Table 5.
Model fitting results of the multilevel analyses for the domains communication with doctor, communication with nurses, communication about new medication, and pain control, and for the global rating of hospitals (standard errors added in parentheses).
| Outcome variable | intercept | β age1 | β gender1 | β education1 | β physical health1 | β mental health1 | Var patients | Var hospital | Var tot2 | PCV3 | ICC4 |
| Doctor (N = 1469) | |||||||||||
| Model 1 | 3.523 (0.022)* | - | - | - | - | - | 0.290 (0.011)* | 0.011 (0.005)* | 0.301 | reference | 0.04 |
| Model 2 | 3.632 (0.048)* | -0.021 (0.032) | -0.059 (0.031) | 0.021 (0.032) | -0.139 (0.045)* | -0.152 (0.034)* | 0.279 (0.010)* | 0.010 (0.004)* | 0.289 | 4.2% | 0.03 |
| Nurses (N = 1507) | |||||||||||
| Model 1 | 3.508 (0.018)* | - | - | - | - | - | 0.205 (0.008)* | 0.007 (0.003)* | 0.212 | reference | 0.03 |
| Model 2 | 3.675 (0.038)* | -0.031 (0.026) | -0.081 (0.026)* | -0.024 (0.026) | -0.148 (0.037)* | -0.158 (0.028)* | 0.193 (0.007)* | 0.005 (0.002)* | 0.198 | 7.1% | 0.03 |
| Medication (N = 630) | |||||||||||
| Model 1 | 2.896 (0.043)* | - | - | - | - | - | 1.142 (0.064)* | 0.000 (0.000) | 1.142 | reference | n.c. |
| Model 2 | 3.082 (0.139)* | -0.169 (0.090) | -0.163 (0.094) | 0.187 (0.100) | -0.331 (0.133)* | -0.171 (0.101) | 1.087 (0.061)* | 0.000 (0.000) | 1.087 | 5.1% | n.c. |
| Pain control (N = 1484) | |||||||||||
| Model 1 | 3.594 (0.018)* | - | - | - | - | - | 0.357 (0.013)* | 0.003 (0.003) | 0.360 | reference | n.c. |
| Model 2 | 3.632 (0.050)* | 0.008 (0.035) | -0.023 (0.035) | 0.028 (0.036) | -0.120 (0.050)* | -0.101 (0.038)* | 0.352 (0.013)* | 0.001 (0.002) | 0.353 | 2.0% | n.c. |
| Rating hospital (N = 1490) | |||||||||||
| Model 1 | 8.422 (0.046)* | - | - | - | - | - | 1.818 (0.067)* | 0.031 (0.018)* | 1.849 | reference | n.c. |
| Model 2 | 8.789 (0.113)* | 0.094 (0.078) | -0.135 (0.078) | -0.228 (0.080)* | -0.265 (0.112)* | -0.440 (0.085)* | 1.744 (0.065)* | 0.023 (0.016)* | 1.767 | 4.6% | n.c. |
*p < 0.05
1Reference group age = younger than 65; reference group gender = males; reference group education = low education; reference group physical health = good physical health; reference group mental health = good mental health.
2Var tot (total variance) = Var patients + Var hospital
3PCV (proportional change in variance) = (Var tot model 1 - Var tot model 2)/(Var tot model 2) * 100%
4ICC (intra-class correlation) = Var hospital/(Var patients + Var hospital)
n.c. = not calculated. The variance explained by the hospital level is not significant and, therefore, the ICC is not calculated.