摘要：Objectives. Because household smoking levels and adoption of domestic smoking rules may be endogenously related, we estimated a nonrecursive regression model to determine the simultaneous relationship between home smoking restrictions and household smoking. Methods. We used data from a May–June 2012 survey of Philadelphia, Pennsylvania, households with smokers (n = 456) to determine the simultaneous association between smoking levels in the home and the presence of home restrictions on smoking. Results. We found that home smoking rules predicted smoking in the home but smoking in the home had no effect on home smoking restrictions. Conclusions. Absent in-home randomized experiments, a quasi-experimental causal inference suggesting that home smoking rules result in lower home smoking levels may be plausible. Secondhand smoke (also known as “passive smoking” or “environmental tobacco smoke”) is a health hazard for children and adults. 1–5 Institutional, city, or national smoking restrictions reduce smoking prevalence and the average consumption of smokers while naturally limiting exposure to secondhand smoke. 6–11 At the household level, the research focus has been on the efficacy of household bans on indoor smoking to reduce nonsmokers’ and children’s exposure. 12–18 Many studies have found that smokers in households with smoking bans or restrictive smoking rules smoke fewer cigarettes than smokers in households with no bans or rules. 19–23 This relationship appears to suggest that household smoking restrictions are effective in reducing household smoking. But do household bans really reduce household smoking? Unfortunately, this situation is not the same as when smoking bans are implemented in bars, 24,25 hospitals, 26 prisons, 27 schools, 28 or countries. 29,30 In all of these examples, the bans are introduced independently of the prevalent smoking levels of the institution, city, or country because passive smoking exposure is seen as an important health hazard that requires an administrative or legislative response. In households, this analogy does not necessarily hold. Household smoking could be negatively related to household smoking bans because smokers who smoke fewer cigarettes or households with little smoking may implement a household smoking ban whereas heavier smokers or households with multiple smokers could not do so. 23,31 In this situation, household smoking bans are endogenous 32 relative to household smoking, which implies that household smoking levels cause the smoking ban, not the reverse. To determine which explanation is correct, an experiment that implemented a household ban randomly in regard to household smoking levels would be appropriate. 14,33,34 Under experimental conditions, the ban would be independent of household smoking levels and the effect of implementing a household smoking ban on smoking could be unambiguously estimated.
