Pega Frank, Gilsanz Paola, Kawachi Ichiro, Wilson Nick, Blakely Tony
Burden of Disease Epidemiology, Equity and Cost-Effectiveness Programme, Division of Health Sciences, University of Otago, PO Box 7343, Wellington, New Zealand.
Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, 677 Huntington Avenue Boston, MA 02115, United States.
Soc Sci Med. 2017 Apr;179:160-165. doi: 10.1016/j.socscimed.2017.03.001. Epub 2017 Mar 2.
The effect of anti-poverty tax credit interventions on tobacco consumption is unclear. Previous studies have estimated short-term effects, did not isolate the effects of cumulative dose of tax credits, produced conflicting results, and used methods with limited control for some time-varying confounders (e.g., those affected by prior treatment) and treatment regimen (i.e., study participants' tax credit receipt pattern over time). We estimated the longer-term, cumulative effect of New Zealand's Family Tax Credit (FTC) on tobacco consumption, using a natural experiment (administrative errors leading to exogenous variation in FTC receipt) and methods specifically for controlling confounding, reverse causation, and treatment regimen. We extracted seven waves (2002-2009) of the nationally representative Survey of Family, Income and Employment including 4404 working-age (18-65 years) parents in families. The exposure was the total numbers of years of receiving FTC. The outcomes were regular smoking and the average daily number of cigarettes usually smoked at wave 7. We estimated average treatment effects using inverse probability of treatment weighting and marginal structural modelling. Each additional year of receiving FTC affected neither the odds of regular tobacco smoking among all parents (odds ratio 1.02, 95% confidence interval 0.94-1.11), nor the number of cigarettes smoked among parents who smoked regularly (rate ratio 1.01, 95% confidence interval 0.99-1.03). We found no evidence for an association between the cumulative number of years of receiving an anti-poverty tax credit and tobacco smoking or consumption among parents. The assumptions of marginal structural modelling are quite demanding, and we therefore cannot rule out residual confounding. Nonetheless, our results suggest that tax credit programme participation will not increase tobacco consumption among poor parents, at least in this high-income country.
反贫困税收抵免干预措施对烟草消费的影响尚不清楚。以往的研究估计了短期影响，未分离税收抵免累积剂量的影响，得出了相互矛盾的结果，并且使用的方法对一些随时间变化的混杂因素（例如，受先前治疗影响的因素）和治疗方案（即研究参与者随时间的税收抵免领取模式）的控制有限。我们利用一项自然实验（行政错误导致税收抵免领取的外生变化）以及专门用于控制混杂、反向因果关系和治疗方案的方法，估计了新西兰家庭税收抵免（FTC）对烟草消费的长期累积影响。我们提取了具有全国代表性的家庭、收入和就业调查的七轮数据（2002 - 2009年），其中包括4404名处于工作年龄（18 - 65岁）的家庭父母。暴露因素是接受FTC的总年数。结果变量是经常吸烟以及在第7轮调查时通常每天吸烟的平均数量。我们使用治疗权重的逆概率和边际结构模型估计平均治疗效果。接受FTC的每增加一年，既不影响所有父母中经常吸烟的几率（优势比1.02，95%置信区间0.94 - 1.11），也不影响经常吸烟的父母的吸烟量（率比1.01，95%置信区间0.99 - 1.03）。我们没有发现接受反贫困税收抵免的累积年数与父母吸烟或烟草消费之间存在关联的证据。边际结构模型的假设要求相当苛刻，因此我们不能排除残余混杂的可能性。尽管如此，我们的结果表明，参与税收抵免计划至少在这个高收入国家不会增加贫困父母的烟草消费。
