onlineF

irst

Applied Research Brief: Tobacco Control; Physical Activity

Living With a Smoker and Physical Inactivity: An UnexploredHealth Behavior PathwayCarole K. Holahan, PhD; Charles J. Holahan, PhD; Xiaoyin Li, MA

Abstract

Purpose. The present study investigated an unexplored health behavior pathway: theassociation between living with a smoker and physical inactivity.

Design. The study performed an analysis of cross-sectional data from the second wave of theSurvey of Midlife Development in the United States (2004–2006).

Subjects. The sample included 1050 women and 945 men, ages 30 to 84 years(mean, 56.5 years).

Measures. In addition to control variables, survey items indexed presence of a smoker in therespondent’s home, and physical inactivity and low health commitment on the part of therespondent.

Analysis. Analysis employed multiple logistic regression analyses, controlling forsociodemographic factors as well as health restrictions on physical activity and respondents’current smoking status.

Results. Living with a smoker was linked to 56% higher odds of physical inactivity. Lowhealth commitment mediated this association. Living with a smoker was linked to lower healthcommitment (B ¼ .31), and low health commitment was linked in turn to increased odds ofphysical inactivity (odds ratio, 1.36).

Conclusions. The increased health risk among individuals living with a smoker is assumedto be a function of exposure to secondhand smoke. We demonstrate an unexplored behavioralpathway involving a link between living with a smoker and physical inactivity. These findingssuggest that household smoking bans may have broad health behavior effects beyond reducingsmoking. (Am J Health Promot 0000;00[0]:000–000.)

Key Words: Smoking, Physical Activity, Health Attitudes, Prevention Research.Manuscript format research; Research purpose: modeling/relationship testing;Study design: nonexperimental; Outcome measure: behavioral; Setting: home;Health focus: smoking control/physical activity; Strategy: policy; Target populationage: adults; Target population circumstances: national sample of men and women

PURPOSE

Secondhand tobacco smoke has beenlinked to increased risk for ischemicheart disease, lung cancer, and otherrespiratory illnesses.1 The increasedhealth risk among individuals living witha smoker is assumed to be a function ofexposure to secondhand smoke.1 How-ever, the possibility that living with asmoker fosters a broader householdculture of negative health behaviors islargely unstudied. Two areas of inquirylay a foundation for such a hypothesis.Increasing evidence from health behav-ior research has documented the clus-tering of different behaviors withinindividuals2 and the contagion of thesame behavior between individuals.3,4

Here, integrating these lines of research,we propose a novel process—that livingwith a smoker provides an unexploredpathway to physical inactivity.

The present study investigated theassociation between living with a smokerand physical inactivity in a U.S. nationalsample of 1995 adults. In addition, weexamined a possible meditational path-way: living with a smoker may under-mine housemates’ commitment tohealth-promoting behavior. A smallbody of studies on living with a smokerand negative dietary behaviors5,6 andhigher body weight7,8 provides a foun-dation for exploring the possible effectsof living with a smoker on physicalinactivity. Data were from the secondwave of the study of Midlife Develop-ment in the United States (MIDUS II),initially undertaken to study behavioral,psychologic, and social factors in mentaland physical health.9 MIDUS II wasfunded by the National Institute onAging. We hypothesized that: (1) livingwith a smoker would be positivelyrelated to physical inactivity, and (2) therelation between living with a smoker

Carole K. Holahan, PhD, and Xiaoyin Li, MA, are with the Department of Kinesiology andHealth Education, University of Texas, Austin, Texas. Charles J. Holahan, PhD, is with theDepartment of Psychology, University of Texas, Austin, Texas.

Send reprint requests to Carole K. Holahan, PhD, Department of Kinesiology and HealthEducation, 2109 San Jacinto Blvd–D3700, University of Texas at Austin, Austin, TX 78712-1415; c.holahan@austin.utexas.edu.

This manuscript was submitted August 20, 2013; revisions were requested November 4, 2013 and March 6, 2014; themanuscript was accepted for publication March 31, 2014.

Copyright � 0000 by American Journal of Health Promotion, Inc.0890-1171/00/$5.00 þ 0DOI: 10.4278/ajhp.130820-ARB-434

American Journal of Health Promotion Month 0000, Vol. 0, No. 0 0

onlineF

irst

and physical inactivity would be medi-ated by low health commitment.

METHODS

DesignThe present study involved an anal-

ysis of cross-sectional data from MIDUSII, collected in 2004–2006. The origi-nal sample was selected through arandom-digit–dialing procedure withmetropolitan oversampling. Eligibleparticipants were noninstitutionalized,English-speaking adults residing in thecontiguous United States. After oralconsent, the data were collected via aphone interview and self-administeredquestionnaires.9 All participants pro-viding complete data on the measureswere included in the present analyses.Institutional Review Board approval forthis data analysis was given by thecurrent investigators’ university.

SampleThe sample included 1050 women

and 945 men (age range, 30–84 years;mean, 56.5 years). The sample com-prised 1827 white individuals, 92 blackindividuals, and 76 individuals report-ing other racial backgrounds. Withrespect to education, 627 respondentshad a high school education or less, 551had some college education, and 817had graduated from college. Mediantotal income was $67,000.

MeasuresSociodemographic Variables. Age (inyears) was a continuous variable. Genderwas coded as male¼ 0, female¼ 1.Income was indexed in $10,000 units.Educational level was dummy coded assome college and college graduate, withhigh school graduate or less as a refer-ence category. Race was dummy coded asblack and other racial background, withwhite as a reference category.

Current Smoking Status. Following pre-vious research,4 participants wereasked, ‘‘Do you smoke cigarettes regu-larly NOW?’’ Regularly was defined as‘‘at least a few cigarettes every day.’’Affirmative responses were coded as 1(n¼ 285; 14%) and negative responseswere coded as 0.

Health Limits. Two items indexed theextent to which participants’ healthlimited moderate and (separately) vig-orous activity. Each item was dummy

coded as ‘‘a little,’’ ‘‘some,’’ and ‘‘a lot,’’with ‘‘not at all’’ as a reference category.The modal response was ‘‘not at all’’ forboth moderate (65%) and vigorous(29%) activity.

Smoker in the Home. Following previousresearch,4 participants were asked, ‘‘Atthe current time, does anyone regularlysmoke cigarettes or other tobaccoproducts INSIDE your home (otherthan yourself)?’’ Affirmative responseswere coded as 1 (n¼ 230; 12%) andnegative responses were coded as 0.

Physical Inactivity. A composite measureof physical inactivity was constructedfrom four items that asked respondentsto indicate how often they engaged inmoderate (e.g., brisk walking, slowswimming, or low-impact aerobics) andvigorous (e.g., running, vigorous swim-ming, or high-intensity aerobics) physi-cal activity during their leisure or freetime during the summer and winter.Current U.S. Centers for Disease Con-trol and Prevention (CDC) aerobicguidelines recommend engaging inmoderate (150 minutes) and/or vigor-ous (75 minutes) activity spread acrossthe week.10 To approximate CDCguidelines, participants who answeredless than several times a week on 5-pointitems ranging from never to severaltimes a week for both moderate andvigorous activity during either summeror winter were coded as 1 (n¼ 1335;67%) for physically inactive. Partici-pants who answered several times aweek or more for either moderate orvigorous activity during both summerand winter were coded as 0.

Low Health Commitment. Following pre-vious research,11 a composite measureof health commitment was developedusing two items. Respondents wereasked their level of endorsement of thestatements, ‘‘I work hard at trying to stayhealthy’’ (scaled 1–7) and ‘‘thought/effort put into health’’ (scaled 0–10).Low health commitment was indexed asthe mean of the two items (reversescored and standardized). The medianon both scales was 2, reflecting relativelyhigh health commitment.

AnalysisAll analyses involved multiple logistic

regression using Mplus (Muthen &Muthen, Los Angeles, California). Pres-

ence of a smoker in the home was thepredictor, low health commitment wasthe mediator, and physical inactivity wasthe outcome. All analyses controlled forage, gender, income, educational level,race, health restrictions on moderateand vigorous exercise, and respon-dents’ current smoking status.

RESULTS

We began by examining the directrelationship between living with asmoker and physical inactivity. Livingwith a smoker was linked to 56% higherodds of physical inactivity (odds ratio,1.56; p , .05; confidence interval [CI],1.06, 2.30). The relationship betweenliving with a smoker and physicalinactivity did not vary by age, gender,income, educational level, race, healthrestrictions on physical activity, or re-spondents’ smoking status (p . .05).

Next, we tested a model with lowhealth commitment mediating therelationship between living with asmoker and physical inactivity (Table).Consistent with the proposed medita-tional model, living with a smoker wassignificantly positively associated withlow health commitment (B ¼ .31; CI,.07, .55; p , .05), which in turn wassignificantly positively associated withphysical inactivity (odds ratio, 1.36; p ,

.01; CI, 1.27, 1.46). The predictedassociations were significant both be-fore and after controlling for covari-ates, although as expected thepredicted associations were attenuatedwith the inclusion of covariates. Sig-nificant mediation from living with asmoker to physical inactivity throughlow health commitment was confirmedby the Baron and Kenny criteria formediation and by bias-corrected boot-strap confidence intervals of the indi-rect effect (95% CI, .01, .10).

DISCUSSION

SummaryWe demonstrate an unexplored be-

havioral pathway involving a link be-tween living with a smoker and physicalinactivity. Living with a smoker waslinked to 56% higher odds of physicalinactivity. Low health commitment me-diated this association. Living with asmoker was linked to lower healthcommitment, and low health commit-

0 American Journal of Health Promotion Month 0000, Vol. 0, No. 0

onlineF

irst

ment was in turn linked to increasedodds of physical inactivity. The analysescontrolled for sociodemographic factorsas well as health restrictions on physicalactivity and respondents’ smoking sta-tus. Housemates who smoke may pro-mote low health commitment throughmodeling, peer pressure, and socialcontrol, and low support for healthybehaviors.3,12,13

LimitationsThe study was cross-sectional and

cannot establish causality. The datawere self-reported and are subject torecall bias, common method variance,and social desirability, and several mea-sures are brief. However, previous stud-ies with the MIDUS sample haveestablished the validity of these mea-sures in predicting health and healthbehavior.4,14 In addition, the samplewas restricted to English-speaking adultsand underrepresented racial and ethnicminorities. The percentage of the sam-ple reporting a smoker in the home wassmall and may restrict generalizability.

SignificancePhysical inactivity is associated with

increased morbidity and mortality.15

Although a link between smoking andphysical inactivity has been establishedwithin individuals,2 the present findingthat one individual’s smoking is asso-ciated with another individual’s physi-cal inactivity is novel.

Acknowledgments

The data for this study were obtained from the Inter-university Consortium for Political and Social Research(ICPSR). The data are drawn from: Carol Ryff, David M.Almeida, John S. Ayanian, Deborah S. Carr, Paul D.Cleary, Christopher Coe, Richard Davison, Robert F.Krueger, Marge E. Lachman, Nadine F. Marks, Daniel K.

Mroczek, Teresa Seeman, Marsha Mailick Seltzer, BurtonH. Singer, Richard P. Sloan, Patricia A. Tun, MaxineWeinstein, and David Williams. MIDLIFE DEVELOP-MENT IN THE UNITED STATES (MIDUS2), 2004–2006 [Computer file]. ICPSR 04652-vl. Madison, WI:University of Wisconsin, Survey Center [producer], 2006.Ann Arbor, MI: Inter-university Consortium for Politicaland Social Research [distributor], 2007-03-22.

References

1. Oberg M, Jaakkola MS, Woodward A, et al.Worldwide burden of disease fromexposure to second-hand smoke: aretrospective analysis of data from 192countries. Lancet. 2011;377:139–146.

2. Heroux M, Janssen I, Lee DC, et al.Clustering of unhealthy behaviors in theaerobics center longitudinal study.Prevention Sci. 2012;13:183–195.

3. Li KK, Cardinal BJ, Acock AC. Concordanceof physical activity trajectories amongmiddle-aged and older married couples:impact of diseases and functionaldifficulties. J Gerontol B Psychol Sci Soc Sci.2013;68:794–806.

4. Holahan CK, Holahan CJ, Li X, Jung S.Social influences on smoking in Americanworkers: the role of the presence of smokersin the workplace and in the home. Am JHealth Promot. 2013;28:105–107.

5. Hampl JS, Taylor CA, Booth CL. Differencesin dietary patterns of nonsmoking adultsmarried to smokers vs. nonsmokers. Am JHealth Promot. 2001;16:1–6.

6. Ortega RM, Lopez-Sobaler AM, Aparicio A,et al. Passive smoking as a conditioner offood habits and nutritional status:repercussions on health. In: Cassady RE,Tidswell EI, eds. Nutrition Research at theLeading Edge. New York, NY: NovaBiomedical; 2008:133–155.

7. Abrevaya J, Tang H. Body mass index infamilies: spousal correlation, endogeneity,and intergenerational transmission. EurEcon. 2011;41:841–864.

8. Koo LC, Kabat GC, Rylander R, et al.Dietary and lifestyle correlates of passivesmoking in Hong Kong, Japan, Sweden,and the USA. Soc Sci Med. 1997:45;159–169.

9. Ryff CD, Almeida DM, Ayanian JS, et al.Midlife Development in the United States(MIDUS II), 2004–2006 [computer file].ICPSR04652-vl. Madison, WI: University ofWisconsin; 2006. Ann Arbor, MI: ICPSR; 2007.

10. Centers for Disease Control andPrevention. How much physical activity doyou need? Available at: http://www.cdc.gov/physicalactivity/everyone/guidelines/index.html/. Accessed August 28, 2013.

11. Holahan CK, Holahan CJ, Velasquez KE, etal. Purposiveness and leisure-time physicalactivity in women in early midlife. WomenHealth. 2012;51:661–675.

12. Umberson D, Crosnoe R, Reczek C. Socialrelationships and health behavior across thelife course. Annu Rev Sociol. 2010;36:139–157.

13. Thoits PA. Mechanisms linking social tiesand support to physical and mental health.J Health Soc Behav. 2011;52:145–161.

14. Ryff CD, Brim OG, Kessler RC, eds. HowHealthy Are We? A Study of Well-Being atMidlife. Chicago, Ill: University of ChicagoPress; 2004.

15. Centers for Disease Control andPrevention. Physical activity and health.Available at: http://www.cdc.gov/physicalactivity/everyone/health/.Accessed August 28, 2013.

SO WHAT? Implications forHealth Promotion Practitionersand ResearchersWhat is already known on this topic?

Existing research has establishedthat diverse health behaviors clusterwithin individuals,2 and that specifichealth behaviors are linked to theenactment of the same behavior inan individual’s social network.3,4

What does this article add?In cross-sectional analyses we dem-

onstrated a previously unexploredassociation between living with asmoker and a health behavior in adifferent domain—physical inactivity.Moreover, we showed how low healthcommitment provides a pathwaythrough which living with a smoker isassociated with physical inactivity.What are the implications for healthpromotion practice or research?

These cross-sectional findings areconsistent with a view that living with asmoker encourages other unhealthybehaviors. They highlight the need toexplore the broader effects of smok-ing and smoking bans on housemates’health practices beyond smoking.

Table

Results of a Multiple Logistic Regression Analysis Examining the Relationship of Living With a Smoker and Low Health

Commitment With Physical Inactivity (N¼ 1995)†

Predictors OR (95% CI)

Age (y) 1.02 (1.01, 1.03)*

Gender (male ¼ 0, female ¼ 1) 0.96 (0.78, 1.19)

Income 0.98 (0.96, 0.99)**

Some college 0.66 (0.50, 0.88)*

College graduate 0.51 (0.39, 0.66)*

Black 3.37 (1.85, 6.12)*

Other race 1.64 (0.94, 2.85)

Smoking status (no ¼ 0, yes ¼ 1) 2.16 (1.47, 3.17)*

A little limitation to moderate activity 1.24 (0.89, 1.72)

Some limitation to moderate activity 1.38 (0.87, 2.18)

A lot of limitation to moderate activity 1.83 (1.03, 3.25)**

A little limitation to vigorous activity 1.88 (1.44, 2.45)*

Some limitation to vigorous activity 1.82 (1.31, 2.52)*

A lot of limitation to vigorous activity 2.25 (1.50, 3.36)*

Living with a smoker (no ¼ 0, yes ¼ 1) 1.45 (0.98, 2.16)

Low health commitment (no ¼ 0, yes ¼ 1) 1.36 (1.27, 1.46)*

† OR indicates odds ratio; and CI, confidence interval.

* p , 0.01.

** p , 0.05.

American Journal of Health Promotion Month 0000, Vol. 0, No. 0 0