Aetiology is the study of the causes of a disorder. ^[1]

This seems simple, but the lecturer immediately emphasises that causes are typically complex. He uses two illustrative examples:

Key teaching point: Even disorders with a single identifiable cause still involve multiple factors at the individual level. Different factors [are] involved in different patients. ^[1]

This is a critical framework. The lecture classifies aetiological factors along three axes:

Classification according to:

• Temporal relationship: Distal (Remote) / Proximal (Recent); Predisposing / Precipitating / Maintaining
• Nature of factor: Intrinsic / Extrinsic; Biological / Psychosocial; Genetic / Environmental
• Relationship to other disorders: Primary / Secondary

The Predisposing / Precipitating / Maintaining Framework

This is THE framework used in psychiatric formulation and is directly tested in SAQs ^[3]:

The lecture references Robert Burton's Anatomy of Melancholy (1621) to show that early writers proposed multiple explanations: Burton separated supposed “supernatural” causes from “natural” congenital/hereditary, bodily, and adverse social causes (including education, fear, loss of liberty, and poverty). This is history of ideas, not a modern causal taxonomy: supernatural explanations and Burton's period language are not evidence-based clinical accounts.

Most disorders are caused by the complex interplay of multiple genetic and environmental factors — Multifactorial ^[1]

A minority of psychiatric disorders have simple (single-factor) causality:

But the vast majority of psychiatric disorders are multifactorial — meaning many genes of small effect plus environmental exposures combine to produce disease.

Just because something is associated with a disorder doesn't mean it causes it. The lecture outlines the methodology:

Strongest demonstration in an appropriate setting:

• Putative cause associated with disorder
• Removing putative cause prevents disorder (interventional study)

Exclusion of confounding via stratified analysis and regression models

Bradford Hill Criteria:

• Strong and consistent association
• Correct temporal relationship
• Dose-response relationship
• Biologically plausible mechanisms

Causal modelling including Mendelian randomization ^[1]

Mendelian randomization (MR) uses genetic variants as instruments to test whether an exposure may causally affect an outcome. It can reduce some conventional confounding and reverse causation, but it is not automatically confounding-free: the instrument must predict the exposure, must not share causes with the outcome, and must affect the outcome only through the exposure. Population stratification, assortative mating, horizontal pleiotropy, and selection can violate these assumptions. Likewise, the Bradford Hill considerations support causal reasoning but are not a mechanical checklist, and an intervention result depends on whether the intervention truly and specifically changes the putative cause.

The same clinical disorder can have different aetiological subtypes.

The lecture uses Alzheimer's disease as the prime example:

Sporadic AD is also associated with PICALM ^[1]

The lecture includes a diagram of the amyloid-processing pathway linking familial and sporadic AD genetics to a common downstream cascade:

(Diagram reproduced in the lecture from Van Es & van den Berg, 2009.) ^[1]

This is a useful historical, amyloid-centred schematic, not a complete modern model of Alzheimer disease. In particular, PS1 and PS2 in the slide mean PSEN1 and PSEN2; APOE, CLU, PICALM, and CR1 are susceptibility loci rather than deterministic diagnoses, and downstream neurodegeneration is more complex than a single linear amyloid cascade.

This is the fundamental quantitative model for understanding multifactorial diseases:

All-or-none traits (diseases) can be modelled as arising from an underlying continuous liability; when liability exceeds a threshold, disease develops. ^[1]

The bell curve on the slide is a modelling assumption: many small genetic and environmental contributions can motivate an approximately normal latent liability. It is not proof that a measurable biological quantity called “liability” is literally Gaussian, and the threshold is a model parameter rather than a sharp natural boundary.

Why does this matter clinically?

• It explains why most people in the population don't get the disease (they're below threshold)
• It explains why relatives of affected individuals have higher risk (they share aetiological factors, so their liability distribution is shifted right)
• It explains the continuum between "normal" and "pathological" — the threshold is somewhat arbitrary

This is the clinical application of the liability-threshold model:

Liability = Diathesis (vulnerability) + Stress (proximal factors) There is a negative correlation between diathesis and stress in patients (collider bias). ^[1]

What this means in plain English:

• To cross the disease threshold, you need some combination of vulnerability (diathesis) and trigger (stress)
• Patients with high diathesis (e.g., strong family history, multiple risk genes) need less stress to become ill
• Patients with low diathesis need more stress to become ill
• When you look only at patients (i.e., people who crossed the threshold), you see an artificial negative correlation between diathesis and stress — this is called collider bias (selecting on the outcome creates a spurious association between the causes)

Clinical implication: When taking a history, if a patient has strong family history (high diathesis), don't be surprised if the precipitating stressor seems trivial. Conversely, a massive trauma can trigger illness even in someone with no family history.

Mediators are the mechanisms through which aetiological factors produce the disorder (A → M → D pathway):

Biological mediators: stress reactivity, brain connectivity Psychological mediators: cognitive function, personality, resilience Social mediators: lifestyle, support network

Potential biomarkers ("endophenotypes") — measurable traits that lie between genotype and clinical phenotype Potential points for intervention ^[1]

Why do mediators matter? Because even if you cannot change an upstream exposure, a genuine mediator may be an intervention point. However, an associated intermediate measure is not necessarily a causal mediator. Distinguish it from a confounder (a common cause of exposure and outcome) and a moderator (something that changes the size or direction of an effect); changing an associated marker does not by itself prove that the clinical outcome will change.

Comorbidity [is] more frequent than expected by chance. ^[1]

The lecture presents three possible explanations (from the Johansson et al. 2013 diagram):

The lecture slide also presents prevalence data for depression-anxiety comorbidity (Johansson et al. 2013) ^[1]:

Key point: The overlap is large — in this dataset, most participants with clinically significant depression also had clinically significant anxiety (8.3% comorbid out of 10.8% total depressed). Shared causes are one explanation, but symptom/measurement overlap, one disorder influencing the other, and selection into a clinical sample can also contribute.

Environmental exposures influenced by behaviour of individual or parents. Behaviours partially genetically determined → gene-environment correlation. ^[1]

This is subtle but important. It means that what we measure as "environmental" exposure may partly be genetically driven:

• Example 1: Exposure to carcinogens in tobacco smoke is an "environment," but the behaviour of smoking is partly genetic (genes affecting nicotine dependence)
• Example 2: Exposure to social adversity associated with a substance use disorder — liability to the substance-use behaviour may itself be partly heritable

Why this matters: It complicates aetiological research. When we find that "childhood adversity" is associated with depression, some of that association may be because parents with psychiatric genetic risk create more adverse environments for their children (passive gene-environment correlation).

Three types of gene-environment correlation (from genetics theory):

1. Passive: Parents provide both genes and environment (e.g., depressed parent provides depression-risk genes AND an emotionally neglectful home)
2. Evocative: Child's genetically influenced behaviour evokes environmental responses (e.g., a temperamentally difficult child provokes harsh parenting)
3. Active: Individual selects environments based on genetically influenced preferences (e.g., sensation-seeking genes → choosing risky peer groups)

Environmental factors increasing disease risk especially in those with specific genetic susceptibility. ^[1]

This is different from correlation — here the effect of the environment depends on the genotype:

Statistically: Gene-environment interaction means the combined effect of gene + environment is greater (or less) than what you'd predict from adding their individual effects.

Interaction is scale-dependent: a pattern can appear on an additive risk scale but not a multiplicative odds scale (or vice versa), and exposure/genotype measurement, population structure, and multiple testing can produce false positives. Replication and a pre-specified model are essential.

History taking: Family/Development, Personality/Lifestyle/Habits, Stressors (acute and chronic), Social resources Formulation: Predisposing/precipitating/maintaining factors + protective factors Management: Promote understanding, reduce contributing factors, strengthen protective factors ^[1]

The slide's genogram is a practical family-history map. Common symbols are: square = male, circle = female, diamond = sex/gender unspecified; a diagonal slash = deceased; an arrow marks the index person; connecting lines represent partnerships and offspring; paired lines may mark consanguinity; broken partnership lines mark separation/divorce; branching from one point identifies twins; and shading or labels record a condition. Conventions vary, so include a legend. A genogram shows family aggregation—not genetic determinism—and sensitive family information should be recorded with consent, relevance, and confidentiality in mind.

The lecture traces the historical arc:

1. 19th century: Familial/social/racial "degeneracy" → stigmatization, eugenics, even genocide
2. Post-WWII: Rejection of eugenics → rise of psychodynamic/psychological theories (Freud)
3. Modern era: Scientific methods to disentangle nature from nurture

Freud: repression of early traumatic memories → psychoanalysis recalls memories into consciousness (catharsis). "Exonerates" patient as victim of external circumstances. Parents/Adults responsible for "traumatizing" children. ^[1]

The deck's older examples require context:

• John Haslam (1798) described a mother and father with severe “melancholia” as an early observation of familial aggregation. The terminology, asylum setting, and coercive care are historical; a two-person pedigree cannot distinguish genes from shared environment.
• The eugenic “burden” and Nazi propaganda images show how claims about heredity were used for stigma, forced sterilisation, and genocide. These practices were scientifically invalid and grave human-rights abuses. Modern psychiatric genetics must be voluntary, non-discriminatory, consent-based, and attentive to uncertainty.
• The photograph labelled “Anna O” refers to Bertha Pappenheim, treated by Josef Breuer rather than being Freud's own patient. Psychoanalytic repression/catharsis is historically influential, but recovered-memory claims and parent-blaming are not established causal explanations. Childhood trauma is clinically important without requiring this historical theory.

The lecture reproduces classic studies summarised by Gottesman and Shields (1976). Across the displayed studies, MZ concordance is roughly 35–58% and DZ concordance roughly 9–26%, depending on study and calculation. Probandwise concordance asks the probability that a co-twin is affected given an affected proband; pairwise concordance is the proportion of affected pairs among all pairs. They are not interchangeable.

The important inference is MZ > DZ, supporting substantial genetic influence, while MZ is far below 100%, demonstrating non-determinism and the importance of non-shared factors and measurement. Modern estimates often place schizophrenia liability heritability around 80%, but the exact value is population- and method-specific.

The lecture's 2017 textbook graphic gives the following illustrative estimates—not a timeless ranking:

The slide then links high heritability with biological intervention and low heritability with psychosocial intervention. Treat that as a historical heuristic, not a treatment rule. Heritability does not tell us which treatment works: phenylketonuria is genetic yet diet-responsive, while highly heritable psychiatric disorders benefit from psychosocial as well as biological care. Treatment should follow clinical evidence, patient needs, and formulation—not a heritability percentage.

Nature 511, 421-427 (24 July 2014) — the landmark PGC2 study identified 128 independent associations spanning 108 loci for schizophrenia ^[1]

GWAS for depression (Howard et al. 2019) identified 103 significant loci and 269 significant genes ^[1]

These are dated snapshots from the slide. A 2022 Psychiatric Genomics Consortium study of 76,755 people with schizophrenia and 243,649 controls reported associations at 287 loci and prioritised 120 genes ^[7]. A 2025 multi-ancestry major-depression study (688,808 cases; 4,364,225 controls) reported 697 associations at 635 loci, 293 of them novel ^[8]. Larger counts do not make either disorder genetically deterministic; phenotype heterogeneity and representation across ancestries remain important.

The Manhattan plots encode chromosome position on the x-axis and −log10(P) on the y-axis; points above the horizontal line meet the conventional genome-wide significance threshold (about P < 5 × 10⁻⁸). The prominent chromosome 6 schizophrenia signal includes the complex MHC region. A peak identifies a region for follow-up, not the mechanism by itself.

Biological mechanisms: key molecules or biological pathways → novel treatment targets → heterogeneity (precision/stratified medicine) Prediction: disease risk, prognosis, treatment response ^[1]

These are research aims rather than current bedside promises. Associations can nominate pathways and targets, but individual prediction of diagnosis, prognosis, or treatment response remains limited.

The lecture specifically names these gene categories and loci:

Dopamine: DRD2 Glutamate: GRM3, GRIN2A, SRR Calcium channels: CACNA1C, CACNA1I, CACNB2 ^[1]

Why these matter clinically:

• DRD2: convergence between an associated locus and a major antipsychotic target is biologically notable, but does not validate a single complete “dopamine hypothesis” or establish that the associated variant changes receptor coding.
• Glutamatergic and calcium-channel loci: support further mechanistic work, but locus-to-gene assignment and therapeutic translation require functional evidence. They do not explain why a particular medicine works in an individual.

The PGC pathway analysis (slide 33) identified three major pathway clusters implicated in schizophrenia:

Teaching point: The slide's pathway analysis groups the top 50 pathways (q < 0.1) into synaptic, histone-methylation, and immune/neurotrophic clusters. Such enrichment and co-expression networks are hypothesis-generating group-level analyses; they do not show that every listed gene is causal or that “immune schizophrenia” can be diagnosed in an individual.

The lecture shows co-expression networks of genes in top pathways with q < 0.1, illustrating that schizophrenia-associated genes cluster in specific biological pathways related to synaptic function, neurotransmission, and neurodevelopment.

Polygenic risk score: Predicts person's disease risk by combining effects of genotypes at multiple loci → Risk stratification → Clinical implications in future ^[1]

A PRS sums weighted effects across many variants. Psychiatric PRS remains primarily a research tool: the International Society of Psychiatric Genetics and Royal College of Psychiatrists state that current scores lack sufficient absolute-risk and treatment-prediction utility for routine psychiatric practice, and performance varies substantially by genetic ancestry ^[9][10].

The lecture illustrates PRS utility with a graph of odds ratios by polygenic-risk decile across three samples (Denmark, MGS, Sweden) ^[1]:

• Individuals in the lowest decile (decile 1) have an odds ratio of ~1 (baseline)
• Odds ratios rise progressively with higher deciles
• In the highest decile (decile 10), odds ratios reach approximately 8 (Denmark), 15 (Sweden), and 20 (MGS)

This shows relative odds between extreme groups in those research samples, with wide and overlapping distributions. It does not mean a person in the top decile has a 20-fold absolute probability, nor does it diagnose schizophrenia. Even a strong top-versus-bottom odds ratio may have limited positive predictive value for a condition with low population prevalence.

Chromosome 22q11.2 deletion syndrome (historically “velocardiofacial/DiGeorge syndrome”):

• Copy number variant (CNV)
• Frequency ~1/4000 live births
• 22-31% develop SZ or schizoaffective disorder
• 1.1% of SZ patients have the deletion ^[1]

The preferred umbrella name is 22q11.2 deletion syndrome (22q11.2DS). The phenotype is highly variable and may include congenital cardiac disease, palatal abnormalities, hypocalcaemia from hypoparathyroidism, immune dysfunction, developmental/learning difficulties, autism-spectrum features, and later psychotic illness. GeneReviews estimates schizophrenia in about 25% of affected adults ^[11]. The facial photographs on the slide illustrate variability and must never be used to diagnose the deletion; clinical suspicion should lead to appropriate genetic testing, counselling, and multisystem surveillance.

The adjacent rare-CNV plot makes a second point: some deletions or duplications have much larger relative effects than common GWAS variants (for example 22q11.2 deletion, NRXN1 deletion, 15q13.3 deletion, 3q29 deletion, or 16p11.2 duplication in the displayed dataset). They remain rare, pleiotropic, and incompletely penetrant, so a carrier's outcome cannot be predicted from an odds ratio alone.

Risk CNVs implicate synaptic proteins:

• CNVs that do not overlap exons: no significant enrichment
• Enrichment depends on how many genes CNVs affect (OR 1.21 per gene)
• Copy number loss (OR 1.40) more enriched than copy number gains (OR 1.12)

Significant enrichment in:

• Presynaptic adhesion molecules: NRXN1, NRXN2
• Post-synaptic scaffolding proteins: DLG1, DLG2, DLGAP1, SHANK1, SHANK2
• Glutamatergic ionotropic receptors: GRID1, GRID2, GRIN1, GRIA4
• Dystrophin and synaptic partners: DMD, DTNB, SNTB1, UTRN ^[1]

Interpretation: The displayed enrichment converges on synaptic structure and function and motivates a synaptic-development hypothesis. It is a group-level pathway signal, not a molecular diagnosis for an individual.

This is an elegant evolutionary genetics argument:

Reduced fecundity in schizophrenia:

• Males FR = 0.23 (95% CI 0.23-0.24)
• Females FR = 0.47 (95% CI 0.46-0.48) (Power et al, JAMA 2013 — Swedish cohort of 2.3M) ^[1]

The puzzle: If schizophrenia has such high heritability (~80%) and reduces reproductive fitness so dramatically, why hasn't natural selection eliminated the risk alleles?

Possible contributors:

1. De novo mutations continuously replenish high-risk alleles before selection can eliminate them
2. Rate of de novo mutations positively associated with paternal age ^[1]
3. Increased risk of schizophrenia in individuals born to older fathers ^[1]
4. Alternative explanation: genetic predisposition associated with later fatherhood (reverse causation — people with schizophrenia-risk genes may have social difficulties that delay fatherhood) ^[1]

The plot's mutation count rises with paternal age (about 70 de novo mutations around age 33 in the displayed data), but an epidemiological paternal-age association is not proof that new mutations caused a given person's illness. Confounding and inherited liability can contribute, the absolute risk for any family remains small, and this evidence should not be used to blame parents.

22q11.2 deletion syndrome associated with intellectual disability, autism AND psychosis ^[1]

Twin studies showed shared genetic aetiology between:

• Schizophrenia, mania and schizoaffective disorder (Cardno et al, 2002)
• Bipolar affective disorder and unipolar depression (McGuffin et al, 2003)
• Anxiety and depression (Kendler et al, 1992) ^[1]

The lecture illustrates shared genetic aetiology using a bivariate twin model diagram ^[1]. For two traits (X and Y) measured in twins, variance is decomposed into:

• A (additive genetic), C (shared environment), E (non-shared environment)
• Cross-trait correlations r_A, r_C, and r_E quantify the degree to which the same genetic, shared-environmental, and non-shared-environmental factors influence both traits
• A high r_A (genetic correlation) between two disorders indicates substantial shared genetic aetiology — this is the statistical basis for the twin-study findings listed above

The diagram's arrows from latent A, C, and E factors to traits X and Y represent variance components; the curved links rA, rC, and rE represent cross-trait correlations. A genetic correlation describes overlap at population level. It does not mean two diagnoses are identical or predict that a person with one disorder will develop the other.

GWAS confirm genetic overlap:

• Substantial shared SNP effects across psychiatric disorders (coheritability)
• Shared genetic aetiology partially explains excess comorbidity
• Challenge to the Kraeplinian dichotomy (the traditional distinction between schizophrenia and bipolar disorder)
• Common variants do not entirely account for heritability estimated from twin studies ("missing heritability") ^[1]

“Missing heritability” refers to the gap between classical family/twin estimates and variance captured by measured variants in a given analysis. Rare variants, incompletely tagged common variants, structural variation, phenotype and measurement differences, and modelling assumptions may contribute. Gene–environment interaction can affect phenotypic variance but should not simply be relabelled as unexplained genetic variance; epigenetic findings likewise require separate causal interpretation.

Schizophrenia associated with obstetric history:

• Birth in winter months / urban areas
• Fetal growth retardation
• Fetal perinatal hypoxia
• Prenatal adversity:
  • Malnutrition: 1944 Dutch famine, and 1958 Chinese famine
  • Infections: 1957 influenza pandemic

Proposed disturbance of development that may increase later psychiatric risk ^[1]

Interpretation: season of birth, urbanicity, fetal growth restriction, hypoxia, famine cohorts, and historical influenza exposure are mostly observational, modest, and non-specific associations. Proposed mechanisms include infection/inflammation, nutrition, vitamin D, placental function, and hypoxic injury, but residual socioeconomic, temporal, and healthcare confounding remains. Do not turn these population signals into a causal story—or blame—about an individual pregnancy.

Why urban birth? Possibly related to higher rates of infection, overcrowding, pollution, and social stress.

Males more prone to intellectual disability, autism, conduct disorder, and substance use disorder Females more prone to childhood emotional disorder, anxiety and depression Age-of-onset of schizophrenia earlier in males than females ^[1]

The lecture includes National Comorbidity Survey (USA) data illustrating gender differences ^[1]:

Key pattern in this historical US binary dataset: overall lifetime prevalence was similar, while depressive/anxiety diagnoses were more common among women and substance-use diagnoses among men; average schizophrenia onset was earlier among men. The slide labels this “gender,” but sex-related biology and gendered exposure, help-seeking, detection, and diagnostic bias are distinct and may interact. Binary averages do not describe every individual and may miss or misclassify transgender and gender-diverse people. They also do not establish a single hormonal mechanism; the oestrogen hypothesis is one proposed contributor, not a complete explanation.

Risk factors:

• Parental loss (death / divorce / separation)
• Maltreatment (Physical / Emotional)
• Sexual abuse

Associated disorders:

• Emotional disorders, Conduct disorders (Delinquency)
• Anxiety and depression, PTSD
• Substance use disorders
• Borderline personality disorder
• Schizophrenia ^[1]

The image is sometimes described as “Bowlby: Attachment,” but it actually depicts Harry Harlow's rhesus-monkey surrogate-mother experiments (cloth versus wire surrogate). These ethically unacceptable animal experiments influenced attachment thinking but are not Bowlby's clinical observations and cannot be directly translated into a deterministic human diagnosis. In people, maltreatment and disrupted caregiving are associated with later mental-health risk, but outcomes vary and safe, stable, supportive relationships are protective ^[12].

Early adversity has wide-ranging effects (e.g. on stress response and neuroplasticity), partially mediated by epigenetic reprogramming, which increases psychiatric disorder risk ^[1]

The proposed mechanism (from the Li et al. 2020 schematic) ^[1]: early adversity may alter HPA-axis regulation, monoamine/BDNF signalling, and plasticity through interacting endocrine, neural, immune, and epigenetic processes. Much of the directional pathway detail comes from animal models or observational human tissue studies. Peripheral methylation is not direct evidence of the same change in living human brain; epigenetic marks are dynamic, and these data do not prove biological destiny or transgenerational inheritance.

The lecture diagram details two parallel pathways affected by early life stress:

HPA Axis Pathway:

• Hypothalamic CRH/AVP signal to pituitary POMC/ACTH, which stimulates adrenal glucocorticoid production; glucocorticoid-receptor feedback regulates the circuit.
• The figure proposes altered GR feedback after early stress. Direction and timing vary across conditions; chronic stress can be associated with hyper- or hypocortisolism, not invariably raised cortisol.

Serotonin / BDNF / Neuroplasticity Pathway:

• The diagram links serotonin transport/metabolism (5-HTT, MAO, 5-HT) with BDNF–TrkB signalling and downstream AKT/CaMK/ERK pathways involved in plasticity.
• These arrows summarise candidate mechanisms; they are not a validated patient-level biomarker panel.

Epigenetic example: studies have reported associations between early adversity and methylation near NR3C1, a glucocorticoid-receptor gene, alongside animal work on caregiving and stress responses. These are important mechanistic clues but do not establish a universal, irreversible pathway from maltreatment to adult illness.

Chronic unremitting stress may dysregulate HPA, neural, immune, and metabolic function ^[1]

Immigrants:

• 2-4 fold increased rates of schizophrenia
• Greater effect in second-generation migrants
• Greater effect in areas with lower density of migrants
• Social exclusion ^[1]

Interpretation: higher psychosis incidence in some migrant and minoritised populations—often stronger in later generations and lower own-group ethnic density—cannot be attributed to migrant status as an intrinsic trait. Social exclusion is one hypothesis; discrimination and structural racism, socioeconomic disadvantage, trauma, neighbourhood context, service access, and diagnostic processes are also relevant. Dopamine “sensitisation” remains a proposed mechanism rather than a settled single explanation.

Recent adverse life events (e.g. bereavement, divorce, injury or illness, loss of employment or wealth, failure or humiliation) may precipitate episodes in vulnerable people ^[1]

Temporal association does not make an event necessary or sufficient: reverse causation (prodromal illness generating stress), recall, and selection can contribute. The “six-fold” figure in the senior note should therefore be treated as a source-specific historical estimate rather than a portable individual risk.

Ability to cope with stress and recover from stressful situations Increased by repeated brief parental separations / appropriate challenges during childhood Partly mediated by greater self-efficacy and coping skills Individual differences in optimal level of challenges ^[1]

The “brief parental separations” statement reflects stress-inoculation research, much of it animal or experimental, and is not a parenting or clinical recommendation. Children should never be deliberately separated or exposed to adversity to “build resilience.” In human care, resilience is dynamic and contextual: manageable age-appropriate challenges with safe, stable relationships and support may help coping and self-efficacy, while maltreatment is harmful ^[12]. Difficulty recovering is not a personal failure.

Good social support:

• Protective against the onset of mental disorders
• Helps maintain remission from illness

High expressed emotions (EE) in family:

• Hostility, over-involvement, critical comments
• Increase risk of relapse of schizophrenia ^[1]

Expressed emotion (EE) is a research construct commonly assessed through criticism, hostility, and emotional over-involvement. Higher EE predicts relapse at group level, but the relationship is bidirectional and must not be used to blame families coping with severe illness. NICE recommends collaborative family intervention for families in close contact with a person with psychosis or schizophrenia; its purpose is support, education, coping, negotiated problem-solving, and relapse reduction—not assigning fault ^[13].

Prolonged and heavy usage of psychoactive substances increase mental disorder risk:

• Alcohol: Psychosis, Dementia
• Amphetamines: Psychosis
• Cannabis: Psychosis ^[1]

Distinguish intoxication or withdrawal, a substance-induced psychiatric disorder, and a primary disorder whose course is worsened by substance use. Amphetamines and cannabis can precipitate psychotic symptoms; alcohol can cause intoxication/withdrawal syndromes, alcohol-induced psychosis, and neurocognitive harm. Associations are also bidirectional because symptoms can influence use. For cannabis, the EU-GEI case-control study found greater psychotic-disorder odds with daily and high-potency exposure ^[14], but an observational odds ratio is not deterministic proof for an individual. Ask non-judgmentally about timing, quantity/frequency, potency, other substances, dependence, and change in symptoms during abstinence.