Deciding whether or not to order a CT for a patient with a mild head injury is a guessing game: Scan to play it safe or trust clinical observation and risk missing something. Clinical tools designed to take the guesswork out of these decisions are controversial, in part because costs have to be considered and because across-the-board rules are never failsafe.
In back-to-back studies published in the Journal of the American Medical Association, independent teams of Dutch and Canadian researchers compared two of the most widely used criteria and found both equally capable of identifying almost every patient who should have CT; although one rule offers more savings and both guidelines would miss a few patients who might benefit from scanning.
Both the Canadian CT head rule (CCHR) and the New Orleans criteria (NOC) were highly sensitive – predicting virtually all cases of intracranial trauma requiring neurosurgery. The CCHR, however, was more specific in predicting important clinical outcomes and was better at reducing unnecessary imaging and related costs.
COMPARING SCALES FOR CT DECISIONS
The NOC call for a CT if a patient has a Glasgow Coma Score (GCS) of 15 and at least one of seven other symptoms: headache, vomiting, drug or alcohol intoxication, age over 60 years, short-term memory deficits, physical evidence of trauma above the clavicles, or seizure.
The GCS is based on eye, motor, and verbal abilities. A score of 13 to 15 indicates mild injury. The CCHR, which applies to patients with a GCS score of 13 to 15, requires CT only if a patient shows amnesia, confusion or a loss of consciousness. Patients are also considered at high risk for neurosurgery if they have any of the following: GCS less than 13 two hours after injury, open or depressed skull fracture, basal skull fracture, two or more episodes of vomiting, or are 65 years or older.
Led by Marion Smits, MD, of Erasmus Medical Center in Rotterdam, in the Netherlands, the first team evaluated the criteria in 3,181 patients. The NOC identified up to 99 percent of important neurological findings compared to 83- to 87-percent with the Canadian guides, but the CCHR had greater specificity (37.2 percent to 39.7 percent) than the NOC (3 percent to 5.6 percent).
“For patients with minor head injury and a GCS score of 13 to 15, the CCHR has lower sensitivity than the NOC … but would identify all cases requiring neurosurgery, and is more likely to reduce the use of CT,” Dr. Smits said.
In the second study, researchers led by Ian Stiell, MD, of the University of Ottawa in Canada, compared the rules in 1,822 adults with GCS scores of 15. Both the NOC and the CCHR had 100 percent sensitivity for predicting the need for surgery, but they too found that the CCHR was far more specific (76.3 versus 12.1 percent for the NOC). The Canadian guidelines could have reduced unnecessary CT to 52.1 percent of injuries compared to 88.0 percent using the NOC guides.
Figure. Dr. Marion S...Image Tools
“Current use of cranial CT for minor head injury is increasing rapidly, but is highly variable and inefficient,” said Dr. Stiell. “Our data show that 90 percent of CTs are negative for clinically important brain injury.”
Figure. CT scans sho...Image Tools
CCHR: MISSING SOME PATIENTS
In an accompanying editorial, Micelle Haydel, MD, who helped develop the NOC, said that while the CCHR guides might reduce CT use, some patients would be missed.
Dr. Haydel, Assistant Director of Emergency Medicine at Louisiana State University Health Sciences Center in New Orleans, pointed out that the NOC identified almost all patients with clinically important injuries while the CCHR missed up to 17 percent of patients who should have received scans.
“Widespread application of the CCHR could certainly reduce CT use but at the cost of missing intracranial injuries that Stiell and colleagues label ‘clinically unimportant,’” she wrote.
As many as one million mild head injuries are treated by US emergency rooms each year, although fewer than one in ten persons with normal consciousness have an abnormal CT scan. Of these, less than 1 percent of patients require neurosurgical treatment, Dr. Haydel noted.
Dr. Haydel added that for decision guidelines to be implemented effectively, physicians must have confidence in the accuracy of the guidelines. “Many physicians find the idea of any ‘missed’ intracranial injuries unacceptable, and a study of patients with intracranial injuries, similar to those considered ‘unimportant’ in the CCHR studies, found that 12 percent of such patients did not need surgery or had poor neurosurgical outcomes.”
Concern over lawsuits is another barrier, she said, citing a study in which 93 percent of surveyed physicians admitted practicing defensive medicine on occasion. “When clinical suspicion is high, further testing is indicated, regardless of the results of decision guidelines.”
Commenting on Dr. Haydel's editorial by e-mail, Dr. Smits agreed that any intracranial finding is considered to be clinically important and generally leads to clinical observation for a minimum of 24 hours.
“An important point in Dr. Haydel's editorial is the fact that the NOC and CCHR were designed for specific study populations and are not directly applicable to all minor head injury patients seen in emergency departments,” Dr. Smits noted.
“As Dr. Haydel points out, the CCHR cannot be applied to patients with posttraumatic seizures, and neither the CCHR nor the NOC can be used in patients taking anticoagulants. In our validation study, we adapted the original decision rules, by adding exclusion criteria as additional risk factors, for use in our entire population of minor head injury patients.
Validation of the decision rules still showed a 100 percent sensitivity of both rules for surgery, with a sensitivity of 99 percent for the NOC and 85 percent for the CCHR to show clinically important CT findings. This adaptation makes validation and eventual implementation of the decision rules possible.
Sherman Stein, MD, Clinical Professor of Neurosurgery at the University of Pennsylvania, is in the final stages of revising the process of his own study of CT costs and mild head injury decision guideline outcomes.
He told Neurology Today that his findings also indicate that CT decision guidelines seem to save money without sacrificing patient care, although he agreed that other factors must also be considered.
“These are fairly sophisticated guidelines. But if a patient has a GCS score of 14 or 15 and there is no loss of consciousness or amnesia, scanning is probably not necessary,” he said. “However, liberal use of CT for patients with mild head injury is not too expensive, and it reduces costs by eliminating skull radiography and unnecessarily prolonged observation in the emergency room or, in some cases, hospital admission.”
CT also provides better results than those obtained by observation alone in many cases. “Even though we don't have any standard guidelines yet, and the CCHR hasn't been tested in a sufficient number of patients, it seems to reduce the need for CT by at least 50 percent. Specificity is a little bit better with the Canadian rules. Instead of scanning 80 percent of mild head injuries, you scan 50 percent. It's clearly more cost-effective than observing patients for a while. It always takes time for rules like these to be accepted, but I believe decision guidelines are the way to go.”
Figure. Dr. Ian Stie...Image Tools
Other CT decision tools “in the pipeline” may further improve on these results as they are revised and refined, he said.
SKEPTICISM ABOUT EITHER GUIDELINE
That said, some experts question whether the current guidelines are adequate. Jonathan Glauser, MD, a physician with the Cleveland Clinic's Department of Emergency Medicine, questioned the applicability of either set of criteria when scanning is available.
In a 2004 article published in the Clinic's journal, Dr. Glauser said that prompt CT actually saves money by obviating the need for hospital admission or prolonged observation in the emergency room. He told Neurology Today that neither the NOC nor CCHR decision tools reflect his experience.
“I'm not satisfied with either the NOC or the CCHR. I don't think the final study has been done, and I doubt there ever will be one,” he said. “I don't believe everyone with a mild head injury needs CT, especially if they are completely lucid, but I also don't think the Glasgow Coma Score or loss of consciousness accurately reflects a patient's status or risk.”
Dr. Glauser says physicians need the flexibility to decide who needs a scan. “Loss of consciousness is probably a red herring,” he said. “A patient's neurological status can't always be determined by their recall of events immediately before or after a head trauma. I don't treat a GCS of 13 or 14 as a mild head injury – I don't even view a 15 as a mild head injury. All it means is that the patients can move their eyes. You can have a GCS of 15 and still be pretty confused after a head injury. It takes a lot more observation by the physician in an emergency situation.
“People with a head injury typically don't give a good history,” Dr. Glauser continued, “even if they are lucid, but ultimately if they appear completely normal I don't recommend CT.”
Reducing scans for mild head injuries by 10 percent could save more than $20 million per year, according to one estimate. Another review and decision analysis of four cost studies concluded that a CT based strategy for mild head injury is less expensive than one based on in-hospital observation, and would reduce costs, on average, by about one third (Emerg Med J 2004; 21:54–58).
ARTICLE IN BRIEF
✓ In back-to-back studies, independent teams of Dutch and Canadian researchers found that the two most widely used criteria for determining when CT should be used for mild brain trauma were equally effective; although one rule offers more savings and both guidelines would miss a few patients who might benefit from CT.