Skip to main content
Fig. 2 | BMC Health Services Research

Fig. 2

From: Investigating the causal effect of socioeconomic status on quality of care under a universal health insurance system - a marginal structural model approach

Fig. 2

Marginal structural modeling conceptual framework. *Baseline covariates (L(0)) were used to predict the low-income group exposure at each wave. The time-varying covariates (L(t)) was used to predict the low-income exposure at each wave, Pinc(t), as the outcome variable in the first stage. Then, during the second stage, Pinc(t) was used as the independent variable to examine the causal relationship between income and preventable hospitalization and comorbidity. All models in the first stage included baseline and prior low-income status to predict Pinc(t)

Back to article page