[Skip to Navigation]
October 23, 2019

Personal Life Events—A Promising Dimension for Psychiatry in Electronic Health Records

Author Affiliations
  • 1Department of Psychiatry, Vagelos College of Physicians and Surgeons, Columbia University, New York, New York
  • 2Division of Translational Epidemiology, New York State Psychiatric Institute, New York
  • 3Department of Psychiatry, Weill Medical College, Cornell University, New York, New York
  • 4Department of Healthcare Policy and Research, Weill Medical College, Cornell University, New York, New York
JAMA Psychiatry. 2020;77(2):115-116. doi:10.1001/jamapsychiatry.2019.3217

The adoption of electronic health records (EHRs) has empowered large-scale research in general medicine through providing clinically relevant data sources at relatively low cost. More recently, EHRs have begun to make substantial inroads in psychiatry, with some notable successes for genomics and for understanding polygenic risk1,2 and prediction of suicidal behavior.3,4 Their use is also critical to accelerate central nervous system innovation and new therapeutic strategies.5

Add or change institution
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    Determining Utility of Sensitive Data Points in the Context of Psychiatric Care
    Jack Lennon, M.A. | Adler University; Rush University Medical Center
    Articles are being published rapidly with increased focus on machine learning (ML) methods in the context of suicide prediction. Weissman et al[1] report that dimensions of psychiatry, such as suicide prediction, could be enhanced if social determinants of health (SDOH) were incorporated into electronic health records (EHRs). Given extensive inclusion of ML in psychiatry,[2] as well as its potential to inform precision treatment rules[3], Weissman et al[1] suggest that questionnaires could be developed to ensure that SDOH are documented. I agree with this claim but believe that such questionnaires may be premature.

    The question does not seem to be
    whether or not the incorporation of SDOH can assist the field of psychiatry, nor does it seem that we should be questioning which ‘groups’ are at greatest risk. While there is room for improvement, we possess knowledge of SDOH and at least cursory knowledge of their relationships with suicide. The first of two important questions should be related to what is done with this information once collected. This information could prove advantageous in collecting multi-site, longitudinal data to support individualized[4] psychiatric care, but what is the cost-benefit ratio when considering patient confidentiality and subsequent healthcare utilization, particularly, for example, among culturally-vulnerable populations?

    Secondly, there must be a sufficient strategy related to standardized forms, with a full understanding of their potential benefits and dangers in the context of both extant literature and the weight the healthcare system places on classifications, diagnostic or otherwise. Timing is a critical component of behavior prediction, such that a combination of SDOH, polygenic scores, and reported symptoms is insufficient to accurately predict suicide without the risk of unnecessary false-positives. Timing and chronicity matter.[2] Specifically, it is not merely timing over the course of a life trajectory but the order of events, such as symptom onset, in the context of social determinants that are often recursive in nature. It is not a combination but a permutation of events that is likely to impact imminent risk of suicide.

    SDOH are overlooked components of EHRs[1] but any useful questionnaire must be cautiously tailored to the time and order of events. These details may be difficult for patients to ascertain if medical visits are not in rapid succession, limiting reliability of, arguably, the most useful component of these questionnaires. Until these components can be confronted and corroborated with real-time data, I question the degree to which questionnaires can meaningfully improve upon our knowledge in the context of suicide prediction in clinical settings and prove useful in ongoing ML strategies.

    [1] Weissman MM, Pathak J, Talati A. Personal life events–a promising dimension for psychiatry in electronic health records. JAMA Psychiatry 2019. Advance online publication. https://doi.org/10.1001/jamapsychiatry.2019.3219
    [2] Fazel S, O’Reilly L. Machine learning for suicide research–can it improve risk factor identification? JAMA Psychiatry 2019. Advance online publication. https://doi.org/10.1001/jamapsychiatry.2019.2896
    [3] Kessler RC, et al. Machine learning methods for developing precision treatment rules with observational data. Behav Res Ther 2019; 120. Advance online publication. https://doi.org/10.1016/j.bra
    [4] Marquand AF, et al. Conceptualizing mental disorders as deviations from normative functioning. Mol Psychiatry 2019; 24: 1415-1424.