Association between DSM-5 and ICD-11 personality dimensional traits in a general medical cohort and readmission and mortality

Sergio A Barroilhet; Alexandra E Bieling; Thomas H McCoy Jr; Roy H Perlis

doi:10.1016/j.genhosppsych.2020.01.003

Association between DSM-5 and ICD-11 personality dimensional traits in a general medical cohort and readmission and mortality

Gen Hosp Psychiatry. 2020 May-Jun:64:63-67. doi: 10.1016/j.genhosppsych.2020.01.003. Epub 2020 Jan 27.

Authors

Sergio A Barroilhet¹, Alexandra E Bieling², Thomas H McCoy Jr³, Roy H Perlis⁴

Affiliations

¹ Center for Quantitative Health, Division of Clinical Research and Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA; Department of Psychiatry, Harvard Medical School, Harvard University, Boston, MA, USA; University Psychiatric Clinic, University of Chile Clinical Hospital, Santiago, Chile.
² Center for Quantitative Health, Division of Clinical Research and Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA.
³ Center for Quantitative Health, Division of Clinical Research and Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA; Department of Psychiatry, Harvard Medical School, Harvard University, Boston, MA, USA.
⁴ Center for Quantitative Health, Division of Clinical Research and Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA; Department of Psychiatry, Harvard Medical School, Harvard University, Boston, MA, USA. Electronic address: rperlis@partners.org.

Abstract

Background: Personality has long been studied as a factor associated with health outcomes. Investigations of large, generalizable clinical cohorts are limited by variations in personality diagnostic methodologies and difficulties with long-term follow-up.

Methods: Electronic health records of a cohort of patients admitted to a general hospital were characterized using a previously developed natural language processing tool for extracting DSM-5 and ICD-11 personality domains. We used Cox regression and Fine-Gray competing risk survival to analyze the relationships between these personality estimates, sociodemographic features, and risk of readmission and mortality.

Results: Among 12,274 patients, 2379 deaths occurred in the course of 61,761 patient-years at risk, with 19,985 admissions during follow-up. Detachment was the most common personality feature. Presence of disinhibition was independently associated with a higher mortality risk, while anankastic traits were associated with a lower mortality risk. Increased likelihood of readmission was predicted by detachment, while decreased likelihood of readmission was associated with disinhibition and psychoticism traits.

Conclusions: Personality features can be identified from electronic health records and are associated with readmission and mortality risk. Developing treatment strategies that target patients with higher personality symptom burden in specific dimensions could enable more efficient and focused interventions.

Keywords: Electronic health record; Machine learning; Mortality; Natural language processing; Personality disorder; Personality domains; Readmission.

Publication types

Research Support, N.I.H., Extramural

MeSH terms

Adult
Aged
Aged, 80 and over
Cohort Studies
Comorbidity
Diagnostic and Statistical Manual of Mental Disorders
Electronic Health Records
Female
Humans
International Classification of Diseases
Machine Learning
Male
Middle Aged
Natural Language Processing
Noncommunicable Diseases / mortality*
Patient Readmission / statistics & numerical data*
Personality Disorders / diagnosis
Personality Disorders / epidemiology*
Personality Disorders / physiopathology*

Grants and funding

R01 MH106577/MH/NIMH NIH HHS/United States