Important mental health history is often present in medical records, but it can be difficult to find, especially when it is missing from the diagnosis codes that clinicians, researchers, and health systems use to search and count symptoms.
A new study led by researchers at the University of New Mexico School of Medicine analyzed the electronic medical records of more than 1.3 million patients served by the Veterans Health Administration (VHA). Researchers highlighted common gaps in how health systems track self-harm, finding that diagnosis codes only capture about a quarter of clinically recorded self-harm history.
“In research and planning, counting only those that are more visible in diagnosis codes can significantly underestimate the need for mental health services,” said Dr. Christophe Lambert, professor and interim chair of the Department of Translational Informatics in the UNM School of Medicine’s Department of Internal Medicine and corresponding author of the study. “Better measurements will help health systems plan better, help researchers study care more precisely, and ultimately help clinicians know when a patient needs further workup.”
This research Medical Internet Research Journal, We used a new machine learning technique previously developed by members of our research team. After expert chart review and statistical calibration, the researchers estimated that documented self-harm was present in approximately 7.9% of patients seen by VHA clinicians. This is more than four times the 1.85% that can be confirmed by diagnosis codes alone. This gap is important because missing history can impact clinical recognition, research results, and planning of mental health services.
The problem list, a provider’s summary of patient health status, showed another visibility gap. Although these are intended to flag conditions of importance to clinical teams, in real-world healthcare they are not always complete or consistently maintained. Among veterans with a diagnosis code for self-harm, 22.6% had self-harm or a history of self-harm listed on the VHA problem list. That is, even if the diagnosis code included self-harm, one of the most prominent summary fields in the record often did not include self-harm.
Past self-harm is clinically important because it is one of the most important predictors of future self-harm and suicide risk. It also determines how care is provided, including how clinicians think about depression, PTSD, bipolar disorder, substance use, traumatic brain injury, and other conditions that may occur with self-harm.
The authors note that VHA already uses specialized suicide and overdose reporting tools and does not rely solely on diagnosis codes or problem lists to monitor suicide risk. This study focused on another related question. To what extent does past self-harm history appear in the parts of records that can be most easily quantified and reviewed at scale by researchers, care teams, and health systems?
“This is a system-level visibility issue,” Lambert said. “Records can be large. In our medical record review, some patient records contained over 500,000 lines of notes. No clinician can be expected to read all of that during a routine office visit.”
The study did not predict future self-harm or reliably determine whether patients had ever self-harmed. Instead, the researchers tested whether a computer model could use patterns in structured electronic health record data to estimate the probability that a history of self-harm was present but missing in the diagnosis code, and compare that probability with an expert review of clinical records.
To do this, the team used a method called . PULSNAR – Positive learning without labels chosen rather than randomly, It was built for messy real-world health data. Most machine learning techniques require clear examples of both “yes” and “no” cases. However, in medical records, a missing diagnosis code does not prove that the patient did not have the disease.
PULSNAR works with that uncertainty in mind. It learns from patients who have a code and estimates how many similar patients exist among those who do not have a code. Its main advantage is that it does not assume that the coded cases are random and takes into account the fact that some cases are more likely to be coded than others.
Dr Praveen Kumar, lead author of the study, said: “Medical records can obscure self-harm in many ways.” “The medical history may be in the clinician’s notes but not in the diagnosis code. Records may also include risk factors, injuries, poisoning, and behaviors consistent with self-harm, even though records alone cannot prove what happened or why.
“Our method helps flag both patterns for reconsideration. This study was able to verify the first pattern because the evidence was already in the notes. The second pattern may be important as well, but we would need to talk to the patient or use information beyond the medical record to confirm it.”
The research team included experts from UNM Health Sciences Center, Raymond G. Murphy Veterans Affairs (VA) Medical Center, Vanderbilt University Medical Center, VA Tennessee Valley Healthcare System, VA Office of Mental Health, Greer Black Company, and UNM Office of Economic Affairs. The team brings together expertise in medical informatics, computer science, psychiatry, biomedical informatics, economics, statistics, and health services research.
Researchers say the self-harm study is part of a broader research program that uses positive, label-free learning to uncover conditions that may be under-recorded in standard medical data. The research team has already published related work using this approach. Detecting undercoded opioid use disorderAnd ongoing research is expanding its application to other conditions where medical records may not provide a complete picture, such as unrecognized PTSD, depression, bipolar disorder, and sleep disorders.
This method has the potential to complement broader VHA mental health and suicide prevention efforts by adding a scalable method to measure conditions that are under-recorded or difficult to ascertain in standard medical data. The researchers emphasized that while the method is still a research tool and not ready for use on its own in clinical care, further development could help health systems better estimate under-recorded mental health conditions, uncover documented medical history that is not clearly visible, and identify records that may require closer consideration.
“History of self-harm is too important to be buried in records that are impractical to review line by line during routine care,” Lambert said. “Our work is to help researchers and health systems find documented histories and clinically relevant patterns in their data, allowing health care teams to see a more complete picture of the people they serve.”
sauce:
University of New Mexico Health Sciences Center
Reference magazines:
