Secondhand smoke exposure is associated with smoke-free laws but not urban/rural status

Abstract

The objective was to determine secondhand smoke (SHS) exposure with and without smoke-free laws in urban and rural communities. The research hypothesis was that SHS exposure in public places could be improved by smoke-free law regardless of urban and rural status. Indoor air quality in hospitality venues was assessed in 53 communities (16 urban and 37 rural) before smoke-free laws; 12 communities passed smoke-free laws during the study period. Real-time measurements of particulate matter with 2.5 µm aerodynamic diameter or smaller (PM_2.5) were taken 657 times from 586 distinct venues; about 71 venues had both pre- and post-law measurements. Predictors of log-transformed PM_2.5 level were determined using multilevel modeling. With covariates of county-level percent minority population, percent with at least high school education, adult smoking rate, and venue-level smoker density, indoor air quality was associated with smoke-free policy status and venue type and their interaction. The geometric means for restaurants, bars, and other public places in communities without smoke-free policies were 22, 63, and 25 times higher than in those with smoke-free laws, respectively. Indoor air quality was not associated with urban status of venue, and none of the interactions involving urban status were significant. SHS exposure in public places did not differ by urban/rural status. Indoor air quality was associated with smoke-free law status and venue type.

Implications: This study analyzed 657 measurements of indoor PM_2.5 level in 53 communities in Kentucky, USA. Although indoor air quality in public places was associated with smoke-free policy status and venue type, it did not differ by urban and rural status. The finding supports the idea that population in rural communities can be protected with smoke-free policy. Therefore, it is critical to implement smoke-free policy in rural communities as well as urban areas.

Introduction

Secondhand smoke (SHS) exposure can be significantly reduced at the population level by effectively implementing smoke-free laws (Ottet al., 1996; Lee et al., 2009). In 2012, 1.1 billion people (16% of the world’s population) lived in countries with comprehensive, national smoke-free laws (World Health Organization, 2013). As of 2012, comprehensive national smoke-free laws were implemented in 11 high-income, 29 middle-income, and 3 low-income countries. At the municipal level, 112 million who live in one of the world’s 100 largest cities are protected by comprehensive smoke-free legislation. However, more than 80% of world population are still exposed to SHS in workplace or public venues.

As of January 1, 2014, in the USA, 24 states, along with the District of Columbia, Puerto Rico, and the U.S. Virgin Islands, had implemented comprehensive smoke-free laws prohibiting smoking in all workplaces, restaurants, and bars (Americans for Nonsmokers’ Rights Foundation, 2013). A total of 36 states have some type of smoke-free law requiring workplaces and/or restaurants and/or bars to be 100% smoke-free. In the USA, a total of 1087 municipalities prohibit smoking in workplaces and/or restaurants and/or bars and 598 have comprehensive laws or regulations.

Although smoke-free laws are gaining popularity in the USA, rural communities have fewer smoke-free laws than their urban counterparts (McMilen et al., 2004; Chriqui et al., 2005). Those living in rural communities are more likely to be exposed to secondhand smoke than those living in urban areas (Yao et al., 2012). Smoking prevalence among adults living in rural communities is higher than in urban areas (Rahilly and Farwell, 2007), although smoking prevalence has recently declined in USA. Smoke-free public places policy may increase the adoption of voluntary smoke-free policies (Borland et al., 2006). However, in rural distressed communities, smoke-free laws may not change smoking behaviors (Hahn et al., 2010).

Kentucky, a largely rural state, does not have a statewide smoke-free law, but there are smoke-free laws at the municipal level. As of January 1, 2014, 38 communities had implemented indoor smoke-free laws or regulations at the city or county level; 22 were comprehensive and 16 had significant exemptions. Given that nearly half of Kentuckians live in rural areas (U.S. Census Bureau, 2012), it is important to study the impact of rural status on exposure to SHS and the differential effects of smoke-free laws. Implementation of and compliance with smoking restrictions is a key strategy to reduce socioeconomic inequalities in smoking-related health indicators. There is limited research on the extent to which SHS exposure at the population level varies based on smoke-free restrictions in urban versus rural localities. The purpose of this paper was to assess differences in the indoor air quality in hospitality venues by smoke-free law status of the community, urban versus rural locality, and venue type, controlling for county-level percent minority population, percent of adults with at least a high school education, adult smoking rate, and venue-level smoker density.

Methods

This study was based on 657 indoor air quality assessments from 53 Kentucky counties; data collection occurred between September 2003 and April 2013. Twelve counties were assessed for indoor air quality before and/or after implementation of a smoke-free ordinance or regulation; there were a total of 276 assessments in 205 venues. Forty-one counties did not have smoke-free policies at any point during the data collection; 381 venues were assessed in these communities. Venues were purposively chosen by the community members so that restaurants, bars, and other public venues would be included.

The concentration of PM_2.5 (particulate matter with an aerodynamic diameter <2.5 μm) was measured using an aerosol monitor called SidePak (model AM510; TSI Inc., Shoreview, MN, USA) with a 2.5-μm impactor. Prior to each measurement, the SidePak monitor was zero-calibrated with a HEPA (high-efficiency particulate air) filter according to manufacturer’s specifications. The air flow rate was set to 1.7 L/min, and the recording of PM_2.5 concentrations was carried out every minute. The measurement was corrected by a conversion factor of 0.295 obtained from calibration against gravimetric measurement (Lee et al., 2008), since secondhand smoke was the major PM source in the monitored indoor space. The monitors were regularly evaluated by collocation for QA/QC (quality assurance/quality control) purpose.

The Sidepak was concealed in either a backpack or purse and set so that automatic 1-min samples were collected continuously before entering the venue and during the visit (mean indoor sampling time = 43 min). When inside the venue, a central location was selected, as far away as possible from direct puffs of cigarettes or cigars. In addition to air quality measurements, venue type (restaurant, bar, or other public place), room size, number of persons present, and number of burning cigarettes and cigars were collected. Community members were trained to collect the data; data were downloaded and analyzed by the research team.

Two county-level characteristics, percent minority population (i.e., non-White) and percent of adults age 25 and older who had graduated from high school, were obtained from the 2010 U.S. Census. County-level adult smoking rate was based on the Behavioral Risk Factor Surveillance System (BRFSS) and was a combined estimate based on data from 2007 to 2009. The urban/rural status of the county was based on the 2003 Rural Urban Continuum Codes (http://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspx); counties with the codes between 1 and 3 were defined as urban, whereas those with the codes between 4 and 9 were defined as rural.

The indoor air quality measurements were log-transformed prior to analysis; this transformation is required so that the outcome more closely approximates a normal distribution, since the untransformed air quality data have right skewed distributions. To assign log-transformed values for all venues, raw scores of 0 for average particulates were recoded to 2 prior to transformation, since this was the smallest nonzero measurement observed. Descriptive statistics, including means and standard deviations or frequency distributions, were used to summarize county-level and venue-level variables. Air quality measurements were summarized descriptively using the geometric mean and standard deviation. The association of smoke-free law status and urban versus rural locality was assessed using Fisher’s exact test. The log-transformed air quality values were compared between venues covered by smoke-free policy and those not covered using a two-sample t test.

Multilevel modeling to determine predictors of log-transformed air quality was done using the MIXED procedure in SAS (SAS, 2011); in this model, venues were nested within county to account for the clustering of locations. This type of model allows the inclusion of both county- and venue-level class variables and covariates. The class variables in the model were smoke-free law status (SF; yes/no), urban locality (U; yes/no), and venue type (V; restaurant, bar, other public place) and their two- and three-way interactions. Covariates included in the model were percent minority population, percent completing at least high school, adult smoking rate, and smoker density in the venue. Post hoc analysis of significant class variables and interaction effects was accomplished using Fisher’s least significant difference procedure for pairwise comparisons. Analysis was conducted using SAS, version 9.3 (SAS, 2011); an alpha level of 0.05 was used throughout.

Results

Nearly one-quarter (23%) of the 53 counties had a smoke-free ordinance or regulation, and 30% of counties were urban (Table 1). The average percent minority among the 53 counties was 7%. The mean percent of adults with at least a high school diploma was 69%. On average, 24% of adults were current smokers in the 53 study counties. There was a significant association between presence of a smoke-free policy and urban status. Of the 16 urban counties in the study, 7 (44%) had a smoke-free policy, compared with 5 of the 37 rural counties (14%; P value, Fisher’s exact test = 0.03). The majority of venues were restaurants (72%), with 12% bars and 16% other public places (e.g., government buildings, bingo halls; Table 1). For the 657 air quality measurements in these venues, the average smoker density was 0.5 per 100 m³, with a range from 0 to 16.

Table 1. Descriptive summary of study variables for 657 air quality assessments in 53 counties

County-level variables (N = 53)
Percent minority race/ethnicity	M = 7.2 (SD = 5.3)	Range: 1.7-27.2
Percent adults over 25 completed high school	M = 69.1 (SD = 9.5)	Range: 52.6–86.5
Adult smoking rate (%)	M = 24.4 (SD = 5.9)	Range: 13.9–40.7
Smoke-free policy
Yes	n = 12	22.6%
No	n = 41	77.4%
Urban
Yes	n = 16	30.1%
No	n = 37	69.8%
Venue-level variables (N = 657)
Smoker density (Number of smoker per 100 m^2)	M = 0.5 (SD = 1.1)	Range: 0.0–15.9
PM2.5 (geometric mean and GSD)	GM = 45.1 (GSD = 3.9)	Range: 2.0–1636.0
Venue type
Restaurant	n = 469	71.4%
Bar	n = 81	12.3%
Other	n = 107	16.3%

The geometric mean of indoor PM_2.5 level was 45.1 µg/m³ (GSD = 3.9), ranging from 0 to 1640 µg/m³. Indoor PM_2.5 levels for 57% of the 657 assessments were higher than the National Ambient Air Quality Standard (NAAQS) of 35 µg/m³ for 24 hr for outdoor air set by the U.S. Environmental Protection Agency (EPA). Whereas 17% of the measurements in communities with smoke-free policies had values higher than the NAAQS, 68% of the assessments in venues not protected by smoke-free policy exceeded the NAAQS. The geometric mean for indoor PM_2.5 level in venues protected by smoke-free policies was 12.9 µg/m³ (GSD = 2.8), compared with 62.2 µg/m³ (GSD = 3.5) for venues in communities without smoke-free laws (t = 15.1 and P < 0.0001 comparing the two smoke-free law status groups).

Among the fixed effects, interaction terms, and covariates included in the model, the significant predictors of indoor air quality were smoker density, the main effects of smoke-free policy status and venue type, and the interaction between smoke-free policy status and venue type (Table 2). The indicator for urban/rural status was not significant, and none of the two- or three-way interactions involving the urban/rural indicator were significant. Since the interaction between smoke-free policy status and venue type was significant, post hoc analysis was applied to this interaction rather than to either main effect.

Table 2. Multilevel model to assess predictors of air quality (N = 657 venues in 53 counties)

Covariate/Fixed Effect	F Test (Degrees of Freedom)	P Value
Percent minority in county*	0.7 (1, 22)	0.4
Percent completed high school in county*	0.8 (1, 37)	0.4
Smoking rate in county*	<0.1 (1, 33)	0.9
Smoker density in venue**	41.2 (1, 626)	<0.0001
Smoke-free policy status (SF) of county*	141.8 (1, 565)	<0.0001
Venue type (V)**	9.5 (2, 639)	<0.0001
Urban status (U) of county*	0.8 (1, 43)	0.4
SF × V	3.1 (2, 621)	0.04
SF × U	0.6 (1, 498)	0.4
V × U	0.2 (2, 639)	0.8
SF × V × U	1.0 (2, 616)	0.4

Notes: *County-level variable. **Venue-level variable.

The post hoc analysis demonstrated that the three types of venues in smoke-free communities did not differ from each other on indoor air quality (P > 0.05 for each pairwise comparison). Indoor air quality in counties without smoke-free policies was significantly lower than the corresponding level obtained from the same type of venue in smoke-free counties. The geometric means for restaurants, bars, and other public places located in counties with smoke-free laws were 2.5, 2.8, and 2.3 µg/m³, respectively. In contrast, the geometric means for restaurants, bars, and other public places located in counties without smoke-free policies were 56.0, 177.3, and 58.6 µg/m³, respectively. In addition, whereas the comparison between restaurants and other public places in counties without smoke-free policies was not significant (P > 0.05), the comparison of the geometric mean for bars and each of the other venue types in counties without smoke-free laws was significant (P < 0.0001 in each case).

The parameter estimate for smoker density was positive, suggesting that poorer indoor air quality was predicted by a greater density of smokers in the venue. Other covariates and fixed effects in the model were not significant predictors of indoor air quality.

Discussion

Indoor air quality was not different in rural and urban communities regardless of smoke-free law status, when controlling for county-level percent minority population, percent completing at least high school education, adult smoking rate, and venue-level smoker density. Further, air pollution in restaurants, bars, and other public places did not differ in smoke-free communities regardless of urban/rural locality. This is encouraging given the fact that smoking rates are generally higher in rural communities (Rahilly and Farwell, 2007), and smoking behaviors may be more resistant to change in these communities (Hahn et al., 2010). However, further analysis with more various communities may be needed to consider political determinants (e.g., party distribution) and economic indicators.

Compliance with smoke-free air laws is critical to protect the population from SHS. Nearly all (94.7%) locations regardless of urban or rural location complied with the smoke-free laws as measured by indoor air quality, similar to a California study where one of the earliest smoke-free air laws was implemented (Weber et al., 2003). Although smoking after smoke-free policies was observed only in two bars, 17% of the monitored locations exceeded the NAAQS of 35 µg/m³ post-law. Some of the locations with high PM_2.5 concentrations may have been due to noncompliance. However, the proportion was relatively small and the lesser number of locations did not allow statistical comparison between urban and rural communities.

Indoor fine particle air pollution in smoke-free communities was significantly lower than in those without smoke-free laws. The finding was consistent with other studies (Lee et al., 2009, 2010). The geometric mean PM_2.5 level in counties without smoke-free laws was nearly 2 times higher than NAAQS of 35 µg/m³. Similarly, indoor air pollution in all three types of venues was greater in communities without smoke-free policies compared with venues located in smoke-free communities. The geometric mean PM_2.5 levels in all three types of venues in counties without smoke-free policies exceeded the NAAQS of 35 µg/m³.

The primary limitation of this analysis was that not all venues were matched at both pre- and post-law time periods. The most venues were measured only once at pre-law period. Although the 71 venues were measured at both pre- and post-law time periods, 134 pre- and/or post-law measurements were not matched. However, the lack of ability to match air quality assessments between pre- and post-law would only bias the analysis toward the null hypothesis of no law difference, so this limitation is minimized. Although venues were purposively chosen by community members so that bars, restaurants, and other types of venues would be included, it is possible that not all venues were considered for inclusion. Some information about venue such as HVAC (heating, ventilation, and air conditioning) was not collected, because we monitored the indoor air quality as customers. Since HVAC can reduce indoor PM level, impact of smoke-free policy could be affected by presence or absence of HVAC. Another potential source of error is lack of relative humidity information. The Sidepak monitor can be sensitive to relative humidity (Day and Malm, 2001). When relative humidity is over 60%, the monitor can overestimate. Since we did not measure relative humidity pre- and post-law, our results could be influenced by the impact of relative humidity on the Sidepak monitor.

Funding

This work was supported by the National Research Foundation (NRF) of Korea Grant funded by the Korean Government (no. 2011-0016022) and BK21 Plus project (22A20130012682).

Acknowledgment

The authors appreciate Mark Travers for training data collectors and analyzing initial data on some of the communities reported here, and Songyi Joo for data management work.

Additional information

Notes on contributors

Kiyoung Lee

Kiyoung Lee is a professor in the Graduate School of Public Health at Seoul National University.

Yunhyung Hwang

Yunhyung Hwang is a doctoral candidate in the Graduate School of Public Health at Seoul National University.

Ellen J. Hahn

Ellen J. Hahn and Mary Kay Rayens are professors in the College of Nursing and direct the Tobacco Policy Research Program, University of Kentucky, Lexington, KY, USA.

Hilarie Bratset

Hilarie Bratset is a former data coordinator and Heather Robertson is a program administrator in the Tobacco Policy Research Program, College of Nursing, University of Kentucky, Lexington, KY, USA.