Poor Separation of Clinical Symptom Profiles by DSM-5 Disorder Criteria

Abstract

Assessment of mental illness typically relies on a disorder classification system that is considered to be at odds with the vast disorder comorbidity and symptom heterogeneity that exists within and across patients. Patients with the same disorder diagnosis exhibit diverse symptom profiles and comorbidities creating numerous clinical and research challenges. Here we provide a quantitative analysis of the symptom heterogeneity and disorder comorbidity across a sample of 107,349 adult individuals (aged 18-85 years) from 8 English-speaking countries. Data were acquired using the Mental Health Quotient, an anonymous, online, self-report tool that comprehensively evaluates symptom profiles across 10 common mental health disorders. Dissimilarity of symptom profiles within and between disorders was then computed. We found a continuum of symptom prevalence rather than a clear separation of normal and disordered. While 58.7% of those with 5 or more clinically significant symptoms did not map to the diagnostic criteria of any of the 10 DSM-5 disorders studied, those with symptom profiles that mapped to at least one disorder had, on average, 20 clinically significant symptoms. Within this group, the heterogeneity of symptom profiles was almost as high within a disorder label as between 2 disorder labels and not separable from randomly selected groups of individuals with at least one of any of the 10 disorders. Overall, these results quantify the scale of misalignment between clinical symptom profiles and DSM-5 disorder labels and demonstrate that DSM-5 disorder criteria do not separate individuals from random when the complete mental health symptom profile of an individual is considered. Greater emphasis on empirical, disorder agnostic approaches to symptom profiling would help overcome existing challenges with heterogeneity and comorbidity, aiding clinical and research outcomes.

Keywords: ADHD; DSM-5; autism spectrum disorder (ASD); comorbidity; depression; diagnosis; heterogeneity; symptom profiles.

Figure 1

Calculation of the symptom dissimilarity between two people. (A) Symptom profile for one person (person 1 in B). Each of the 47 MHQ elements (representing a comprehensive set of possible mental symptoms) are coded as 1 if it is a symptom (i.e., the rating selected is above the threshold where it is considered a symptom, generally in this paper ≥8 for problem items and ≤1 for spectrum items), and 0 if it is not a symptom (i.e., does not meet the threshold). (B) Symptom profile for two people. Symptom dissimilarity is calculated as the sum of the absolute difference between the symptom matrix for each person/total number of symptoms.

Figure 2

Symptom prevalence in the sample. (A) Prevalence of number of clinical mental health symptoms in the sample as defined by a threshold of severity in the MHQ (≥8 for problem items and ≤1 for spectrum items; black line). Dotted line denotes the prevalence when the threshold is shifted by one point (≥7 for problem items and ≤2 for spectrum items). The percentage with each higher number of clinical symptoms decreases exponentially. (B) The proportion of respondents who reported severe problems with each of the 47 elements of mental well-being included in the MHQ, as defined by a threshold of severity in the MHQ (≥8 for problem items and ≤1 for spectrum items; black bars). Prevalence of specific symptoms across the sample ranged from 24.4 to 1.1%. Gray bars denote the proportion when the threshold is shifted by one point (≥7 for problem items and ≤2 for spectrum items).

Figure 3

Mapping of MHQ symptoms to DSM-diagnostic criteria. MHQ items corresponding to each DSM-5 defined disorder (left; red squares) and associated prevalence of each disorder following application of diagnostic rules (right). Full mapping and rules shown in Table 2.

Figure 4

Prevalence and comorbidity among DSM-disorders. (A) Percentage of individuals by number of disorders to which they map. Inset shows the mapping of the 12.3% with one or more disorder. (B) Percentage of individuals with each disorder who also map to each other disorder (numbers in Supplementary Table 3). (C) Distribution of number of clinical symptoms across individuals who do not map to any disorder criteria (black line) and those who map to at least one (red line).

Figure 5

Heterogeneity of symptom profiles within and between disorders. (A) Distribution of symptom profile dissimilarity across all individuals who map to criteria for PTSD (gray), ASD (black) and a randomly selected groups of individuals from the pool of individuals with at least one of the 10 disorders (red). (B) Distribution of the dissimilarity of symptom profiles of all individuals mapping to the diagnostic criteria for addiction (black) and the dissimilarity of symptom profiles between those with addiction and those with ADHD excluding comorbid individuals. (C) Symptom dissimilarity within and between each pair of disorders ranges from 37.9% (within PTSD) to 45.6% (between ASD and Schizophrenia). (D) Symptom dissimilarity as in (C) but with comorbid individuals excluded ranged from 37.8 to 45.7%.

Figure 6

Symptom profile of 2 individuals who map to DSM-criteria for depression. Comparison of symptom profiles of 2 individuals mapping to diagnostic criteria for depression is an example of the within disorder symptom heterogeneity. Higher numbers toward the outside of the circle represent a greater problem.