Skip to main content

Table 2 Mixed-effects regression models of the association of hospital bed supply and inequality with maternal mortality ratio, China, 2004–2016

From: Hospital bed supply and inequality as determinants of maternal mortality in China between 2004 and 2016

Variables

Log of maternal mortality ratio

Model 1

Model 2

Model 3

Model 4

Model 5d

Fixed effects, β (95% CI)

 Hospital beds per 1000 population

 

−0.201c (−0.268, −0.134)

−0.178c (−0.245,-0.110)

−0.112c (− 0.191,-0.033)

−0.112b (− 0.210, − 0.013)

 Gini coefficient

 

2.036c (0.941, 3.131)

1.931c (0.840, 3.023)

1.837c (0.762, 2.911)

1.354b (0.123, 2.584)

 Birth rate, ‰

  

−0.005 (− 0.035, 0.026)

−0.010 (− 0.041, 0.020)

−0.012 (− 0.046, 0.023)

 Female illiteracy, %

  

0.020c (0.008, 0.031)

0.018c (0.006, 0.030)

0.022c (0.009, 0.035)

 Log of GDP per capita

   

−0.382c (− 0.645, − 0.118)

−0.406c (− 0.710, − 0.101)

 Year

No

Yes

Yes

Yes

Yes

Random effects, variance (SE)

 Variance between provinces

0.506 (0.135)

0.405 (0.108)

0.319 (0.089)

0.243 (0.071)

0.222 (0.071)

 Variance within provinces

0.239 (0.018)

0.055 (0.004)

0.055 (0.004)

0.055 (0.004)

0.062 (0.006)

 Residual covariance

    

0.401 (0.058)

  ICC

0.679

0.881

0.854

0.816

0.783

 -2 Residual log likelihood

671.36

157.92

162.86

157.95

109.70

  AIC

675.36

161.92

166.86

161.95

115.7

  1. Note: CI: confidence interval, SE: standard error, ICC: intraclass correlation, AIC: Akaike info criterion. c, b and a denote 1, 5 and 10% significance levels, respectively. d, Compared with model 4, model 5 further took the serial covariance in MMR across time within provinces into account when measuring the residual effects