A study examining chest CT exams ordered using a decision support system has found studies with a high utility score in the clinical decision support (CDS) system also have a high probability of positive findings, according to a research abstract accepted at the Society for Imaging Informatics in Medicine (SIIM) annual meeting.

“The main reason to perform this study was to actually look at the scores which the decision support system provides to the referring physician and look at the exam outcomes following those scores,” the study’s lead author, Supriya Gupta, MBBS, MD, of the department of radiology at Massachusetts General Hospital (MGH), told Health Imaging. The study was a first of its kind pilot study to correlate utility scores with outcomes at Gupta’s organization.

Using a natural language processing tool, data from MGH’s RIS and order entry system were retrospectively mined to retrieve chest CT exams from 2005 to 2010. Fields such as exam code and patient demographics were selected from RIS, while ordering indications and utility scores (on a scale of zero to nine) were selected from the decision support system. Common indications from ordering physicians were recorded, and exams were grouped into high, intermediate and low score groups based on utility score.

Overall, of the nearly 94,000 chest CT reports analyzed, 76 percent had positive findings. The proportion of exams with high, intermediate and low scores were 87.45, 0.42 and 12.13 percent respectively. Breaking the numbers down further, exams with a high utility score had a positive finding rate of 76.6 percent, compared with 72 percent for exams with a low score.

Gupta and colleagues also looked at some of the most commonly selected indications, and found outcomes varied for each. For example, the positive finding rate for highly scored reports when one of the selected indications was “abnormal chest x-ray” was 86 percent, while the positive finding rate when one of the indications was “lymphoma” was 65 percent.

“In general, more than one-third exams with high decision support scores (7-9) had presence of report findings. While the reports with low scores (score 0-3) had only few reports with positive report findings,” wrote the authors in the study abstract.

Gupta said that it’s good for physicians to understand how likely positive findings are when given a certain utility score by the decision support system, but the implications extend beyond the referring physician. Administrators, healthcare policy makers and developers of such systems could all find value in the outcome data. “It can emphasize the importance of the current guidelines and can even help modify these guidelines to include other things based on the outcomes,” she said.