My favorite acronym in medicine has always been “Victim of Modern Imaging Technology”, as it so practically fits with our day to day clinical practice. How often have we gotten images or reads that over-call a given problem leading us down a rabbit hole of more testing and (heaven forfend) more surgery or treatment that might hurt the person for whom we’re caring?
It is for this reason that we praise the efforts of Lee and coworkers from University of Chicago who have taken a good look at this.
Magnetic resonance imaging (MRI) is generally considered the most sensitive imaging for diagnosis of osteomyelitis; however, it is associated with significant cost and is at times ordered as initial screening imaging when a less resource-intensive test would suffice. The purpose of this retrospective cohort study was to examine the differences between patients with osteomyelitis of the foot and ankle, and their subsequent treatment course, who underwent MRI compared with those who did not. Financial impact of MRI as it relates to clinical decision-making was also calculated. Patients treated for a diagnosis of osteomyelitis of the foot and ankle from 2009 to 2015 were retrospectively identified. Demographics, imaging modalities, and operative procedures for each patient were collected. An “impact MRI” was defined as one that led to a subsequent operative procedure within the same admission. The impact cost of an MRI was estimated using the equation: (average MRI cost) × (total MRIs/impact MRIs). A total of 144 patients underwent 220 MRIs, and 399 patients did not have MRIs. The operative rate between the 2 groups was similar (70.8% versus 70.4%, p = .93). Multiple linear regression showed that MRI was not a significant predictor of operation (p = .50). However, we found a significant correlation between MRI use and operative intervention for patients with increased comorbidities. From 2011 to 2015, there was a significant increase in impact cost, while controlling for average MRI cost ($8172 to $15,292, p ≤ .05). Over the study period, the impact cost of an MRI significantly increased from 1.8 to 5.0 times the average cost.
Dr. Armstrong:
No one denies that MRIs are often overused in the clinical setting, and that their often high case specificity is often misinterpreted at the expense of a consideration of their case specificity (NB, the ROC curve.) The “art” of any statistical analysis strongly depends on a correct and necessarily subjective use of Bayes theorem. IMHO, A cursory review of the impact cost equation seems incorrect as the impact MRIs should be in the numerator and the total MRIs should be in the denominator. Also, being a retrospective review, the analysis suffers from result knowledge confounders not found in prospective analyses. Please respond.
Respectfully,
A statistically minded clinician
I couldn’t agree more with your stressing the importance of understanding pre-test probability to any clinical question1 especially this one. Blindly ordering tests without that understanding is a recipe for overuse.
What about antibiotic courses for osteo averted by negative MRIs? So many things other than surgery go into this…
Agreed!