Creating an electronic system to monitor blood pressure control among kidney-disease patients was no easy task in a study from Brigham and Women's Hospital in Boston. Such a system forms the basis of performance measures for its practitioners and for population health management efforts.
But relying on EHR problem lists to initially identify patients with chronic kidney disease for the study proved inadequate because physicians often don't log vitals there, according to the paper published at the Journal of the American Medical Informatics Association. In fact, just 281 patients were found from problem lists during the two-year study period, while an additional 1,377 were identified from billing codes. Of them, 1276 patients had blood pressure not at goal, set at 130/80 mm Hg.
In gleaning the number whose blood pressure was at goal, the researchers also employed natural language processing (NLP) to glean data from doctors' free-text notes and resorted to a paper worksheet given to doctors at the point of care. There they could explain other circumstances, including BP readings taken at home and the reasoning behind physician action or inaction.
Using an algorithm that incorporated multiple BP readings increased the rate of control from 37.1 percent to 42.3 percent. That grew to 52.6 percent using NLP to capture BP readings from notes. Data from point-of-care worksheets indicated that in 52 percent of visits, patients identified as not having controlled BP actually were at goal based on BP readings taken at home or on that day in the office.
"Despite our goal of developing an electronic method to assess performance, we were limited in that we could not integrate all information necessary to fully assess BP control," the authors wrote, adding that that included assessing whether the physician had taken appropriate action.
The emphasis on structured fields in the Meaningful Use program will help, but will not be sufficient because not all needed information will be in structured fields. While NLP can parse free-text data, it does not identify all the nuances of care that go into determining whether the care was appropriate, the authors said.
Capturing the nuances of care remains a vexing problem in the push to digital records. Though EHR problem lists can get crowded, a study published at BMC Medical Informatics and Decision Making found that doctors say more information is better in that field.
A study of rule-authoring tools used to convert medical knowledge into machine-executable clinical decision support rules across Partners Healthcare in Boston, however, found many limitations--and frustrations.
Meanwhile, Ohio State's Department of Biomedical Informatics, with a $1.3 million grant, is working on a "fusion" technology to piece together data stored in narrative text fields and multiple data sets to speed up recruitment of patients for clinical trials.
To learn more:
- read the abstract