Table 3

Multivariate logistic regression showing variables influencing the odds of using services or drugs and the choice of providers when being ill


Using vs. not use services or drugs
Providers vs. Self-treatment
Public vs. private

Explanatory variables
OR (CI 95%)
OR (CI 95%)
OR (CI 95%)


(N = 8380)
(N = 8125)
(N = 6123)

R2: 0.085
R2: 0.036
R2:0.049

No of household members (continuous)
1.07(0.98–1.16)
1.05(0.01–1.11)
0.98(0.88–1.07)
Perceived seriousness of illness



     Can work (reference category)
1.00
1.00
1.00
     Miss work
2.70(1.95–3.73)
1.31(1.15–1.5)
1.31(0.89–1.91)
     Confine to bed
4.55(3.0–6.98)
1.85(1.59–2.16)
3.11(2.23–4.35)
Male (reference group female)
0.49(0.34–0.71)
0.96(0.81–1.13)
0.87(0.72–1.04)
Age group



     15–60 (reference category)
1.00
1.00
1.00
     Under 15
1.59(94–2.67)
1.13(0.89–1.43)
1.41(0.94–2.09)
     Over 60
0.71(0.45–1.10)
1.17(0.94–1.58)
0.41(0.25–0.78)
Education level



     No schooling (reference category)
1.00
1.00
1.00
     Only primary school
2.10(1.29–3.42)
1.22(0.94–1.58)
0.80(0.50–1.28)
     Higher education
3.59(1.65–6.79)
1.51(1.05–2.16)
1.12(0.61–2.05)
Equalized household expenditure quintiles



     Poorest (reference category)
1.00
1.00
1.00
     2nd
1.55(0.96–2.46)
1.22(0.95–1.56)
1.20(0.68–2.08)
     3rd
1.67(1.02–2.72)
1.31(1.02–1.66)
1.34(0.79–2.23)
     4th
1.66(1.11–2.83)
2.51(1.92–3.28)
1.89(1.11–3.23)
     Richest
4.78(2.42–8.43)
2.77(2.10–3.66)
2.09(1.22–3.32)

Adjusted standard errors for correlation between episodes for each person (corrections for clustering). See earlier the method section; bold represents significance.

Thuan et al. BMC Health Services Research 2008 8:162   doi:10.1186/1472-6963-8-162