As EHRs become more interoperable, medical professionals can leverage the “vast amount” of data in them to better determine screening, diagnoses and outcomes, said Reis, an assistant professor at Harvard Medical School and researcher in Children's informatics program. “We want to look at all conditions and ask which human diseases are predictable.”
The study of EHRs and domestic abuse has indicated information technology can help detect such cases much earlier. Domestic abuse is the most common cause of nonfatal injury to women in the U.S., and it's a condition that is hard for doctors to detect, Reis said. Patients cover up the reasons they come in with bruises and frequently alternate providers to mask repeated injuries. While there are screening protocols doctors can use to ask questions of patients, domestic abuse cases continue to fall through the cracks, he said.
The research includes medical records from more than 500,000 non-identifiable patients who had at least four years worth of hospital admissions and emergency department visits data. The computer model developed by Reis and the research team integrates all the ICD-9 codes used to document medical information and scores them to determine which conditions lead to a higher indication of domestic abuse. Out of thousands of codes for things like contusions, lacerations, and psychological issues, hundreds have been flagged.
The model was able to predict domestic abuse cases in those medical records an average of 10 to 30 months earlier than the time that the patients received their diagnosis, Reis said.
Reis said he sees the model—currently a prototype being turned into a working application through a four-year, $1.3 million grant from the National Library of Medicine—as becoming a screening support system. Doctors will be able to get a broad perspective of a medical history before they meet with a patient and use that snapshot to ask better questions and determine accurate diagnoses faster, he said.
It's similar to genetic profiling of patients, viewed in the past several years as a way to predict when a person might develop certain conditions. By using medical history and IT, medical professionals can gain that same picture, Reis said. “Fewer people have looked at the clinical side.”
What do you think? Write us with your comments at [email protected]. Please include your name, title and hometown.