A new analysis is challenging a common interpretation of psychiatric comorbidity—the co-occurrence of two or more mental disorders. The authors argue that apparent links between broad forms of psychopathology may be driven less by people experiencing multiple disorders and more by people reporting little or no psychopathology.
The proposed explanation centers on “nomorbidity,” which the authors use to describe the absence of a diagnosis of a single mental disorder. The idea does not dispute that people can experience more than one condition. Instead, it questions whether statistical relationships found in broad community samples should automatically be interpreted as evidence that disorders commonly occur together or share the same underlying cause.
What the analysis examined
Research spanning nearly half a century has repeatedly identified two broad dimensions of psychopathology. Externalizing involves difficulties with behavioral and emotional control, while internalizing involves intense negative emotions.
A related concept, the p-factor, has been proposed as a general liability to mental disorder. It has been used to account for shared features of psychopathology, including comorbidity, persistence and symptom severity.
The authors propose a different interpretation: covariance between these dimensions may be strongest among people with very low symptom levels. If so, the resulting statistical pattern could describe the shared absence of symptoms rather than the co-occurrence of clinically significant problems.
Patterns in the data
In the first wave of the Adolescent Brain Cognitive Development Study, 81% of participants had nonclinical levels of both internalizing and externalizing. By comparison, 5.3% showed some combination of at-risk or clinical levels in both areas.
The association between the two dimensions also changed at a threshold, or “change point.” Before that point, a 1-point increase in internalizing corresponded to a 32% increase in externalizing. After the change point, the corresponding increase was 7%.
Across four waves, the change points for the relationship between internalizing and externalizing fell between internalizing scores of 3 and 5. The authors report the same general pattern—stronger associations at lower symptom levels than at higher levels—for relationships involving externalizing, internalizing, thought problems and attention problems.
In the authors’ interpretation, these results indicate that the covariance between psychopathology dimensions was driven by nomorbidity rather than comorbidity. They also report that conventional linear models overpredicted associations at symptomatic and clinical levels because the data showed nonlinearity, skewness and kurtosis.
Why the distinction matters
When diagnoses are considered across the population, most people do not meet criteria for a disorder during a given period. Around 70% show no diagnosis of a single mental disorder when the past year is assessed, and about 50% do so when lifetime diagnoses are considered. Over long periods of follow-up, approximately 17% to 40% are not diagnosed with a single mental disorder at any time.
At the same time, receiving one diagnosis is associated with higher odds of receiving additional diagnoses. That pattern is one reason comorbidity has been treated as an important feature of mental-health research and classification.
The new analysis suggests that these two observations need to be separated more carefully. A structure calculated from covariance in a general-population sample may help describe overall symptom burden or support population screening. It should not automatically be taken as evidence of clinical comorbidity, a shared cause, or treatment-relevant co-occurrence among people with substantial symptoms.
A challenge to one-size-fits-all models
Modern approaches, including the Hierarchical Taxonomy of Psychopathology, have already questioned whether standard diagnostic categories divide mental disorders at the most informative points. The authors’ findings add another caution: there may not be one structure of psychopathology that applies equally to asymptomatic people and those experiencing clinically meaningful symptoms.
They suggest that distinct subpopulations may differ in their susceptibility to psychopathology, patterns of comorbidity and overall organization of symptoms. The p-factor, in their view, appears largely irrelevant to clinical levels of psychopathology and to the subgroup experiencing psychopathology in an appreciable form.
That conclusion is presented as a proposed interpretation, not a settled replacement for existing models. The source also notes potential criticisms involving Berkson’s bias and the effects of conditioning on severity, while arguing that those concerns apply only under particular assumptions.
What readers should take from it
The findings do not mean that co-occurring mental disorders are unreal or unimportant. Rather, they suggest that the statistical tools used to study large populations can blur important differences between people with few symptoms and people experiencing significant psychological difficulties.
For researchers, the message is to test for nonlinear relationships and avoid treating population-wide correlations as direct evidence about clinical groups. For the public, the practical takeaway is more limited: broad research terms such as “comorbidity” and “general psychopathology” may describe statistical patterns across a population, not a diagnosis or a prediction about any one person.
AI tools were used to assist with the preparation of this article.
Leave a Reply