Table 2 Multilevel logistic regression models for the probability of having a preventable hospitalization, 2011–2014
Model 1 Empty model | Model 2 individual-level | Model 3 both individual- and county-level | |
|---|---|---|---|
Estimate(SE) | Estimate(SE) | Estimate(SE) | |
Intercept | −1.62(0.03)*** | −1.69(0.06)*** | −1.62(< 0.001)*** |
Individual-level | |||
DSCP participation | |||
None(ref) | |||
Low | −0.01(0.04) | −0.01(0.04) | |
Medium | −0.09(0.04)* | −0.09(0.04)* | |
Full | 0.29(0.81) | 0.35(0.80) | |
Sociodemographic characteristics | |||
Gender | |||
Female(ref) | |||
Male | 0.13(0.02)*** | 0.13(0.02)*** | |
Age | |||
19–40(ref) | |||
41–64 | −0.30(0.05)*** | −0.30(0.05)*** | |
65+ | 0.12(0.05)* | 0.11(0.05)* | |
Income level | |||
Non-low income level(ref) | |||
Low income level | 0.34(0.07)*** | 0.34(0.07)*** | |
Living region | |||
Taipei division(ref) | |||
Northern division | 0.06(0.05) | 0.07(0.05) | |
Central division | 0.08(0.05) | 0.09(0.05) | |
Southern division | 0.12(0.05)* | 0.13(0.05)* | |
Kaoping division | 0.06(0.06) | 0.07(0.06) | |
Eastern division | −0.03(0.09) | −0.03(0.09) | |
Health related factors | |||
Comorbidity | 0.08(0.01)*** | 0.08(0.01)*** | |
Catastrophic illness | |||
No(ref) | |||
Yes | −0.27(0.05)*** | −0.27(0.05)*** | |
County-level: Area health resources | |||
Hospitals to population ratio | |||
Low(ref) | |||
High | −0.02(0.06) | ||
Physicians to population ratio | |||
Low(ref) | |||
High | −0.07(0.07) | ||
Clinics to population ratio | |||
Low(ref) | |||
High | −0.09(0.07) | ||
Variance (s.e.) | 0.0190 (0.0071) | 0.0145 (0.0060) | 0.00987 (0.0044) |
z value for covariance parameter estimates | 2.66** | 2.44** | 2.26* |
ICC% | 0.575% | 0.439% | 0.299% |
-2 Log Likelihood | 53,040.48 | 52,563.17 | 52,557.70 |
AIC | 53,044.48 | 52,595.17 | 52,595.70 |
AICc | 53,044.49 | 52,595.18 | 52,595.71 |
BIC | 53,046.67 | 52,612.62 | 52,616.43 |
Pearson Chi-Square | 60,878.27 | 60,908.46 | 60,913.13 |
Pearson Chi-Square/DF | 1.00 | 1.00 | 1.00 |