Skip to main content

Advertisement

Table 2 Multivariate analyses.

From: Should policy-makers and managers trust PSI? An empirical validation study of five patient safety indicators in a national health service

Mortality in Low-mortality DRGs Model 1
Hospital random effect
Model 2
adds age-sex fixed-effect
Model 3
adds comorbidity fixed-effect
Patient variables (OR 95% CI)    
Constant 0.001 (0.0005 to 0.007)   
Measures of clustering    
Hospital level variance (SE) 0.20 (0.07)   
Rho (95% CI) 0.06 (0.03 to 0.11)   
MOR (95% CI) 1.53 (1.35 to 1.81)   
Decubitus Ulcer Model 1
Hospital random effect
Model 2
adds age-sex fixed-effect
Model 3
adds comorbidity fixed-effect
Patient variables (OR 95% CI)    
Constant 0.008 (0.007 to 0.009) 0.00015 (0.00013 to 0.0002) 0.00018 (0.00015 to 0.0002)
Age   1.053 (1.052 to 1.055) 1.046 (1.045 to 1.047)
Sex   1.11 (1.08 to 1.14) 1.14 (1.09 to 1.17)
Paralysis    5.05 (4.81 to 5.31)
Other neurological disorders    3.74 (3.59 to 3.89)
Diabetes w chr. complications    1.87 (1.79 to 1.99)
Weight loss    5.16 (4.85 to 5.47)
Fluid And electrolyte disorders    2.97 (2.83 to 3.13
Measures of clustering    
Hospital level variance (SE) 0.54 (0.43 to 0.68) 0.46 (0.37 to 0.58) 0.38 (0.29 to 0.47)
Rho (95% CI) 0.14 (0.12 to 0.17) 0.12 (0.10 to 0.15) 0.10 (0.08 to 0.13)
MOR (95% CI) 2.01 (1.86 to 2.18) 1.91 (1.78 to 2.06) 1.79 (1.68 to 1.92)
Catheter-related infection Model 1
Hospital random effect
Model 2
adds age-sex fixed-effect
Model 3
adds comorbidity fixed-effect
Patient variables (OR 95% CI)    
Constant 0.001 (0.0009 to 0.0012) 0.0012 (0.001 to 0.0014) 0.0012 (0.001 to 0.0015)
Age   1.017 (1.016 to 1.018) 1.014 (1.013 to 1.015)
Sex   0.47 (0.44 to 0.49) 0.49 (0.46 to 0.51)
Peripheral vascular disease    2.05 (1.84 to 2.29)
Paralysis    2.20 (1.93 to 2.56)
Weight loss    3.63 (3.03 to 4.35)
Fluid And electrolyte disorders    2.36 (2.09 to 2.66)
Measures of clustering    
Hospital level variance (SE) 1.10 (0.84 to 1.44) 1.11 (0.85 to 1.45) 1.05 (0.80 to 1.38)
Rho (95% CI) 0.25 (0.20 to 0.31) 0.25 (0.20 to 0.31) 0.24 (0.20 to 0.30)
MOR (95% CI) 2.71 (2.39 to 3.13) 2.72 (2.39 to 3.14) 2.65 (2.38 to 3.06)
Postoperative PE or DVT Model 1
Hospital random effect
Model 2
adds age-sex fixed-effect
Model 3
adds comorbidity fixed-effect
Patient variables (OR 95% CI)    
Constant 0.0046 (0.0043 to 0.005) 0.00065 (0.00055 to 0.0007) 0.00059 (0.00052 to 0.0007)
Age   1.031 (1.029 to 1.032) 1.029 (1.028 to 1.03)
Sex   0.96 (0.92 to 1.002) 1.02 (0.82 to 1.06)
Pulmonary circulation disease    2.39 (2.16 to 2.66)
Paralysis    2.32 (2.07 to 2.58)
Lymphoma    2.14 (1.69 to 2.69)
Metastatic cancer    2.80 (2.56 to 3.03)
Solid tumor w/o metastasis    1.84 (1.65 to 2.03)
Coagulopthy    2.89 (2.56 to 3.25)
Weight loss    2.58 (2.25 to 2.94)
Measures of clustering    
Hospital level variance (SE) 0.26 (0.19 to 0.33) 0.24 (0.19 to 0.31) 0.20 (0.15 to 0.26)
Rho (95% CI) 0.07 (0.06 to 0.09) 0.07 (0.05 to 0.09) 0.06 (0.04 to 0.07)
MOR (95% CI) 1.62 (1.53 to 1.73) 1.59 (1.51 to 1.70) 1.53 (1.45 to 1.63)
Postoperative sepsis Model 1
Hospital random effect
Model 2
adds age-sex fixed-effect
Model 3
adds comorbidity fixed effect
Patient variables (OR 95% CI)    
Constant 0.014 (0.013 to 0.015) 0.006 (0.005 to 0.007) 0.0068 (0.0059 to 0.0077)
Age   1.022 (1.021 to 1.023) 1.019 (1.018 to 1.02)
Sex   0.63 (0.61 to 0.66) 0.64 (0.61 to 0.66)
Congestive heart failure    2.51 (2.36 to 2.69)
Paralysis    2.16 (1.95 to 2.39)
Weight loss    3.39 (2.91 to 3.89)
Measures of clustering    
Hospital level variance (SE) 0.28 (0.21 to 0.37) 0.31 (0.24 to 0.41) 0.30 (0.23 to 0.39)
Rho (95% CI) 0.08 (0.06 to 0.10) 0.09 (0.07 to 0.11) 0.08 (0.07 to 0.11)
MOR (95% CI) 1.65 (1.55 to 1.78) 1.70 (1.59 to 1.84) 1.69 (1.58 to 1.82)
  1. Estimates for hospital (clustering) and individual effects
  2. A Rho statistic value different to 0 represents the existence of cluster effect -the propensity of having an outcome is more similar among the patients within a hospital, that among patients from different hospitals; as for the magnitude of rho, the more the value, the greater the clustering