Background
Thirdhand smoke (THS) is a recent discovery that contributes to indoor pollution and compromises health [
1]. The 3 R definition of THS describe it as the residual pollutant that
remains on a variety of indoor surfaces and in dust, is
re-emitted back into a gas phase, and
reacts with other compounds in the air [
2]. THS exposure may occur long after secondhand smoke appears (SHS) [
3]. It can linger on surfaces long after cigarettes have been extinguished [
4‐
6]. Matt (2011) [
7] found that even weeks and months after a cigarette has been smoked, harmful particulates remain on countertops, floors, upholstery, carpets, clothing, and other surfaces. Furthermore, removal of nicotine residues from carpet and walls has been found to be nearly impossible [
8], leading to continual exposure to THS. In addition, when the nicotine in tobacco smoke sorbed to indoor surfaces reacts with nitrous acid, a common component of indoor air pollutant, substantial levels of hazardous carcinogens called tobacco-specific nitrosamines (TSNAs) were formed, including NNA and NNK [
5,
7,
9,
10]. NNA is absent in freshly emitted SHS, but a main TSNA formed in this process. In a recent study, Bo et al. demonstrated for the first time that exposure to THS, acute or chronic, increased DNA damage (genotoxicity) in human cell lines, which could lead to formation of cancer. While research on human exposure to THS and its effects on health, behavior and social cultural consequences warrants further study [
6,
8‐
12], THS poses a likely health hazard to non-smokers who are exposed.
Infants and small children are especially susceptible to THS exposure because of their immature respiratory and immune systems and likelihood to crawl and play on, inhale, touch and hand-to-mouth contaminated surfaces, such as floors and upholstery [
11,
12]. There also has been research that suggests that THS is potentially hazardous to the health of fetuses [
13]. The poor are also more likely to be exposed to THS because smoking and SHS are more prevalent in low-income households. A recent study found the nicotine concentrations in the dust are higher in low-income non-smoking households than those with income above the median [
14]. Matt et al.’s study [
7] indicated that THS accumulates in smokers’ homes and persists even long after the smokers move out and the homes are cleaned and repainted for new residents. Non-smokers living in homes (houses, condos, apartments) formerly occupied by smokers are involuntarily exposed to THS [
7]. Therefore, those with incomes that do not allow much free choice of rental housing might be more likely to be exposed to THS [
15].
Knowledge and beliefs about SHS are correlated with smoking cessation and reduction; however, few studies have examined similar constructs about THS and how they may impact preventive smoking-related behaviors. Two studies have found no to limited awareness about THS compared to SHS [
16,
17], while one of them showed that beliefs about THS’ harm on children was independently associated with home smoking bans [
17]. Recently, education about THS has been incorporated into interventions to promote home smoking bans [
18,
19]. Drehmer [
20] found that THS harm beliefs were related to more strict enforcement of smoke free bans in homes and cars and increased numbers of quit attempts, which is encouraging evidence for inclusion of THS education in interventions aiming to decrease the impact of tobacco use [
21]. Based on this literature, more accurate knowledge and beliefs about THS might be associated with being a nonsmoker, having a smoke-free home, and having been exposed to education regarding the harms of THS.
A reliable and valid scale of THS beliefs might enable more precise assessment of the belief that THS is harmful. It may enable assessment of the degree to which such beliefs contribute to smoke-free home bans, avoidance of SHS in homes and other indoor spaces. The purpose of this article is to present the development of a THS Beliefs Scale and its initial psychometric properties tested among participants of a survey panel.
Results
Participant characteristics in the pilot sample
Due to the sampling design, 50% of participants were current smokers living with at least one non-smoker, and 50% were current non-smokers living with at least one smoker. Most participants were married or living with a partner (67.2%) and white (78.8%) with at least some college (68.8%) and an annual household income over $25,000 (76.0%).
Beliefs about Thirdhand Smoke scale development
Descriptives of the 19 item scale
Cronbach’s alpha for the entire 19-item scale was 0.95 (both raw and standardized). Exclusion of any item would reduce reliability. Item means were reasonably close to the scale mean and standard deviations were adequate (around 1.0). The scree plot and the eigenvalues were both supporting the conclusion of a two factor solution that explained 97.0% of the variance. Oblique varimax rotation yielded strong factor loadings onto two distinct factors. The first factor can be described as measuring beliefs about THS persistence in the built environment. The second factor includes items that assess beliefs about the impact of THS on health.
Scale reduction
Traditional suggestions for scale reduction start with eliminating items with low factor loadings with less than .30 [
29]. No items had factor loadings this small. Hence, we sequentially reduced the scale by excluding items with the lowest factor loadings such that the final scale’s reliability as measured with Cronbach’s alpha was greater than 0.90. This approach reduced the scale to nine items: four items related to THS persistence in the environment and five items related to THS impact on health. The final items, their means, standard deviations and factor loadings are presented in Table
2. Factor 1 includes items that describe THS in the built environment, capturing persistence of smoke particles, accumulation of THS, and ineffectiveness of THS reduction by means other than not smoking in the house. Factor 2 includes health impact of THS and transmission of THS through means other than the air. This reduced scale had excellent overall reliability (Cronbach’s alpha = .91) and strong reliability in the sub-scales (Cronbach’s alpha = .88 for both factors).
Table 2
Means, standard deviations, and factor loadings of final scale items in the validation sample
Smoke particles can remain in a room for weeks. | 3.65 | (0.95) | 0.759 | |
Smoke particles can remain in a room for days. | 3.89 | (0.85) | 0.749 | |
Smoke particles get absorbed into furniture and walls. | 4.29 | (0.74) | 0.582 | |
Opening windows or using air conditioners does not eliminate all smoke particles in a room. | 3.86 | (0.87) | 0.491 | |
Breathing air in a room today where people smoked yesterday can harm the health of adults. | 4.02 | (0.84) | | 0.721 |
Particles in rooms where people smoked yesterday can cause cancer. | 3.56 | (0.98) | | 0.675 |
Breathing air in a room today where people smoked yesterday can harm the health of infants and children [ 17]. | 4.21 | (0.85) | | 0.576 |
After smoking a cigarette, smoke particles on skin, hair, and clothing can be passed on to others through touch. | 3.67 | (0.99) | | 0.571 |
After touching surfaces where cigarette smoke has settled, particles can enter the body through the skin. | 3.47 | (0.87) | | 0.559 |
Scale invariance and differences in responses
MIMIC models tested indicate scale invariance for smoking status (daily and non-daily smoker versus non-smoker) as well as home smoking ban status (partial versus no ban) making the scale equally suitable for smokers and nonsmokers and those with partial and no smoking ban in their homes. In addition, there were no significant differences in THS beliefs by smoking status or home smoking ban status. However, older participants in the pilot sample were less concerned with the impact of THS on health (β = −.009, p = .02) indicating a reduction in the THS on health score by .09 for each 10 years increase in age.
Participant characteristics in the validation sample
Demographics of participants in the validation sample are shown in Table
3. The validation sample participants had a mean age of 41.1 years (
SD = 12.65), were mostly female (83.6%), African American (69.9%), unemployed (71.6%) and reported an annual household income of $25,000 or less (85.7%) with 46.0% reporting a household income of $10,000 or less. Most participants were renting their home (77.3%) and were smokers themselves (60.3%). About half reported a full smoking ban in their home and an additional 29.0% reported a partial ban.
Table 3
Demographic characteristics of study participants in the validation sample and differences in THS overall scores and sub-scale scores
Full sample | 335 | 100 | 3.92 | 0.63 | | 3.79 | 0.66 | | 3.86 | 0.59 | |
Age group (N/%) |
18–24 | 25 | 7.5% | 3.8 | 0.44 | 0.68 | 3.8 | 0.55 | 0.58 | 3.8 | 0.54 | 0.77 |
25–39 | 124 | 37.0% | 3.8 | 0.55 | 3.9 | 0.59 | 3.8 | 0.66 |
40–65 | 174 | 51.9% | 3.9 | 0.63 | 4.0 | 0.67 | 3.8 | 0.69 |
65 and older | 12 | 3.6% | 3.7 | 0.59 | 3.8 | 0.70 | 3.6 | 0.62 |
Gender (N/%) |
Male | 55 | 16.4% | 3.8 | 0.68 | 0.46 | 3.8 | 0.72 | 0.2853 | 3.8 | 0.75 | 0.78 |
Female | 280 | 83.6% | 3.9 | 0.57 | 3.9 | 0.62 | 3.8 | 0.64 | |
Marital Status (N/%) |
Married/living with partner | 186 | 55.5% | 3.8 | 0.59 | 0.49 | 4.0 | 0.65 | 0.48 | 3.8 | 0.67 | 0.58 |
Single/divorced/separated/widowed | 148 | 44.2% | 3.9 | 0.59 | 3.9 | 0.63 | 3.8 | 0.65 | |
Race (N/%) |
White | 48 | 14.3% | 3.8 | 0.56 | 0.90 | 4.0 | 0.61 | 0.57 | 3.7 | 0.60 | 0.44 |
African American | 234 | 69.9% | 3.9 | 0.60 | 3.9 | 0.64 | 3.8 | 0.68 |
Latina/o | 41 | 12.2% | 3.9 | 0.58 | 3.9 | 0.66 | 3.9 | 0.60 |
Other | 12 | 3.6% | 3.8 | 0.52 | 3.8 | 0.50 | 3.7 | 0.69 |
Education (N/%) |
High school graduate/GED or less | 208 | 62.1% | 3.8 | 0.59 | 0.19 | 3.9 | 0.65 | 0.04 | 3.8 | 0.65 | 0.71 |
Some college/vocational/technical school | 109 | 32.5% | 3.9 | 0.56 | 4.0 | 0.59 | 3.8 | 0.66 |
College graduate or higher | 18 | 5.4% | 4.1 | 0.68 | 4.3 | 0.62 | 3.9 | 0.79 |
Employment (N/%) | | | | | | | | | | | |
Employed | 91 | 27.2% | 3.9 | 0.54 | 0.51 | 3.9 | 0.61 | 0.93 | 3.9 | 0.61 | 0.21 |
Not employed | 240 | 71.6% | 3.8 | 0.61 | 3.9 | 0.64 | 3.8 | 0.68 |
Income (N/%) |
$10,000 or less | 154 | 46.0% | 3.8 | 0.57 | 0.91 | 3.9 | 0.62 | 0.61 | 3.8 | 0.65 | 0.99 |
$10,001 to $25,000 | 133 | 39.7% | 3.9 | 0.62 | 4.0 | 0.66 | 3.8 | 0.68 |
$25,001 to $50,000 | 35 | 10.4% | 3.8 | 0.61 | 3.9 | 0.62 | 3.8 | 0.69 |
$50,001 to $75,000 | 6 | 1.8% | 3.9 | 0.58 | 4.1 | 0.70 | 3.7 | 0.55 |
More than $75,000 | 4 | 1.2% | 3.9 | 0.17 | 4.1 | 0.55 | 3.8 | 0.25 |
Home ownership |
Owner | 70 | 20.9% | 4.0 | 0.62 | 0.04 | 4.1 | 0.66 | 0.02 | 3.9 | 0.68 | 0.13 |
Resident | 259 | 77.3% | 3.8 | 0.58 | 3.9 | 0.62 | 3.8 | 0.68 |
Home smoking ban |
No ban | 70 | 20.9% | 3.6 | 0.60 | 0.001 | 3.8 | 0.62 | 0.02 | 3.5 | 0.69 | 0.0004 |
Partial ban | 97 | 29.0% | 3.8 | 0.58 | 3.9 | 0.63 | 3.8 | 0.65 |
Full ban | 168 | 50.1% | 4.0 | 0.57 | 4.0 | 0.64 | 3.9 | 0.63 |
Smoking status |
Smoker | 202 | 60.3% | 3.8 | 0.59 | 0.02 | 3.9 | 0.64 | 0.09 | 3.7 | 0.65 | 0.01 |
Non-smoker | 133 | 39.7% | 4.0 | 0.58 | 4.0 | 0.62 | 3.9 | 0.66 |
Group assignment |
Intervention group | 158 | 47.2% | 3.97 | 0.59 | | 4.04 | 0.66 | | 3.89 | 0.65 | |
Control group | 177 | 52.8% | 3.76 | 0.57 | 0.001 | 3.82 | 0.59 | 0.002 | 3.69 | 0.66 | 0.005 |
Beliefs about Thirdhand Smoke scale psychometric properties
Confirmatory factor analysis showed the good model fit for the two factor model (Χ
2(24) = 65.546, p < .0001, RMSEA = 0.072, 90%CI (RMSEA) = [0.051; 0.093], CFI = 0.957, TLI = 0.936, SRMR = 0.048). Error covariances between items about THS being harmful to children and to adults were allowed to covary as were items asking about smoke particle transfer through touch and smoke particles entering the body through the skin. Both adjustments improved the a-priori two factor model without error covariances constrained (Χ
2(26) = 140.086, p < .0001, RMSEA = 0.114, 90%CI(RMSEA) = [0.096; 0.133], CFI = 0.883, TLI = 0.838, SRMR = 0.061). The a-priori specified model had statistically significantly better fit than a one factor model (ΔΧ2(1) = 38.532, p < .0001). The high correlation between the factors of 0.82 might indicate low convergent validity. However, no factor loadings were weak (i.e. below .40) and most were strong (i.e. greater than .60).
The factor scores as well as the total scale score differed significantly by key participant criteria indicating good construct validity. Those who were in the intervention group and thus were exposed to educational materials on THS scored significantly higher on the total THS scale score as well as the sub-scores than participants in the control group (p = .0001 for the full score, p = .002 for THS persistence score, p = .005 for THS health score). Furthermore, a similar pattern was observed when comparing those who had a full smoking ban in their home compared to those who had a partial or no ban (p = .001, p = .002, p = .001, respectively). Participants who owned their homes had higher THS persistence (p = .04) and total scores (p = .02), but had comparable scores on THS health. Smokers had, on average, lower total THS scores (p = .02) and lower THS health scores (p = .01). However, there was no statistically significant difference between smokers and non-smokers on the THS persistence score. When looking at demographics, the only differences found in THS scores were by educational attainment where those with more education had higher THS persistence scores (p = .04). There were no significant differences in any of the scores by age, gender, race, marital status, income, and employment status.
Discussion
While prior research [
17,
30‐
32] has used one specific item to measure this construct, the current study is the first to our knowledge to develop a valid and reliable measure of beliefs regarding THS. This is a critical step in research aimed at addressing THS exposure. As smoke-free air policies are more commonly being implemented both in public and in private spaces [
17,
31,
32], individual understanding of THS and its impact will be critical in further reducing the harm of tobacco smoke. In particular, intervening on beliefs about THS may prove to be beneficial in reducing THS and SHS exposure. Consequently, the creation of a valid and reliable scale of beliefs regarding THS is critical in assessing these intervention efforts.
This study yielded a valid and reliable 9-item scale assessing this construct, reduced from originally 19 items. The final scale with scoring instructions is available in the Additional file
1. While we had conceptualized five domains, the exploratory factor analysis of the full scale indicates that there are only two distinct domains measured by these 19 items. Interestingly, the two factors that emerged related to: 1) how THS operates within the built environment, i.e. the persistence of smoking particles in indoor spaces, and 2) the impact of THS on health. While the former subscale may demonstrate how THS functions within the concrete setting in which smoking occurs, the second measures perceived risk to human health, which is an important theoretical construct related to smoking behavior [
33]. In the context of multiunit housing managers or owners, the former may be of particular interest, as they consider the impact of smoking on rental cleanup/deposits or property values [
34], and might not be concerned with the negative health effects of smoking on renters.
The scale scores differed, as expected, by smoking status and smoke-free ban status in the home. In addition, we found that those who owned their home scored higher on the THS persistence sub-scale. This might be due to home owners living longer in their home than those who rent and thus observing the persistence of the smell and discoloration of walls or due to homeowners increased interest in preserving property values. However, future studies might explore reasons for those differences. In addition, the intervention participants had higher THS sub- and total scores. Thus, having exposure to educational materials may create awareness of THS, its persistence in the environment and its harms.
There are limitations to the current study. While participants for the pilot study were recruited from across the United States, the validation sample was limited in location to Houston, TX. Thus, findings cannot be extended to those living in different situations. Moreover, this scale should be tested in different populations in varied geographic locations and in those with higher income to explore its external validity. Future studies should test a larger number of possible validity variables. Future research examining how this scale operates alongside measures of perceived risk or harm of smoking, SHS, and receptivity to policies to reduce tobacco smoke exposure could enhance our understanding of the validity of this measure.
Acknowledgements
We thank all members of the smoke-free homes team for their contributions to this study.