Table 3.
Multivariate model analysis of delayed cerebral ischemia with predictors
| Predictora | Univariate analysis | Multivariate analysisb | ||||
|---|---|---|---|---|---|---|
| DCI (n = 46) | non‐DCI (n = 80) | OR (95% CI) | p value | OR (95% CI) | p value | |
| Components of Model 1 | ||||||
| WFNS grade | 4 (3–5) | 1 (1–3) | 2.19 (1.64–2.91) | <.001 | 1.81 (1.33–2.47) | <.001 |
| Modified Fisher grade | 3 (2–4) | 2 (2–2) | 2.89 (1.78–4.68) | <.001 | – | – |
| Clipping | 42 (91.3%) | 63 (78.8%) | 2.83 (0.89–9.01) | .078 | – | – |
| Ngb (day 1), ng/ml | 9.9 ± 3.7 | 7.2 ± 1.5 | 1.56 (1.26–1.93) | <.001 | – | – |
| Ngb (day 2), ng/ml | 11.2 ± 4.0 | 8.6 ± 1.7 | 1.47 (1.20–1.80) | <.001 | – | – |
| Ngb (day 3), ng/ml | 9.5 ± 2.2 | 7.8 ± 1.3 | 1.94 (1.44–2.62) | <.001 | 1.57 (1.12–2.19) | .009 |
| Ngb (day 5), ng/ml | 8.5 ± 2.1 | 7.1 ± 1.4 | 1.57 (1.24–2.00) | <.001 | – | – |
| Ngb (day 7), ng/ml | 7.5 ± 1.8 | 6.4 ± 1.2 | 1.72 (1.30–2.33) | <.001 | – | – |
| mean Ngb, ng/ml | 9.3 ± 2.5 | 7.4 ± 1.1 | 2.19 (1.49–3.21) | <.001 | – | – |
| Components of Model 2 | ||||||
| WFNS grade | 4 (3–5) | 1 (1–3) | 2.19 (1.64–2.91) | <.001 | 1.81 (1.33–2.47) | <.001 |
| Ngb (day 3), ng/ml | 9.5 ± 2.2 | 7.8 ± 1.3 | 1.94 (1.44–2.62) | <.001 | 1.57 (1.12–2.19) | .009 |
| WFNS grade × Ngb (day 1) | 40.0 ± 25.2 | 15.1 ± 11.9 | 1.08 (1.05–1.11) | <.001 | – | – |
| WFNS grade × Ngb (day 2) | 44.4 ± 28.2 | 18.0 ± 14.2 | 1.07 (1.04–1.10) | <.001 | – | – |
| WFNS grade × Ngb (day 3) | 36.6 ± 18.7 | 16.3 ± 12.5 | 1.09 (1.05–1.12) | <.001 | – | – |
| WFNS grade × Ngb (day 5) | 33.0 ± 17.4 | 15.2 ± 12.3 | 1.08 (1.05–1.11) | <.001 | – | – |
| WFNS grade × Ngb (day 7) | 29.3 ± 15.6 | 13.3 ± 10.5 | 1.09 (1.06–1.13) | <.001 | – | – |
| WFNS grade × mean Ngb | 36.7 ± 20.5 | 15.6 ± 12.1 | 1.08 (1.05–1.11) | <.001 | – | – |
Abbreviations: DCI, delayed cerebral ischemia; Ngb, neuroglobin; OR, odds ratio; WFNS, World Federation of Neurosurgical Societies.
Predictors of Model 1 included the preoperative and operative items listed in Table 1 that had p < .10 (except the interaction variables “WFNS grade × Ngb”). Predictors of Model 2 included the variables which were significant in the multivariate model analysis of Model 1, and the interaction variables.
Multivariate analysis: all variables having p < .05 from univariate analyses were included in multivariate analysis. The backward stepwise multivariate regression was performed to create the final model whereby the least nonsignificant variable was removed from the model one by one, until all remaining variables had p < .05.