Table 2.
Comparing analytical methods to adjust for confounding by indication in proportional odds logistic regression models with the Glasgow Outcome Scale as outcome
| Approach | POCON dataset ICP monitoring OR (95% CI) | Tirilazad dataset Intracranial operation OR (95% CI) | EBIC dataset Primary referral OR (95% CI) |
|---|---|---|---|
| Unadjusted model | 0.51 (0.32–0.81) | 1.04 (0.70–1.54) | 0.85 (0.66–1.10) |
| Covariate adjustment* | 0.91 (0.48–1.74) | 0.92 (0.59–1.42) | 0.85 (0.64–1.15) |
| Propensity score matching** | 0.80 (0.42–1.54) | 0.89 (0.53–1.50) | 0.89 (0.76–1.18) |
| Hospital-level approach*** | 1.17 (1.01–1.42) | 1.42 (0.95–1.97)ⱡ | 0.91 (0.81–1.03)Ⱡ |
Notes:
Model was adjusted for the following confounders: age, GCS motor score, pupillary reaction, hypoxia, hypotension, CT classification, tSAHs, EDHs, glucose and hemoglobin.
A propensity score was calculated based on the following variables: age, GCS motor score, pupillary reaction, hypoxia, hypotension, CT classification, tSAHs, EDHs, glucose and hemoglobin. For ICP monitoring, matching resulted in 67 patients receiving the intervention (propensity score 0.47, probability on survival 0.46, probability on favorable outcome 0.28) and 67 patients not receiving the intervention (propensity score 0.46, probability on survival 0.43, probability on favorable outcome 0.32). For craniotomy, matching resulted in 96 patients receiving the intervention (propensity score 0.83, probability survival 0.63, probability favorable outcome 0.42) and 96 patients not receiving the intervention (propensity score 0.83, probability survival 0.63, probability favorable outcome 0.42). For primary referral, matching resulted in 312 patients being primary referred (propensity score 0.46; probability survival 0.65; probability favorable outcome 0.49) and 312 patients being secondary referred (propensity score 0.47, probability survival 0.65, probability favorable outcome 0.48).
Per 10% change; model was adjusted for the following confounders: age, GCS motor score, pupillary reaction, hypoxia, hypotension, CT classification, tSAHs, EDHs, glucose and hemoglobin.
Analyses in seven centers with a total of 172 patients.
Analyses in 12 centers with a total of 350 patients.
Abbreviations: CT, computed tomography; EBIC, European Brain Injury Consortium; EDHs, epidural haemorrhages; GCS, Glasgow Coma Scale; ICP, intracranial pressure; POCON, Prospective Observational Cohort Neurotrauma; tSAHs, traumatic subarachnoid hemorrhages.