Table 3.
OR (95% CI) | P Value | ARR (95% CI), %a | |
---|---|---|---|
Level I centers (n = 159 511) | |||
Unadjusted mortality | 0.88 (0.85–.90) | <.001 | 1.6 (1.4–1.7)b |
| |||
Logistic regression | |||
Standard | 1.31 (1.27–1.38) | <.001 | 2.9 (2.6–3.6) |
| |||
Controlled for clusteringc | 1.32 (1.20–1.45) | <.001 | 3.0 (1.9–4.2) |
| |||
After propensity score matchingd | 1.16 (1.14–1.17) | <.001 | 1.5 (1.4–1.6) |
| |||
Level II centers (n = 63 964) | |||
Unadjusted mortality | 0.86 (0.83–0.91) | <.001 | 1.7 (1.4–1.8)b |
| |||
Logistic regression | |||
Standard | 1.37 (1.28–1.48) | <.001 | 4.3 (2.6–4.5) |
| |||
Controlled for clusteringc | 1.37 (1.23–1.53) | <.001 | 3.4 (2.2–5.0) |
| |||
After propensity score matchingd | 1.15 (1.13–1.17) | <.001 | 1.4 (1.3–1.6) |
Abbreviations: ARR, absolute risk reduction; OR, odds ratio.
Absolute RR calculations were made after calculating the number needed to treat (NNT), based on the adjusted ORs.25 NNT=1− (PEER × [1−absolute OR])/(1−PEER) × PEER × (1−OR)]. PEER is the patient expected event rate (eg, the event rate in the ground transport group [control group]. ARR=1/NNT.
Indicates greater mortality for transportation by helicopter.
Generalized estimating equations with robust variance calculations were performed. Independent variables included systolic blood pressure, heart rate, respiratory rate, motor component of Glasgow Coma Score, e-code, age, type of trauma, injury severity score, and sex. E-codes refer to types of external injuries according to the International Classification of Disease, Ninth Edition, Clinical Modification.
Logistic regression was performed following estimation of a propensity score with the subclassification matching method. The covariates used to estimate the propensity score were systolic blood pressure, heart rate, respiratory rate, motor component of Glasgow Coma Scale score, e-code, age, type of trauma, injury severity score, facility identifier, and sex.