Using Marginal Structural Modeling to Estimate the Cumulative Impact of an Unconditional Tax Credit on Self-Rated Health
| Autor: | Tony Blakely, Kristie Carter, Ichiro Kawachi, Frank Pega, M. Maria Glymour |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Adult
Male Epidemiology Health Status Marginal structural model 010501 environmental sciences 01 natural sciences Young Adult 03 medical and health sciences 0302 clinical medicine Linear regression Humans Medicine 030212 general & internal medicine Poverty 0105 earth and related environmental sciences Self-rated health Models Statistical business.industry Confounding Regression analysis Middle Aged Taxes Confidence interval Cohort Regression Analysis Female Self Report business New Zealand Demography Cohort study |
| Zdroj: | American Journal of Epidemiology. 183:315-324 |
| ISSN: | 1476-6256 0002-9262 |
| DOI: | 10.1093/aje/kwv211 |
| Popis: | In previous studies, researchers estimated short-term relationships between financial credits and health outcomes using conventional regression analyses, but they did not account for time-varying confounders affected by prior treatment (CAPTs) or the credits' cumulative impacts over time. In this study, we examined the association between total number of years of receiving New Zealand's Family Tax Credit (FTC) and self-rated health (SRH) in 6,900 working-age parents using 7 waves of New Zealand longitudinal data (2002-2009). We conducted conventional linear regression analyses, both unadjusted and adjusted for time-invariant and time-varying confounders measured at baseline, and fitted marginal structural models (MSMs) that more fully adjusted for confounders, including CAPTs. Of all participants, 5.1%-6.8% received the FTC for 1-3 years and 1.8%-3.6% for 4-7 years. In unadjusted and adjusted conventional regression analyses, each additional year of receiving the FTC was associated with 0.033 (95% confidence interval (CI): -0.047, -0.019) and 0.026 (95% CI: -0.041, -0.010) units worse SRH (on a 5-unit scale). In the MSMs, the average causal treatment effect also reflected a small decrease in SRH (unstabilized weights: β = -0.039 unit, 95% CI: -0.058, -0.020; stabilized weights: β = -0.031 unit, 95% CI: -0.050, -0.007). Cumulatively receiving the FTC marginally reduced SRH. Conventional regression analyses and MSMs produced similar estimates, suggesting little bias from CAPTs. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|