Chart review tougher than claims analysis in utilization analysis of lumbar MRI

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - BackPain2

Utilization reviewers seeking to ascertain whether MRI was inappropriately ordered for low-back pain might assume that direct review of patients’ charts would tilt toward patient empathy—that is, largesse—while aloof analysis of claims data would reflect a higher bar to clear.

They would be wrong, as chart review classified more scans as inappropriate than did the claims-analysis method in a study comparing the two.

The study report is running in the September Journal of the American College of Radiology.

Researchers at Stanford University and the VA Palo Alto Health Care System in Menlo Park, Calif., reviewed the cases of the first 146 veterans who underwent lumbar spine MRI (LS MRI) in the outpatient setting in fiscal year 2012.

The team determined the sensitivity and specificity of the low-effort administrative (claims review) method, regarding inappropriateness as the positive result and considering labor-intensive chart review to reflect the true state.

Using McKesson’s InterQual criteria for chart review and CMS’s National Quality Forum-endorsed system for evaluating the administrative data, they found that, compared with chart review, the administrative approach had specificity (correctly assigning scans that were appropriate) of 82 percent and sensitivity (not finding the full number of inappropriate scans) of 27 percent.

More surprisingly, 59 percent of the 146 scans were flagged as inappropriate upon chart review while the administrative method identified only 23 percent as inappropriate.

The researchers conclude that the administrative method is “economically feasible for identifying the overuse of LS MRI, but it underestimates the true extent of inappropriate ordering.”

In their discussion, Lead author Tigran Avoundjian, MPH, of the VA and colleagues cite an economic analysis showing that more appropriate orders of LS MRI could lower annual U.S. healthcare expenditures by $300 million.

In fact, the exam has often been called a poster child for unneeded health spending, and it rose to the top in a recent study of low-value health services based on data on nearly 1.5 million privately insured patients.

In the present study, Avoundjian and team conclude that analyzing administrative data can be a low-effort, and thus relatively thrifty, way to identify inappropriate LS MRI orders. The problem is that this approach may underestimate the proportion of inappropriate scans.

For its part, chart review takes into account the more detailed information contained in medical records, but it’s probably too pricey a way to routinely measure performance, much less evaluate a large enough sample to compare providers with statistical precision.

Avoundjian et al. acknowledge as limitations their study’s use of data from just one facility and over just one fiscal year.

“Findings might be different if we had evaluated orders over time or across multiple facilities,” they write. “The patients in the sample are similar to the general VA population, a mostly elderly, white and male population, but they differ from patients in other parts of the U.S. healthcare system.”