Skip to main content

Table 2 Risk estimates from alternative models for associations between hospital and surgeon volume and prolonged LOS

From: Factors predicting hospital length-of-stay after radical prostatectomy: a population-based study

Model Variable (comparison) Adjusted OR1 95% CI2 p-value
Model 1 (primary analysis): Hospital and surgeon volume both fitted in model (without interaction term)1 Hospital volume: (higher vs lower) 0.34 0.26-0.45  
Surgeon volume: (higher vs lower) 0.55 0.42-0.71  
Model 2: Surgeon volume omitted Hospital volume: (higher vs lower) 0.37 0.29-0.48 p < 0.0013
Model 3: Hospital volume omitted Surgeon volume: (higher vs lower) 0.63 0.49-0.81 p < 0.0013
Model 4: Hospital and surgeon volume fitted with an interaction term4 Hospital volume main effect: (higher vs lower) 0.28 0.20-0.40 p = 0.1054
Surgeon volume main effect: (higher vs lower) 0.46 0.33-0.64
  Interaction effect 1.55 0.92-2.65  
  1. 1Final multivariate model as shown in Table 1, 295% confidence intervals, 3p-value for likelihood ratio tests compared to baseline model (model 1), 4Model 4 is the baseline model with the addition of an interaction term; the p value is for the interaction term.