Article Text
PDF
Abstract
Background In 2011, New York City (NYC) parks and beaches became smoke-free. There is currently little research evaluating the impact of such laws on smoking behaviour at the population level.
Methods We used an interrupted time-series study design to analyse data from the New York State Adult Tobacco Survey to assess the law's impact using the rest of New York State as a comparison. Trends in how frequently respondents noticed people smoking in parks and beaches were analysed between the third quarter of 2009 and the fourth quarter of 2012, comparing NYC to the rest of the state.
Results The trend in the frequency of NYC residents noticing people smoking in local parks and beaches decreased significantly over the six quarters after the law took effect. There was no comparable decline among residents in the rest of the state. An increase in the number of respondents who never noticed people smoking in NYC contributed to this decline.
Conclusions These results are consistent with previous studies and provide population-level evidence that suggest the law has reduced smoking in parks and on beaches.
- Environment
- Denormalization
- Secondhand smoke
Statistics from Altmetric.com
Introduction
Laws prohibiting smoking in parks and on beaches have been adopted by municipalities throughout North America over the last decade. In May 2011, New York City (NYC) joined hundreds of cities and counties in the USA by making all city parks and beaches smoke-free.1 ,2 Visitors can be fined $50 for a violation. There is currently little research evaluating the effectiveness of smoke-free air laws on smoking behaviour in large outdoor public spaces like parks and beaches. Previous research on the effectiveness of outdoor smoking bans has tended to focus on restaurants3 or college campuses.4 ,5 A recent study of a British Columbian bylaw is one of the few to look at smoking restrictions in parks and on beaches.6
Two previous studies evaluating the NYC law described in Johns et al7 suggest it was effective in reducing smoking in the targeted areas. A litter audit found that smoking-related litter decreased by about two-third on beaches the first summer of implementation. An observational study found a 60% decrease in the number of people smoking in parks 5 months after the law. Although these results provide converging evidence, they have several limitations. Neither study included an external comparison area, making it difficult to rule out secular trends or other history effects as alternative explanations. The studies also examined relatively short-term outcomes in the summer and fall immediately after the law took effect. It is unknown whether the decrease in smoking persisted beyond the first year of the law.
To address these limitations, we analysed data from the New York State Adult Tobacco Survey (ATS) to assess the law's impact over a longer time period while using the rest of New York State, which did not have a comprehensive smoke-free parks and beaches law at that time, as a comparison. Since 2009, the ATS has included an item assessing how frequently respondents noticed people smoking in parks and beaches. We analysed trends in this item through 2012 and predicted a decrease in the number of respondents noticing smoking in NYC in the period following the law compared with the rest of the state (ROS).
Methods
The New York ATS is a quarterly cross-sectional telephone survey of randomly selected state residents ages 18 years and older. The sample includes smokers and non-smokers, and uses a stratified dual-frame design that included landline and cellphone users. Respondent place of residence was based on self-reported zip code. Responses are weighted to adjust for non-response and poststratified to state population totals for phone usage, age, race, sex and education. Starting in the third quarter (July) of 2009, respondents were asked how frequently they had noticed people smoking in the parks or on the beaches of their local municipality in the past 12 months. Responses were recorded as never, occasionally, frequently and almost every time. Data for this item were available through Q4 of 2012; the law went into effect in Q2 of 2011, which is the 8th quarter in the available data series. A total of 15 839 participants responded to this item; 5212 (33%) from NYC and 10 627 (67%) from the ROS.
We used segmented regression to conduct analyses for interrupted time-series studies.8 ,9 The key variables in the model were time, which is a continuous variable that counts the quarters (14) from the start of the observation period (Q3 2009) to the end (Q4 2012), a dummy variable representing the law (prelaw=0; postlaw=1), a dummy variable representing region of respondent residence (NYC=1; ROS=0) and a variable representing the trend after the law, coded as 0 before the law (Q3 2009 to Q1 2011) and (time−8) for Q2 2011 through Q4 2012. The model also included three interactions: region * time to control for regional secular trends; region * law to test for a mean level change in NYC after the law took effect; and region * trend-after-law to test if the trend (ie, slope) in NYC following the law decreased compared with ROS.
In the first model, we analysed item responses as a continuous outcome ranging from 0 (never) to 3 (almost every time). We next analysed the prevalence of ‘never’ and ‘almost every time’ responses in separate models to examine how the effects were distributed across response categories. In preliminary analyses, we included respondent age, race, sex and smoking status along with the focal variables (ie, law, region, time, trend-after-law and all two-way interactions). Smoking status and sex were not significant variables in the model and therefore excluded from the final models. The final models controlled for respondent age and race, and included dummy variables representing quarters to control for seasonal effects. A Durbin–Watson test confirmed that responses were not autocorrelated. Analyses were conducted using survey procedures in SAS V.9.2 to account for the complex survey design and weights.
Results
The results of the regression analyses are presented in table 1. The first model produced a significant region * trend-after-law interaction, indicating that the frequency of noticing smoking in parks and beaches in NYC decreased over the six quarters following the law compared with the ROS, B=−0.08, p=0.005. The predicted values from the model are displayed in figure 1. In NYC, the predicted frequency decreased from 1.86 (95% CI 1.74 to 1.99) in Q2 of 2011 to 1.57 (95% CI 1.45 to 1.69) in Q4 of 2012. By comparison, the predicted frequency in the rest of state did not change over this period (1.76 in Q2 of 2011 vs 1.74 in Q4 of 2012). Respondent smoking status did not moderate this effect (data not shown).
Segmented regression models for noticing smoking in parks and beaches, comparing NYC to the rest of the state before and after the smoke-free law: New York State Adult Tobacco Survey, Q3 2009–Q4 2012
Predicted frequency of noticing people smoking in parks and on beaches between Q3 2009 and Q4 2012, comparing New York City with the rest of the state.
Analysis of the prevalence of never seeing people smoking in parks and beaches (model 2) produced only a marginal region * trend-after-law interaction, indicating a positive trend in NYC following the law, B=0.01, p=0.067. In NYC, the predicted per cent of never noticing smokers increased from 12.3% (95% CI 8.6 to 16.1) in Q2 of 2011 to 22.3% (95% CI 18.0 to 26.6) in Q4 of 2012. In the ROS, the predicted per cent of respondents reported never seeing smokers showed a small upward trend from 12.4% (95% CI 9.6 to 15.1) in Q2 of 2011 to 16.9% (95% CI 13.8 to 20.0) in Q4 of 2012. There was also a marginal region * trend-after-law interaction for the prevalence of almost always seeing people smoking, B=−0.03, p=0.073 (model 3). In Q2 of 2011, 36.8% (95% CI 31.4 to 42.4) of respondents in NYC and 29.7% (95% CI 25.5 to 33.9) of respondents in the ROS reported almost always noticing smoking in parks and beaches. In Q4 of 2012, 28.1% (95% CI 23.3 to 32.9) of NYC respondents said they almost always noticed smoking, while 31.9% (95% CI 27.8 to 35.9) in the ROS said they almost always noticed smoking.
Discussion
The trend in the frequency of NYC residents noticing people smoking in parks and on beaches decreased significantly across the quarters after those areas became smoke-free; there was no comparable change in the ROS. An increase in the trend in never seeing people smoking contributed to the overall decrease, with the prevalence in NYC nearly doubling over the six quarters following the law. The results provide further evidence that the law was successful in reducing smoking in NYC parks and beaches, and shows a gradual but persistent impact over a year after implementation. They are also consistent with the recent British Columbian bylaw study and build on them by demonstrating population-level impact using a stronger study design.6
The data do not allow us to determine whether the decrease occurred equally on beaches and in parks. Previous findings from NYC indicate that smoking declined at similar rates in both areas.7 Because smoking was measured indirectly, the results could be influenced by recall biases. The 12-month timeframe of the survey item in particular could have diluted the estimated impact of the law immediately after it took effect; however, this would make our findings conservative. Using the rest of New York State as a comparison area also has limitations. A number of municipalities outside of NYC had smoke-free parks and beaches laws at the time the law took effect in NYC. However, this would make it more difficult to detect changes in NYC, which would also make our results conservative.
Further research is needed to examine the impact of such laws on distal outcomes, like perceived antismoking norms.10 Given the association between perceived norms and smoking behaviours,11 policies that contribute to antismoking norms have the potential to help reduce the prevalence of smoking via denormalisation.2 ,12
Taken together with previous studies,6 ,7 these results show that smoke-free air laws in outdoor public spaces can be effective and durable. Such policies could offer jurisdictions a simple strategy for reducing exposure to tobacco pollutants (ie, smoke and litter) in outdoor areas.
What this paper adds
There is currently little research examining the effect of laws prohibiting smoking in parks and on beaches on smoking behaviour. This study is the first to assess a smoke-free parks and beaches law at the population level. The use of a quasi-experimental interrupted time-series design is an improvement over previous studies.
The results show a significant decrease in the trend of New York City residents noticing people smoking in parks and on beaches after the law took effect. They decrease over a year. There was no significant change in trend in the rest of New York State in the postlaw period.
These findings suggest self-enforced smoking bans in large outdoor areas can be effective and durable.
Acknowledgments
We thank Gene Glass, Kevin Konty, Alfredo Morabia and Dean Keith Simonton for their willingness to offer comments on the analytic methods used in this study.
References
Footnotes
Contributors MJ designed the study, performed the analysis and drafted and edited the manuscript. SMF and DTR supported the analysis and helped draft and edit. SMK edited the paper. HRJ supported conceptualisation of the study design and analysis, and interpretation of the results.
Competing interests None.
Ethics approval The New York State Department of Health Institutional Review Board provided ethical review and approval.
Provenance and peer review Not commissioned; externally peer reviewed.