Skip to main content

Table 4 Independent risk factors for short-term (2 months) health and functional outcomes following a motor vehicle crash injury

From: Restriction in functioning and quality of life is common in people 2 months after compensable motor vehicle crashes: prospective cohort study

Independent variable General health status (Likert scale 1 to 5) SF-12 PCS SF-12 MCS EQ-5D summary score EQ-5D VAS
β a (95 % CI) β a (95 % CI) β a (95 % CI) β a (95 % CI) β a (95 % CI)
Pre-injury health status      
 Excellent/very good 0   0 0 0
 Good −0.55 (−0.83, −0.27)   −3.4 (−6.3, −0.5) 0.03 (−0.02, 0.08) −6.5 (−12.3, −0.6)
 Fair/poor −1.46 (−1.89, −1.03)   −11.6 (−16.2, −7.0) −0.10 (−0.18, −0.02) −16.2 (−25.5, −6.9)
Education level (≤ secondary)      −5.1 (−9.3, −0.8)
Chronic illness   −5.9 (−8.2, −3.6)   −0.05 (−0.10, −0.01) −4.8 (−9.6, 0.0)
Chronic pain     −0.08 (−0.14, −0.02)  
Self-perceived threat to life −0.33 (−0.55, −0.11) −3.2 (−5.5, −0.8) −3.8 (−6.1, −1.5) −0.04 (−0.08, −0.01) −10.5 (−14.9, −6.1)
NISS (4–8)   −4.4 (−7.6, −1.3)    
Fracture     −0.10 (−0.17, −0.03)  
Whiplash −0.23 (−0.45, −0.00)   −3.0 (−5.3, −0.7) −0.07 (−0.11, −0.03)  
Body mass index (kg/m2)      
 <20 0     0
 20–24 −0.17 (−0.58, 0.24)     −0.8 (−9.1, 7.5)
 25–29 −0.45 (−0.87, −0.04)     −3.4 (−11.8, 4.9)
 ≥30 −0.53 (−0.95, −0.10)     −8.9 (−17.6, −0.3)
Admitted to hospital (≥1 night) −0.38 (−0.65, −0.10) −4.0 (−6.9, −1.0)   −0.10 (−0.15, −0.06)  
 Model R 2 0.22 0.13 0.11 0.19 0.19
  1. Wald chi-square tests for logistic models and partial F tests for linear regression models. Backward elimination approach was used to determine independent risk factors. The underlying assumption for the above model is normal distribution of the score data. To check the impact of non-normality, the models were re-run after Blom’s normalising transformation of the dependent variables. The great majority of findings were confirmed, the notable differences being that the coefficients fair/poor health in models for EQ-5D outcomes became marginally non-significant (P values 0.060 and 0.082), the coefficient for chronic illness in the model for EQ-5D VAS became more clearly significant (P = 0.018), and the coefficient for BMI >30 in the model for EQ-5D VAS became non-significant (P = 0.12)
  2. aAdjusted for other covariates in the column