Skip to main content

Table 3 Estimation results of the three-period difference-in-differences analysis

From: Does disease management for diabetic nephropathy reduce medical expenditure? Evidence from a three-period difference-in-differences analysis

Dependent Variable Dialysis cost Total medical claims costs (excluding dialysis costs)
Estimation methodRandom EffectFixed EffectFixed Effect (robust)Fixed Effect (robust)
 Estamated coefficient [z-statistics]Estimated coefficient [t -statistics]Estimated coefficient [t -statistics]Estimated coefficient [t -statistics]
T: treatment group dummy variable (β1)−0.0098 [− 0.50]− 0.0086 [− 0.44]−0.0080 [−1.77]0.0432 [1.31]
A: after-intervention dummy variable (β2)0.0160 [3.00]**0.0161 [3.02]**0.0160 [1.63]0.0383 [3.87]**
TA: treatment group and after-intervention dummy variable (β3)−0.0310 [−1.25]− 0.0300 [− 1.21]−0.0299 [−2.13]*−0.0886 {−2.49]*
TB: treatment group and before-intervention dummy variable (β4)0.0397 [1.82]0.0411 [1.88]0.0410 [2.72]**−0.0158 [− 0.47]
B: before-intervention dummy variable (β5)−0.0474 [−8.02]**−0.0475 [− 8.04]**−0.0475 [−3.31]**−0.0477 [− 4.63]**
Female sex dummy variable− 0.0274 [− 0.36]OmittedOmittedOmitted
Age−0.0449 [− 9.56]**OmittedOmittedOmitted
Charlson comorbidity index score0.0049 [3.66]**0.0043 [3.24]**0.0040 [1.16]0.0812 [20.37]**
Ondo dummy variable−0.0894 [− 0.089]OmittedOmittedOmitted
Kamagari dummy variable0.1206 [0.91]OmittedOmittedOmitted
Primary income earner dummy variable−0.0584 [− 0.75]OmittedOmittedOmitted
Secondary income earner dummy variable0.6652 [6.30]**OmittedOmittedOmitted
Constant3.1846 [9.75]**0.2479 [65.88]**0.2479 [28.49]**8.1396 [882.56]**
Diagnostic Test
 Number of observations80,29780,29780,29780,297
 Number of individuals2856285628562856
 R2 within0.00120.00120.00120.0253
  between0.03870.00440.00440.0851
  overall0.04240.00140.00140.0512
 sigma_u1.60071.64571.64570.7619
 sigma_e0.52360.52350.52350.7176
 rho0.90330.90800.90800.5298
  1. Notes: All estimation models are panel data estimation models. “Omitted” means that Stata statistical software automatically omitted several independent variables. * indicates significance at the 5% level, and ** indicates significance at the 1% level. The“sigma_u” is the standard deviations of residuals within groups, and the “sigma_e” is the standard deviation of residuals overall. The rho is the interclass correlation of the above two values. The rho value (0.908) tells us that 90.8% of the variance is because of difference across time (within units)
\