The year 2019 marks the 10th anniversary of the publication of the Pediatric Emergency Care Applied Research Network clinical decision instrument for the identification of children at very low risk of clinically important traumatic brain injury (PECARN TBI).¹ It is the largest, but hardly the only, entrant in the field. The United Kingdom contributes the children’s head injury algorithm for the prediction of important clinical events (CHALICE), and Canada contributes the Canadian assessment of tomography for childhood head injury (CATCH).^2,3

These instruments arose from recognition of the increasing use of computed tomography (CT) in children as well as harms from radiation often exceeding the risks from missed injuries. These decision instruments, PECARN TBI in particular, have been widely promoted and used since their introduction.

The question remains, however, are they really necessary? As previously discussed, decision instruments are rarely validated against usual practice.⁴ More frequently, the best comparison for their performance comes from the derivation cohort. Looking at the original studies, the added value of these decision instruments remains a matter of reasonable question. The derivation of CHALICE is anachronistic enough to not be relevant, with a 3 percent rate of CT and a 23 percent rate of skull X-ray. In PECARN TBI, the base rate for imaging is 35 percent; in CATCH, 53 percent. In these contexts, it is less impressive that PECARN TBI classifies only 53 to 58 percent of the population as very low risk, while CATCH performs similarly, excluding just 48 percent.

Almost by serendipity, illumination arrives from below the equator in the form of a large comparative study of the three decision instruments.⁵ The authors of this study state, “In some countries, such as Australia and New Zealand, no clinical decision rules predominate.” And the purpose of their study is most reasonably framed by the question, which of these instruments ought we generally disseminate and encourage in our practice? The answer, according to their headliner analysis published in The Lancet, is any of them.

Direct, head-to-head comparisons are difficult because each instrument is tested in a slightly different population. For example, PECARN TBI contains specific language regarding the exclusion of trivial mechanisms of injury. CATCH has its own specific inclusion criteria limiting its application, while CHALICE was derived from a population without exclusions and can be applied to any child with a minor head injury. In their study, with slightly more than 20,000 children, the instruments all performed similarly, separated only by narrow differences. The most sensitive instrument was PECARN TBI, followed by CATCH and then CHALICE, but injuries were so rare that the 95 percent confidence intervals all overlap. Logically, the negative predictive value of each tool was between 99.7 and 100 percent. Comparing specificity and inferring the number of children in whom a CT would be performed are difficult because PECARN TBI creates an explicit observation or CT cohort, but effectively, the balance between sensitivity and specificity is as expected—one waxes as the other wanes.

Almost a footnote in their original study, however, was the performance of underlying clinical judgment.⁶ In these modern practice settings, the rate of CT was far from the feared 30 to 50 percent witnessed in the derivation studies. Clinicians obtained advanced imaging a mere 8 percent of the time, and in spite of this low rate, their performance was not inferior to the most sensitive decision instrument. In these 20,000 children, there were 160 clinically important brain injuries, and clinicians missed only two of them. Each injury missed by clinicians was diagnosed several days later due to persistent symptoms, and each patient recovered without intervention. In contrast, PECARN TBI missed one injury, and both CHALICE and CATCH each missed at least 10.

So while PECARN TBI has the best sensitivity of the decision instruments, we ought to ask ourselves, why bother utilizing one at all? Clinical judgment in these settings performed statistically identically and at a fraction of the imaging rate. These studies offer some value because the risk factors identified here are almost certainly being incorporated into the judgment of clinicians involved. However, rather than fully relinquishing diagnostic decisions to these instruments, we probably best serve our patients by retaining our independence.

The harm from overreliance on these decision instruments is not just from potential excess imaging but from pervasive misapplication. For example, the Canadian Head CT Rule is intended for alert patients with loss of consciousness or disorientation following trauma. These patients have a non-zero, but also not high, incidence of serious intracranial injury. Bizarrely, a recent serious academic exercise evaluated its use in those with minimal head injury, reflecting indication creep beyond its intended use.⁷

These decision instruments already deliver poor positive predictive values in low-risk populations, and although they certainly can be applied in very-low-risk populations, doing so will increase the frequency of false positives. This pitfall especially afflicts PECARN TBI, which includes a whole host of trivial head injuries in its exclusion criteria. This instrument is almost routinely, and inappropriately, applied to the exceedingly low-risk population, a rabbit hole leading inevitably to unnecessary imaging.

Rather than using these instruments as a crutch, think of them as a guide. In each decision instrument, those elements associated with high risk for intracranial trauma are reasonable indications for advanced imaging. Those children who are obviously low risk on clinical evaluation will clearly meet the PECARN TBI very-low-risk criteria as well, and it is reasonable to document these criteria as justification for added certainty.

For the remainder of patients in the nebulous undefined area of risk, the best test is time, which is radiation-free and inexpensive.

The opinions expressed here are solely those of Dr. Radecki and do not necessarily reflect those of his employer or academic affiliates.

References

1. Kuppermann N, Holmes JF, Dayan PS, et al. Identification of children at very low risk of clinically-important brain injuries after head trauma: a prospective cohort study. Lancet. 2009;374(9696):1160-1170.
2. Dunning J, Daly JP, Lomas JP, et al. Derivation of the children’s head injury algorithm for the prediction of important clinical events decision rule for head injury in children. Arch Dis Child. 2006;91(11):885-891.
3. Osmond MH, Klassen TP, Wells GA, et al. CATCH: a clinical decision rule for the use of computed tomography in children with minor head injury. CMAJ. 2010;182(4):341-348.
4. Radecki RP. Published clinical decision aids may lack validation. ACEP Now. 2018;37(2):26-28.
5. Babl FE, Borland ML, Phillips N, et al. Accuracy of PECARN, CATCH, and CHALICE head injury decision rules in children: a prospective cohort study. Lancet. 2017;389(10087):2393-2402.
6. Babl FE, Oakley E, Dalziel SR, et al. Accuracy of clinician practice compared with three head injury decision rules in children: a prospective cohort study. Ann Emerg Med. 2018;71(6):703-710.
7. Davey K, Saul T, Russel G, et al. Application of the Canadian computed tomography head rule to patients with minimal head injury. Ann Emerg Med. 2018;72(4):342-350.