Evaluation of Secondhand Smoke Exposure in New York City Public Housing After Implementation of the 2018 Federal Smoke-Free Housing Policy

This cohort study examines whether the 2018 implementation of a federal ban on smoking in public housing settings was associated with a decrease in indoor levels of secondhand smoke in New York City (NYC) public housing developments 12 months later.


Introduction
Tobacco and involuntary exposure to secondhand smoke (SHS) remain leading causes of preventable death in the US, with more than 480 000 deaths annually. 1 A large body of research has documented health hazards associated with SHS exposure. These include cancers, cardiovascular disease, and respiratory complications among adults and sudden infant death syndrome, respiratory infections, and asthma among children. 2,3 Population-based studies using serologic cotinine biomarkers performed during the past 30 years have confirmed that exposure to SHS has decreased owing in great part to reductions in cigarette smoking and laws prohibiting smoking in public settings such as bars, restaurants, and workplaces. Among US residents who do not smoke, the prevalence of SHS exposure between 1988 and 2014 decreased from 87.5% to 25.2%. 4 Despite this progress, reducing SHS exposure remains a public health priority. Nationally, approximately 1 in 4 individuals who do not smoke (58 million individuals) are still exposed to SHS routinely, primarily at home. 5 Persistent disparities in SHS exposure are widely documented, with exposure being highest among racial/ethnic minority populations and persons with low income, reflecting higher smoking prevalence [6][7][8][9] and differential risks across housing and social environments. [10][11][12][13] Pursuant to a recent rule passed by the US Department of Housing and Urban Development (HUD), all public housing authorities were required to implement smoke-free housing (SFH) policies in their developments beginning July 30, 2018. 14 Few studies have prospectively evaluated the association of SFH policies with SHS exposure using objectively measured airborne nicotine concentration or particulate matter less than 2.5 μm in diameter (PM 2.5 ). [15][16][17][18][19][20] These studies have had mixed results, mostly showing initial or no associations with SHS exposure, but follow-up duration and sample sizes have been limited.
As part of a larger natural experiment study to assess consequences of the SFH policy on health outcomes, we prospectively measured air quality before and 12 months after SFH policy implementation in 21 large public housing and private sector high-rise buildings housing predominantly low-income residents in New York City (NYC). This city is an important setting in which to examine consequences of the SFH policy because the NYC Housing Authority (NYCHA) is the largest housing authority in the US, with roughly 400 000 residents. Nearly 100 000 additional NYC households receive housing vouchers to subsidize apartment rental costs in private-sector buildings (herein referred to as Section 8). The goal of this analysis was to empirically assess the association of the SFH policy with SHS exposure in its first year of implementation using large samples of apartments and common areas.

Study Population
For this cohort study, we selected a purposeful sample of 21 high-rise buildings located in Manhattan and the Bronx, including 10 NYCHA buildings subject to the SFH policy and 11 Section 8 buildings.
The following building eligibility criteria were established to ensure homogeneity: (1) buildings were high-rise (>15 floors); (2) buildings had a large resident population (>150 units); (3) at least 80% of the resident population were Black or Hispanic individuals (mirroring distribution in most NYCHA buildings); and (4) at least 20% of the resident population was younger than 18 years. The NYU Grossman School of Medicine institutional review board approved the study protocol, and all study participants signed a written informed consent form. This study followed the Transparent Reporting of Evaluations With Nonrandomized Designs (TREND) reporting guideline. 21 Detailed information on the study population and procedures has been described elsewhere. 22 conducted the survey among NYCHA residents and the subsequent prepolicy air-monitoring data collection from April to July 2018, immediately before the federal SFH rule went into effect, and in the 11 selected Section 8 buildings immediately thereafter, from August to November 2018, before the onset of cold weather. We targeted households that were above the third floor and had working telephone numbers to avoid biasing influences of ambient outdoor PM 2.5 . Adults aged 18 years or older who spoke English or Spanish were eligible for participation; we invited 1 member per household to participate in a 15-minute telephone survey that included questions about demographic characteristics, cigarette smoking behavior, and SHS incursions. We conducted doorto-door enrollment in person using tablets among households not reached by telephone. The survey yielded 1030 responses (559 NYCHA residents and 471 Section 8 residents), for a response rate of 35.3% among NYCHA residents and 32.1% among Section 8 residents.
At the conclusion of the survey, we invited a volunteer sample of 275 nonsmoking households (157 NYCHA households and 118 Section 8 households) to enroll in a longitudinal air-monitoring study to track air quality in 7-day stretches inside their homes every 6 months for 2.5 years (5 waves in total). This sample size was selected a priori based on statistical power considerations detailed in Cardozo et al. 22 We excluded 12 households from this analysis based on evidence of probable indoor smoking by residents or guests (8 households) or incomplete air quality data at baseline (4 households), for an analytic sample size of 263 nonsmoking households (153 in NYCHA buildings and 110 in Section 8 buildings) (eFigure in the Supplement). A mean of 16 (range, [14][15][16][17][18][19] households per NYCHA building and 11 (range, 5-18) households per Section 8 building participated.
After conducting air-monitoring sessions among the 263 nonsmoking households before implementation of the SFH policy, we repeated air monitoring 6 months (December 2018 to March 2019) and 12 months (May 2019 to September 2019) after policy implementation. We compared findings before vs 12 months after policy implementation using comparable summer months to avoid seasonal influences. Of the 263 households enrolled, we completed air monitoring 12 months after the policy in 230 households, for a 1-year response rate of 87.5%. In addition to nonsmoking households, we also monitored air quality in 2 randomly selected hallways and 2 randomly selected stairwells per building for 7-day stretches in each wave (45 measurements in NYCHA buildings and 46 in Section 8 buildings).

Objective Air Quality and SHS Measures
Experts have emphasized the superiority of air-quality monitoring over biological cotinine to measure SHS exposure in housing. 17,[24][25][26] Therefore, the primary measure of SHS incursions in this study was airborne nicotine concentration measured over a 7-day period using passive, bisulfatecoated filters placed in living rooms of enrolled nonsmoking households and common areas.
Airborne nicotine concentration has been widely recognized in studies of SHS as a highly specific marker of tobacco smoking. 11,16,27,28 Passive samplers were prepared and analyzed at Johns Hopkins University Bloomberg School of Public Health using the school's Secondhand Smoke Exposure Assessment Laboratory standard operating procedures. 29 The nicotine detection limit for 7-day samples was 0.017 μg/m 3 . We also measured ambient PM 2.5 inside homes (secondary outcome) using AirBeam monitors (HabitatMap), which are novel low-cost particle sensors, 30 and counted cigarette butts in common areas twice per wave in 4 randomly selected hallways and 2 randomly selected stairwells in the bottom 10 floors per building, averaging the numbers of cigarette butts counted. Detailed information on the study methods is described elsewhere. 22

Statistical Analysis
We tabulated demographic variables, smoking prevalence, and prevalence of SHS incursions from baseline survey responses and compared differences among NYCHA and Section 8 residents using χ 2 statistics. For air nicotine concentration, we calculated geometric means and estimated the percentage of filters with detectable nicotine. 22 Estimates were computed separately for common areas (stairwells and hallways) and nonsmoking apartments. For PM 2.5 , we calculated means and the We used a difference-in-difference (DID) approach to compare within-group changes in air quality between NYCHA vs Section 8 buildings before vs 12 months after SFH policy implementation using mixed linear regression models for the 2 air quality measures (nicotine concentration and PM 2.5 concentration) in stairwells and hallways as well as for cigarette butt counts in stairwells and hallways. We modeled outcomes for nicotine concentration by taking the natural log of the original nicotine concentration to account for skewedness. All regression models included fixed effects for study arm (NYCHA vs Section 8), data collection wave (before vs 12 months after policy implementation), and their interaction and adjusted for clustering of units nested within buildings and repeated measurements over time. Model-based mean differences with 95% CIs before SFH policy implementation to 12 months after SFH policy implementation were calculated for each outcome measure. The DID estimates compared the magnitude of differential changes in NYCHA vs Section 8 buildings. Exact P values were reported; the significance level was set at P < .05 using a 2-sided test. To account for the possible influence of systematic variation in outdoor air quality between groups, we repeated the DID model adjusting for outdoor ambient PM 2.5 levels. Because ambient air pollution varies spatially more than temporally across NYC over short durations, we opted to use spatially granular zip code data from the 2017 NYC Community Air Survey. 32 All analyses were performed using SAS statistical software, version 9.4 (SAS Institute).

Smoking Prevalence, SHS Exposure, and Indoor Air Quality Before Policy Implementation
Telephone survey results suggested that residents across both building types were similar in demographic and smoking-related characteristics. Among 1030 residents living in NYCHA and  (Table 1).
In the summer months immediately before policy implementation, measures of air nicotine and PM 2.5 were higher in NYCHA buildings than in Section 8 buildings (

Changes in Indoor Air Quality 1 Year After Policy Implementation
In summer 2019, one year after policy implementation, air quality was measured again and field staff repeated systematic cigarette butt counts. In NYCHA common areas and in nonsmoking apartments, introduction of the SFH policy was not associated with differential reductions in air nicotine compared with similar settings in Section 8 buildings (Table 3). Overall, nicotine concentrations remained higher in NYCHA settings (Figure). In apartments and stairwells, temporal changes in air nicotine levels were similar between NYCHA and Section 8 buildings, and the magnitude of differential change between arms was negligible and not statistically significant (apartment DID: , but the differences were not statistically significant. Before adjustment for ambient air quality, indoor PM 2.5 levels were decreased in both NYCHA and Section 8 apartments, but decreases were larger in Section 8 apartments (Table 2). After adjustment for ambient air quality, the difference in indoor PM 2.5 levels in Section 8 apartments compared with NYCHA apartments was even larger. Variability in mean stairwell cigarette butt counts across buildings was not statistically significant, but reductions were larger for Section 8 buildings (DID, 5.18; 95% CI, −3.15 to 13.51; P = .21) (Table 3).

Discussion
This study evaluated whether a federal policy to ban smoking in public housing authority buildings was associated with reductions in air nicotine and PM 2.5 levels 1 year after the policy went into effect with use of a natural experiment design and extensive monitoring in a sample of NYC public housing and Section 8 high-rise buildings. We found no differential change in air nicotine levels in NYCHA apartments or stairwells compared with Section 8 buildings. Although modest reductions in air nicotine levels were observed in NYCHA vs Section 8 hallways over time, a finding corroborated by fewer observed cigarette butts, we cannot rule out the possibility of chance in these modest changes. Continued monitoring may confirm whether NYCHA residents are smoking less frequently in visible common areas, such as hallways, owing to the SFH policy and whether they continue to smoke in stairwells and inside homes, where neighbors may be less likely to notice or complain.
The findings of this study are consistent with those of other studies examining the association of SFH policies with SHS exposure. 15,16,18 One recent study 18  studies, these authors found that Boston SFH policy implementation was associated with greater decreases in PM 2.5 and airborne nicotine concentration 1 year after the policy implementation compared with housing that had no policy, although the association was not equal at all sites. 16

Limitations
This study has limitations. First, buildings selected for monitoring were not randomly chosen, which may limit generalizability of findings to all NYC public housing buildings. However, building criteria were determined a priori, and buildings were purposefully selected from a small list in consultation with NYCHA managers, taking into account factors likely to affect recruitment (eg, events affecting resident trust) or air quality (eg, nearby construction). In addition, similarity in resident characteristics, smoking prevalence, e-cigarette use, other household behaviors associated with PM 2.5 levels, and the buildings' structural characteristics between final samples of NYCHA and Section 8 buildings, as well as high response rates in both groups, suggests high internal validity and applicability to other buildings. Second, air quality monitoring equipment is subject to variability. To address this, PM 2.5 monitors were calibrated before each sampling wave using cigarette smoke.
Passive nicotine monitors have been extensively validated in previous studies, 15,40-42 although a lower limit of detection would have permitted more detailed evaluation of SHS exposure in nonsmoking households. Third, the DID analysis included only 1 round of prepolicy measurements, limiting the ability to verify the parallel-trends assumption in comparison groups before policy intervention.

Conclusions
In this cohort study of air quality, HUD SFH policy was not associated with reduced indoor air nicotine levels. These findings suggest that in its first year, the HUD-issued SFH policy was not associated with SHS levels in the largest housing authority in the US. Additional efforts to support policy enforcement, provide cessation services, and raise awareness of health benefits of reduced SHS exposure are needed.