Computed tomography is inaccurate in estimating the severity of adult splenic injury

Computed tomography (CT) is increasingly utilized in evaluation of adult splenic injury (SI). CT correlation with operative findings, CT relationship to successful nonoperative (NO) management, and CT reading reproducibility were examined. Records of patients > or = 15 years old admitted over a 3-year period were reviewed. Computed tomography scans were graded by two radiologists blinded to clinical results. Computed tomography scans were performed on 49 of 77 patients with SI. Eighteen underwent initial operation (OR) and 31 initial NO. Operative patients had higher Injury Severity Scores and Abdominal Abbreviated Injury Scale scores (p < 0.0001). Grade II readings predominated in the NO group (55%). Nonoperative management was successful for 9 grade III and 3 grade IV readings. Computed tomography matched OR grade in 10 readings, underestimated it in 18, and overestimated it in 6. Computed tomography missed SI in five patients. Radiologists disagreed on 9 of 45 (20%) scans. Computed tomography poorly predicted operative findings. Interobserver variability was common. SI management should not be based solely on CT severity.