Table 3.
Procedural Features | Whole Population (n = 25) |
Suboptimal Result (n = 12) |
Optimal Result (n = 13) |
p |
---|---|---|---|---|
PASCAL ACE | 15 (60%) | 7 (58%) | 8 (61%) | 0.9 |
XTW | 9 (36%) | 5 (42%) | 4 (31%) | 0.9 |
PASCALE ACE + P10 | 1 (4%) | 0 | 1 (8%) | 1 |
Posterior grasping | 15 (60%) | 9 (75%) | 6 (46%) | 0.2 |
Number of implanted devices | ||||
1 | 7 (28%) | 3 (25%) | 4 (31%) | 0.9 |
2 | 15 (60%) | 8 (67%) | 7 (54%) | 0.9 |
3 | 3 (12%) | 1 (8%) | 2 (15%) | 0.6 |
Fluoroscopy duration (min) | 26 (19–39) | 36 (24–51) | 24 (12–28) | 0.6 |
ROC curve analysis showed that AUC for AP-LAI was 0.769 (95% CI 0.51–0.93, p < 0.05) and Youden test identified the best cut-off value < 0.5 (sensitivity 50%, specificity 100%) for a suboptimal result. SL-LAI showed lower AUC in predicting suboptimal result (0.675, 95% CI 0.42–0.872). Comparing the two ROC curves, we showed that AUC difference is significant, with AP-LAI showing the best association to the outcome (p < 0.05), as detailed in Figure 4.