If your institution adopts clinical decision-support (CDS) software, will you drive away referring physicians who may become frustrated with the system? Fortunately, there's no need to worry, if the experience of Massachusetts General Hospital (MGH) is any indication.
An MGH team analyzed orders for imaging studies that had low appropriateness scores in its CDS system and were then canceled; the researchers found that only a handful of exams were performed later at an unaffiliated hospital or imaging center. This indicates that CDS doesn't lead to "leakage," which occurs when physicians go outside their existing network to have services performed.
"Since CDS does not drive leakage, organizations can feel comfortable implementing clinical decision support for Medicare and non-Medicare populations without fear of driving studies outside of the institution," said lead author Dr. Anand Prabhakar.
He presented the findings of the study, a large retrospective review over a three-year period, during a scientific session at the RSNA 2015 meeting in Chicago.
Evidence-based ordering
Appropriateness criteria are useful for helping clinicians make evidence-based choices when ordering imaging studies. Order-entry CDS systems bring these criteria to the point of care, and they have been shown to reduce the number of low-utility exams and limit the growth of imaging, Prabhakar said. Importantly, the U.S. Centers for Medicare and Medicaid Services (CMS) will also mandate the use of CDS at some point in 2017.
Dr. Anand Prabhakar from MGH.There are some downsides to CDS, however. For example, a 2007 study from Oregon Health and Science University reported that CDS systems were rigid and caused fatigue among the providers using them. In addition, other evidence has suggested that physicians could be antagonized by software that questions their judgment, according to Prabhakar.
"Thus, providers could theoretically opt to avoid CDS and send their imaging to other locations that perhaps don't have CDS," he said.
This potential referral of imaging out of network could be considered leakage, he added. Leakage would be defined as diagnostic imaging that's performed outside of an institution on patients who typically get their care at that location. These studies could be performed at an institution that is either affiliated or nonaffiliated with the local enterprise.
Uncertain appropriateness
Out-of-network leakage of imaging studies could have very important implications, he said. For example, studies that have been ordered without CDS software could have uncertain appropriateness. There could also be impaired coordination of care, as images might not be transferred back to the local institution.
"Perhaps there's [also] lower quality of imaging done at centers that may not be able to answer the clinical question [for] why the imaging was ordered to begin with," Prabhakar said.
What's more, these studies may increase out-of-pocket costs for patients and affect financial incentives under risk-sharing contracts and accountable care organizations, he said.
As a result, the researchers set out to determine whether CDS systems contributed to out-of-network leakage of advanced outpatient imaging studies. The study included patients who have a primary care provider at their institution and were participating in one of three risk-sharing insurance contracts. The team mined insurance payment data and obtained the outpatient MR/CT imaging payment data for these patients from January 1, 2011, to December 31, 2013.
Imaging CDS was first implemented at MGH in 2001, and it's used by 98% of the institution's referrers, Prabhakar said. The system is based on the American College of Radiology's Appropriateness Criteria and flags low decision-support scores in red on an ordering screen. The researchers queried the CDS database for CT/MR orders that were flagged red and then ultimately canceled by the referring provider. They then cross-matched these low CDS scores with all imaging studies performed within 60 days of the initial order.
The study population of 63,378 patients had received a total of 36,022 CT and MRI studies over the three-year period. Of these, 11,234 (31.2%) were performed outside of their home institution. Patients received 3,513 of those imaging studies at affiliates, while 7,721 (21.4%) were performed at nonaffiliated hospitals or imaging centers. The amount of imaging leakage was relatively consistent over the three-year period.
The total leakage rate was comparable to other nationwide estimates of 24% to 30%, according to Prabhakar.
"This has many implications for population health and risk-shared contracts," he said.
No impact from CDS
Turning to the CDS data, the researchers found that 4,866 studies were initially flagged as low value and were subsequently canceled by the referring provider. Of these, 111 were ultimately performed anyway within 60 days, but only five (4.5%) were provided at outside facilities.
The findings led Prabhakar to conclude that "clinical decision support does not contribute to leakage out of network."
He did acknowledge a number of limitations to the research, including its reliance on a single institution.
"However, this analysis could be performed at your own hospital," he said.
The study also used a 60-day time interval to assess for leakage; however, the team felt that this was a reasonable time frame to determine whether a referring provider was attempting to get around the CDS system, according to Prabhakar.
In addition, the researchers did not know if the outside imaging centers were using CDS. Also, only six (60%) of MGH's 10 affiliated practices have CDS, he noted.
