RESEARCH DOSSIER
Comorbidity and Transdiagnostic Psychopathology Liability
Abstract. Comorbidity is the occurrence of more than one clinically defined disorder in the same individual across a specified observation period. Transdiagnostic psychopathology liability refers to the relatively general distribution of vulnerability, symptom covariance, developmental continuity, and shared risk processes that cross established diagnostic boundaries. Their scientific relationship poses a central question for psychiatric classification: why do apparently different disorders repeatedly occur together, and which portions of that co-occurrence arise from shared causes, overlapping definitions, reciprocal symptom processes, exposure histories, or patterns of ascertainment? This fundamental research monograph integrates the history of psychiatric comorbidity, population epidemiology, developmental longitudinal studies, phenotypic and genomic factor analysis, dimensional nosology, symptom network theory, causal inference, and clinical intervention research. A critical narrative synthesis of original studies and methodological reviews, rather than a protocol-driven systematic review or meta-analysis, identifies robust evidence for extensive cross-disorder association and shared genetic liability alongside unresolved questions concerning the causal interpretation, temporal stability, cultural transportability, and individual-level utility of a single general psychopathology factor. The monograph proposes a disciplined evidentiary framework that distinguishes statistical summaries, mechanistic hypotheses, risk prediction, and treatment decision-making. It concludes that comprehending psychiatric comorbidity requires coordinated measurement of diagnostic syndromes, dimensional processes, developmental exposures, functioning, and time-dependent clinical trajectories.
Keywords: Comorbidity; and; Transdiagnostic; Psychopathology; Liability
Introduction: The Research Object
Psychiatric diagnoses rarely exist in conceptual or clinical isolation. A person may experience persistent anxiety and depressive episodes, a developmental disorder alongside compulsive symptoms, or traumatic stress accompanied by problematic substance use. These combinations pose practical difficulties for assessment, treatment selection, prognosis, and interpretation of research. At the level of populations, the distribution of such combinations also challenges the expectation that each diagnostic label corresponds to an independent disease process. The scientific object of this dossier is the patterned co-occurrence of disorders and the cross-cutting liability structures that may help explain that pattern.
The problem is older and broader than contemporary discussions of the general psychopathology factor, often abbreviated p. Shared symptoms can arise because two categories overlap in their definitions; common exposures can increase the probability of several outcomes; one disorder can alter the future risk of another; a disposition can influence multiple pathways; and research designs can create or amplify apparent associations. The same observable diagnosis pair can therefore have several explanations. The task of a fundamental synthesis is to establish the distinct kinds of evidence required to adjudicate among them, rather than to assume that statistical covariance is itself a causal finding.
The relevant literatures include population surveys, registries, birth cohorts, family and twin studies, genome-wide association studies, psychiatric interviews, symptom-level assessments, factor analytic models, symptom networks, and randomized treatments. The editorial boundary follows the scientific object: specific anxiety, mood, psychotic, developmental, and substance-related disorders are discussed where their overlap illuminates general questions. Their comprehensive syndrome-specific treatment belongs to neighboring clinical dossiers. The primary disciplinary home is Clinical Psychology and Psychopathology, with direct connections to personality psychology and mathematical and computational psychology.
A central evaluative premise governs this synthesis. Evidence that disorders covary is evidence of covariation. Evidence of a reproducible higher-order statistical factor establishes a model-based description of shared variance under specified assumptions. Evidence that a process causes several disorders requires temporal identification, experimental manipulation when feasible, or sufficiently justified quasi-experimental inference. This distinction is indispensable when a convenient aggregate score risks being mistaken for a unitary biological disease.
Definitions, Boundaries, and Levels of Explanation
Comorbidity conventionally indicates that two or more disorders coexist within a defined person and time window. Lifetime comorbidity means that criteria for multiple disorders have been met at some points across life; the disorders need not have occurred simultaneously. Twelve-month or current comorbidity imposes a narrower window, while concurrent episode overlap requires additional temporal evidence. These alternatives measure different things. A published percentage or association is interpretable only when the reference population, diagnostic criteria, interview method, observation interval, and weighting are known.
Transdiagnostic liability is a construct representing predispositions and processes relevant to more than one category of psychopathology. A liability can be measured at several levels: a statistical factor extracted from diagnostic covariance, a continuous temperament trait, a shared environmental exposure, a polygenic influence, or a dynamic process such as perseverative worry and avoidance. The word liability denotes susceptibility under conditions and over time, not destiny, and it acquires scientific meaning only through an explicit operational definition. A factor score is one estimate; a causal exposure or biological mechanism requires a different design.
Diagnostic overlap is particularly important because some psychiatric criteria share symptom content, severity thresholds, duration rules, and impairment requirements. A person might satisfy multiple categories through a relatively small set of overlapping complaints. By contrast, true clinical multimorbidity can involve distinct symptom mechanisms, separate courses, and different intervention needs. A third issue is co-occurrence at the level of dimensions: high negative affect, disinhibition, and intrusive cognition may cluster without satisfying any diagnostic threshold. The three should be analyzed together while retaining their different denominators and inferential uses.
The term transdiagnostic has at least four meanings in this field: processes occurring across diagnoses, dimensional constructs spanning diagnoses, interventions deliberately addressing more than one disorder, and designs sampling across categories. These meanings often overlap but are not interchangeable. For example, a treatment can work for several disorders without showing that all disorders share one etiology; a common genetic component can underlie several disorders without implying the same symptom trajectory in every individual. The central objects are thus diagnosis combinations, dimensions of vulnerability, and processes that connect them.
Historical Development of the Comorbidity Problem
The history begins with the longstanding observation that psychiatric symptoms do not distribute neatly among a small number of exclusive clinical kinds. The development of increasingly explicit diagnostic categories made it possible to describe and count co-occurrence, but it also made overlap more visible. In child and adolescent research, Achenbach, T. M. (1966) applied factor analysis to psychiatric symptoms, anticipating a later shift from categories toward empirical dimensions. Internalizing and externalizing groupings subsequently became central to developmental psychopathology and child assessment.
The word comorbidity was given a particularly influential methodological definition by Feinstein, A. R. (1970)90054-8) in medicine. Its psychiatric adoption raised questions about which condition should count as an index disorder, how additional conditions should be measured, and what a joint diagnosis actually signifies. Caron, C., & Rutter, M. (1991) organized conceptual and research strategies around comorbidity in child psychopathology, recognizing the roles of symptom overlap, classification, referral pathways, and common causal influences. This remains a useful warning against treating every observed association as evidence for an additional independent disease.
A major epidemiological transition came with representative diagnostic surveys. Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M., E (1994) reported National Comorbidity Survey estimates based on structured assessment in the United States. Their work helped establish that joint occurrence was a central quantitative feature of population psychopathology. Rather than being a rare exception confined to specialty clinics, associations between psychiatric conditions became one of the main phenomena that any classification system needed to explain.
The next stage sought a structural account of the correlations. Krueger, R. F. (1999) showed that common disorders could be modeled in terms of broad internalizing and externalizing dimensions. In the twenty-first century, research expanded to a general psychopathology factor, hierarchical dimensional systems, network theory, genetic covariance, and experimental process-based interventions. Each development answered a different part of the historical question: whether disorders group together, how they group, what produces those groupings, and whether the information can guide care.
What Counts as Co-Occurrence?
In clinical records, two diagnostic codes may represent two independently assessed disorders, a provisional diagnosis and its later replacement, a syndrome and an underlying medical condition, or an administrative choice made for reimbursement. A source that counts diagnostic codes cannot automatically establish two confirmed disorders. An interview administered by trained investigators may better standardize criteria, but recall, cultural meaning, informant differences, exclusion rules, and the age of respondents remain important.
Time is equally consequential. If an individual experiences childhood attention difficulties, adolescent anxiety, and adult depression, lifetime joint diagnoses can be established without an episode of simultaneous symptoms. This pattern may reflect a general developmental liability, a sequence of reciprocal risk, or independent exposures occurring at different life stages. Conversely, two diagnoses recorded during one hospitalization can reflect shared acute stressors, assessment intensity, or intertwined symptoms. “Comorbid” should therefore be accompanied by a temporal statement whenever mechanisms or treatment choices are being discussed.
The distinction also governs comparisons across studies. Studies may report odds ratios for cross-sectional joint diagnosis, relative hazards of a later diagnosis given an earlier one, correlations among dimensional scores, or genetic correlations derived from population association statistics. These estimates are not measures of the same quantity. Even with consistent criteria, different age ranges and observation periods alter opportunities for a second disorder to emerge, while selection into treatment settings changes the visible distribution of disorder combinations.
Diagnostic Classification and Construct Overlap
The DSM and ICD systems organize clinically significant experiences into categories using combinations of symptoms, duration, exclusions, and functional consequences. They support communication, research eligibility, health statistics, and care coordination. Their categories also contain shared symptoms and symptoms that vary continuously across diagnostic boundaries. This combination ensures that diagnostic co-occurrence partly reflects the architecture of the classification system itself.
Anxiety and depression provide a familiar illustration: impaired concentration, disturbed sleep, fatigue, negative expectations, and physical agitation may contribute to more than one clinical description, while persistent apprehension and loss of interest remain distinguishable phenomena. In compulsive disorders, repetitive behavior may be driven by fear reduction or by other goals; in psychosis and mood disorders, the temporal relationship between mood episodes and psychotic features can determine classification. Substance-induced symptoms, medication effects, medical conditions, and developmental context can further complicate category assignment. Exact criteria should be checked against the operative editions of WHO ICD-11 and DSM-5-TR updates.
A cross-diagnostic factor estimated from these categories inherits some of their measurement decisions. A particular interview might permit multiple simultaneous diagnoses, apply hierarchical exclusions, or use categories designed for a specific age range. The resulting covariance matrix can change accordingly. A convincing research program therefore triangulates diagnostic models against item-level symptom measurements, independent informants, functional outcomes, and longitudinal observations instead of treating categories as measurement-free access to independent natural kinds.
Conditional Risk and the Base-Rate Problem
The prevalence of two disorders influences how often they will be observed together even if they were statistically independent. If A and B were independent within a specified population and period, their expected joint prevalence would equal the product of their individual prevalences. The difference between observed and expected overlap, together with confidence intervals and adjustment for measurement and confounding, is more informative than the raw number of comorbid cases. Associations are then estimated with conditional probabilities, odds ratios, risk ratios, hazards, or correlations chosen according to the design.
This elementary reasoning matters because a common diagnosis will appear alongside many other conditions simply by virtue of frequency. Elevated odds of a second disorder may also conceal heterogeneity across age, sex, cohort, exposure, access to care, and severity. Relative associations can be substantial while absolute risk remains modest in a low-base-rate population; conversely, a small relative elevation can carry considerable population burden when exposure and outcome are common.
For scientific and clinical interpretation, risk statistics need their denominator and time scale. A hazard ratio from a national registry represents an association in event rates under the registry's definitions and models. It should not be read as a patient's personal probability, an experimentally established causal effect, or a universal number transferable between health systems. The work of Plana-Ripoll, O., Pedersen, C. B., Holtz, Y., Benros, M. E., Dalsgaard, (2019) provides a model of how large-scale cross-disorder associations can be mapped while preserving the interpretive importance of age, sex, timing, and study design.
Research Questions and Evidence Strategy
The research is organized around six linked questions. Which psychiatric conditions co-occur more frequently than their base rates predict? Which dimensional models describe that covariance across cohorts, age ranges, and measurement systems? To what extent do genetic and environmental influences operate across diagnostic categories? How does liability change or persist across development? Which methodological artifacts can produce seemingly general factors? And what evidence connects cross-diagnostic models to outcomes meaningful for people receiving care?
The evidentiary hierarchy is question-dependent. Representative surveys are suitable for estimating population patterns; longitudinal cohorts address time-ordered incidence and diagnostic transitions; genetic studies estimate molecular association and shared inheritance under particular assumptions; experiments and randomized trials test manipulation of selected processes or treatments; and mathematical comparisons evaluate the fit, identifiability, and predictive value of candidate structures. None of these designs independently resolves every causal question. Convergence across distinct methods increases credibility when the methods have sufficiently different biases.
This is a structured critical narrative review anchored to identified original studies, authoritative methodological work, and official diagnostic sources through November 2026. Searches and source checks were conducted for the central concepts and landmark studies, with deliberate inclusion of critical contemporary assessments. There was no preregistered exhaustive search, dual-screened selection procedure, risk-of-bias table for all eligible studies, quantitative pooling, or new human-participant data collection. The resulting synthesis is therefore judged by traceability, argumentative completeness, and accurate distinction between evidence classes, rather than by a claim of systematic-review coverage.
Internalizing and Externalizing Spectra
Internalizing and externalizing dimensions emerged as a practical way to explain why particular diagnostic groups share substantial variance. Internalizing generally encompasses fear, anxious-misery, depressive distress, and other forms of inwardly experienced dysregulation. Externalizing encompasses behavioral disinhibition, impulsive rule violation, and certain substance-related difficulties. These are empirical organizations of symptoms and syndromes: their exact indicators depend on the sample, the set of disorders included, the scoring system, and the age range investigated.
In Krueger, R. F. (1999), the joint distribution of common disorders could be organized through broad dimensions, with internalizing subdivided into correlated but distinguishable components. Watson, D. (2005) subsequently argued for a quantitative hierarchical model of mood and anxiety disorders grounded in their empirical relatedness. Such findings established that diagnostic co-occurrence has structure. They also encouraged the study of continuous severity distributions instead of an exclusive focus on whether people fall above or below multiple thresholds.
The spectra perform useful descriptive work because a continuous score can retain information that disappears when several related symptoms are each dichotomized. A high internalizing score can indicate a broad concentration of distress and anxiety-related vulnerability even when the precise diagnostic combination shifts over time. Nevertheless, a statistical spectrum is a measured construct, and a person at the same estimated location on that spectrum may have different symptoms, functional impairments, causes, and care needs. Factor scores summarize patterns of variation; they do not eliminate meaningful variation within their indicators.
Some disorders also participate in more than one broad domain. Attention-related problems, irritability, eating pathology, and substance use may load differently across samples and model specifications. Greater dimensional coverage can increase ecological validity while making the model more sensitive to choice of indicators. The question is whether a classification retains distinct information about prognosis, mechanisms, treatment response, and functioning beyond broad factor scores. The capacity to distinguish cross-domain overlap from domain-specific pathways is therefore a primary criterion for evaluating the internalizing–externalizing model.
General Psychopathology and the p Factor
The general psychopathology factor, conventionally represented by p, is a statistical dimension intended to summarize the positive correlations among diverse forms of psychopathology. Lahey, B. B., Applegate, B., Hakes, J. K., Zald, D. H., Hariri, A. R., & (2012) investigated a general factor in prevalent adult psychopathology, while Caspi, A., Houts, R. M., Belsky, D. W., Goldman-Mellor, S. J., Harringto (2014) examined the possibility of a general factor in the structure of psychiatric disorders in a longitudinal cohort. The concept reflects a recurring empirical observation: individuals high on one cluster of symptoms tend, on average, to be at elevated risk for others.
There are several ways to estimate p. A higher-order model posits broad dimensions such as internalizing and externalizing with a higher general factor accounting for their correlations. A bifactor model places the general factor and domain-specific factors as simultaneous predictors of observed indicators, often under orthogonality constraints. A one-factor model ignores specialized clustering; an exploratory or network approach offers still other summaries. These models can yield different general-factor scores and different meanings even when fitted to the same data. The term p is therefore a family of statistical operationalizations rather than a single laboratory measurement.
Smith, G. T., Atkinson, E. A., Davis, H. A., Riley, E. N., & Oltmanns, J (2020) reviewed interpretations of the general factor and its possible relations to severity, functional impairment, cognitive and behavioral regulation, and broad vulnerability. Explanatory proposals are valuable because they make predictions, but no single proposal follows directly from obtaining a significant general-factor loading. A general factor could reflect shared causes, common symptoms, shared response tendencies, unequal item reliability, the overall burden of mental-health problems, or a combination of these.
One consequence is particularly consequential for communication with non-specialists: p is not a diagnosis in the DSM or ICD. It is not a biomarker that can be used as a stand-alone clinical test. It is also not a demonstrated single source of all mental illness. Its legitimate current uses are to describe common variance, compare models, generate hypotheses, and potentially improve prediction when tested against meaningful alternatives. The question of causal interpretation remains open and scientifically productive.
Bifactor Modeling and Its Interpretive Problems
Bifactor models are attractive because they apparently separate shared psychopathology from domain-specific variation. In a conventional specification, each indicator loads on a general factor and one designated group factor; the factors are usually constrained to be uncorrelated. This makes it possible to ask whether a domain has reliable residual coherence after broad covariance is removed. However, the arrangement also changes the meaning of the group factors: they are residual constructs conditional on the general dimension, not necessarily the ordinary internalizing or externalizing traits one would measure without that adjustment.
The statistical flexibility of bifactor models creates an interpretive hazard. A model with many freely estimated parameters may fit observed covariances well even when its theoretical interpretation is weak. An estimated general factor may be dominated by a particularly prevalent or severe syndrome, while domain-specific factors may become unstable, weak, or difficult to interpret. Factor-score determinacy, reliability, identification, convergence, and out-of-sample replication must therefore be examined rather than inferred from one favorable global fit index.
Watts, A. L., Greene, A. L., Bonifay, W., & Fried, E. I. (2024) identify five interconnected challenges in the p literature: confusing statistical fit with construct validity; evaluating insufficiently specified theories; weak replication of some purported general factors; violations of latent-variable assumptions; and the reification of statistical factors into causal entities. Their critique should be taken as a substantive methodological intervention, because any model claiming a general source of psychopathology must show stable measurement and make predictions that discriminate it from competing explanations.
An informed appraisal therefore reports more than comparative fit. It asks which indicators define the factor, how strongly loadings vary across samples, whether general-factor scores retain meaning across populations and time, whether correlations with external variables were evaluated using non-overlapping measures, and whether conclusions persist under alternative reasonable model specifications. Correlated-factor, hierarchical, bifactor, and symptom network specifications should be compared on their actual empirical and theoretical merits. A general factor with an appealing name is not a substitute for the work of construct validation.
Hierarchical Taxonomy of Psychopathology
The Hierarchical Taxonomy of Psychopathology, HiTOP, organizes mental-health phenomena through nested empirical dimensions, ranging from specific symptom components and syndromes through broader spectra. It was proposed as a research-oriented dimensional alternative to the limitations of rigid categorical nosologies. Kotov, R., Krueger, R. F., Watson, D., Achenbach, T. M., Althoff, R. R., (2017) presented a consensual hierarchical framework and argued that symptom heterogeneity, diagnostic instability, and frequent comorbidity are partly predictable consequences of imposing categories on correlated dimensions.
HiTOP has a different emphasis from treating p as the primary endpoint. A hierarchical taxonomy preserves information at multiple resolutions, making it possible to examine broad internalizing burden, narrower fear or distress factors, and distinct symptoms within the same framework. The value of a hierarchy depends on whether these levels predict outcomes, capture reliable differences, and support communication across clinical settings. A broad score may efficiently summarize population variation, whereas a narrower level may better identify mechanisms or intervention targets.
The Kotov, R., Krueger, R. F., Watson, D., Cicero, D. C., Conway, C. C., DeY (2021) review of HiTOP development discussed its integration with clinical research and the challenges of implementing a dimensional approach. An empirically estimated taxonomy needs reliable instruments, normed interpretations, functioning measures, clinician usability, and tests in representative populations. Its validity also depends on how symptoms are sampled: an inventory concentrated on mood and anxiety cannot establish the full organization of psychosis, developmental symptoms, eating pathology, and substance-related disorders.
Dimensional taxonomy and categorical diagnosis can coexist in clinical work. Categories remain operationally relevant for eligibility, documentation, administrative systems, and treatment evidence based on syndromic trials. Dimensional scores can characterize cross-cutting severity and atypical combinations. The scientifically productive question is whether adding a hierarchy increases diagnostic precision, temporal prediction, and treatment planning beyond categories and patient-reported symptoms. Formal taxonomy is justified by those gains, not merely by elegance of mathematical structure.
Research Domain Criteria and Process Dimensions
The National Institute of Mental Health introduced the Research Domain Criteria, or RDoC, as a framework for investigating mental-health processes across conventional diagnostic boundaries. Insel, T., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D. S., Quinn, K (2010) described the programmatic move toward constructs anchored in behavioral systems and multiple units of analysis. The contemporary NIMH framework includes domains concerning negative valence, positive valence, cognitive systems, social processes, arousal and regulatory systems, and sensorimotor processes, with associated constructs and measurement approaches.
RDoC and a statistical model of co-occurring diagnoses ask different questions. RDoC starts with candidate functional systems and investigates their links to clinical manifestations; factor models start with observed covariance and seek a parsimonious structure. A process such as threat sensitivity might contribute to several diagnoses, but its relationship to symptoms, environmental context, and neural activity must be empirically established. Biological recordings cannot be treated as privileged causal proof when the associated processes remain underdetermined or their clinical relevance has not been demonstrated.
RDoC invites research designs that measure behavior, physiology, neural function, and subjective experience along continuous dimensions while sampling people across diagnostic boundaries. Its strengths include explicit mechanistic hypotheses and the capacity to investigate intermediate processes. Its vulnerabilities include uncertain correspondence between laboratory tasks and real-world distress, high measurement noise, and possible overgeneralization from narrow research samples. Functional systems may also vary by developmental stage and context.
The most productive relationship between RDoC and HiTOP is a demanding one: structural maps propose what covaries, while process research asks why particular covariations occur. A viable bridge requires repeated measurement, discriminant predictions, reliable tasks, and prospective tests linking the proposed mechanism to specific symptom trajectories. Agreement between a dimension and a biological correlate is a starting point for such work rather than the endpoint of explanation.
Symptom Networks and Reciprocal Processes
Network theory offers an account of psychopathology in terms of interacting symptoms and processes. Borsboom, D. (2017) proposed that mental disorders can arise from systems in which symptoms influence one another. For example, insomnia may worsen concentration, fatigue may reduce activity, withdrawal may weaken social support, and loss of rewarding experiences may increase depressed mood. Under this account, the same pattern of covariance can develop through mutually reinforcing processes without requiring one latent disease variable to cause every symptom.
Cross-sectional symptom networks usually estimate conditional associations between variables. Those edges should not automatically be read as directional causal connections. Measurement error, unmeasured common causes, differences in symptom frequency, item redundancy, and sample selection affect estimated network structure. A strong cross-sectional edge between sleep disturbance and worry is compatible with several causal stories, including reciprocal influence, shared stress exposure, and overlap in reporting. Directed intervention studies and intensive longitudinal data provide more relevant evidence about temporal processes.
Fried, E. I., & Cramer, A. O. J. (2017) identified methodological challenges and directions for psychopathological networks, including replication, measurement, and interpretation. Individual-level time-series designs add a further consideration: an association estimated across people need not describe the dynamics operating within a particular person's daily life. Population networks, between-person networks, and within-person networks must be separated in analysis and interpretation.
The network and latent-factor approaches should be treated as empirically testable families of models. A latent liability can influence several symptoms that subsequently reinforce one another; symptom interactions can also generate positive correlations that are summarized by a general factor. These accounts are compatible in some circumstances. The decisive scientific objective is to identify which mechanisms, parameters, and intervention targets best explain changes over time, not to select a model solely for its philosophical appeal.
Comparing Explanatory Models
The range of candidate models can be organized by the question each is best suited to answer. A correlated-factor model asks which symptom or diagnosis groups share variance. A hierarchical model asks whether broader dimensions explain correlations among narrower ones. A bifactor model separates an estimated general factor from constrained residual group factors. A symptom network models conditional dependencies and, with appropriate longitudinal design, potential dynamic interactions. A causal developmental model asks how exposures, traits, and symptoms influence one another over time. A genomic factor model summarizes covariance in genetic association statistics rather than symptoms themselves.
Model comparison should combine in-sample fit with external prediction, replication, stability, and construct interpretation. Better fit alone can reward complexity, while penalized criteria are sensitive to sample size and model specification. Cross-validation can help evaluate predictive performance, but predictive success does not determine the correctness of a causal story. Estimation across independent datasets is valuable when cohort composition, diagnostic interviews, and measurement instruments differ in known ways.
An additional challenge is causal equivalence. Different directed graphs can imply the same observational associations, and statistically equivalent factor parameterizations can support different verbal interpretations. When investigators make a causal claim, they should state the counterfactual or intervention implied by that claim, the assumptions necessary for identification, and the plausible competing structures. A covariance structure is consistent with a hypothesis but ordinarily does not prove it.
Taken together, these considerations support a plural, constrained scientific synthesis. Broad factors provide efficient summaries, narrower dimensions recover meaningful content, longitudinal data establish patterns of sequence, genomic designs distinguish genetic covariation, and interventions test modifiable processes. The ultimate evaluation is pragmatic and empirical: does the model clarify mechanisms, improve valid prediction, guide effective care, and remain robust when researchers change instruments, samples, or analytical assumptions?
Community Epidemiology of Cross-Disorder Associations
Community epidemiology established psychiatric comorbidity as a population phenomenon rather than an artifact restricted to the most severely affected patients in treatment. The original Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M., E (1994) used structured diagnostic interviewing to estimate both individual disorders and combinations across American adults. The survey's historical importance lies in converting clinical impressions into estimates with explicit denominators and sampling procedures. Such surveys were crucial in demonstrating that the statistical relationships among disorders warranted independent scientific explanation.
Later National Comorbidity Survey Replication work examined lifetime and twelve-month distributions of psychiatric diagnoses under a defined set of DSM-IV criteria. Kessler, R. C., Chiu, W. T., Demler, O., Merikangas, K. R., & Walters, E (2005) analyzed twelve-month prevalence, severity, and comorbidity, while a companion analysis addressed lifetime prevalence and age of onset. These designs made it possible to analyze the relation between lifetime vulnerability and current burden. They also revealed why the choice of time window matters: a lifetime history captures multiple developmental opportunities to meet criteria, whereas a twelve-month estimate is more closely tied to contemporaneous service demand.
Cross-national and administrative studies broadened the geographic range of observation, although study populations and health systems remain heterogeneous. Plana-Ripoll, O., Pedersen, C. B., Holtz, Y., Benros, M. E., Dalsgaard, (2019) analyzed Danish nationwide registers involving approximately 5.9 million persons, finding extensive associations across different categories of mental disorders. National registers provide very large observation frames and opportunities for longitudinal linkage; their diagnoses, however, are shaped by treatment contact, coding practices, and service structures. They describe diagnosed population pathways under a particular health system, not an unbiased census of every symptom ever experienced.
The general empirical result is consistent across several methodologies: disorders occur together more often than independence would predict, and the associations form a patterned rather than uniform matrix. Magnitudes differ substantially by diagnostic pair, cohort, age, and analytic definition. It is therefore inappropriate to repeat a single percentage as “the” prevalence of psychiatric comorbidity. A defensible synthesis reports the study's jurisdiction, interval, included disorders, case definition, and outcome metric, while distinguishing survey prevalence from registry incidence and symptom-level dimensional covariance.
Ascertainment, Selection, and Collider Bias
Psychiatric samples are often drawn from clinics where the probability of admission is influenced by severity, functional impairment, the number of complaints, insurance access, service availability, and referral behavior. If two disorders independently increase the probability of entering treatment, conditioning on treatment contact can create or change the association observed between them. This is a form of selection or collider bias. It is particularly relevant when researchers infer general population patterns from tertiary specialty samples.
A related problem is surveillance intensity. A patient already receiving care for one disorder may undergo more clinical assessments and consequently receive a second diagnosis earlier than an otherwise similar person outside care. Conversely, diagnostic overshadowing can prevent clinicians from recognizing another disorder when the first diagnosis dominates the encounter. Registers can thus contain both increased detection due to service contact and missed cases due to narrow clinical focus. Neither process necessarily reflects a change in underlying biological susceptibility.
These problems were anticipated by Caron, C., & Rutter, M. (1991) account of comorbidity research strategies. Their underlying methodological insight is enduring: observed co-occurrence may be a joint product of disease processes, assessment systems, and sample selection. Modern causal diagrams can make this explicit by representing variables that influence both exposure ascertainment and outcome observation.
Possible safeguards include community-based sampling, validation of administrative codes, assessment of detection probability, sensitivity analyses for referral patterns, independent record review, and comparisons between treated and untreated cohorts. Statistical adjustment does not automatically cure selection bias, especially when eligibility depends on unobserved symptoms or downstream variables. Researchers should describe selection mechanisms as part of the scientific interpretation, not merely as a generic limitations paragraph.
Longitudinal Course and Diagnostic Migration
Cross-sectional comorbidity describes what occurs together within a measurement window; longitudinal study asks how a person's diagnostic profile changes. The same individual may cross criteria thresholds for different disorders at different stages, with changing severity and predominant symptom content. Such movement can reflect heterotypic continuity, in which an underlying vulnerability manifests differently with age, development, context, and opportunities. It can also reflect genuine new processes, environmental shocks, or ordinary uncertainty in diagnosis.
Caspi, A., Houts, R. M., Ambler, A., Danese, A., Elliott, M. L., Hariri, (2020) examined psychopathology over four decades in the Dunedin birth cohort. Their longitudinal analysis showed substantial movement among diagnostic categories across development and a high degree of lifetime comorbidity among people meeting criteria for mental disorders. Repeated assessments make an essential contribution here: rather than projecting contemporary categories backward from a single retrospective report, the study observed changing symptoms and diagnoses through multiple waves. The result challenges the expectation that a lifetime psychiatric course is adequately described by one enduring disorder label.
Longitudinal research also distinguishes heterotypic continuity from homotypic continuity. In the latter, the same symptom pattern or disorder recurs or persists; in the former, manifestations change across categories. Both may arise within one life course. The possibility of persistent general susceptibility does not imply that the determinants of each transition are the same. Specific transitions may depend on sleep, substance exposure, interpersonal disruption, developmental stage, treatment, and social resources.
A scientific model of diagnostic migration must specify time and transition rules, account for censoring and missing observations, and distinguish state change from altered interview criteria. Markov and hidden-state models, multistate survival analysis, longitudinal latent-variable models, and trajectory models are useful under different assumptions. When an individual stops satisfying one diagnostic threshold and begins satisfying another, there may be meaningful continuity in symptoms, functional impairment, or liability. The task is to measure that continuity directly instead of reading it into category succession.
Newer national research adds an age- and sex-sensitive view of diagnostic shifts after entry into care. Wandall, E. N. S., Andersen, P. K., Christensen, R. H. B., & Benros, M. (2026) used Danish register linkages to follow 273,400 people after first psychiatric contact across seven broad diagnostic groups and 2.1 million person-years. The investigators observed diagnostic shifts in 66,474 people (24.3%) and multiple shifts in 30,558 (11.8%), with meaningful differences by age, sex, and initial diagnosis. This expands the empirical picture beyond static pairwise overlap and shows that later trajectories depend on both initial clinical classification and demographic context. Because the sample was ascertained through psychiatric contact, its estimates characterize registered service users under Danish coding and care conditions rather than all untreated population psychopathology.
Operational Measurement and Interview Design
The empirical study of comorbidity depends on the reliability and conceptual coverage of measurement. Structured and semi-structured interviews standardize symptom criteria, duration, exclusions, onset information, and impairment. Symptom questionnaires provide economical continuous scores but often measure narrower subsets of clinically relevant domains. Clinician ratings, self-report, caregiver report, and administrative codes capture partially different evidence. Each method can contribute, provided its inferential role is stated.
A common design compares binary diagnosis indicators. This facilitates estimates of joint prevalence and certain classification questions, but dichotomization sacrifices information about severity, subthreshold symptoms, and fluctuations around the cutoff. Item-level dimensional assessment can preserve such information, although it may include redundant indicators or rely on shared response styles. Some symptoms are normatively age-dependent, and some instruments were developed for particular languages or treatment populations. Measurement choice influences both the observed magnitude of comorbidity and the mathematical form of extracted factors.
Psychometric models require explicit assumptions about indicator distributions. Continuous symptom scores may need treatment different from ordinal or binary indicators; robust estimation and appropriate link functions matter. Item response approaches can assess thresholds and discrimination, while longitudinal measurement models investigate whether indicators retain comparable meanings through development. Where respondents are drawn from a range of countries, modes of administration, and translation contexts, calibration requires substantive linguistic and cultural evidence rather than a presumption of equivalence.
A practical measurement architecture for transdiagnostic psychopathology includes the diagnostic interview and several complementary levels: symptom severity by domain, functional impairment, distress, exposure histories, major contextual factors, and repeated observations when temporal mechanisms are relevant. The purpose is to prevent a broad total score from erasing the distinct dimensions of experience that generated it. Transparent item inventories and openly specified scoring rules are particularly important for independent replication.
Reliability, Invariance, and Missing Data
Construct reliability is necessary but insufficient for explanatory validity. A general-factor score may show high internal consistency because items share content, method variance, or a broad severity dimension. That does not establish that the score measures one natural mechanism. Reliability must be evaluated at the level intended for interpretation, including the factor score, domain-specific residuals, and change over time. A model that yields a precise general score while producing unstable specific factors may be useful for one task but inappropriate for another.
Measurement invariance is central to claims about a universal transdiagnostic factor. At a minimum, the researcher must consider whether the same configuration of indicators, comparable loadings, and appropriate thresholds hold across groups or time periods. Tests are sensitive to sample size and model assumptions; partial invariance, differential item functioning, and substantive differences in symptom expression may require alternative interpretations. The validity of comparisons between adolescents and adults, or between culturally distinct populations, is an empirical achievement.
Missingness poses a different threat. Participants with greater impairment may be less likely to attend follow-up, and death, hospitalization, homelessness, migration, and service disengagement can shape what remains observable. Routine complete-case analyses can distort longitudinal associations when the probability of being measured depends on prior symptoms or unmeasured outcomes. Multiple imputation, inverse-probability weighting, and joint modeling may improve inference under explicit assumptions; their performance requires sensitivity analysis when missingness may depend on unobserved values.
Transparent research reporting should provide instrument versions, item inclusion decisions, diagnostic exclusion rules, scoring transformations, attrition patterns, estimators, factor loadings, reliability measures, and alternative model results. Strong conceptual conclusions require reproducible analytical choices. The methodological benchmark for a general liability claim is not that one dataset yields a broad factor, but that the meaning and external predictive performance of the estimated construct remain defensible across appropriately varied and independently assessed settings.
Genomic Factors and Disorder-Specific Signals
Genomic structural equation modeling estimates latent patterns from genetic covariance matrices and offers a way to compare genetic dimensions with phenotypic models. Grotzinger, A. D., Mallard, T. T., Akingbuwa, W. A., Ip, H. F., Adams, M (2022) analyzed shared genetic architecture across psychiatric disorders and reported several broad correlated factors. Such results suggest that pleiotropic structure can have substantial organization at intermediate levels, rather than being adequately summarized by either completely independent disorders or one homogeneous source.
A more extensive study, Grotzinger, A. D., Werme, J., Peyrot, W. J., Frei, O., de Leeuw, C., Bic (2026), analyzed genetic association data spanning fourteen psychiatric disorders and more than one million cases. Published online in December 2025 and in a 2026 journal issue, it identified five genomic factors—compulsive disorders, schizophrenia–bipolar disorder, neurodevelopmental disorders, internalizing disorders, and substance use disorders—accounting for a substantial share of common-variant genetic variation in the modeled traits. The analysis found both broad pleiotropic overlap and distinct factor-level biological enrichments, while retaining disorder-specific genetic variance.
This study demonstrates why the level of aggregation matters. A general dimension can summarize widespread overlap, but intermediate factors can recover differentiated genomic architectures. The reported factors also cannot be directly equated to phenotypic syndrome factors: measured diagnoses, genomic liabilities, and brain-relevant expression patterns occupy different inferential levels and may be affected by different forms of selection and error. Genetic factors are not themselves psychiatric syndromes or personalized diagnoses.
Genome-wide evidence supports a plural account of transdiagnostic liability: many risk variants contribute to several outcomes, often through developmental and biological pathways that are only partly characterized. Polygenic scores may eventually improve population-level risk prediction when developed and validated appropriately, but their performance varies with ancestry, phenotype, and setting. They do not explain all shared liability, identify a clinically sufficient mechanism, or replace direct clinical assessment. External validation and careful evaluation of utility are mandatory before individualized use.
Environmental Exposure and Developmental Risk
Environmental explanations of comorbidity range from discrete exposures to the cumulative effects of enduring contexts. Childhood adversity, social deprivation, unstable housing, discrimination, chronic illness, family stress, interpersonal violence, and exposure to harmful substances may each influence the risk of several outcomes. The same exposure may operate through multiple mediators, including physiological stress responses, learned threat expectations, disrupted support, altered opportunities, and constraints on access to treatment. Exposure histories are therefore candidate causes or modifiers of liability only to the extent that study designs support their temporal and causal interpretation.
Risk factors may be broad or disorder-specific, and both patterns deserve investigation. A shared early adversity exposure can contribute to anxiety, depressive symptoms, and substance-related coping, while a particular traumatic event or developmental insult may increase risk for a narrower symptom constellation. Even when adversity is associated with several diagnoses, a common association does not establish one pathway. Differences in timing, intensity, duration, resources, subsequent exposures, and genetic sensitivity can alter the trajectory. Prospective assessment is preferable to purely retrospective reports when exposure reconstruction is vulnerable to current mood or recall.
Genetic and environmental liability are intertwined. Genetic propensity may influence exposure to some environments; caregiver traits and conditions shape family contexts; environmental exposures may amplify or buffer inherited sensitivity; and the outcome itself may change later environments. These possibilities require designs capable of separating confounding from mediation and selection. Sibling comparisons, adoption and twin approaches, quasi-experiments, natural experiments, and repeated-measures cohorts each address some assumptions but leave others unresolved.
The general conclusion from these literatures is that vulnerability is distributed across the life course and embedded in contexts. An environmental account should specify what is being measured and when, which symptoms or diagnoses follow, and which alternative pathways could explain the association. It should also retain the distinction between population-level risk and a claim about the causes of a particular person's distress. Responsible application avoids deterministic narratives about individuals with adverse histories.
Neurobiological Research and Its Limits
Neurobiological studies investigate whether patterns of brain activity, anatomy, connectivity, endocrine function, inflammation, reward learning, or cognitive control correspond to multiple forms of psychopathology. Cross-disorder association is a legitimate research target: a single system may participate in more than one clinical syndrome, and different syndromes may recruit partially overlapping circuits. However, many brain processes are shared features of ordinary adaptive functioning, and a neural correlate of broadly measured distress need not be a pathology-specific mechanism.
Cross-sectional imaging associations are sensitive to clinical heterogeneity, medication exposure, motion, data preprocessing, analytic flexibility, sampling, and differences in symptom measurement. Findings associated with a general psychopathology factor may reflect overall illness burden, substance exposure, treatment histories, demographic characteristics, or another measured or unmeasured variable. Demonstrating reproducible associations across independent datasets is therefore distinct from determining whether a process is causal or clinically actionable.
DeYoung, C. G., Blain, S. D., Latzman, R. D., Grazioplene, R. G., Haltig (2024) reviewed neurobiological substrate research in relation to hierarchical psychopathology models. The need for a cross-level bridge is fundamental: symptom factors are psychometric abstractions, brain recordings are physical measurements, and mechanistic explanations specify how particular neural processes influence cognition, affect, and behavior under particular conditions. Statistical relationships between these levels can be informative while remaining multiply interpretable.
Greater evidentiary strength would come from longitudinal neural assessments, preregistered multivariate prediction, reliable experimental manipulations, patient-level out-of-sample validation, and measures of how changes in proposed mechanisms precede clinical changes. Such methods can test whether a candidate process is transdiagnostic in a consequential sense. In their absence, a broad brain signature should be described as a candidate correlate or predictor, not an established common cause of psychiatric disorders.
Temperament, Neuroticism, and Negative Affect
Temperament and personality provide an important bridge between psychiatric comorbidity and relatively stable individual differences. Neuroticism, or the broad disposition toward negative emotionality, is associated with elevated reports of anxiety, distress, emotional reactivity, and many internalizing complaints. This pattern makes it a candidate transdiagnostic risk marker. Nevertheless, the construct and psychiatric symptoms overlap in assessment content, creating a problem of criterion contamination when instruments repeatedly ask about similar experiences.
The interpretive question is whether a trait measured independently and early predicts later multi-disorder outcomes beyond baseline symptoms, prior episodes, familial adversity, and relevant demographic or contextual covariates. Prospective designs offer better leverage than cross-sectional associations. A trait may function as a marker of vulnerability, an intermediary through which experiences influence later symptoms, or a dynamic disposition that changes with clinical episodes. Those roles require different causal evidence.
The general-factor literature also discusses possible links with personality organization, self-regulation, and overall functioning. Smith, G. T., Atkinson, E. A., Davis, H. A., Riley, E. N., & Oltmanns, J (2020) summarize how p has been associated with broad correlates and how competing interpretations can yield similar findings. If p and neuroticism share item content or global evaluative variance, their correlation may be less independent than the names suggest. Careful construct measurement is therefore required before attributing p to a fundamental personality process.
Broad traits can be clinically useful when they contextualize stress sensitivity, risk perception, or persistence of distress. They are not a stand-alone diagnosis and should not displace the description of actual symptoms and functional needs. At the individual level, context, acquired skills, social support, physical health, and treatment can change the course of problems despite elevated trait-related vulnerability. Liability must be conceived probabilistically and developmentally.
Emotion Regulation, Avoidance, and Threat Learning
Many anxiety, depressive, obsessive-compulsive, and trauma-related symptoms involve difficulty responding flexibly to aversive experiences. Threat anticipation, repetitive worry, rumination, safety behaviors, suppression, and experiential avoidance can maintain distress when they prevent corrective learning, reduce valued activity, or create recurrent cues for alarm. These are plausible transdiagnostic processes because they appear in several clinical categories, though their functional roles and prevalence differ across disorders.
Avoidance can provide immediate relief while preserving longer-term expectations of threat. Rumination can keep negative content accessible and reduce engagement with alternative activities; suppression may alter how emotional experiences are expressed; and rigid efforts to control internal states may increase perceived failure. Yet the causal status of each process depends on temporal sequencing, context, measurement, and experimental manipulation. The same outward behavior can serve different purposes. A patient who avoids a particular setting because of immediate danger presents a different functional circumstance from someone whose avoidance is maintained by an inaccurately generalized threat expectation.
Research design must distinguish shared process labels from common mechanisms. Two disorders might both involve “avoidance,” but exposure targets, reinforcement histories, cognitive appraisals, and physiological responses may differ. Intensive daily measurement can track when avoidance increases and when symptoms change, whereas randomized behavioral interventions can assess whether modifying the hypothesized process improves outcomes. General process models become more informative when they specify which manifestations, moderators, and time scales they predict.
Transdiagnostic emotion regulation research thus offers a credible level of mechanistic inquiry between broad factors and individual symptoms. It connects patient-experienced problems to manipulable behavior while leaving open the possibility that several distinct mechanisms produce similar clinical appearances. Intervention effects may also arise through nonspecific processes such as social support, expectancy, and behavioral activation. Multiple measurements and comparator treatments help interpret the active ingredients.
Reward, Impulsivity, and Externalizing Pathways
Externalizing comorbidity often involves correlated patterns of behavioral disinhibition, impulsive decision-making, rule-violating behavior, reward sensitivity, and substance-related problems. These processes are developmentally shaped and can interact with social opportunity, exposure to substances, peer networks, and environmental constraints. Broad externalizing scores often summarize robust associations, but each indicator has its own measurement challenges and normative context.
Impulsivity is not one homogeneous trait. Urgency, attentional control difficulties, delay discounting, novelty seeking, and weak response inhibition can produce different behaviors and depend on different tasks. An observational link between a self-reported impulsivity scale and several diagnoses may be driven partly by shared content or common distress. Experimental measures of action cancellation, value updating, or delayed reward can help distinguish candidate mechanisms, provided their reliability and ecological relevance are assessed.
Substance use presents especially complicated causal relations. Repeated intoxication or withdrawal may cause psychiatric symptoms; psychiatric distress can influence substance consumption; shared predispositions may increase risk for both; and local availability or social circumstances may affect exposure. A single externalizing dimension cannot by itself allocate causality among these pathways. Longitudinal study is essential for determining onset order, evaluating potentially reversible exposure effects, and distinguishing primary and substance-induced syndromes.
The broader lesson is that shared liability and syndrome-specific pathways operate on different levels. A person may show elevated externalizing vulnerability yet develop only one condition, while another person may develop multiple conditions through a mixture of shared factors and distinct exposures. A credible explanation must account for both patterns. The goal is to understand conditional developmental trajectories and identify reliable intervention opportunities, not to assign an immutable identity from a factor score.
Trauma, Substance Use, and Bidirectional Relations
Traumatic exposure, stress-related symptoms, substance-related disorders, and depressive or anxiety symptoms can form interrelated clusters. The occurrence of one can precede, exacerbate, or complicate another, but multiple pathways remain plausible. Exposure to traumatic events may increase distress and alter threat processing; substance consumption can change sleep, arousal, cognition, and social safety; and subsequent social disruptions may maintain both. Underlying vulnerabilities may increase the chance of encountering certain risks without reducing responsibility for harmful environments.
Studies that record only whether a person ever experienced trauma, a psychiatric diagnosis, and substance use may fail to identify temporal relations. More informative designs track exposure timing, symptom emergence, duration, quantity and pattern of substance use, remission, relapse, and shifts in social conditions. Sensitivity analyses should consider reverse causation and confounding. Clinical histories can include self-medication narratives, yet such narratives are hypotheses about individual mechanisms that require careful evaluation.
A second important principle is to distinguish the impact of clinical co-occurrence on safety and functioning from an explanation of its cause. Concurrent traumatic stress symptoms and heavy substance use may affect treatment engagement, risks of injury, sleep, adherence, and access to stable care regardless of whether one condition caused the other. A comprehensive assessment addresses both, while decisions about causal mechanisms depend on evidence and context.
The comorbidity literature can thereby inform integrated, coordinated treatment without implying universal treatment uniformity. For some patients, a shared behavioral target is useful; others need an intervention that addresses a specific syndrome or exposure. The presence of multiple diagnoses is a prompt for broad assessment and coordinated planning, rather than an automatic indication to apply an undifferentiated protocol.
Developmental Transitions and Contextual Variation
Psychiatric vulnerability is dynamic across childhood, adolescence, adulthood, and later life. Development changes the environments to which people are exposed, the kinds of symptoms they can report, the social roles whose impairment becomes visible, and the diagnostic categories that can be applied. A childhood pattern of attention or behavioral regulation difficulties may be followed by adolescent anxiety, substance exposure, or adult mood symptoms. Different trajectories may be linked by shared liability, accumulated stress, reciprocal consequences, or evolving social opportunity.
Population-based developmental studies are critical because specialty care samples often overrepresent persistent or complex problems. Merikangas, K. R., He, J.-P., Burstein, M., Swanson, S. A., Avenevoli, S (2010) assessed the lifetime prevalence of mental disorders in U.S. adolescents, highlighting the substantial early burden of diagnosable psychopathology. However, cross-sectional lifetime estimates still rely on retrospective information for onset and cannot establish each transition mechanism. Repeated prospective assessments better distinguish changing clinical states from stable dispositions.
The Caspi, A., Houts, R. M., Ambler, A., Danese, A., Elliott, M. L., Hariri, (2020) illustrates why sustained observation modifies the scientific question. A person classified by one diagnosis at a particular time may exhibit other kinds of difficulty over decades. This creates a rationale for longitudinal risk profiles and functional outcomes that remain relevant as diagnostic labels change. It also suggests studying periods in which vulnerabilities become amplified by transitions in schooling, employment, intimate relationships, and caregiving.
Developmental research should avoid assuming a fixed meaning of psychopathology liability across age. Measurement invariance, informant shifts, and changing symptom expression must be evaluated. Interventions may alter trajectories without changing the general-factor score in a simple linear way, and a stable broad score may conceal meaningful improvements in everyday life. The life course is an organizing dimension of the phenomenon, rather than a peripheral covariate.
Culture, Social Inequality, and Diagnostic Transport
Psychiatric comorbidity is observed through languages, institutions, diagnostic instruments, care systems, and social expectations. A symptom questionnaire developed in one language may not preserve its conceptual structure after translation. Explanatory models of distress, idioms of suffering, social norms surrounding substance use, and access to professional assessment can affect both symptom reporting and who receives a diagnosis. These differences can alter the covariance structure from which transdiagnostic factors are estimated.
Social disadvantage can also influence multiple forms of mental-health difficulty through chronic insecurity, housing instability, exposure to violence, discrimination, and barriers to support. Such exposures are not reducible to individual weakness. A factor derived from clinical symptoms may summarize their downstream effects without representing the external forces that produced them. Olthof, M., Lichtwarck-Aschoff, A., & Fried, E. I. (2025) specifically warned that reifying p can shift attention away from external causes of psychopathology.
Comparative research must test whether relationships among symptoms and factors are invariant across relevant populations. Appropriate designs include item-level translation and adaptation, culturally informed interviews, local construct validation, and stratified sensitivity analyses. Population comparisons should not be interpreted as differences in intrinsic vulnerability when group differences in measurement, adversity, or service access have not been adequately examined.
The wider scientific implication is that transdiagnostic patterns arise at several levels. Individual psychological processes interact with households, communities, institutions, and historical conditions. Cross-cultural robustness would strengthen confidence in a proposed broad dimension, while meaningful differences could reveal context-dependent mechanisms. Both outcomes advance theory. Valid scientific synthesis preserves the distinction between a factor common to selected indicators and a factor presumed universal across every human social setting.
Functioning, Severity, and Disability
Comorbidity matters clinically partly because the presence of multiple syndromes can coincide with greater functional impairment, service complexity, and persistent distress. Yet diagnosis counts are imperfect proxies for burden. One severe syndrome can profoundly restrict functioning, while several mild or partly overlapping diagnoses may have different implications. Disability, social participation, occupational functioning, daily activity, perceived quality of life, and patient-defined goals provide complementary information.
A transdiagnostic factor might capture something close to broad severity, but this is a hypothesis requiring discriminant testing. If a factor score strongly correlates with an impairment scale containing similar items, some predictive performance may be circular. Researchers should test associations with independently assessed functional outcomes, evaluate whether broad factors add information beyond baseline symptom burden, and determine how prediction varies across follow-up periods and care settings.
Longitudinal changes in functioning are especially important. A patient's diagnostic profile may remain formally unchanged while sleep, relationships, work participation, and quality of life improve. Conversely, a person may cross below a diagnostic threshold while retaining substantial impairment. The use of multiple outcomes is a safeguard against interpreting categorical remission as complete recovery and against presenting overall symptom reduction as sufficient evidence of durable well-being.
The most defensible clinical integration of comorbidity research therefore combines structured assessment with an individualized account of functioning. It asks which symptoms and circumstances create the greatest impairment, what changes would matter to the patient, what risks require coordinated attention, and how progress will be measured. A broad liability score may assist this formulation if valid; it cannot replace the functional and relational context in which clinical problems are lived.
Causal Inference from Observational Data
The existence of comorbidity can be established descriptively; the causal explanation of that comorbidity is a harder scientific problem. Consider an observed association between persistent anxiety and later depression. It may arise because anxiety contributes to depressive symptoms through avoidance and social loss, because a shared predisposition increases risk for both, because early adversity influences each, or because clinical selection changes the sampled relationship. These explanations can yield similar correlations while implying different prevention strategies and different responses to intervention.
A causal research question should specify an exposure or process, an outcome, a time horizon, and the comparison of potential outcomes that would be relevant if the exposure or process changed. Prospective observation is useful for establishing that a putative cause precedes an outcome, but temporal priority alone does not remove confounding. Cross-lagged panel models can represent reciprocal associations yet may combine stable between-person differences with within-person change unless carefully specified. Dynamic models should also account for time-varying confounders and the possibility that symptoms influence subsequent exposures.
Genetically informed designs and instrumental variable approaches can provide additional leverage where their assumptions are defensible. However, genetic variants can influence multiple outcomes through pleiotropic pathways, threatening exclusion restrictions in some instrumental variable analyses. Sibling comparisons can help control for familial factors while introducing other limitations. No design becomes causal merely because it is computationally sophisticated. Assumptions, alternative explanations, negative controls, and sensitivity analyses should accompany causal language.
Experiments offer a more direct test of modifiable processes. Random assignment to interventions can support causal claims about the assigned intervention under trial conditions, provided randomization, outcome measurement, adherence, and loss to follow-up are handled appropriately. A successful transdiagnostic intervention nevertheless does not prove that one common mechanism caused every targeted disorder. To establish mediation, process change and symptom change must be measured with appropriate temporal ordering and designs that address mediator–outcome confounding.
Clinical Assessment and Differential Formulation
A comorbidity-informed assessment begins by establishing the phenomenology, timing, severity, and functional significance of each presenting problem. It examines whether symptoms are distinct or overlapping, whether they reflect concurrent disorders or diagnostic alternatives, and whether medication effects, medical illness, intoxication, withdrawal, sleep disruption, or developmental conditions affect interpretation. Structured instruments can improve consistency, while clinical interviews supply chronology, context, meaning, and opportunities to test competing formulations.
The clinician should distinguish at least three questions: Which diagnoses are supported under current criteria? Which broader dimensional vulnerabilities and processes are relevant? Which immediate clinical risks and functional priorities require action? These are complementary decisions. A person may have high internalizing severity without meeting several separate diagnoses; another may meet multiple diagnoses because of genuinely distinct clinical syndromes. A diagnostic checklist alone may understate the importance of either person's life context.
Case formulation can map proximal and distal influences: established vulnerabilities, precipitating circumstances, maintaining behaviors, protective relationships, and available resources. It should be revisable as new evidence appears. A transdiagnostic hypothesis can be tested within care by tracking whether changes in a suspected process are followed by changes across multiple symptoms, while recognizing that ordinary clinical observation cannot fully identify causation. Treatment planning may target a broad mechanism, a particular syndrome, or both in a sequenced or integrated manner.
Risk assessment remains specific to the individual problem rather than deduced from a general psychopathology score. For example, the presence of multiple diagnoses may justify careful inquiry into self-harm, substance intoxication, violence exposure, medical interactions, and barriers to care, but no statistical factor produces a precise individual prediction of these outcomes by itself. Individual clinical decisions require current assessment, context, and relevant professional standards.
Transdiagnostic Psychological Interventions
Transdiagnostic psychological interventions aim to address processes or skill deficits relevant to several diagnostic categories. Common targets include emotion awareness, avoidance, cognitive rigidity, behavioral withdrawal, maladaptive expectations, and responses to distress. The clinical logic is straightforward: if several problems are maintained in part by related processes, one coherent intervention may improve multiple symptom clusters and reduce the burden of learning separate protocols.
This treatment rationale should be tested independently from the broader p-factor theory. A transdiagnostic intervention can succeed because it contains effective ingredients common across several disorders, because it improves nonspecific functioning, or because it allows flexible adaptation. Such success does not demonstrate that all psychiatric diagnoses share a unique latent cause. The evidence for a treatment is therefore the comparative clinical outcome under a specified design, not the label applied to its conceptual framework.
The Unified Protocol for Transdiagnostic Treatment of Emotional Disorders is an influential example. It integrates techniques targeting emotional responses, avoidance, behavioral patterns, and exposure across anxiety-related presentations. Its clinical research provides a more direct test of transdiagnostic strategy than correlational factor analysis because participants can be assigned to treatment approaches. The appropriate evaluation includes blinded outcome assessment where possible, prespecified endpoints, treatment completion, adverse experiences, maintenance of change, and comparisons to established disorder-specific protocols.
A broader range of transdiagnostic approaches also includes process-focused cognitive behavioral methods, interventions addressing rumination and worry, and programs for integrated treatment of co-occurring conditions. The term does not establish efficacy by itself. Each protocol must be evaluated for the populations and disorders it includes, the intensity and quality of usual care, the qualifications of providers, and the durability of outcomes. Particular conditions may require targeted care even when broad processes are relevant.
Treatment Evidence and Clinical Generalization
Barlow, D. H., Farchione, T. J., Bullis, J. R., Gallagher, M. W., Murray (2017) conducted a randomized clinical equivalence trial comparing the Unified Protocol with diagnosis-specific protocols and a waitlist control condition in adults with four principal anxiety disorder presentations. The study randomized 223 participants, assessed outcomes after treatment and at six-month follow-up, and found reductions in principal-diagnosis severity under the Unified Protocol that met its prespecified equivalence comparison with established single-disorder protocols. This is strong evidence for a practical transdiagnostic treatment option within the study's clinical scope.
Several details prevent overgeneralization. The trial was conducted at a specialized outpatient center, used defined eligibility criteria, and compared interventions within a limited range of anxiety-related principal diagnoses. It supports conclusions about those conditions and outcomes, rather than an inference that the same treatment is equally effective for psychotic, developmental, addictive, eating, or all other psychiatric disorders. Equivalence under a trial's specified margin also differs from universal interchangeability across patient preferences and clinical settings.
The results can be examined at several levels. A principal-diagnosis outcome assesses the focal problem; additional symptom measures can detect broader effects; completion rates capture acceptability; and follow-up evaluates persistence. The trial also offers an important model for interpreting transdiagnostic evidence: one must compare treatments against credible alternatives and examine whether benefits remain clinically meaningful over time. Statistical non-difference without an appropriate equivalence or noninferiority design would be weaker evidence for comparable outcomes.
Treatment studies commonly enroll patients whose main diagnosis is clear enough for eligibility, even when comorbid symptoms are present. Their results may generalize unevenly to patients with severe multimorbidity, cognitive impairment, substance dependence, unstable housing, or acute medical risk. Research needs independent pragmatic replication in representative services, longer-term follow-up, and person-valued functioning measures. Across mental-health care, the central question is whether a transdiagnostic approach provides safe, reliable, and accessible benefit in the settings where it is actually used.
Medication, Polypharmacy, and Multimorbidity
Pharmacological research intersects with comorbidity through shared treatment indications, co-occurring medical conditions, combined medication exposure, and differential response. One medication class may be studied for more than one syndrome; this does not imply that all relevant disorders have a single pathophysiology. A shared pharmacological effect can influence several symptoms through one or more neurochemical processes, while syndromes retain distinct clinical trajectories and benefit–risk profiles.
Psychiatric multimorbidity can increase the complexity of medication reconciliation, adherence, side-effect attribution, discontinuation, and recognition of interactions. Multiple diagnoses do not by themselves justify multiple medications. The clinical issue is whether each treatment has a current indication, an evidence-supported expected benefit for the person, acceptable risk, and a coherent monitoring plan. Medication effects may also influence the very symptoms used to assign diagnoses, complicating observational analyses of comorbidity.
Research claiming treatment implications for a cross-diagnostic factor should distinguish symptom improvement from changes in the latent construct. If a factor score decreases after treatment, that may reflect simultaneous changes in several indicators; it does not establish pharmacological action on a singular p mechanism. Individual-level biological mechanisms require appropriately targeted evidence. Safety assessment should follow current approved labeling, professional recommendations, and the patient's specific clinical context.
This dossier addresses pharmacological complexity as a scientific consequence of multimorbidity. It offers no medication selection or dosing recommendations. Detailed therapeutic efficacy, interactions, and regulatory indications belong to drug- and disorder-specific research with current official label verification. These boundaries protect the inference from a broad epidemiological pattern to a concrete medical decision.
Prediction, Personalization, and Model Translation
For a cross-diagnostic construct to change clinical practice, it must contribute information that clinicians and patients can use. Prediction is one potential path. Researchers can ask whether a general or hierarchical liability score forecasts later disorder onset, persistence, functional deterioration, treatment response, or service needs more accurately than existing diagnostic, severity, exposure, and functioning measures. The decisive benchmark is incremental performance against a plausible baseline model.
Good prediction requires calibration, discrimination, reliability over time, and external validation across settings. A model that performs well in a large hospital registry may lose accuracy in community samples or in a system where diagnostic coding differs. Data leakage, inappropriate cross-validation, and overlapping samples can inflate estimated performance. Results should be evaluated for meaningful decisions and potential harms of false positive or false negative classifications. A nominally impressive machine-learning score has limited value without reproducible benefit in a well-defined application.
General liability may also support a dynamic view of patient trajectory, in which a broad vulnerability estimate is combined with domain-specific symptoms, current conditions, patient preferences, and changing context. Personalization then becomes an empirical question about moderators of intervention response. Subgroup findings should be preregistered or independently replicated when possible, since high-dimensional searches can produce unstable apparent differences.
Artificial systems can assist with longitudinal data integration, identification of missed symptom domains, measurement quality checks, or transparent decision support. Their reliability depends on validated input data, appropriate human oversight, privacy safeguards, and independent monitoring of error. Cross-diagnostic automated inference must not treat unverified text patterns as diagnoses or confuse predictive associations with causal mechanisms. This is the substantive interdisciplinary point of contact with psychology in the Artificial Era: computational systems can extend assessment and research capacities while their psychological classifications remain answerable to evidence and to patient welfare.
Critical Scientific Discussion
The evidence supports several conclusions with different degrees of force. First, psychiatric diagnostic categories show widespread and patterned comorbidity in population surveys, longitudinal cohorts, and clinical datasets. Second, broad dimensional structures and higher-order general factors repeatedly summarize aspects of this covariance, although their exact definition and stability vary. Third, molecular genomic studies identify substantial cross-disorder overlap alongside intermediate factor structure and disorder-specific signals. Fourth, psychological interventions can effectively target processes that operate across multiple diagnoses, with direct randomized evidence available for particular protocols and clinical populations.
Each conclusion has important limits. Comorbidity can be partly generated by overlapping criteria, differential access to assessment, and clinical selection. A factor does not disclose its own cause. Molecular overlap does not imply a unitary disorder. A successful transdiagnostic treatment does not prove that the treatment acts on a single general liability. Cultural, developmental, and methodological variation place boundaries on claims of universality.
The controversy concerning p is especially informative. Caspi, A., Houts, R. M., Fisher, H. L., Danese, A., & Moffitt, T. E. (2024) address competing models and interpretations of the general factor, while Watts, A. L., Greene, A. L., Bonifay, W., & Fried, E. I. (2024) argue that some factor-based claims have outpaced construct validation, model robustness, and causal justification. Their disagreement is substantive because the same label can designate a statistical tendency toward positive covariance, a broad burden-of-illness measure, a hypothesized developmental liability, or a proposed common causal source. These versions entail different empirical tests, and they should not be merged into one proposition.
Pettersson, E. (2025) further examined the psychometric properties and genomic etiology of the general factor. Recent discussion emphasizes that both the factor's mathematical behavior and its relationship to genetic architecture must be evaluated rather than presumed. Olthof, M., Lichtwarck-Aschoff, A., & Fried, E. I. (2025) additionally caution that an exclusively person-centered latent explanation can divert attention from external structural causes. Together, these contributions illustrate why a mature field needs both confirmatory and critical programs.
The synthesis is therefore a layered rather than a unitary explanatory framework. Classification describes clinically significant patterns; dimensional measurement quantifies variation; genomic research estimates portions of shared inherited association; developmental and ecological designs examine changing contexts; experiments identify modifiable processes; and clinical trials assess practical intervention outcomes. These levels can inform one another through explicit links and falsifiable hypotheses. None is entitled to replace the others by definitional fiat.
Research Priorities for the Next Generation
The first priority is stronger measurement. Multicenter studies should preregister symptom coverage, factor specification, invariance tests, and independent validation strategies. Diagnostic and dimensional measures should be reported alongside functioning and patient-defined outcomes. The field should compare general and intermediate factors using more than model fit, including stability across instruments, age, populations, and study sites.
The second priority is prospective temporal design. Frequent measurements and long follow-up intervals can clarify when broad vulnerability precedes symptoms, when one symptom cluster influences another, and how environmental changes modify trajectories. Designs should distinguish between-person associations from within-person dynamics. Well-specified hypotheses about direction and timing are more informative than repeated discovery of covariance alone.
The third priority is triangulation of genetic, biological, psychological, and social evidence. Genomic factor models must be compared with phenotypic ones without assuming equivalence. Brain measures should be validated for reliability and generalizability. Social determinants deserve prospective and quasi-experimental study as potential causes and moderators. Analyses should explicitly quantify uncertainty introduced by selection, measurement error, and lack of representation across cultural and ancestry groups.
The fourth priority is translational evaluation. The field needs randomized comparative studies of transdiagnostic and diagnosis-specific approaches in routine settings, especially for people with complex multimorbidity and varying resources. Trials should report outcomes for comorbid domains, functioning, adverse effects, treatment burden, and durability. Prediction models should be assessed for added clinical usefulness and potential harms before deployment.
The fifth priority is conceptual transparency. Authors should define whether they mean p as a score, an inferred liability, an explanatory mechanism, or a hypothesized cause, and then choose tests appropriate to that meaning. Falsifiable distinctions between shared cause, symptom interactions, classification effects, and common exposure are necessary for theoretical progress. Scientific maturity in this field will be expressed through better discrimination among plausible accounts rather than simply a growing number of studies using the same transdiagnostic vocabulary.
Conclusion
Psychiatric comorbidity is a foundational empirical property of the distribution and development of mental-health problems. It is observable through diagnostic interviews, symptom dimensions, clinical records, and long-term cohorts, and its structure is illuminated by correlated spectra, hierarchical factors, genomic covariance, and changing symptom networks. The persistence of cross-disorder associations across methods gives the research problem enduring importance. At the same time, each method measures a different level of organization and permits a different range of inference.
Transdiagnostic psychopathology liability is most defensible as a precisely operationalized family of shared vulnerability and process constructs. A broad statistical factor can be useful without being a single disease, biological essence, or universal clinical test. Multiple pathways—shared predisposition, environmental exposures, reciprocal symptoms, diagnostic overlap, developmental transition, and ascertainment—can produce partly similar observations. Valid explanation requires designs capable of separating these possibilities and quantifying where separation remains incomplete.
The scientific synthesis supports an integrated research program in which general and specific dimensions are measured together, temporal evidence is prioritized, external causes are included, genomic and behavioral levels are connected cautiously, and intervention claims are judged by randomized and pragmatic outcomes. This framework preserves the practical value of diagnosis while advancing the study of shared mechanisms and the full continuity of mental-health trajectories across a life course. The resulting knowledge belongs to a broader psychology concerned with how vulnerability, adaptation, social organization, and clinical care interact as human environments and computational capabilities develop.
References
- Achenbach, T. M. (1966). The classification of children's psychiatric symptoms: A factor-analytic study. Psychological Monographs: General and Applied, 80(7), 1–37. https://doi.org/10.1037/h0093906
- Barlow, D. H., Farchione, T. J., Bullis, J. R., Gallagher, M. W., Murray-Latin, H., Sauer-Zavala, S., Bentley, K. H., Thompson-Hollands, J., Conklin, L. R., Boswell, J. F., Ametaj, A., Carl, J. R., Boettcher, H. T., & Cassiello-Robbins, C. (2017). The Unified Protocol for Transdiagnostic Treatment of Emotional Disorders compared with diagnosis-specific protocols for anxiety disorders: A randomized clinical trial. JAMA Psychiatry, 74(9), 875–884. https://doi.org/10.1001/jamapsychiatry.2017.2164
- Borsboom, D. (2017). A network theory of mental disorders. World Psychiatry, 16(1), 5–13. https://doi.org/10.1002/wps.20375
- Brainstorm Consortium. (2018). Analysis of shared heritability in common disorders of the brain. Science, 360(6395), eaap8757. https://doi.org/10.1126/science.aap8757
- Caron, C., & Rutter, M. (1991). Comorbidity in child psychopathology: Concepts, issues and research strategies. Journal of Child Psychology and Psychiatry, 32(7), 1063–1080. https://doi.org/10.1111/j.1469-7610.1991.tb00350.x
- Caspi, A., Houts, R. M., Belsky, D. W., Goldman-Mellor, S. J., Harrington, H., Israel, S., Meier, M. H., Ramrakha, S., Shalev, I., Poulton, R., & Moffitt, T. E. (2014). The p factor: One general psychopathology factor in the structure of psychiatric disorders? Clinical Psychological Science, 2(2), 119–137. https://doi.org/10.1177/2167702613497473
- Caspi, A., Houts, R. M., Ambler, A., Danese, A., Elliott, M. L., Hariri, A., Harrington, H., Hogan, S., Poulton, R., Ramrakha, S., Rasmussen, L. J. H., Reuben, A., Richmond-Rakerd, L., Sugden, K., Wertz, J., Williams, B. S., & Moffitt, T. E. (2020). Longitudinal assessment of mental health disorders and comorbidities across 4 decades among participants in the Dunedin Birth Cohort Study. JAMA Network Open, 3(4), e203221. https://doi.org/10.1001/jamanetworkopen.2020.3221
- Caspi, A., Houts, R. M., Fisher, H. L., Danese, A., & Moffitt, T. E. (2024). The general factor of psychopathology (p): Choosing among competing models and interpreting p. Clinical Psychological Science, 12(1), 53–82. https://doi.org/10.1177/21677026221147872
- Cross-Disorder Group of the Psychiatric Genomics Consortium. (2019). Genomic relationships, novel loci, and pleiotropic mechanisms across eight psychiatric disorders. Cell, 179(7), 1469–1482.e11. https://doi.org/10.1016/j.cell.2019.11.020
- DeYoung, C. G., Blain, S. D., Latzman, R. D., Grazioplene, R. G., Haltigan, J. D., Kotov, R., Michelini, G., Venables, N. C., Docherty, A. R., Goghari, V. M., Kallen, A. M., Martin, E. A., Palumbo, I. M., Patrick, C. J., Perkins, E. R., Shackman, A. J., Snyder, M. E., & Tobin, K. E. (2024). The hierarchical taxonomy of psychopathology and the search for neurobiological substrates of mental illness: A systematic review and roadmap for future research. Journal of Psychopathology and Clinical Science, 133(8), 697–715. https://doi.org/10.1037/abn0000903
- Feinstein, A. R. (1970). The pre-therapeutic classification of co-morbidity in chronic disease. Journal of Chronic Diseases, 23(7), 455–468. )90054-8 https://doi.org/10.1016/0021-9681(70
- Fried, E. I., & Cramer, A. O. J. (2017). Moving forward: Challenges and directions for psychopathological network theory and methodology. Perspectives on Psychological Science, 12(6), 999–1020. https://doi.org/10.1177/1745691617705892
- Grotzinger, A. D., Mallard, T. T., Akingbuwa, W. A., Ip, H. F., Adams, M. J., Lewis, C. M., McIntosh, A. M., Grove, J., Dalsgaard, S., Lesch, K.-P., Strom, N., Meier, S. M., Mattheisen, M., Børglum, A. D., Mors, O., Breen, G., Lee, P. H., Kendler, K. S., Smoller, J. W.,. Nivard, M. G. (2022). Genetic architecture of 11 major psychiatric disorders at biobehavioral, functional genomic and molecular genetic levels of analysis. Nature Genetics, 54(5), 548–559. https://doi.org/10.1038/s41588-022-01057-4
- Grotzinger, A. D., Werme, J., Peyrot, W. J., Frei, O., de Leeuw, C., Bicks, L. K., Guo, Q., Margolis, M. P., Coombes, B. J., Batzler, A., Pazdernik, V., Biernacka, J. M., Andreassen, O. A., Anttila, V., Børglum, A. D., Breen, G., Cai, N., Demontis, D., Edenberg, H. J.,. Smoller, J. W. (2026). Mapping the genetic landscape across 14 psychiatric disorders. Nature, 649(8096), 406–415. https://doi.org/10.1038/s41586-025-09820-3
- Insel, T., Cuthbert, B., Garvey, M., Heinssen, R., Pine, D. S., Quinn, K., Sanislow, C., & Wang, P. (2010). Research domain criteria (RDoC): Toward a new classification framework for research on mental disorders. American Journal of Psychiatry, 167(7), 748–751. https://doi.org/10.1176/appi.ajp.2010.09091379
- Kendler, K. S., Prescott, C. A., Myers, J., & Neale, M. C. (2003). The structure of genetic and environmental risk factors for common psychiatric and substance use disorders in men and women. Archives of General Psychiatry, 60(9), 929–937. https://doi.org/10.1001/archpsyc.60.9.929
- Kessler, R. C., McGonagle, K. A., Zhao, S., Nelson, C. B., Hughes, M., Eshleman, S., Wittchen, H.-U., & Kendler, K. S. (1994). Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: Results from the National Comorbidity Survey. Archives of General Psychiatry, 51(1), 8–19. https://doi.org/10.1001/archpsyc.1994.03950010008002
- Kessler, R. C., Chiu, W. T., Demler, O., Merikangas, K. R., & Walters, E. E. (2005). Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 617–627. https://doi.org/10.1001/archpsyc.62.6.617
- Kotov, R., Krueger, R. F., Watson, D., Achenbach, T. M., Althoff, R. R., Bagby, R. M., Brown, T. A., Carpenter, W. T., Caspi, A., Clark, L. A., Eaton, N. R., Forbes, M. K., Forbush, K. T., Goldberg, D., Hasin, D., Hyman, S. E., Ivanova, M. Y., Lynam, D. R., Markon, K.,. Zimmerman, M. (2017). The Hierarchical Taxonomy of Psychopathology (HiTOP): A dimensional alternative to traditional nosologies. Journal of Abnormal Psychology, 126(4), 454–477. https://doi.org/10.1037/abn0000258
- Kotov, R., Krueger, R. F., Watson, D., Cicero, D. C., Conway, C. C., DeYoung, C. G., Eaton, N. R., Forbes, M. K., Hallquist, M. N., Latzman, R. D., Mullins-Sweatt, S. N., Ruggero, C. J., Simms, L. J., Waldman, I. D., Waszczuk, M. A., & Wright, A. G. C. (2021). The Hierarchical Taxonomy of Psychopathology (HiTOP): A quantitative nosology based on consensus of evidence. Annual Review of Clinical Psychology, 17, 83–108. https://doi.org/10.1146/annurev-clinpsy-081219-093304
- Krueger, R. F. (1999). The structure of common mental disorders. Archives of General Psychiatry, 56(10), 921–926. https://doi.org/10.1001/archpsyc.56.10.921
- Lahey, B. B., Applegate, B., Hakes, J. K., Zald, D. H., Hariri, A. R., & Rathouz, P. J. (2012). Is there a general factor of prevalent psychopathology during adulthood? Journal of Abnormal Psychology, 121(4), 971–977. https://doi.org/10.1037/a0028355
- Merikangas, K. R., He, J.-P., Burstein, M., Swanson, S. A., Avenevoli, S., Cui, L., Benjet, C., Georgiades, K., & Swendsen, J. (2010). Lifetime prevalence of mental disorders in U.S. adolescents: Results from the National Comorbidity Survey Replication–Adolescent Supplement (NCS-A). Journal of the American Academy of Child & Adolescent Psychiatry, 49(10), 980–989. https://doi.org/10.1016/j.jaac.2010.05.017
- National Institute of Mental Health. (n.d.). Research Domain Criteria (RDoC). https://www.nimh.nih.gov/research/research-funded-by-nimh/rdoc/about-rdoc
- Olthof, M., Lichtwarck-Aschoff, A., & Fried, E. I. (2025). Reification of the p factor draws attention away from external causes of psychopathology. Journal of Psychopathology and Clinical Science, 134(4), 339–340. https://doi.org/10.1037/abn0000961
- Pettersson, E. (2025). On the psychometric properties and genomic etiology of the general factor of psychopathology. Molecular Psychiatry, 30(11), 5523–5533. https://doi.org/10.1038/s41380-025-03151-5
- Plana-Ripoll, O., Pedersen, C. B., Holtz, Y., Benros, M. E., Dalsgaard, S., de Jonge, P., Fan, C. C., Degenhardt, L., Ganna, A., Greve, A. N., Gunn, J., Iburg, K. M., Kessing, L. V., Lee, B. K., Lim, C. C. W., Mors, O., Nordentoft, M., Prior, A., Roest, A. M.,. McGrath, J. J. (2019). Exploring comorbidity within mental disorders among a Danish national population. JAMA Psychiatry, 76(3), 259–270. https://doi.org/10.1001/jamapsychiatry.2018.3658
- Smith, G. T., Atkinson, E. A., Davis, H. A., Riley, E. N., & Oltmanns, J. R. (2020). The general factor of psychopathology. Annual Review of Clinical Psychology, 16, 75–98. https://doi.org/10.1146/annurev-clinpsy-071119-115848
- Wandall, E. N. S., Andersen, P. K., Christensen, R. H. B., & Benros, M. E. (2026). Nationwide transdiagnostic age and sex patterns in mental disorder disease development. JAMA Psychiatry, 83(8), 789–798. https://doi.org/10.1001/jamapsychiatry.2026.0882
- Watson, D. (2005). Rethinking the mood and anxiety disorders: A quantitative hierarchical model for DSM-V. Journal of Abnormal Psychology, 114(4), 522–536. https://doi.org/10.1037/0021-843X.114.4.522
- Watts, A. L., Greene, A. L., Bonifay, W., & Fried, E. I. (2024). A critical evaluation of the p-factor literature. Nature Reviews Psychology, 3, 108–122. https://doi.org/10.1038/s44159-023-00260-2
- World Health Organization. (2024). Clinical descriptions and diagnostic requirements for ICD-11 mental, behavioural and neurodevelopmental disorders. World Health Organization. https://www.who.int/publications/i/item/9789240077263