PEDIATRICS Vol. 124 No. 1 July 2009, pp. e145-e154 (doi:10.1542/peds.2009-0075)
REVIEW ARTICLE |
Should a Head-Injured Child Receive a Head CT Scan? A Systematic Review of Clinical Prediction Rules
a Division of Pediatric Medicine and the Pediatric Outcomes Research Team (PORT)
e Division of Pediatric Emergency Medicine
f Hospital Library, Hospital for Sick Children, Toronto, Ontario, Canada
Departments of b Pediatrics
d Health Policy, Management and Evaluation
h Medicine, University of Toronto, Toronto, Ontario, Canada
c Child Health Evaluative Sciences, Hospital for Sick Children Research Institute, Toronto, Ontario, Canada
g Keenan Research Centre, Li Ka Shing Knowledge Institute of St Michael's Hospital, Toronto, Ontario, Canada
CONTEXT: Given radiation- and sedation-associated risks, there is uncertainty about which children with head trauma should receive cranial computed tomography (CT) scanning. A high-quality and high-performing clinical prediction rule may reduce this uncertainty.
OBJECTIVE: To systematically review the quality and performance of published clinical prediction rules for intracranial injury in children with head injury.
METHODS: Medline and Embase were searched in December 2008. Studies were selected if they included clinical prediction rules involving children aged 0 to 18 years with a history of head injury. Prediction-rule quality was assessed by using 14 previously published items. Prediction-rule performance was evaluated by rule sensitivity and the predicted frequency of CT scanning if the rule was used.
RESULTS: A total of 3357 titles and abstracts were assessed, and 8 clinical prediction rules were identified. For all studies, the rule derivations were reported; no study validated a rule in a separate population or assessed its impact in actual practice. The rules differed considerably in population, predictors, outcomes, methodologic quality, and performance. Five of the rules were applicable to children of all ages and severities of trauma. Two of these were high quality (
11 of 14 quality items) and had high performance (lower confidence limits for sensitivity >0.95 and required 
56% to undergo CT). Four of the 8 rules were applicable to children with minor head injury (Glasgow coma score 13). One of these had high quality (11 of 14 quality items) and high performance (lower confidence limit for sensitivity = 0.94 and required 13% to undergo CT). Four of the 8 rules were applicable to young children, but none exhibited adequate quality or performance.
CONCLUSIONS: Eight clinical prediction-rule derivation studies were identified. They varied considerably in population, methodologic quality, and performance. Future efforts should be directed toward validating rules with high quality and performance in other populations and deriving a high-quality, high-performance rule for young children.
Key Words: systematic review • craniocerebral trauma [MeSH] • brain injuries [MeSH] • decision-support techniques [MeSH] • decision trees [MeSH] • clinical prediction rules • clinical decision rules • predictive value of tests [MeSH] • human [MeSH] • tomography • x-ray computed [MeSH] • infant • child • preschool • adolescent
Abbreviations: CT—computed tomography • ICI—intracranial injury • MeSH—medical subject heading • NPV—negative predictive value • PPV—positive predictive value • GCS—Glasgow coma scale • CI—confidence interval
Accepted Mar 11, 2009.
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
