Influence of Mirrored Computed Tomograms on Decision-Making for Revising Surgically Treated Orbital Floor Fractures

This study evaluated whether intraoperative imaging with computer-assisted virtual reconstruction would be advantageous in reconstructions of orbital floor fractures. The surgeon's intention to revise a reconstructed primary orbital floor fracture by evaluating a postoperative mirrored computed tomographic (CT) scan was analyzed intraoperatively before wound closure, during inpatient hospitalization, and after hospitalization. The inter-rater agreement and the match of intention to revise and actual revision were analyzed.

Fifty-one anonymized postoperative CT scans of patients with a unilateral orbital floor fracture were mirrored using software. These computer-assisted virtual reconstructions were consecutively examined by 4 examiners. Seven of these patients underwent a revision. In the first part, the inter-rater agreements for all 3 times were analyzed. In the second part, the examiners' intentions to revise were compared with the actual performed revisions.

The overall inter-rater agreements were 0.69 for the intraoperative phase, 0.55 for the in-hospital phase, and 0.39 for the post-hospital phase. The intraoperative inter-rater agreement for each examiner was 0.58 to 0.80. The Fleiss κ value for the in-hospital and post-hospital phases was lower. The comparison of the examiners' intention to revise and the actual revisions showed that 15 to 24 additional would have been revised. In contrast, 6 of 7 actual revisions would have been revised intraoperatively. The missed actual revision was the same case by all 4 examiners. The accordance of intention to revise with the actual revisions decreased during hospitalization and even more after hospitalization. This study showed strong agreement among examiners for revising anatomically incorrectly reduced orbital floor fractures intraoperatively by evaluating postoperative mirrored CT scans. During the in-hospital and post-hospital phases, the restraints against revision seemed to increase, thus leading to poorer inter-rater agreement. This analysis of postoperative CT scans with computer-assisted virtual reconstructions of the orbit would have led to considerably more revisions intraoperatively, but all actual revisions were detected except for 1 case. This case was the same for all 4 examiners. Operation time would have been prolonged in the additional revised cases, but a better anatomic reconstruction would have been achieved. Furthermore, the intraoperative result of the reconstruction would have been controlled instantly and corrected immediately, if needed.

This study showed that of 6 of 7 actual revisions, implant placement would have been revised intraoperatively by all 4 examiners, if intraoperative imaging with computer-assisted virtual reconstruction of the orbit would have been applied. Therefore, the authors suggest that intraoperative imaging with computer-assisted virtual reconstruction could be advantageous in the prevention of later revisions of orbital floor fractures. In this study, the threshold to revise implant placement intraoperatively seemed to be lower when using intraoperative imaging with virtual reconstructions, because considerably more cases would have been revised intraoperatively by the examiners. In the in-hospital and post-hospital phases, this threshold increased, suggesting the more important role of clinical findings. It is uncertain whether the actual surgeons would have revised the same cases as the examiners if they had used intraoperative imaging with virtual reconstructions for their deliberation. However, the intraoperative inter-rater agreement was good and cost-intensive postoperative revisions might be prevented.