Higher depressive symptom burden was associated with all-cause mortality, and exploratory analyses suggested symptom-level heterogeneity, but item-level modeling did not meaningfully improve discrimination and symptom-specific findings require independent replication.
Each 1-point increase in PHQ-9 total score was associated with higher all-cause mortality.
When affective and somatic domains were modeled simultaneously, only the somatic domain remained modestly associated with all-cause mortality.
Four individual PHQ-9 items — anhedonia, fatigue, appetite or weight disturbance, and thoughts of death or self-harm — were associated with all-cause mortality in separate item-level analyses after false discovery rate correction.
In a mutually adjusted model including all items simultaneously, anhedonia, fatigue, and thoughts of death or self-harm remained independently associated with all-cause mortality.
Item-level models improved in-sample statistical fit but did not meaningfully improve mortality risk discrimination compared to the total score model.
This research examined whether all depressive symptoms carry equal risk for death or whether some symptoms are more dangerous than others. Using data from nearly 18,000 U.S. adults followed for mortality through 2019, the study found that higher overall scores on a standard depression questionnaire (the PHQ-9) were linked to a greater risk of dying. When the researchers looked at physical (somatic) symptoms of depression versus emotional (affective) symptoms separately, the physical symptoms — things like fatigue, sleep problems, and appetite changes — appeared to drive most of the mortality risk. At the level of individual symptoms, anhedonia (loss of interest or pleasure), fatigue, appetite or weight disturbance, and thoughts of death or self-harm each showed statistically significant associations with higher mortality risk on their own. When all symptoms were considered together in one model, anhedonia, fatigue, and thoughts of death or self-harm remained independently linked to mortality risk, with thoughts of death or self-harm showing the strongest association (about 51% higher risk). However, while these item-level models fit the data better statistically, they barely improved the ability to actually predict who would die compared to simply using the total depression score. This research suggests there may be meaningful differences among depressive symptoms in their relationship to mortality risk, with somatic and certain specific symptoms appearing particularly important. However, the authors caution that the item-level findings were exploratory, and the practical value of tracking individual symptoms over a total score for predicting death risk was minimal. These findings would need to be confirmed in other independent studies before drawing firm conclusions.
Jin Z. (2026). Depressive symptom heterogeneity and mortality in a prospective U.S. cohort.. European archives of psychiatry and clinical neuroscience. https://doi.org/10.1007/s00406-026-02368-x