
Adjusted Mean (SE)*

95%Confidence Interval

Adjusted pvalue**

Equality of adjusted means (pvalue)***
 
Hospital A (n = 375)

.735 (.033)

(.670, .800)

<0.001

A vs C(<0.001)

A vs B(<0.001)

Hospital B (n = 601)

.503 (.027)

(.451, .555)
   
Hospital C (n = 1002)

.550 (.020)

(.510, .591)
 
C vs B (0.488)
 
 * Adjusted age of patients, total FIMS motor and cognitive scores at admission (all p < 0.001) Using analysis of covariance
 ** Test for overall equality of adjusted means in the 3 institutions from analysis of covariance
 ***Adjustment for multiple comparisons using Bonferroni test.
 The distribution of total FIM efficiency gain, total FIM motor score at admission and total FIM cognitive score at admission were slightly skewed (skewness statistic were 0.676, 0.371 and 0.275 respectively). As the sample size in each hospital was large enough and the robustness property of the parametric test, ANCOVA is appropriate.