Confirming diagnostic categories within a depression continuum: Testing extra-linearity of risk factors and a latent class analysis.

BACKGROUND

Dimensions are recommended as replacements for diagnostic categories of depression, but clinicians continue to use categories. Categories are appropriate if major, underlying changes in symptom structure occur above a clinical cut-off on a depression continuum.

METHODS

Cross-sectional surveys of Chinese undergraduates (n = 39,446) 2014-2018 measured self-reported depressive symptoms, associated psychopathology and etiological risk factors using standardised instruments. We created a continuum using PHQ-9 scores and tested linear and extra-linear contrasts in associated psychopathology, and etiology. We carried out latent class analyses (LCA).

RESULTS

Most symptoms showed linear increase, but depressed mood, anhedonia, and suicidal ideation showed marked increase at the severe end of the continuum. There was extra-linear increase in associated psychotic symptoms, other psychopathology, age, low family income, chronic pain and physical illness, childhood physical and sexual abuse, and neglect. Four LCs corresponding to Melancholic, Severe melancholic, Non-melancholic, and Mild depression were confirmed, but only above a clinical cut-off along the continuum. Etiological risk factors did not differentiate between classes but showed overall dramatic increase in impact above threshold of clinical severity.

LIMITATIONS: Only one self-report instrument was used (PHQ-9) to measure depression and diagnoses were not validated by clinical interviews.

CONCLUSIONS

Categories are necessary to describe the dramatic changes in underlying structure and symptom associations above a clinical threshold of severity. These result from extra-linear impact of etiological risk factors at the severe end of the depression continuum. Although the study confirmed melancholic and non-melancholic subtypes, further investigation should investigate etiological factors that determine this subdivision.