and psychiatric diagnosis are not synonymous with maltreatment.

Administrative outcomes require particular caution. A report to child protective services (CPS) records an allegation or concern; an investigation, substantiation decision, service episode, and out-of-home placement reflect different administrative processes. None alone identifies the precise exposure, perpetrator, severity, or developmental meaning of what occurred. Conversely, absence of administrative contact does not establish absence of adversity.

Throughout the review, observed associations, statistical mediation, causal effects, and demonstrated intervention effects are kept separate. A mediator is not established because it correlates with both exposure and outcome, and a randomized intervention does not randomize the mediator it changes. Hypothetical examples are used to clarify clinical reasoning; they are not cases, prevalence estimates, or evidence of typical outcomes.

3. An integrative developmental framework

The framework separates four levels that are frequently collapsed in discussions of intergenerational risk: (1) familial liability and prenatal influences; (2) parental symptoms and functional capacity; (3) caregiving tasks and the child’s experienced environment; and (4) developmental outcomes. These levels are linked by probabilistic pathways, not by a deterministic chain. Each transition is modified by social conditions, treatment, other caregivers, developmental timing, and the child’s characteristics.

3.1 Familial liability, gene–environment correlation, and indirect genetic effects

Genetic transmission refers to inherited alleles that contribute probabilistically to offspring characteristics. Gene–environment correlation (rGE) describes covariation between genetic differences and environmental exposure. Passive rGE can occur when parents provide both inherited alleles and correlated rearing conditions; evocative rGE can occur when a child’s genetically influenced characteristics elicit responses from others; and active rGE concerns selection into environments. These are explanatory categories, not assignments of moral responsibility. A child’s temperament or behavior never transfers responsibility for abusive conduct away from an adult.

Gene–environment interaction (G×E) addresses a different question: whether the association or effect of an environmental exposure differs across genetic backgrounds, or vice versa, on a specified statistical scale. Correlation does not demonstrate interaction, and interaction does not establish mediation. The distinction matters because a large difference between jointly exposed groups and a reference group is not, by itself, evidence of biological synergy.

Warrier and colleagues’ genome-wide study of childhood maltreatment used population and family analyses in 185,414 participants and found evidence consistent with a complex relationship between genetic differences and measured maltreatment [3]. Such findings caution against treating measured exposure as independent of familial liability, but they do not identify parental psychiatric illness or impaired mentalization as the predominant route. Likewise, a child polygenic score can reflect multiple pathways and is not a molecular measure of caregiving.

Indirect parental genetic effects, sometimes termed genetic nurture, provide another important distinction. Kong and colleagues showed that nontransmitted parental alleles predicted offspring educational attainment [4], demonstrating that parental genotype can be associated with offspring outcomes through environmental pathways. However, magnitude and mechanism cannot be imported from educational attainment into psychiatric transmission. Recent evidence is especially informative here: in a large Dutch family cohort, nontransmitted parental polygenic scores for depression showed little robust association with offspring depression, anxiety, symptoms, or related traits, whereas transmitted scores were consistently associated [27]. This does not rule out indirect genetic effects from other traits or in other populations, but it argues against treating genetic nurture as a generic explanation for parent–offspring resemblance in depression and anxiety.

3.2 From symptoms to caregiving tasks

The clinically useful question is not whether a symptom is 'parenting related' in the abstract, but whether it changes a concrete caregiving task. Severe withdrawal may reduce responsiveness to a frightened child; disorganization may interrupt supervision or medication management; irritability may make ordinary child protest more likely to be interpreted as hostile; insomnia may reduce patience and executive control. In other families, the same diagnosis may produce little observable disruption because symptoms are well treated, the parent retains functional capacity, or another caregiver provides continuity.

This function-first formulation also prevents psychological explanations from obscuring material constraints. A parent can understand a child’s needs accurately while being unable to meet them because of housing instability, food insecurity, unsafe work schedules, lack of childcare, or fragmented treatment. Material conditions can affect both parental symptoms and caregiving and therefore act as confounders, mediators, or intervention targets depending on the question.

A recent systematic review and meta-analysis comparing parents with and without depression or anxiety found differences in parenting stress, discipline, warmth, withdrawal, relationship quality, and self-efficacy, while also emphasizing substantial cross-sectional evidence and potential confounding by child and socioeconomic factors [31]. These findings support measuring parenting and family function rather than inferring them from diagnosis; they do not imply that every parent with depression or anxiety shows the same pattern.

3.3 From caregiving to the child’s experienced environment

The child is exposed to experiences, not constructs. The developmentally relevant exposure may be repeated unavailability for comfort, frightening or hostile interactions, unsafe supervision, inconsistent meals, abrupt caregiver changes, conflict, or chronic uncertainty about who is available. The same parental symptom can therefore produce different child exposures, and different symptoms can converge on the same exposure.

Measurement should reflect this specificity. Repeated assessment within families can compare periods of symptom exacerbation and relative stability, while direct observation, multiple informants, and administrative records can reduce reliance on a single reporting process. Within-family change does not remove all confounding, but it narrows the mechanism: if symptom change is followed by task-specific caregiving change and then by change in child functioning, the proposed pathway becomes more credible.

3.4 Protection is distributed across relationships and systems

Risk is not located solely within the parent–child dyad. Another trusted adult can maintain routines and emotional availability during episodes of illness. Effective treatment can restore functioning. Stable housing and income can prevent symptom exacerbation from becoming a caregiving crisis. Pediatric, maternity, adult mental health, and family services can either coordinate these supports or create gaps between them. The relevant unit of prevention is therefore often a caregiving system rather than an individual parent.

This broader view also helps distinguish etiologic importance from intervention leverage. A pathway may explain only a modest proportion of population variance yet still be an important target because it is identifiable and changeable. Conversely, a strong statistical predictor may have little clinical value if it does not improve decisions or outcomes.

Direct genetic effects and caregiving effects should not be treated as competing explanations that must divide a fixed share of the same causal pathway. A child’s inherited liability can influence neurodevelopment and psychopathology through mechanisms that do not operate through caregiving, while caregiving can simultaneously exert environmental effects or alter the developmental expression of that liability. The practical rationale for assessing caregiving therefore does not depend on showing that caregiving accounts for most familial transmission. It depends on whether a specific caregiving difficulty is identifiable, consequential for the child, and potentially modifiable.

Table 1. Evidence domains, defensible inferences, and common overinterpretations

4. What population and genetically informed evidence establish

4.1 Familial risk is substantial and transdiagnostic

The largest contemporary synthesis of familial high-risk and registry studies included 211 independent studies, more than 3.1 million offspring of affected parents, and over 20 million control offspring [1]. Across parental diagnoses, offspring had elevated risks not only for the same disorder but also for other disorders. This supports a broad familial-risk model rather than a one-disorder/one-outcome model. It also argues for prevention strategies that attend to general development and functioning instead of screening only for a parent-matched diagnosis.

Absolute-risk estimates from such syntheses should nevertheless be interpreted with caution. Study designs, follow-up ages, diagnostic procedures, and ascertainment vary substantially, and most included studies in the meta-analysis were judged at high risk of bias [1]. Familial risk identifies a population in whom prevention may be valuable; it does not tell a clinician which child is currently unsafe or which mechanism is operating in a particular family.

4.2 Administrative records identify concentrations of service involvement

In a Swedish cohort of 1,249,463 children, Nevriana and colleagues reported out-of-home-care incidence rates of 6.3 versus 0.8 per 1,000 person-years in children with versus without recorded parental mental illness; the fully adjusted hazard ratio was 4.42 (95% CI, 4.24–4.61) [5]. Associations varied by diagnosis, parent sex, child age, and socioeconomic circumstances. The study did not contain detailed reasons for placement or direct measures of parenting or abuse, so it supports attention to family support needs but does not identify a uniform pathway from psychiatric diagnosis to caregiving failure.

In California’s 2006 birth cohort of 551,232 infants, Hammond and colleagues found that 34.6% of infants whose mothers had a mental health disorder documented during delivery hospitalization were reported to CPS within one year, compared with 4.4% of other infants [6]. Adjusted relative rates were 2.60 (95% CI, 2.47–2.73) without documented maternal substance use disorder and 5.69 (95% CI, 5.51–5.87) with co-occurring substance use disorder. These figures concern reports of alleged maltreatment in a specific historical, administrative, and clinical context. They are not current population probabilities for all parents receiving mental health care.

Taken together, these cohorts demonstrate concentrated service involvement, not a single 'trauma rate.' Their outcomes, denominators, thresholds, and surveillance differ. Diagnosis can also increase professional contact and therefore opportunities for concerns to be recorded, while serious difficulties can remain unseen in families with little service contact. Research should examine ascertainment directly rather than assume either that recorded risk is merely surveillance bias or that every record represents confirmed harm.

4.3 Genetic confounding attenuates associations but does not eliminate environmental harm

Evidence that childhood adversity is genetically correlated with later psychopathology does not imply that adversity has no causal effect. A meta-analysis of 34 quasi-experimental studies including 54,646 participants found that the association between childhood maltreatment and mental health problems attenuated from Cohen’s d = 0.56 to d = 0.31 after quasi-experimental adjustment [8]. The attenuation is evidence that wider familial and individual factors contribute; the remaining association is compatible with an environmentally mediated effect. No single design removes all confounding, and the synthesis does not isolate parental mental illness as the initiating cause.

Adoption designs provide complementary leverage by separating biological relatedness from rearing relationships. The Finnish adoption study of family functioning among adoptees with and without familial schizophrenia risk illustrates this strategy [9]. Yet adoption is not randomization of family functioning. Selective placement, prenatal exposures, restricted environmental range, and reciprocal child effects remain. Apparent subgroup differences should not be described as genetic susceptibility unless interaction is formally tested.

4.4 Prospective and retrospective adversity measures should not be treated as interchangeable

Baldwin and colleagues’ meta-analysis of 16 studies involving 25,471 participants found low agreement between prospective and retrospective measures of childhood maltreatment (κ = 0.19; 95% CI, 0.14–0.24) [7]. This discrepancy is not proof that one method is valid and the other invalid. Prospective records may miss unreported experiences; retrospective reports are influenced by memory, interpretation, and current state. The measures identify partly different groups and answer different questions.

For early-childhood research, the implication is methodological: direct observation, caregiver and partner report, child-protection records, clinical records, and later self-report should be modeled as distinct indicators rather than collapsed into a single adversity variable. Disagreement itself may be informative about surveillance, disclosure, or changing interpretation.

5. Parenting, sensitivity, and parental mentalization

5.1 Parenting sensitivity is relevant, but it is not the whole pathway

Observed parental sensitivity has consistent but generally modest associations with child behavior. A meta-analysis of 108 studies involving 28,114 children found small associations between greater parental sensitivity and fewer internalizing (r = −0.08) and externalizing (r = −0.14) problems [33]. Such findings support sensitivity as one relational process among many, not as a sufficient explanation for intergenerational psychopathology.

More directly, a 2024 meta-analytic structural equation model tested sensitive-responsive parenting as a mediator between maternal depression and offspring psychopathology across 68 studies for internalizing outcomes and 92 for externalizing outcomes [24]. The synthesized indirect pathways were small (approximately b = .01 for internalizing and b = .02 for externalizing). This is an important corrective to mechanistic narratives: even when a parenting process is statistically supported, its average mediated contribution can be modest.

The clinical implication is not that sensitivity is unimportant. A small average indirect effect can still be relevant at population scale, but an aggregate meta-analytic estimate does not establish the magnitude of the mediated pathway within individual families. Average effects may also obscure heterogeneity across populations, contexts, and measurement conditions, which the meta-analysis was not designed to estimate at the individual-family level. The implication is that intervention should be matched to measured difficulty. Improving sensitivity cannot be expected to solve unsafe housing, severe parental relapse, lack of childcare, or a child’s independent neurodevelopmental disorder.

5.2 Mentalization and parental reflective functioning are promising relational constructs

Parental mentalization refers broadly to understanding behavior in terms of mental states. Parental reflective functioning (PRF), coded representations, mind-mindedness, and related measures capture overlapping but nonidentical aspects of this capacity. They differ in method, dependence on language, context, and scoring. Treating them as interchangeable indicators of a single fixed trait obscures both measurement error and construct heterogeneity.

A systematic review of 16 studies found generally positive associations between PRF and observed parenting, with variation across parenting dimensions and contexts [10]. A meta-analysis of parental mentalization and infant attachment reported an association with attachment security of r = 0.30 across 20 effect sizes involving 974 dyads; an association remained after accounting for sensitivity [11]. A 2025 scoping review of 73 studies further found that most studies reported at least one association between parental mentalization and child mental health or functioning, although the evidence base was heterogeneous and often observational [28].

These findings establish relational relevance, not causal primacy. They do not show that low PRF causes maltreatment, that it mediates most of the effect of parental mental illness, or that increasing a mentalization score will necessarily improve safety. A 2025 study comparing mothers with borderline personality disorder, mothers with depressive or anxiety disorders, and healthy comparison mothers found group differences in parental mentalizing measures [12], but cross-sectional group differences cannot establish temporal mediation and should not be generalized to all parents with psychiatric illness.

5.3 Mentalization may be a mechanism, moderator, marker, or treatment target

The strongest version of the mentalization hypothesis is temporal: symptom or stress change alters how the parent interprets the child; interpretation changes alter observable caregiving; caregiving change alters the child’s exposure and subsequent outcome. A weaker but still clinically useful possibility is that PRF marks broader capacities such as emotion regulation, attention, cognitive flexibility, or relationship quality. A third possibility is moderation: mentalization may buffer the effect of stress without being the main route through which illness exerts risk.

These possibilities require different tests. Cross-sectional mediation cannot establish temporal ordering. Longitudinal mediation must measure the mediator before the downstream behavior and outcome, and it must address mediator–outcome confounding. Component trials must show that adding a reflective intervention produces incremental benefit over otherwise comparable support. Until these tests are met, PRF should not be presented as a validated screening test for maltreatment or parental fitness.

Two hypothetical families illustrate the distinction. In one, a distressed parent repeatedly interprets a toddler’s crying as manipulation and responds with escalating anger; reflective work may help generate alternative explanations and safer responses. In another, the parent accurately understands the child’s distress but becomes too unwell to maintain meals, nursery attendance, or overnight care; continuity planning and practical support are the immediate intervention. Both are compatible with parental mental illness, but the active mechanism and appropriate support differ.

6. Interaction and epigenetics: useful hypotheses with strict evidentiary limits

Differential susceptibility proposes that some individuals are more responsive to favorable as well as adverse environments. Demonstrating it requires reliable environmental measurement, an adequate range of exposure, sufficient power, explicit interaction testing, and replication. Large-sample work has failed to support many historical candidate-gene and candidate-gene-by-environment claims for major depression [13]. That evidence challenges specific candidate-gene narratives; it does not disprove all G×E effects. Nor is a genetic interaction necessary to justify supportive caregiving: a beneficial environmental main effect is sufficient.

Epigenetic findings require similar restraint. A systematic review of 100 empirical studies of childhood maltreatment and DNA methylation described a heterogeneous literature with differences in tissue, design, loci, and analytic strategy [14]. A methylation difference in peripheral tissue does not identify the direction of causation, establish a corresponding brain change, or prove functional consequences. Genotype, cell composition, medication, smoking, diet, stress, developmental timing, and other exposures can affect methylation.

Accordingly, methylation should not be treated as a molecular archive that identifies a specific parent–child relational history, and a relationship intervention should not be claimed to 'reverse trauma biologically' without prospective evidence linking intervention, molecular change, and clinically meaningful function. Molecular measures can enrich mechanistic studies; they cannot replace direct assessment of safety, symptoms, caregiving, or development.

7. Intervention evidence and its boundaries

7.1 Preventive and whole-family programs show benefit, but effects are heterogeneous

The prevention literature is broader than mentalization-focused programs. An early meta-analysis of 13 trials involving 1,490 children reported a lower incidence of mental disorder in intervention groups (relative risk 0.60; 95% CI, 0.45–0.79) [15]. A later synthesis of randomized trials likewise reported reduced incidence and small improvements in internalizing symptoms [16]. These findings support targeted prevention, but the interventions, ages, parental diagnoses, outcomes, and follow-up periods varied considerably.

More recent syntheses temper any claim of a single proven family model. A 2024 systematic review and meta-analysis of whole-family programs identified 41 independent studies covering 30 interventions and found small effects across parent, child, and family outcomes, alongside a lack of high-quality evidence [25]. A 2025 meta-analysis of 27 randomized trials for adolescents and young adults with a seriously ill parent reported small-to-moderate short-term benefits for psychological and positive adjustment but no sustained long-term effect on psychological adjustment and no clear post-intervention effect on externalizing or interpersonal outcomes [30]. These older-youth findings are informative about program heterogeneity but should not be generalized directly to infancy.

A 2025 randomized trial of a family-based preventive intervention for children of parents with severe mental illness also examined family functioning and the home environment, illustrating the field’s movement toward outcomes that sit between parent symptoms and child diagnosis [32]. Such outcomes are theoretically valuable because they test whether the family system actually changes, but they still require linkage to sustained child benefit.

7.2 Early-childhood multicomponent interventions strengthen the case for integrated support

Evidence focused specifically on the earliest years is particularly relevant. A 2024 systematic review and meta-analysis of randomized interventions that explicitly targeted both parenting behavior and parental mental health during pregnancy or the first three years of life included 25 articles and 8,520 children and caregivers [26]. The interventions produced small improvements in child cognitive development (SMD 0.19; 95% CI, 0.04–0.34) and social-emotional development (SMD 0.26; 95% CI, 0.17–0.34), but the pooled effect on depressive symptoms in female caregivers did not reach conventional statistical significance. Most programs devoted more intervention time to parenting than to parental mental health.

This pattern is conceptually important. Child benefit can occur without parallel evidence of large change in parental depressive symptoms, which argues against assuming that adult symptom remission is the sole route to improved child outcomes. Conversely, improvement in a parent’s symptoms should not be assumed to resolve every caregiving or child difficulty.

7.3 Treating parental illness is necessary care, but offspring benefit must be measured

The STAR*D-child study associated remission of maternal depression with improvements in offspring symptoms and diagnoses [17]. Because parental remission was not randomized, families in which remission occurred may have differed in other ways, and the offspring were aged 7–17 years. The study therefore supports coordinated attention to parent and child outcomes but does not establish that adult treatment alone prevents early-childhood adversity.

A 2026 systematic review identified a specific but consequential evidence gap. After screening 938 randomized studies of psychological treatment for adult depression beyond the postnatal period, Lawrence and colleagues found no eligible trials that both treated the parent’s depression as the primary intervention target and measured child mental health or wellbeing [29]. This finding should not be interpreted as an absence of intervention research involving depressed parents and their children: parenting, dyadic, perinatal, and child-focused interventions address related questions but fell outside the review’s eligibility criteria. Rather, the result shows that the intergenerational consequences of treating adult depression itself have rarely been incorporated into adult-treatment trials. Adult mental health trials therefore cannot be assumed to deliver intergenerational benefit when offspring outcomes are not measured.

7.4 Reflective and relationship-focused interventions require mechanism-specific claims

The ATTACH program reported improvements in reflective functioning and selected family and child measures in phase 2 pilot work [18]. The publication structure includes small randomized pilots followed by quasi-experimental treatment of waitlist controls; it should not be interpreted as multiple independent large randomized replications. The findings justify confirmatory testing but do not establish reduced maltreatment incidence or prove that PRF is the mediator of benefit.

Child–Parent Psychotherapy was tested in a randomized trial involving 75 preschool child–mother dyads exposed to marital violence and produced improvements in a defined clinical population [19]. As a multicomponent relationship-focused intervention, it may work through several pathways. Efficacy therefore does not identify PRF as the active ingredient, and transfer to different parental diagnoses or service settings requires additional evidence.

Recent null or equivocal trials are equally informative. In the 2026 multicenter CHIMPS trial, the primary analysis did not demonstrate a statistically significant benefit on the primary diagnostic outcome, and clinically relevant advantages on secondary outcomes were not established [20]. Engagement and missing follow-up data complicated interpretation. The result does not disprove family intervention; it demonstrates why theoretical plausibility, pilot improvement, and favorable uncontrolled change must be followed by adequately powered comparative trials with implementation data.

8. Implications for families, practitioners, and services

The evidence supports treating parental mental illness as a family-relevant condition without treating diagnosis as a proxy for parenting capacity. Professional guidance already points in this direction. NICE perinatal guidance includes assessment of the parent–infant relationship, social circumstances, caregiving responsibilities, and collaborative planning [21]. The American Academy of Pediatrics links recognition of perinatal depression with referral, follow-up, and support for the parent–child relationship [22]. NICE child-abuse guidance emphasizes assessment of the wider care environment and safe engagement with families [23]. The recommendations below translate the review’s conceptual model into practice; they are not a newly validated protocol.

8.1 For families: convert diffuse inherited-risk concerns into a concrete care plan

Parents can be told two things at once: familial psychiatric risk is real, and it is not destiny. A useful planning conversation shifts from 'What did my child inherit?' to 'What becomes harder when I am unwell, what remains stable, and who can help?' This framing preserves the relevance of genetic liability while identifying modifiable tasks.

A family plan should identify early warning signs of deteriorating functioning, the caregiving tasks most likely to be affected, at least one safe adult who can assume agreed responsibilities when needed, and a route back to treatment or crisis support. In severe perinatal illness, collaborative written planning is already recommended by NICE [21]. Extending a task-based continuity plan beyond the perinatal period is a logical application of the present framework, although it is not itself a proven maltreatment-prevention program.

Young children also benefit from developmentally appropriate explanations. The aim is to make care understandable without making the child responsible for monitoring the adult. Explanations should distinguish the parent’s illness from the child’s behavior and identify which adults remain available. Existing strengths, cultural practices, and trusted relationships should be treated as resources rather than as afterthoughts.

8.2 For practitioners: assess function, exposure, and the child separately

Begin with the parent’s children and caregiving responsibilities, then move from diagnosis to function: Which symptoms affect which tasks? At what times? What happens on a difficult day? Who notices when care is interrupted? What is the child exposed to? Which parts of care remain dependable? This sequence distinguishes symptom severity from caregiving impact.

Assess the child in their own right. Developmental functioning, behavior, sleep, school or nursery attendance, somatic symptoms, relationships, and the child’s perspective may reveal needs that are not captured by the parent’s clinical status. Observation of interaction can be useful when it answers a defined question, but no single observation or PRF score should be treated as a test of parental fitness.

Match intervention to the identified mechanism. Recurrent hostile interpretations of the child’s distress may warrant reflective or relationship-focused work alongside symptom treatment and safety assessment. Inability to maintain routines may require alternative caregiving and practical support. Housing or food insecurity requires material assistance. A child with established symptoms needs assessment and treatment regardless of whether the parent is improving.

Parents with serious mental illness have described fear, concealment, and mistrust in support relationships [2]. Practitioners should therefore explain why questions are being asked, what confidentiality does and does not cover, and what safeguarding responsibilities apply. This is not merely an ethical courtesy: service acceptability and engagement are part of whether an intervention can reach the families for whom it is intended.

8.3 For service leaders: build a pathway, not a referral handoff

Adult mental health, maternity, pediatrics, early-years services, and family support often hold different pieces of the same risk pathway. Coordination should specify who treats the parent, who assesses the child, who addresses practical needs, and who helps maintain dependable care during periods of relapse. A referral is not a completed intervention until the family can access and use the service.

Evaluation should follow the entire pathway: families identified, referrals completed, support received, caregiving difficulties changed, child safety maintained, developmental outcomes improved, and burden acceptable. Reduced CPS reports or out-of-home placement can be favorable only when child safety and wellbeing are independently maintained; administrative reduction is not itself proof of benefit.

Current evidence does not support using polygenic scores or methylation assays to decide which families receive maltreatment-prevention services. Such tools would require evidence of incremental prediction, fairness across ancestry groups, and improved decisions and outcomes. Service allocation can instead be grounded in observed symptoms, functioning, caregiving demands, child needs, material conditions, and available support.

9. Research agenda: designs that can distinguish mechanisms

First, prospective family cohorts should begin in pregnancy or infancy and repeatedly measure parental symptoms, treatment, sleep, substance use where relevant, caregiving tasks, PRF or other mentalization measures, social conditions, child characteristics, direct adversity indicators, and child development. The key gain is temporal resolution: whether changes in symptoms precede changes in caregiving and whether those changes precede child outcomes.

Second, family-based genetic designs should combine parent–offspring genotypes with measured environments. Transmitted and nontransmitted alleles can help distinguish direct from indirect genetic associations; sibling, adoption, and children-of-twins designs offer complementary leverage. Analyses should address assortative mating, population structure, prenatal pathways, selection, ancestry limits in polygenic scores, and reciprocal child effects. Neither a polygenic covariate nor a within-family coefficient automatically identifies a caregiving effect.

Third, intervention research should move from package-versus-usual-care comparisons toward component questions. For example, a trial could compare high-quality family support with and without a reflective component, while keeping contact time and practical assistance comparable. To demonstrate mediation, change in the proposed mediator must precede change in caregiving, which in turn must precede the child outcome. Randomization of the treatment does not randomize the mediator.

Fourth, adult-treatment trials should measure offspring outcomes when participants are parents. The 2026 empty systematic review of parent-depression psychotherapy trials demonstrates how much mechanistic and preventive information is currently lost [29]. A pragmatic minimum dataset could include child age, caregiving responsibility, brief child functioning measures, and family-level adverse events, with more detailed assessment in nested cohorts.

Fifth, mechanism studies should incorporate implementation. Engagement, treatment dose, transportation, childcare, housing instability, clinician turnover, and trust can determine whether an efficacious intervention produces population benefit. These factors are not peripheral to causal inference: an intervention that families cannot access or tolerate cannot change the proposed mechanism in real-world practice.

Finally, fathers, nonbirth parents, kin caregivers, culturally diverse families, parents with disabilities, and families outside high-income service systems require intentional inclusion. Measures of mentalization, sensitivity, and 'appropriate' caregiving may perform differently across language, culture, neurotype, and context. Fairness requires distinguishing communication style from caregiving impairment and testing measurement invariance rather than assuming it.

10. Limitations of this review

The review’s targeted selection may privilege studies that fit its conceptual questions, and no formal certainty grading was undertaken. The addition of recent systematic reviews and trials improves currency but does not make the search exhaustive. A systematic review would be required to estimate pooled effects or formally compare risk of bias across every relevant study.

Much of the evidence concerns mothers, depression, selected clinical samples, older children, or high-income service systems. Findings about maternal depression and sensitive parenting cannot simply be generalized to fathers, psychosis, bipolar disorder, personality disorder, substance use, or every phase of early childhood. Likewise, intervention syntheses often combine programs with different targets and outcomes.

Observational studies remain vulnerable to shared familial causes, reverse directionality, measurement overlap, and differential surveillance. Genetically informed designs reduce some biases while introducing others. Mentalization measures are heterogeneous and often correlated with broader cognitive or relational capacities, making construct-specific causal claims difficult.

The framework gives caregiving explicit attention because it is developmentally proximal and potentially modifiable. It may therefore underemphasize prenatal biology, direct genetic effects, exposures outside the household, and the child’s emerging disorder. Direct genetic effects can influence developmental outcomes through pathways that are partly or wholly independent of caregiving; their presence does not negate environmental effects, nor does evidence for caregiving effects imply that inherited liability has been explained away. The framework’s value should therefore be judged not by whether caregiving explains most familial risk, but by whether separating inherited liability, symptoms, tasks, exposures, and outcomes leads to stronger causal tests and more useful support.

11. Conclusion

Parental mental illness is a robust marker of elevated offspring psychiatric risk, but it is not a developmental mechanism in itself. Intergenerational risk becomes actionable only when research and services identify what changes in the child’s care, which exposures matter, what protects the child, and which intervention produces improvement. Genetic liability, prenatal influences, caregiving, material conditions, and service responses should therefore be modeled together rather than placed in competition.

The most defensible practical response is function-first and family-centered: treat parental illness; identify task-specific caregiving disruption; maintain safe and dependable care during symptomatic periods; address material barriers; and assess the child directly. Parenting sensitivity and parental mentalization are credible relational processes, but their average mediated effects appear modest and their role in preventing maltreatment remains unproven. They should be assessed and treated when clinically relevant, not elevated into universal explanations or prediction tools.

This position combines urgency with epistemic restraint. Families do not need researchers to identify a single dominant pathway before offering support. They do need researchers and services to state what problem an intervention is intended to change, measure whether that change occurs, and verify that the child’s safety, development, and functioning improve. That is the standard by which an intergenerational prevention framework should ultimately be judged.

References

1. Uher R, Pavlova B, Radua J, Provenzani U, Najafi S, Fortea L, et al. Transdiagnostic risk of mental disorders in offspring of affected parents: a meta-analysis of family high-risk and registry studies. World Psychiatry. 2023;22(3):433–448. https://doi.org/10.1002/wps.21147

2. Harries CI, Smith DM, Gregg L, Wittkowski A. Parenting and serious mental illness (SMI): a systematic review and metasynthesis. Clinical Child and Family Psychology Review. 2023;26:303–342. https://doi.org/10.1007/s10567-023-00427-6

3. Warrier V, Kwong ASF, Luo M, Dalvie S, Croft J, Sallis HM, et al. Gene–environment correlations and causal effects of childhood maltreatment on physical and mental health: a genetically informed approach. Lancet Psychiatry. 2021;8(5):373–386. https://doi.org/10.1016/S2215-0366(20)30569-1

4. Kong A, Thorleifsson G, Frigge ML, Vilhjalmsson BJ, Young AI, Thorgeirsson TE, et al. The nature of nurture: effects of parental genotypes. Science. 2018;359(6374):424–428. https://doi.org/10.1126/science.aan6877

5. Nevriana A, Kosidou K, Hope H, Wicks S, Dalman C, Pierce M, Abel KM. Parental mental illness and the likelihood of child out-of-home care: a cohort study. Pediatrics. 2024;153(3):e2023061531. https://doi.org/10.1542/peds.2023-061531

6. Hammond I, Eastman AL, Leventhal JM, Putnam-Hornstein E. Maternal mental health disorders and reports to child protective services: a birth cohort study. International Journal of Environmental Research and Public Health. 2017;14(11):1320. https://doi.org/10.3390/ijerph14111320

7. Baldwin JR, Reuben A, Newbury JB, Danese A. Agreement between prospective and retrospective measures of childhood maltreatment: a systematic review and meta-analysis. JAMA Psychiatry. 2019;76(6):584–593. https://doi.org/10.1001/jamapsychiatry.2019.0097

8. Baldwin JR, Wang B, Karwatowska L, Schoeler T, Tsaligopoulou A, Munafò MR, Pingault JB. Childhood maltreatment and mental health problems: a systematic review and meta-analysis of quasi-experimental studies. American Journal of Psychiatry. 2023;180(2):117–126. https://doi.org/10.1176/appi.ajp.20220174

9. Myllyaho T, Siira V, Wahlberg KE, Hakko H, Läksy K, Roisko R, Niemelä M, Räsänen S. Interaction of genetic vulnerability to schizophrenia and family functioning in adopted-away offspring of mothers with schizophrenia. Psychiatry Research. 2019;278:205–212. https://doi.org/10.1016/j.psychres.2019.06.017

10. Stuhrmann LY, Göbel A, Bindt C, Mudra S. Parental reflective functioning and its association with parenting behaviors in infancy and early childhood: a systematic review. Frontiers in Psychology. 2022;13:765312. https://doi.org/10.3389/fpsyg.2022.765312

11. Zeegers MAJ, Colonnesi C, Stams GJJM, Meins E. Mind matters: a meta-analysis on parental mentalization and sensitivity as predictors of infant–parent attachment. Psychological Bulletin. 2017;143(12):1245–1272. https://doi.org/10.1037/bul0000114

12. Zitzmann J, Georg A, Rosenbach C, Renneberg B. “If I could read your mind…”: parental mentalizing in mothers with borderline personality disorder. Borderline Personality Disorder and Emotion Dysregulation. 2025;12:20. https://doi.org/10.1186/s40479-025-00290-7

13. Border R, Johnson EC, Evans LM, Smolen A, Berley N, Sullivan PF, Keller MC. No support for historical candidate gene or candidate gene-by-interaction hypotheses for major depression across multiple large samples. American Journal of Psychiatry. 2019;176(5):376–387. https://doi.org/10.1176/appi.ajp.2018.18070881

14. Parade SH, Huffhines L, Daniels TE, Stroud LR, Nugent NR, Tyrka AR. A systematic review of childhood maltreatment and DNA methylation: candidate gene and epigenome-wide approaches. Translational Psychiatry. 2021;11:134. https://doi.org/10.1038/s41398-021-01207-y

15. Siegenthaler E, Munder T, Egger M. Effect of preventive interventions in mentally ill parents on the mental health of the offspring: systematic review and meta-analysis. Journal of the American Academy of Child & Adolescent Psychiatry. 2012;51(1):8–17.e8. https://doi.org/10.1016/j.jaac.2011.10.018

16. Thanhäuser M, Lemmer G, de Girolamo G, Christiansen H. Do preventive interventions for children of mentally ill parents work? Results of a systematic review and meta-analysis. Current Opinion in Psychiatry. 2017;30(4):283–299. https://doi.org/10.1097/YCO.0000000000000342

17. Weissman MM, Pilowsky DJ, Wickramaratne PJ, Talati A, Wisniewski SR, Fava M, et al. Remissions in maternal depression and child psychopathology: a STAR*D-child report. JAMA. 2006;295(12):1389–1398. https://doi.org/10.1001/jama.295.12.1389

18. Letourneau N, Anis L, Novick J, Pohl C, Ntanda H, Hart M. Impacts of the Attachment and Child Health (ATTACH™) parenting program on mothers and their children at risk of maltreatment: phase 2 results. International Journal of Environmental Research and Public Health. 2023;20(4):3078. https://doi.org/10.3390/ijerph20043078

19. Lieberman AF, Van Horn P, Ghosh Ippen C. Toward evidence-based treatment: child–parent psychotherapy with preschoolers exposed to marital violence. Journal of the American Academy of Child & Adolescent Psychiatry. 2005;44(12):1241–1248. https://doi.org/10.1097/01.chi.0000181047.59702.58

20. Paumen T, Busmann M, Daubmann A, Höller A, Ozga AK, Kilian R, et al. Effectiveness of a family-oriented intervention for children of mentally ill parents on children’s mental health and health-related quality of life: a prospective, rater-blinded, cluster-randomized controlled multicenter trial. Child and Adolescent Psychiatry and Mental Health. 2026;20:135. https://doi.org/10.1186/s13034-026-01174-6

21. National Institute for Health and Care Excellence. Antenatal and postnatal mental health: clinical management and service guidance. Clinical guideline CG192. Recommendations. Accessed October 19, 2026. https://www.nice.org.uk/guidance/cg192/chapter/Recommendations

22. Rafferty J, Mattson G, Earls MF, Yogman MW; Committee on Psychosocial Aspects of Child and Family Health. Incorporating recognition and management of perinatal depression into pediatric practice. Pediatrics. 2019;143(1):e20183260. Reaffirmed May 2025. https://doi.org/10.1542/peds.2018-3260

23. National Institute for Health and Care Excellence. Child abuse and neglect. NICE guideline NG76. Recommendations. Accessed October 19, 2026. https://www.nice.org.uk/guidance/ng76/chapter/Recommendations

24. Borairi S, Deneault AA, Madigan S, Fearon P, Devereux C, Geer M, et al. A meta-analytic examination of sensitive responsiveness as a mediator between depression in mothers and psychopathology in children. Attachment & Human Development. 2024;26(4):273–300. https://doi.org/10.1080/14616734.2024.2359689

25. Moltrecht B, Lange AMC, Merrick H, Radley J. Whole-family programmes for families living with parental mental illness: a systematic review and meta-analysis. European Child & Adolescent Psychiatry. 2024;33(9):3203–3246. https://doi.org/10.1007/s00787-024-02380-3

26. Al Sager A, Goodman SH, Jeong J, Bain PA, Ahun MN. Effects of multi-component parenting and parental mental health interventions on early childhood development and parent outcomes: a systematic review and meta-analysis. Lancet Child & Adolescent Health. 2024;8(9):656–669. https://doi.org/10.1016/S2352-4642(24)00134-2

27. Trindade Pons V, Oldehinkel AJ, van Loo HM. Genetic nurture effects in depressive and anxiety disorders and symptoms, and in related traits. Molecular Psychiatry. 2025;30(12):5694–5700. https://doi.org/10.1038/s41380-025-03265-w

28. Moreira A, Vives J, Ballespí S. Transgenerational association of mentalization with child mental health: a scoping review. Family Process. 2025;64(1):e13085. https://doi.org/10.1111/famp.13085

29. Lawrence PJ, Dunn A, Agarwal M, Bowen C, Can B, Castle VE, et al. What happens to children’s mental health when we treat their parent’s depression? We have no idea. An empty systematic review. Behavioural and Cognitive Psychotherapy. 2026;54(1):36–46. https://doi.org/10.1017/S1352465825100970

30. Landi G, Pakenham KI, Bao Z, Cattivelli R, Crocetti E, Tossani E, Grandi S. Efficacy of psychosocial interventions for young offspring of parents with a serious physical or mental illness: systematic review and meta-analysis. Clinical Psychology Review. 2025;118:102569. https://doi.org/10.1016/j.cpr.2025.102569

31. Bennett M, McClure A, Khalesi H, Vu MTN, Feng J, Reupert A, Yap MBH. Differences in parental factors between parents with and without depression or anxiety issues: a systematic review and meta-analysis. Journal of Affective Disorders. 2026;392:120085. https://doi.org/10.1016/j.jad.2025.120085

32. Gjøde ICT, Müller AD, Hjorthøj C, Hemager N, Ingversen S, Moszkowicz M, et al. Effects on family functioning and the home environment of a family-based preventive intervention for children of parents with severe mental illness: a randomized controlled trial. Journal of Consulting and Clinical Psychology. 2025;93(4):267–280. https://doi.org/10.1037/ccp0000924

33. Cooke JE, Deneault AA, Devereux C, Eirich R, Fearon RMP, Madigan S. Parental sensitivity and child behavioral problems: a meta-analytic review. Child Development. 2022;93(5):1231–1248. https://doi.org/10.1111/cdev.13764