Rory Spiegel, MD, is the chief resident in emergency medicine at Newark Beth Israel Medical Center in New Jersey. In his free time he enjoys riding in TARDISes, defending the Wall, and “chasing the wind.” He is a self-described nerd who blogs regularly on nihilism and the art of doing nothing at EMnerd.com, where a version of this article originally appeared.

MedPageToday.com
Published: Aug 19, 2014
By Rory Spiegel MD
Story Source

Action Points

• This editorial suggests that the high-sensitivity of MRI may lead to clinically irrelevant diagnoses when the modality is used to evaluate cervical tenderness after trauma.
• The study under discussion examined outcomes in patients who had received imaging for neck trauma, but was not structured to perform both CT and MRI in all patients.

In this post, Spiegel examines a recent study of the Clinical Relevance of Magnetic Resonance Imaging in Cervical Spine Clearance from JAMA Surgery.

The Enigma

Removing a cervical collar in the early aftermath of a traumatic injury is becoming an increasingly difficult task. With ever more sensitive imaging modalities we have progressively devalued the traditional methods used to evaluate the integrity of the spinal column in favor of more technologically advanced ones.

Despite decades of success in treating this pathology and clear evidence that clinically relevant spinal injuries present with obvious clinical signs, we have let anecdotal evidence get the best of us. With this in mind, we now turn to the enigma that is the neurologically intact patient with persistent midline tenderness with no evidence of pathology on cervical CT.

As we concluded in our previous post, in the neurologically intact patient with persistent midline tenderness, MRI identifies far more injuries than CT. In a cohort of 178 prospectively gathered patients with isolated persistent midline tenderness and a negative CT, Auckland et al. reported 78 (44%) with injuries identified on MRI. Although the majority required no intervention, 33 (18.5%) required use of a collar and five (2.8%) required surgical management.

These findings taken at face value are concerning to say the least, and do not fit with our clinical experience. In fact, there is reasonable evidence demonstrating this increased signal found on MRI is merely the noise of an overly sensitive test applied to an extraordinarily low-risk population.

MRI is prone to overcalling pathology. Even a surprising number of asymptomatic healthy controls, with no history of acute trauma, will have radiologically significant pathology found on MRI.

Furthermore, when findings on MRI are compared to the injuries identified during surgical exploration, MRI demonstrates a propensity for identifying lesions where none exist(specificities ranging from 59.0% to 80.5%). Given this, it no longer seems appropriate to consider MRI the gold standard for defining disease in acute spinal trauma. Rather we should examine clinical follow-up and functional patient-oriented outcomes.

Simply put, what would happen to these patients if we just left well enough alone?

The Investigation

A recent article published in JAMA Surgery by Resnick et al. attempted to examine this very question. The authors investigated the utility of MRI in patients with persistent midline tenderness or sensory deficits and normal CT findings. Only instead of using MRI as the gold standard they used the patients’ discharge diagnosis.

In this prospectively gathered cohort the authors included all patients with a GCS of 15, who were not intoxicated, and had no distracting injuries. Of the 830 patients included in this trial, 164 (19.8%) had cervical spine injuries. Twenty-three (2.8%) of these were deemed clinically significant, all were identified on the initial CT scan. Only 15 (2.2%) of the patients had injuries identified exclusively on the MRI, none of which were deemed clinically relevant.

Unfortunately, due to the pragmatic nature of this trial, not all patients received an MRI. The decision was left up to each individual treating physician. Ultimately, 100 of the 830 patients received an MRI during their hospital stay. The most common reasons an MRI was ordered were equivocal findings on CT followed by persistent midline tenderness, or sensory deficits concerning enough for the treating physician to require further investigations.

Similar to the Ackland study, 46% of the patients who underwent MRI imaging were found to have additional findings that were not seen by CT. The majority of these were ligamentous and soft tissue injuries and none altered clinical management. Like the Ackland study, the MRI identifies far more pathology, very little of which is clinically relevant.

Compared to MRI, CT was 90.9% (CI from 85.3%-94.8%) sensitive for identifying cervical spine injury. When discharge diagnosis was used as the gold standard, Resnick et al. assert 100% sensitivity and specificity for diagnosing clinically important cervical spine injuries.

Unfortunately, long-term follow-up to test the validity of these findings was not performed. Nor were there a sufficient number of patients with serious cervical injuries in this cohort to claim 100% sensitivity with any certainty (confidence interval as low as 85.1%). A total of five patients were discharged home wearing C-collars for comfort, the rest of the patients had their collars removed before discharge.

Of the patients with negative CTs and persistent midline tenderness, removing the collar prior to discharge did not result in catastrophic injury. There were not any reports of patients readmitted to these medical centers with obvious cervical spinal injuries. We are unable to determine how these patients did in the short term after discharge.

There may, though unlikely, have been a catastrophic injury missed that presented to a different hospital. It is also unclear if any minor injuries that may have benefited from earlier intervention went undetected, though this later scenario is even less likely as C-collar use for comfort has for the most part been debunked as a useful therapy.

Clinical Judgment

How we use this information is still not entirely clear. All midline tenderness is not created equal.

There is a certain degree of clinical judgment that should be applied when evaluating these patients. Maybe patients with persistent tenderness who are unable to actively range their necks through 45 degrees of rotation (a retrospective application of the Canadian S-Spine Rule) are more concerning. Maybe those with bilateral paresthesia are those who merit further investigation. Maybe, like the Ackland study demonstrated, patients with severe cervical spondylosis on CT scan cannot be cleared by this modality.

Performing MRIs on the majority of these patients will lead to a significant increase in pathological diagnoses. Most of these will be of little clinical significance and the few true positives are likely to reveal themselves clinically during the patient‘s stay in the emergency department. If we insist on imaging all patients with persistent pain or tenderness, we risk exposing a group of patients, the large majority of which are without true clinical disease, to potentially harmful interventions.

Some will be asked to follow-up with spine surgeons for further downstream testing. Some will be given a hard collar for 10 weeks and exposed to all the associated morbidity. Others will be exposed to surgical procedures that may very well not be clinically required. All will be turned into patients, given a label, diagnosed with a disease that’s major determinants of long-term prognosis are patients’ mental well being and financial security.

Dx: Overdiagnosis

We live in a world of ever–advancing medical technology. A world where the boundaries between states of disease and health are becoming increasingly less defined. It is easy to demonize nonspecific laboratory investigations like D-Dimer or procalcitonin for their intellectual dishonesty. Likewise, the CT scan is an equally natural scapegoat because of its accessibility and the obvious concerns of radiation.

Although each of these culprits is responsible in its own way for the crisis we currently face, the real perpetrator of overdiagnosis is information and the ambiguity it hurls at us. We have clearly demonstrated that modern medicine in its current form is incapable of standing idle. Our desire to act far overwhelms our powers of reason.

Though the current data cannot definitively negate the utility of MRI in the neurologically intact patient with persistent midline tenderness, we can say its indications are few and far between. Used empirically, it will surely lead to far more harm than good.