When physicians indicate a medical situation is an exception to a guideline, most of the time these medical decisions are correct. These findings have implications for improved decision-support tools, according to a study published in the Feb. 16 issue of the Annals of Internal Medicine.
In the observational study of patient medical records, Stephen D. Persell, MD, of Northwestern University in Chicago, and colleagues evaluated whether EHRs can correctly identify when a physician's order does not meet the quality guidelines.
“An increasing proportion of apparent quality deficits—identified by clinical decision-support systems or on performance reports—may actually be incorrect,” wrote the authors. “If physicians believe that the tools used to measure their performance are incorrect, they may not heed decision support or they may dismiss the results of performance reports, even if they are not achieving the desired results.
Persell and colleagues evaluated 650 standardized medical exceptions recorded at a single center between Feb. 9, 2008, and Sept. 10, 2008, by “creating decision-support tools in EHRs that allowed providers to enter medical or patient reasons for not allowing point-of-care alerts.”
During the study, a peer-review panel analyzed and judged the occurrences of medical misconceptions to review feedback for how often inappropriate medical exceptions led to changes in patient care. The researchers also analyzed decision support for 16 chronic diseases and preventive care quality measures relating to coronary heart disease, heart failure, diabetes mellitus and prevention and screening.
Of the 650 records evaluated, physicians used medical-exception reporting tools 5.5 percent of the time when the reasons for not following decision support were not related to a medical reason. Of the remaining 614 patient records, 93.6 percent were deemed appropriate, 3.1 percent were inappropriate and 3.3 percent were of “uncertain appropriateness."
The frequencies of inappropriate and uncertain exceptions were:
- Coronary heart disease -- seven and 10, respectively,
- Heart failure -- zero and two, and
- diabetes -- 10 and eight.
Of the medical exceptions reported by physicians, a total of 78 (12 percent) “were instances in which a clinician recorded that a diagnosis that triggered a quality alert was not present.”
According to the study, peer-reviewers disagreed with a physician’s medical exception 10.2 percent of the time and were uncertain of the exception 2.6 percent of the time.
Of the records under review for preventive services, 334 were judged as appropriate, while 2 were inappropriate. In the cases which received feedback, eight of 19 resulted in a change in management decision. Two of the 12 records deemed "uncertain exceptions" resulted in a management change after feedback.
“Our study shows that the reasons physicians entered are valid most of the time,” the authors wrote. “We can use this information to deliver more accurate performance measurement to physicians, make subsequent clinical decision support more accurate and make the rationale for deviating from guidelines visible to providers so it may be used in clinical care.”
Because this review could be useful to educators and practice guideline developers, the authors suggested that a more efficient, practical approach to the review process is necessary for future evaluations. Using EHR-based decision-support tools or omitting peer review for exceptions to quality measures in which “physicians rarely recorded inappropriate exceptions,” could be useful, the researchers wrote.
The Agency for Healthcare Research and Quality funded the study.