Abstract 3326: Reliability Of Hematoma Volume Calculation At Local Sites In The Setting Of A Multicenter Intracerebral Hemorrhage Clinical Trial

Abstract

Background: In the clinical trials evaluating new therapies for acute management of intracerebral hemorrhage (ICH), rapid hematoma volume (HV) measurements from computed tomographic (CT) scan using previously described ‘ABC/2’ method at study sites is frequently required for either determining eligibility or stratification. However, the accuracy of this method by multiple observers using variable visual display modes is not known. Method: We performed a post hoc analysis of a multicenter clinical trial that recruited ICH patients within 6 hours of symptom onset at ten sites in the United States. HV was measured by local site investigators to determine eligibility for enrollment (<60 cc) into the trial using the ABC/2 method (manual method) on the initial CT scan. CT scan images were subsequently sent to the core imaging laboratory for blinded computer based analysis of HV using ‘Medical Image Processing, Analysis, and Visualization (MIPAV)’ software (computer method). Paired t-test was used to identify any differences between the manual and the computer methods of HV measurements. Spearman’s correlation was used to study the agreement between the two methods. Results: HV measurements by both the manual and the computer methods were available in 51 patients enrolled in the trial (mean±standard deviation [SD] age 62±15 years; 26 were women.) The mean±SD HV using as calculated by he computer method (14.6±15.3 mL) was not statistically different from the values calculated by the manual method (16.5±15 mL; p=0.6). Using Spearman’s correlation, there was a significant correlation between the two methods (rho 75%, p<0.01). Manual calculation under estimated the HV by an average of 1.5 mL in the lower 50 percentile of computer-calculated HV (<9.4 ml), while overestimated the HV in the upper 50 percentile by an average of 3.1 mL. None of these differences were statistically significant. Conclusion: Rapid measurements at study sites for either determining eligibility or stratification provides reliable HV measurements despite measurements by multiple observers using variable visual CT scan display modes.