Accuracy of the Orlando Head CT Criteria in detecting clinically significant abnormalities in patients without head trauma: A systematic review and meta-analysis.

Abstract

Background: This systematic review and meta-analysis evaluated the accuracy of the Orlando Head CT Criteria (focal deficit, altered mental status, age≥60, headache with vomiting) in predicting clinically significant abnormalities (CSAs) in ED patients without trauma.

Methods: This study followed PRISMA-DTA guidelines. Searched databases included PubMed, EMBASE, Web of Science, CINAHL, Google Scholar, and gray literature. Studies were included if they contained ED patients without trauma and analyzed accuracy of the Orlando Head CT Criteria. Summary estimates of sensitivity, specificity, diagnostic odds ratio (DOR), and likelihood ratios (LRs) were calculated using a bivariate mixed model with a random-effects approach. Risk of bias and applicability were assessed via QUADAS-2 and certainty of evidence via the GRADE framework. Heterogeneity was assessed via forest plots, a hierarchical summary receiver operating characteristic curve and subgroup analysis. Sensitivity analysis was performed using the leave-one-out method. Threshold effects were investigated via Spearman's correlation and publication bias using Deeks' funnel plot.

Results: Seven studies (8914 patients) were included: four prospective and three retrospective. Orlando Head CT Criteria were 98.4% (95% confidence interval [CI] 97.1%-99.1%) sensitive, 17.9% (95% CI 12.3%-25.3%, 95% CI) specific with a DOR of 13.4 (95% CI 6-29.6) and negative LR (-LR) of 0.09 (95% CI 0.04-0.19) in predicting CSA CT scans. Implementation of these criteria would have decreased CT use by 16.2% (95% CI 9.7%-23.9%). GRADE certainty of evidence was moderate. No subgroup or single study was a source of heterogeneity. There were no threshold effects (Spearman ρ = -0.26). We found no publication bias (Deeks' funnel plot asymmetry test, p = 0.71).

Conclusion: Orlando Head CT Criteria were sensitive in detecting CSA CT scans with a low -LR. Further studies are required to assess whether these criteria can decrease CT use or augment clinical judgment for patients undergoing nontrauma cranial CT.