Phenome-wide diagnostic comparison among suicide deaths and living individuals with chronic pain diagnoses

Chronic pain, regardless of its type, is a significant risk factor for suicide. However, not all individuals with chronic pain also experience suicidal thoughts and behaviors. Better characterization of clinical risk profiles and comorbidities across the medical spectrum among people with chronic pain who die by suicide is urgently needed to aid treatment and prevention strategies.

This case-control study leverages population-based data from the Utah Suicide Mortality Risk Study. Specifically, we identify clinical phenotypes from diagnostic data that differentiate between individuals that died by suicide with chronic pain diagnoses (N = 1,410) and living control individuals who also had chronic pain diagnoses (N = 4,664). Medical diagnostic codes were aggregated via phecodes to perform a phenotype-based phenome-wide association study. Using multivariable logistic regression analysis adjusting for covariates and multiple testing, differences in 1,727 common clinical phenotypes (phecodes) were assessed between suicide deaths and controls with chronic pain diagnoses. Models were also stratified by sex.

Chronic pain diagnoses were nearly three times more prevalent in individuals who died by suicide compared with those who did not. Sixty-five phecodes were significantly overrepresented among suicide deaths with chronic pain diagnoses compared with controls with chronic pain diagnoses. Utah suicide deaths with chronic pain had significantly more psychiatric diagnoses (mood disorders, anxiety disorders, attention deficit hyperactivity disorder, posttraumatic stress disorder, personality disorders, schizophrenia/psychosis, substance use related traits and prior overdoses, and diagnoses related to previous suicidal thoughts and behaviors) in addition to insomnia and specific pain related diagnoses compared to Utah controls with chronic pain (odds ratios ranged from 1.40-7.10). Twenty-five phecodes were overrepresented in controls with chronic pain compared to suicides. These were related to preventative care, cancer, obesity and other conditions (odds ratios ranged from 0.16-0.73). Sex-specific analyses largely replicated the combined analyses, yet the strength of the association was stronger for women with phecodes related to prior self-harm.

Results identified multiple clinical comorbidities with chronic pain that differentiate suicide deaths from living control individuals with a history of diagnosed chronic pain. Our findings may help discern individuals with chronic pain who may be at greater risk for suicide death.