Smoke-free policies can effectively protect nonsmokers from secondhand smoke (SHS) exposure in multiunit housing. We surveyed all affordable multiunit housing properties in North Carolina to determine the statewide prevalence of smoke-free policies and to identify predictors of smoke-free policies.
Representatives of affordable housing properties in North Carolina completed a mailed or online survey during June through October 2013. The primary outcome measure was presence of a smoke-free policy, defined as prohibiting smoking in all residential units. We used χ2 analysis and multivariate logistic regression to identify correlates of smoke-free policies.
Of 1,865 eligible properties, responses were received for 1,063 (57%). A total of 16.5% of properties had policies that prohibited smoking in all residential units, while 69.6% prohibited smoking in indoor common areas. In multivariate analysis, an increase in the number of children per unit was associated with a decrease in the odds of having a smoke-free policy at most properties. Newer properties across all company sizes were more likely to have smoke-free policies. Accessing units from interior hallways predicted smoke-free policies among medium-sized companies.
More smoke-free policies in affordable multiunit housing are needed to protect vulnerable populations, particularly children, from SHS exposure. Public health professionals should continue to educate housing operators about SHS and the benefits of smoke-free policies at all properties, including older ones and ones where units are accessed from outside rather than from an interior hallway.
The home is a primary source of secondhand smoke (SHS) exposure (
Smoke-free policies can protect nonsmokers from SHS exposure in MUH (
This study was a cross-sectional, observational study. Our sample of affordable housing properties included all multiunit properties in North Carolina that receive site-based federal subsidies or state low-income housing tax credits. This sample comprised properties owned by for-profit and nonprofit private owners and did not include public housing properties. We defined MUH as housing with more than 1 living unit and excluded properties classified as group homes.
This study was an establishment survey, with the unit of analysis being the property itself and not the individual answering the survey. In some cases, a single respondent completed multiple surveys because he or she managed more than 1 property; however, all survey questions were objective questions about policies and events occurring at the property and none assessed respondent attitudes.
We created a sampling frame of affordable housing properties in North Carolina by obtaining lists from the US Department of Agriculture’s (USDA’s) Rural Development program, the US Department of Housing and Urban Development (HUD), and the North Carolina Housing Finance Agency (NCHFA), the state’s purveyor of low-income housing tax credits. We cross-checked this frame with lists of properties from 8 large affordable housing management companies and added missing properties. After removing duplicates, we identified 1,865 properties, all of which were eligible to participate in this census survey.
We developed the survey instrument by reviewing previous MUH operator surveys and obtaining input from smoke-free MUH researchers. We also consulted affordable housing property managers, who provided feedback regarding use of industry-specific terms and appropriateness of survey response categories. We pretested the survey with 18 property managers and made revisions to eliminate confusing wording and skip patterns and reduce respondents’ difficulty with finding information needed to answer the questions.
The Office of Human Research Ethics at the University of North Carolina at Chapel Hill (UNC-CH) ruled that this study did not constitute human subjects research under federal regulations and did not require its approval. We called properties to confirm the appropriate respondents’ names and contact information and their preference for an online or paper survey. For large companies, we contacted corporate headquarters to determine the most appropriate survey respondent for their properties.
The Carolina Survey Research Laboratory (CSRL) at UNC-CH collected data from June to October 2013. Respondents who chose to complete the survey online received a mailed letter containing the URL and password for the survey. Respondents who preferred paper surveys were mailed a questionnaire in booklet form with a prepaid return envelope. The CSRL contacted nonrespondents by telephone during July and August 2013 and sent them either online survey information or mailed questionnaires by August 30, 2013. Several medium and large companies (defined below) instructed their property managers to participate in the survey and contacted nonrespondents twice during the data collection period to encourage participation. Respondents who chose to be compensated received a check for $25.
To ensure representation among small companies and single owners, we stratified and weighted properties by management company size. Using tertiary cut points, we defined as small those companies managing 1 to 27 properties in North Carolina, as medium those managing 28 to 79, and as large those managing 80 to 171. We produced weights as the inverse stratum-specific sampling rates and then adjusted for differential response based on stratum-specific response rates. SAS software (version 9.2) (SAS Institute, Inc) was used in all statistical analyses and reported significance at the .05 level.
We used χ2 goodness-of-fit tabulations to compare characteristics such as year built and location, subsidy type and source, and type of access to units in smoke-free and smoking-allowed properties. We used logistic regression to identify relationships between a smoke-free policy and possible predictor variables. Variables included in the logistic regression model were those that were significant in bivariate analysis or identified by North Carolina housing operators as being important in determining whether to adopt smoke-free policies. The model included the following predictor variables: year built (in or before 2001, or after 2001); company size; type of access to residential units (from an interior hallway, from outside, or a combination); number of stories in residential buildings (1 or 2 stories or ≥3 stories); and number of children per unit (number of children per property divided by the number of residential units per property). The quadratic term for number of children per unit was also included because it indicated curvilinearity in the relationship between the number of children per unit and the log odds of a property’s being smoke-free. Family subsidy status was included in an early version of the model but was not significant and not included in the final model.
A total of 1,063 completed surveys were received for the 1,865 eligible properties, for a response rate of 57%. Almost two-thirds (62.3%) of the respondents completed the survey online. Most surveys (87%) were completed by on-site managers.
The responding properties had an average of 55.2 units and 8.0 buildings per property (weighted numbers) (
| Characteristic | Unweighted | Weighted |
|---|---|---|
|
| ||
| No. of units (n = 1,063) | 51.2 | 55.2 (51.7–58.6) |
| No. of residential buildings (n = 1,061) | 7.9 | 8.0 (7.0–8.9) |
|
| ||
| ≤2001 | 74.7 | 75.0 (72.2–77.8) |
| >2001 | 25.3 | 25.0 (22.2–27.8) |
|
| ||
| Family subsidies (n = 1,051) | 41.9 | 40.2 (37.1–43.3) |
| Elderly subsidies (n = 1,054) | 58.6 | 57.3 (54.2–60.5) |
| Disability subsidies (n = 1,055) | 66.5 | 64.6 (61.5–67.6) |
|
| ||
| USDA Rural Development | 36.6 | 31.4 (28.8–33.9) |
| HUD | 38.4 | 43.7 (40.7–46.6) |
| NCHFA tax credits | 45.8 | 43.2 (40.1–46.2) |
|
| ||
| Small | 33.8 | 47.1 (43.8–50.3) |
| Medium | 29.5 | 29.8 (26.9–32.6) |
| Large | 36.7 | 23.2 (20.9–25.4) |
|
| ||
| MSA | 62.8 | 64.9 (62.0–67.9) |
| Non-MSA | 37.2 | 35.1 (32.1–38.0) |
Abbreviations: CI, confidence interval; HUD, US Department of Housing and Urban Development; MSA, metropolitan statistical area; NCHFA, North Carolina Housing Finance Agency; USDA, US Department of Agriculture.
All multiunit properties in North Carolina that receive site-based federal subsidies or state low-income housing tax credits. This sample comprised properties owned by for-profit and nonprofit private owners and did not include public housing properties.
Because of rounding of weighted percentages, some totals may not sum to 100%.
Small companies were defined as managing 1–27 properties, medium companies were defined as managing 28–79 properties, and large companies were defined as managing 80–171 properties.
Office of Management and Budget (
Bivariate analysis indicated a relationship between subsidy source and year built. HUD properties were older than NCHFA properties (
Most properties (81.2%; 95% CI, 78.8%–83.5%) allowed smoking in all residential units, while 16.5% (95% CI, 14.2%–18.7%) did not allow smoking in any residential units, and 2.4% (95% CI, 1.4%–3.3%) allowed smoking in some residential units. Of the very small percentage of properties that allowed smoking in some residential units, 73.6% (95% CI, 62.7%–84.6%) allowed current residents to smoke under a “grandfather” policy, with the intention of being completely smoke-free once all grandfathered smokers had moved out. Among properties prohibiting smoking in residential units, 58.3% (95% CI, 52.4%–64.3%) had converted to smoke-free and 41.7% (95% CI, 35.7%–47.6%) had opened as smoke-free. Most properties banned smoking in indoor common areas (69.6%; 95% CI, 66.7%–72.6%), while few properties prohibited smoking in outdoor common areas (10.8%; 95% CI, 8.9%–12.6%) or outdoor private areas (7.9%; 95% CI, 6.3%–9.5%).
For purposes of comparing smoke-free and smoking-allowed properties, we categorized properties as having a smoke-free policy if smoking was not allowed in any residential units, and smoking-allowed if smoking was allowed in some or all residential units. Children were less likely than adults to live in properties with smoke-free policies, with 10.5% (95% CI, 5.6%–15.4%) of children living in smoke-free properties and 15.1% (95% CI, 10.6%–19.6%) of adults doing so.
In bivariate analysis, the following variables were significantly associated with smoke-free policy status (
| Characteristic | Smoking-Allowed (n = 886), Weighted % (95% CI) | Smoke-Free (n = 177), Weighted % (95% CI) |
|
|---|---|---|---|
|
| 83.5 (81.3–85.8) | 16.5 (14.2–18.7) | NA |
|
| |||
| <1981 | 89.3 (85.3–93.3) | 10.7 (6.7–14.7) | .01 |
| 1981–1989 | 89.8 (86.0–93.6) | 10.2 (6.4–14.0) | |
| 1990–2001 | 86.8 (82.5–91.1) | 13.2 (8.9–17.5) | |
| >2001 | 70.3 (64.6–76.0) | 29.7 (24.0–35.4) | |
|
| |||
| Yes | 85.0 (81.4–88.7) | 15.0 (11.3–18.6) | .36 |
| No | 82.8 (80.0–85.7) | 17.2 (14.3–20.0) | |
|
| |||
| Yes | 89.5 (86.6–92.5) | 10.5 (7.5–13.4) | <.001 |
| No | 78.9 (75.6–82.1) | 21.1 (17.9–24.4) | |
|
| |||
| Yes | 78.8 (75.0–82.6) | 21.2 (17.4–25.0) | <.001 |
| No | 87.1 (84.4–89.9) | 12.9 (10.1–15.6) | |
|
| |||
| Yes | 88.1 (84.9–91.3) | 11.9 (8.7–15.1) | .01 |
| No | 80.3 (77.1–83.5) | 19.7 (16.5–22.9) | |
|
| |||
| Yes | 85.1 (82.3–87.9) | 14.9 (12.1–17.7) | .11 |
| No | 81.3 (77.5–85.1) | 18.7 (14.9–22.5) | |
|
| |||
| Yes | 82.5 (79.6–85.4) | 17.5 (14.6–20.4) | .24 |
| No | 85.4 (81.7–89.2) | 14.6 (10.8–18.3) | |
|
| |||
| Small | 85.4 (81.8–89.0) | 14.6 (11.0–18.2) | .01 |
| Medium | 77.9 (73.7–82.1) | 22.1 (17.9–26.3) | |
| Large | 86.9 (83.6–90.2) | 13.1 (9.8–16.4) | |
|
| |||
| Interior hallway | 77.0 (71.6–82.4) | 23.0 (17.6–28.4) | <.001 |
| Outside | 86.7 (84.2–89.3) | 13.3 (10.7–15.8) | |
| Both | 74.9 (65.4–84.4) | 25.1 (15.6–34.6) | |
|
| |||
| Yes | 82.4 (76.8–88.0) | 17.6 (12.0–23.2) | .65 |
| No | 83.8 (81.3–86.3) | 16.2 (13.7–18.7) | |
|
| |||
| MSA | 84.1 (80.3–88.0) | 15.9 (12.0–19.7) | .71 |
| Non-MSA | 83.2 (80.4–86.0) | 16.8 (14.0–19.6) | |
Abbreviations: CI, confidence interval; HUD, US Department of Housing and Urban Development; MSA, metropolitan statistical area; NA, not applicable; NCHFA, North Carolina Housing Finance Agency; USDA, US Department of Agriculture.
Properties were defined as “smoking-allowed” if they allowed smoking in some or all residential units, and “smoke-free” if smoking was banned in all residential units.
All multiunit properties in North Carolina that receive site-based federal subsidies or state low-income housing tax credits. This sample comprised properties owned by for-profit and nonprofit private owners and did not include public housing properties.
χ2 test comparing characteristic categories of smoking-allowed and smoke-free properties.
Small companies were defined as managing 1 to 27 properties, medium companies were defined as managing 28 to 79 properties, and large companies were defined as managing 80 to 171 properties.
Office of Management and Budget (
In the multivariate model of predictors of smoke-free policies, examination of 2-way interactions (likelihood ratio test
| Variables | Small Company | Medium Company | Large Company | |||
|---|---|---|---|---|---|---|
| Odds Ratio (95% CI) |
| Odds Ratio (95% CI) |
| Odds Ratio (95% CI) |
| |
|
| ||||||
| Number of children per unit | 0.7 (0.2–2.1) | .50 | 0.1 (0.0–0.6) | .02 | 0.1 (0.0–0.4) | .003 |
| Quadratic term for the number of children per unit | 1.1 (0.7–1.7) | .67 | 3.9 (0.9–15.2) | .05 | 2.7 (1.2–6.0) | .02 |
|
| ||||||
| ≤2001 | 0.3 (0.1–0.5) | <.001 | 0.4 (0.2–0.8) | .005 | 0.2 (0.1–0.3) | <.001 |
| >2001 | 1 [Reference] | |||||
|
| ||||||
| Interior hallway | 0.6 (0.3–1.3) | .22 | 3.4 (1.7–7.0) | <.001 | 1.1 (0.4–3.0) | .85 |
| Outside or both | 1 [Reference] | |||||
|
| ||||||
| 1 or 2 | 1.2 (0.5–2.5) | .70 | 3.7 (1.4–10.1) | .01 | 0.3 (0.1–0.7) | .006 |
| 3 or more | 1 [Reference] | |||||
Abbreviation: CI, confidence interval.
Properties were defined as having a smoke-free policy if they did not allow smoking in any residential units.
All multiunit properties in North Carolina that receive site-based federal subsidies or state low-income housing tax credits. This sample comprised properties owned by for-profit and nonprofit private owners and did not include public housing properties.
Small companies were defined as managing 1 to 27 properties in North Carolina.
Medium companies were defined as managing 28 to 79 properties in North Carolina.
Large companies were defined as managing 80 to 171 properties in North Carolina.
Logistic regression model of predictors of smoke-free policies by management company size.
Number of children per property divided by the number of units per property.
An increasing number of children per unit was associated with decreasing odds of a smoke-free policy at properties managed by both medium and large companies. The number of children per unit was inversely related to the presence of a smoke-free policy until reaching 1.9 children per unit among properties managed by medium companies and 2.5 children per unit among properties managed by large companies, shifting to a positive relationship at higher numbers. Properties where an increasing number of children per unit was associated with decreasing odds of a smoke-free policy included 76.7% of properties managed by medium companies and 89.8% of properties managed by large companies.
Several other variables predicted smoke-free policy status in multivariate analysis. Properties managed by all company sizes had lower odds of having a smoke-free policy if they were built before 2002. Type of access to residential units predicted smoke-free policy status among properties managed by medium companies: smoke-free policies were more likely among properties with units accessed by interior hallways (odds ratio, 3.4).
This study was the first statewide survey at the property level to assess the availability of smoke-free policies in affordable MUH. While most properties banned smoking in indoor common areas, few prohibited smoking in residential units, suggesting significant potential for exposure to SHS for residents. Although North Carolina is a tobacco-growing state, this prevalence of smoke-free policies is similar to findings from operator surveys in other areas of the country (
We found that children in North Carolina affordable housing are at increased risk of SHS exposure. An increasing concentration of children was associated with decreasing odds of having a smoke-free policy at most properties. Also, children were more likely to live in smoking-allowed properties than adults; however, the strength of this latter finding is in question because of overlapping confidence intervals. This study adds to the small body of literature regarding children’s SHS exposure in MUH. Children living in MUH have significantly higher levels of exposure to SHS than children living in detached houses (
Older properties across all management company sizes were less likely to have smoke-free policies. This finding, also observed in a New York study of MUH (
Our study also found an association between the existence of a smoke-free policy at a property and its source of subsidy: HUD-subsidized properties were less likely to be smoke-free, while properties subsidized by NCHFA tax credits were more likely to be smoke-free. This difference may be attributed to the age of the property: HUD properties were older than properties that received tax credits, which are used to finance new construction or renovations (
We examined variables not tested in other studies, including management company size and type of access to residential units. Regarding management company size, small and large companies were less likely to have smoke-free policies than medium companies. Small companies may lack knowledge about trends in smoke-free policies, and large companies may be reluctant to convert a large portfolio of properties to smoke-free. More research is needed to determine reasons for this finding and to guide targeted outreach to promote smoke-free policies among small and large management companies. Regarding type of access to residential units, buildings with units accessed from the outside were less likely to have smoke-free policies than units accessed from an interior hallway; interior hallway access predicted smoke-free policies among medium companies. Housing operators may believe there is less smoke transfer between units that do not share a common indoor hallway, so that properties of this style are a lower priority for smoke-free conversion. North Carolina housing operators suggested adding this question to the survey, demonstrating the importance of engaging stakeholders to identify relevant fields of inquiry. Public health professionals should emphasize in their communications to housing operators that residents of apartments accessed from outdoors would benefit from smoke-free policies as well, because smoke can travel between these units through shared walls and ceilings and inside from outdoor balconies and patios.
Strengths of this study include incorporating input from housing industry professionals into the survey design, sampling at the property level, using government lists to create the sampling frame, and having a large sample size and a high response rate in comparison to previous operator surveys. However, this study has several limitations. First, the study examined only affordable housing properties and the results may not be generalizable to market-rate housing. It is possible that market-rate housing has a higher prevalence of smoke-free policies because of lower smoking rates among higher income individuals (
More smoke-free policies in affordable MUH are needed to protect vulnerable populations, particularly children, from SHS exposure. The availability of smoke-free policies in affordable housing properties in North Carolina is low, though increasing. After data collection was completed, 2 large management companies converted their properties to smoke-free in 2014, increasing the percentage of smoke-free properties in North Carolina from 16.5% to 27.9% in the period of a year. To ensure this trend continues in North Carolina and is replicated throughout the country, public health professionals should continue to educate housing operators about the benefits of smoke-free policies at all multiunit properties.
This article was supported in part by a cooperative agreement with the Centers for Disease Control and Prevention (FOA CDC-RFA-DP11-1115PPHF11). Portions of this project’s work involve the Community Transformation Grants initiative supported by Centers for Disease Control and Prevention funding. Users of this document should be aware that every funding source has different requirements governing the appropriate use of those funds. Under US law, no federal funds are permitted to be used for lobbying or to influence, directly or indirectly, specific pieces of pending or proposed legislation at the federal, state, or local levels. Organizations should consult appropriate legal counsel to ensure compliance with all rules, regulations, and restriction of any funding sources. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
The authors thank Ruth Petersen, Sally Herndon, Jim Martin, Sharon Nelson, and Jill Rushing from the North Carolina Division of Public Health for their leadership and assistance in making this survey possible. We also thank Brian King of the Office on Smoking and Health at the Centers for Disease Control and Prevention and Andrea Licht and Mark Travers of the Roswell Park Cancer Institute for providing expert advice on the design of the survey instrument; Anna Hoffmeyer and the Carolina Survey Research Laboratory team for their dedication in implementing this survey; and Nina Collins for providing assistance with reviewing literature and editing manuscript drafts. Finally, we acknowledge the following individuals and organizations for partnering on this study: Scott Alderman of Landura Management Associates; Rick Allen and Becky Stevens of Partnership Property Management; Leah Lyerly and Jeff Gagnon of Westminster Company; Tim Morgan of Evergreen Construction Company; Debbie Honeycutt of GEM Management; Rita Blackmon of Excel Property Management; and Nicole Kiefer of The Arc of North Carolina.
The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors' affiliated institutions.