secondary decompressive craniectomy for refractory cerebral edema. Five patients (3 in the decompressive cra- niectomy group and two in the hematoma evacuation only group) who experienced rebleeding underwent sec- ondary operation. Six of the 8 patients with reoperation died within 1 month. There was no significant difference in the causes of death between the groups. Six-month follow-up information was available in all cases. By that time, 41 of the 84 patients (49%) had achieved functional survival although only one of them showed good recov- ery (GOS score 5).
In unadjusted analyses, the 30-day mortality was 32% in decompressive craniectomy group versus 43% in hematoma evacuation only group. The 6-month func- tional survival was 55% in decompressive craniectomy group versus 45% in hematoma evacuation only group. No statistically significant difference could be found in either comparison between groups (P 5 .26 and P 5 .28, respectively) (Tables 2 and 3).
However, it was noted that the baseline characteristics of patients were not well balanced. We therefore used multi- variate regression to evaluate the independent effect of the intervention method on outcomes, adjusting for baseline differences. Cerebral herniation and ICH score were found as confounders to treatment method. After control of these factors, decompressive craniectomy group was indepen- dently associated with a lower mortality. The estimated odds ratios for 30-day mortality and 6-month functional survival were 0.12 (95% confidence interval 0.02-0.64, P5.01) and 23.23 (95% confidence interval 2.13-252.86, P 5.01), respectively (Tables 4 and 5).