Skip to main content
. 2021 May 25;18(11):5657. doi: 10.3390/ijerph18115657

Table 5.

Two-part model—relationship between disability and OOP health expenditures.

Level of Disability Inpatient Care Outpatient Care Self-Treatment Any Care
Part 1: logit a (coefficients)
None (Ref.) 0 0 0 0
Mild 0.548 *** 0.637 ** 0.525 *** 0.827 ***
Moderate 1.207 *** 0.759 *** 0.531 ** 1.149 ***
Severe 0.856 ** 0.366 0.513 * 1.000 **
Part 2: GLM b (coefficients)
None (Ref.) 0 0 0 0
Mild 0.059 0.333 *** 0.167 0.311 *
Moderate 0.208 0.286 0.397 * 0.567 **
Severe 1.056 ** 1.622 *** 0.930 ** 1.352 ***
Predicted value combining logit and GLM (OOP expenditures in VND)
None (Ref.) 0 0 0 0
Mild 33,079 * 66,115 *** 26,790 *** 108,879 ***
Moderate 78,107 *** 68,903 *** 44,047 *** 181,682 ***
Severe 116,449 *** 164,857 *** 83,196 *** 353,283 ***
n 3967 3967 3967 3967
N 63,947,541 63,947,541 63,947,541 63,947,541

Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. All models include age, gender, completed education, employment, health insurance, (log) household income, household size, rural/urban place of residence, and province of residence. a Logit link is used. b Generalized linear models (GLM) with log links and gamma distribution are used. Survey data analyses and robust standard errors are used.