Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Diagnostic Testing for Acute Head Injury in Children: When are Head Computed Tomography and Skull Radiographs Indicated?

Quayle KS, Jaffe DM, Kuppermann N, Kaufman BA, Lee BC, Park TS, McAlister WH

Pediatrics 1997;99:E11 [abstract] [full-text]

Reviewed by Hanna S. Sahhar, MD, Hope Children's Hospital, Oak Lawn, IL, and The University of Illinois at Chicago School of Medicine

Review posted September 28, 1999

I. What is being studied?

Study objective:

The purpose of the study was to evaluate clinical features associated with head injury that could influence the decision to obtain imaging studies.

Study design

The study was a prospective cohort study of patients under 18 years of age seen in an urban pediatric emergency department with nontrivial, nonpenetrating head injury. Patients with nontrivial head injuries were defined as children who had symptoms related to head injury (headache, amnesia, vomiting, drowsiness, loss of consciousness, seizure, or dizziness) or significant physical findings (altered mental status, neurologic deficit, or altered skull anatomy).

Patients less than 1 year of age were included with any alteration of scalp surface anatomy, and between 12 and 24 months of age with a scalp hematoma. Each patient was to undergo skull radiographs and non-contrast head CT scans. Patient disposition was recorded, and discharged patients were followed up by telephone.

II. Are the results of the study valid?

Note: These questions follow from Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612. [abstract]

1. Was a representative group of patients completely followed up?

Yes. A representative group was obtained, i.e., all patients with nontrivial injury rather than a subset, but no, they were not completely followed up. Although very few patients were lost to follow up, the follow up that was obtained was not complete because the information collected reflected the general well-being of the child in the week after the injury. More detailed information regarding the neuropsychologic functioning of these children was not obtained. Furthermore, even children with normal computed tomograms (CT) of the head may have subtle brain injuries better detected by magnetic resonance imaging or perhaps with a follow-up CT scan, neither of which were obtained in these patients. In addition, 35 children with severe head injuries were transferred to the operating room or the intensive care unit without skull radiographs done in the emergency room, which may have led to an underreporting of skull fractures.

2. Were all potential predictors included?

Yes. The investigators included every variable that could possibly predict intracranial injury including: headache, dizziness, vomiting, drowsiness, seizure, amnesia, loss of consciousness, abrasion, contusion, laceration, hematoma, skull depression, signs of basilar fracture, altered mental status, focal deficit, and skull fracture as seen on skull radiograph.

3. Did the investigators test the independent contribution of each predictor variable?

Yes, using chi-square to identify univariate predictors of intracranial injury, and multiple logistic regression analysis to identify independent predictors of intracranial injury. Altered mental status, focal neurologic deficit, signs of a basilar skull fracture, loss of consciousness for more than 5 minutes, and skull fracture were determined to be significant univariate variables, with positive predictive values greater than 20%, negative predicitve values greater than 90%, and odds ratios greater than 3.51 with 95% confidence intervals between 1.28 and 71.63.

The significant independent variables from the multivariate analysis depended on whether the skull radiographs were used as predictors, or as positive findings.: As a predictor of intracranial injury, having a skull fracture carried an odds ratio of 92.4 (95% CI 10.8, 793). Other predictors included a focal neurologic deficit (OR 63.2; 95% CI 4.7, 846) and seizures (OR 19.2; 95% CI 1.5, 243). Excluding skull radiographs as a predictor, and including depressed skull fractures as intracranial injuries, a palpable skull depression (OR 17.9; 95% CI 3.6, 88.9), signs of basilar skull fracture (OR 10; 95% CI 1.7, 57.3), focal neurological deficit (OR 5.7; 95% CI 1.4, 23.5), and depressed mental status (OR 4.3; 95% CI 1.8, 10) were independent predictors.

4. Were outcome variables clearly and objectively defined?

The outcome assessed was the presence or absence of abnormality on head CT scan, as determined by an unblinded pediatric neuroradiologist. Exact specifications on how abnormalities were defined is not given, and no attempt to measure interrater reliability was made. It is arguable whether CT abnormality is a surrogate outcome (compared to measured neurologic outcome), but since the study's objective was to assist in the decision on whether to perform imaging in children with head trauma, this is clearly an appropriate outcome to measure.

III. What are the results?

1. What is(are) the prediction tool(s)?

The study identified possible univariate and multivariate predictors for intracranial injury in children with nontrivial, nonpenetrating head injuries. Although the results were encouraging, the authors could not develop a prediction tool to determine which diagnostic tests would be indicated in this population. Since only 23% of the children with at least one of these predictors had intracranial injuries, and 59% of all intracranial injuries occurred in neurologically normal children (or 6% of neurologically normal children had intracranial injury), it is hard to build confidence in any particular model. Nonetheless, the authors shared in the article clinical practice guidelines they developed for their own practice based on the results of the study.

They state that "head CT is recommended for head-injured children with altered mental status, focal neurologic deficits, signs of a basilar skull fracture, seizure, or a palpable depression of the skull. Because intracranial injuries occur in the absence of these findings, head CT should be considered for neurologically normal children with histories of loss of consciousness, vomiting, headache, drowsiness, or amnesia. "

2. How well does the model categorize patients into different levels of risk?

Not very well. Based on the results of the study a prediction tool could not be developed that could predict which patients needed imaging studies.

3. How confident are you in the estimates of risk?

Very confident. All odds ratios for the univariate variables were greater than 3.51 with 95% confidence intervals between 1.28 and 71.63. The independent variables had odds ratios greater than 3.3 with confidence intervals between 1.3 and 846.

IV. Will the results help me in caring for my patients?

1. Does the tool maintain its prediction power in a new sample of patients?

The investigators did not validate their prediction tool in a new sample of patients. In general, there are three ways to test a clinical prediction rule or model on a new set of patients. The best method is to validate the model in an entirely independent sample of patients. The second validation method is to randomly split the initial sample of patients into two groups and use one group to develop the clinical prediction rule of the model and the other group to validate the model. The third method is to use complex statistical techniques that repeatedly sample patients from the population and repeatedly test the accuracy of the prediction model.

2. Are your patients similar to those patients used in deriving and validating the tool(s)?

Yes. Our hospital is an urban, tertiary care center with similar pediatric demographic characteristics.

3. Does the tool improve your clinical decisions?

No overall prediction tool was offered, although the individual predictors, i.e., altered mental status, focal neurologic deficit, signs of a basilar skull fracture, seizure, and skull fracture, certainly strongly suggest the need for further imaging. The preponderance of injuries in children without these predictors, however, leaves us with continued uncertainty in these decisions.

 

Comments

Back to the EB Journal Club Index