Association of Secondhand Smoke With Stroke Outcomes
Background and Purpose—Approximately half of never smokers are exposed to secondhand smoke (SHS). Smoking is a well-established stroke risk factor, yet associations between SHS, stroke, and poststroke mortality remain uncertain. We aimed to determine the prevalence of exposure to SHS among those with and without stroke and its impact on mortality.
Methods—Data were obtained from the US National Health and Nutrition Examination Surveys for 27 836 never smokers with/without self-reported stroke aged ≥18 years, sampled from 1988 to 1994 and 1999 to 2012, with linked mortality through 2010. Household exposure to SHS was determined by self-report; exposure severity was quantified by serum cotinine level. Independent relationships between SHS and all-cause mortality were assessed using Cox regression models, before and after adjusting for sociodemographics and comorbidities.
Results—From 1988 to 1994 to 1999 to 2012, age-adjusted prevalence of exposure to SHS declined from 11.5% to 6.6% among survivors of stroke (P=0.08), and 14.6% to 5.9% among persons without stroke (P<0.01). Factors associated with high exposure to SHS were male sex, black race, ≤12th-grade education, poverty income ratio ≤200%, high alcohol intake, and history of myocardial infarction (all P<0.05). High exposure to SHS was associated with higher odds of previous stroke (odds ratio, 1.46; P=0.026). There was a dose-dependent relationship between exposure to SHS and all-cause mortality after stroke.
Conclusions—Individuals with previous stroke have 50% greater odds to have been exposed to SHS; SHS is associated with a 2-fold increase in mortality after stroke. This study highlights the importance of obtaining exposure to SHS history and counseling patients and their families on the potential impact of SHS on poststroke outcomes.
Cigarette smoking, an independent modifiable risk factor for cardiovascular disease, doubles the risk of stroke.1 Up to 1 in 4 cardiovascular disease deaths in the United States each year are attributable to smoking2; the risk is strongly dose related.3 Although smoking is a well-established stroke risk factor, the associations between secondhand smoking, stroke, and mortality after stroke remain uncertain.
The Centers for Disease Control and Prevention found that nearly half of US nonsmokers are exposed to secondhand smoke (SHS),4 and no risk-free level of exposure to SHS exists.5,6 Despite a quarter century of multipronged efforts to reduce active smoking, exposure to SHS continues to cause >41 000 deaths among nonsmoking adults and 400 deaths in infants each year, and ≈$5.6 billion annually in terms of lost productivity.7
The aims of the current study were (1) to describe trends in the prevalence of household exposure to SHS among never smokers over the past two and half decades, (2) to examine the association between household exposure to SHS and stroke, and (3) to explore the dose effect of household exposure to SHS on all-cause mortality after stroke.
Data were obtained from the US National Health and Nutrition Examination Surveys (NHANES), which are a series of cross-sectional, national, stratified, multistage probability surveys constituting representative samples of the civilian, noninstitutionalized US population. Each survey participant completed a household interview and underwent a physical examination. Detailed descriptions of the plan and operation of each survey have been published.8 The study received approval from the National Center for Health Statistics Research Ethics Review Board, and participants were asked to sign an informed consent form. Our sample included adult (≥18 years) never smokers from survey participation in 1988 to 1994 and 1999 to 2012 to mortality assessment in 2010.
Definition of SHS
Household exposure to SHS was measured in 2 ways: self-report and serum cotinine levels. NHANES provides data on exposure to SHS at home in family smoking questionnaires. The specific question that provided data on exposure to SHS at home was: “Does anyone who lives here smoke cigarettes, cigars, or pipes anywhere inside this home?” If the answer to this question was yes, the respondent was considered to be exposed to SHS at home.
Exposure to SHS severity was quantified by serum cotinine level, a metabolite of nicotine that has been validated in previous human studies with a sensitivity of 91% and specificity of 87% to differentiate active smokers from nonsmokers not exposed to SHS and those exposed to SHS (area under the curve 0.902).9–11 Cotinine levels were categorize by undetectable (<0.05 ng/mL), median, and across quartiles.
Outcome Measures: SHS Prevalence, Stroke, and Mortality
Prevalence of exposure to SHS and the geometric mean of serum cotinine levels were computed and compared across time (1988–1994 and 1999–2012) in people with and without stroke. History of stroke was obtained by self-report (“Has a doctor or other health professional ever told you that you had a stroke”). Mortality was reported as 10-year cumulative mortality and as mortality rates per 1000 person-years.
Demographic and comorbid covariates included age, sex, race/ethnicity, poverty income ratio (PIR; which represents the ratio of family income to their appropriate poverty threshold. Ratios <100% indicate that the income for the respective family below the official definition of poverty. PIR >200% is a commonly used threshold in NHANES studies, indicates that income was 200% above the appropriate poverty threshold), education level (≤12th grade versus >12th grade), hypertension (blood pressure >140/90 mm Hg, on antihypertensive medications, or self-report), diabetes mellitus (hemoglobin A1C ≥6.5%, on diabetes mellitus medications, or self-report), hypercholesterolemia (total serum cholesterol ≥240 mg/dL or on cholesterol-lowering medications), history of myocardial infarction (MI, by self-report), chronic kidney disease (CKD, urine albumin to urine creatinine ratio >30 mg/g), chronic obstructive pulmonary disease (forced expiratory volume 1/forced vital capacity ratio <0.7, or self-report), and alcohol intake in the past 12 months. Time was categorized as early (1988–1994) and late (1999–2012).
We estimated the survey-weighted prevalence of the baseline sociodemographic and comorbidities by self-reported history of stroke. We compared the crude and age-adjusted prevalence of self-reported household exposure to SHS for a given time interval according to the level of each baseline covariate in people with and without stroke using logistic regression analyses taking into account the NHANES survey design. Crude and age-adjusted geometric means of serum cotinine levels were compared by level of each covariate over time in people with and without stroke using linear regression analyses taking into account the NHANES survey design. Mean serum cotinine–level comparisons were determined on the log scale to reduce skewness and remove the impact of outliers. Age adjustment was performed using the direct method based on 2 age groups (<65 years and ≥65 years) to the overall age distribution in the early time interval under the regression models.
To evaluate the relationship between exposure to SHS and the odds of previous stroke, multivariable logistic regression analyses before and after adjusting for covariates were performed. Model 1 adjusted for sociodemographic factors (age, sex, race/ethnicity, PIR, and education level). Model 2 adjusted for sociodemographic factors (as above) and comorbidities (hypertension, diabetes, hypercholesterolemia, MI, CKD, chronic obstructive pulmonary disease, and alcohol intake).
For mortality, we computed the crude and the age-adjusted mortality rates per 1000 person-years by each cotinine category using the Poisson regression model adjusting for the survey design after stratifying by previous stroke. In addition, we computed the hazard ratios and the corresponding 10-year adjusted mortality risks by each cotinine category using the Cox regression model taking into account the survey design after stratifying by previous stroke (adjusted models 1 and 2).
To examine the association between exposure to SHS and all-cause mortality, univariable and multivariable analyses were performed using Cox regression models after adjusting for the survey design variables. Age-adjusted hazard ratios of all-cause mortality comparing highest quartile of serum cotinine versus undetectable serum cotinine were computed in people with previous stroke versus those without previous stroke. Data were analyzed using STATA software version 11.2 (StataCorp, Inc, College Station, TX) and R (version 3.0.2).
A total of 27 836 never smokers aged ≥18 years were included in the study. Survivors of stroke were more likely to be women, age ≥65 years, have ≤12th-grade education, have PIR ≤200%, and have hypertension, diabetes mellitus, hypercholesterolemia, MI, and CKD (Table 1).
Prevalence of Exposure to SHS
From 1988 to 1994 to 1999 to 2012, age-adjusted prevalence of household exposure to SHS declined from 11.5% (95% confidence interval [CI], 5.6%–17.4%) to 6.6% (95% CI, 4.4%–8.8%) among survivors of stroke (P=0.08) and from 14.6% (95% CI, 13.0%–16.2%) to 5.9% (95% CI, 5.3%–6.5%) among people without stroke (P<0.01; Table 2). Although SHS prevalence among all subgroups without previous stroke declined over time, SHS prevalence among survivors of stroke who were aged <45 years, ≥80 years, men, were associated with PIR >200% and with CKD, or had reported high alcohol intake increased without achieving statistical significance.
The geometric mean of serum cotinine levels decreased over time from 0.26 to 0.06 ng/mL (P<0.01) among survivors of stroke and from 0.24 to 0.05 ng/mL (P<0.01) among people without previous stroke. Similar trends were observed after adjusting for age. There was a temporal shift toward more individuals having undetectable serum cotinine levels, from 21.84% in 1988 to 1994 to 64.78% in 1999 to 2012 (inversely, prevalence of exposure to SHS declined from 78.16% to 35.22% over time; Table I in the online-only Data Supplement).
Characteristics associated with high exposure to SHS (higher mean serum cotinine level) were younger age, male sex, black race, ≤12th-grade education, PIR ≤200%, high alcohol intake (>12 drinks per year), and history of MI (all P<0.05 in 1999–2012 survey; Table II in the online-only Data Supplement).
Association Between SHS and Stroke History
Stroke prevalence was higher among participants with self-reported household SHS (2.69±0.48%) compared with those without exposure to SHS (2.24±0.14%) in 1999 to 2012. High exposure to SHS (cotinine level above median) was associated with higher odds of stroke (odds ratio, 1.46; 95% confidence interval [CI], 1.05–2.3; P=0.026) in the NHANES 1999 to 2012 cohort (Table 3). This relationship was not observed in the 1988 to 1994 cohort.
Dose-Dependent Relationship Between Exposure to SHS and Mortality Among Survivors of Stroke
Survivors of stroke with self-reported exposure to SHS had higher all-cause mortality compared with survivors of stroke without exposure to SHS (age-adjusted mortality rate: 96.4±20.8 versus 56.7±4.8 per 1000 person-years; P=0.026; Table 4). Survivors of stroke were at higher odds of dying from exposure to SHS (adjusted hazard ratio=1.72; 95% CI, 1.02–2.91; P=0.041). This correlation was not observed in participants in the 1988 to 1994 survey or among participants without a history of stroke.
There was a dose-dependent relationship between the quantity of exposure to SHS and all-cause mortality after stroke. Age-standardized mortality rates per 1000 person-years were 48.4±4.9, 47.8±11.5, 75.9±23.6, 103.1±31.1 across increasing cotinine quartiles (trend P=0.01; Figure [A]). Similarly, adjusted 10-year cumulative mortality rates were 39.0±5.9%, 40.7±9.8%, 58.3±14.2%, 65.4±13.3% across increasing cotinine quartiles (trend P=0.019; Table 4). Adjusted mortality rate ratios were 1.04, 1.74, and 2.11 across increasing cotinine quartiles (trend P=0.019). This relationship was not observed in people without stroke (Table II in the online-only Data Supplement).
There was a differential effect of SHS on mortality by history of stroke (Figure [B]). Among individuals with previous stroke, people with serum cotinine levels in the highest quartile had 2.3 times greater mortality rates compared with those with undetectable cotinine levels (hazard ratio=2.34; 95% CI, 1.36–4.01; P<0.01). Among individuals without a history of stroke, the corresponding age-standardized hazard ratio was 1.28 (95% CI, 0.98–1.67; P=0.07). Mortality rates associated with SHS were nearly 2 times greater for people with stroke compared with those without stroke, after adjusting for age (rate ratio=1.82; 95% CI, 0.95–3.50; P=0.07).
Over the past two and a half decades, the prevalence of household exposure to SHS among never smokers decreased considerably from 12% to 7% in participants with previous stroke and from 15% to 6% in participants without previous stroke. Household exposure to SHS remained higher among younger individuals, men, non-Hispanic blacks, and those living in poverty. Individuals with previous stroke were more likely to be exposed to SHS than those without previous stroke. There was a dose-dependent relationship between exposure to SHS and all-cause mortality after stroke. This dose–response relationship was not observed in people without previous history of stroke.
Our finding of a temporal decline in SHS prevalence is consistent with other population-based studies,11–13 meta-analyses,14,15 and a recent CDC Mortality and Morbidity Weekly Report.13 In an analysis of nonsmokers aged ≥3 years who participated in NHANES 1999 to 2012, Homa et al13 found that exposure to SHS declined from 52.5% in 1999 to 2000 to 25.3% in 2011 to 2012. The 2011 to 2012 prevalence noted in the study by Homa et al was similar to the prevalence observed in the REGARDS study (Reasons for Geographic and Racial Differences in Stroke) cohort (23%). Our design differed from that of Homa et al in that we included a longer time frame, only never smokers, adults ≥18 years, and reported outcomes based on self-reported household exposure to SHS and cotinine levels.
We identified factors associated with higher household SHS exposure: younger age, male sex, black race, lower education, poverty, high alcohol intake, and history of MI. Similarly, a US study demonstrated that during 2011 to 2012, SHS was highest among children aged 3 to 11 years (40.6%), non-Hispanic blacks (46.8%), people living below the poverty level (43.2%), and people living in rental housing (36.8%).13 Our study also provides insights into temporal trends in household SHS exposure. Although declines occurred in all age, sex, and racial/ethnic categories among participants without previous stroke, self-reported exposure to SHS increased among survivors of stroke who were 18 to 44 years, ≥80 years, men, had a PIR >200%, had a history of CKD, and had reported high annual alcohol intake. These findings have not been reported previously.
Since the 1980s, smoking bans have been implemented in a variety of settings, likely contributing to the temporal decline in SHS.14 Comprehensive smoke-free legislation has been associated with lower rates of hospital admissions (and deaths) for stroke (relative risk [RR], 0.84; 95% CI, 0.75–0.94), coronary events (RR, 0.85; 95% CI, 0.82–0.88), and other heart disease (RR, 0.61; 95% CI, 0.44–0.85).15 Numerous studies have shown a dose-dependent relationship between SHS and stroke incidence.6,9,16,17 We further demonstrated a dose-dependent relationship between household exposure to SHS and all-cause mortality after stroke, a relationship that was not observed in participants without stroke.
The lack of association between SHS and mortality among those without stroke is hypothesis-generating and may imply a differential physiological effect of exposure to SHS on individuals with different comorbid conditions. Perhaps, SHS can accelerate atherosclerosis in those with underlying vascular risk factors but does not have an effect on those without atherosclerosis. Indeed, SHS may act synergistically with other vascular risk factors such as diabetes mellitus and hypertension. Howard et al16 quantified the progression of carotid artery atherosclerosis using 10 914 participants from the ARIC study (Atherosclerosis Risk in Communities) and found a greater impact of SHS smoking on atherosclerotic progression in participants with diabetes mellitus than those without diabetes mellitus (interaction P=0.004). Chronic smoke exposure can lead to endothelial dysfunction, impaired arterial wall elasticity leading to dysfunctional cerebral autoregulation as have been demonstrated in animal and human studies.18–21
The prevalence of stroke was greater in those who self-reported exposure to SHS than those without this exposure in the later cohort (1999–2012) but not in the earlier cohort (1988–1994). A similar pattern was observed for those with cotinine levels above the median when compared with people with undetectable cotinine levels across time eras (Table 3). The difference in findings between time eras may be attributable to improved sensitivity of instruments used in cotinine detection in 2001. The limit of detection for serum cotinine was 0.05 ng/mL and changed to 0.015 mg/mL because of improvements in the method. It is also plausible that the difference is because of chance as crude and adjusted odds ratios for self-reported SHS were nonsignificant in either crude or adjusted models. The observation that the odds ratios for those with cotinine levels below the median (but not undetectable) seem to have better outcomes than those with undetectable levels of cotinine may provide some support that these are chance findings.
This study has several strengths: we used a large nationwide sample of individuals in the United States, with both survey- and laboratory-based measurements for exposure to SHS and comorbid conditions. Use of serum cotinine levels reduced recall bias intrinsic to self-reported SHS; nevertheless, one cannot completely exclude miss-reporting or laboratory errors. This study has several limitations. First, because stroke was assessed by self-report, we lacked information about stroke type, duration since stroke, stroke severity, and functional status, factors which may have had an effect on mortality. Second, NHANES only captures noninstitutionalized individuals and those who can comprehend and respond to surveys, resulting in a possible bias toward a healthier population. Third, we lacked detail on the duration and location of exposure to SHS. Serum cotinine levels could be influenced by places outside of the household, such as in the workplace, restaurants, and transportation system. Although smoke-free laws may have contributed to decline in total serum cotinine as exposure to smoking in public places have declined, they may not necessarily have direct impact on reducing household SHS.
In summary, the prevalence of exposure to SHS among never smokers with and without history of stroke declined between 1988 to 1994 and 1999 to 2012. High exposure to SHS was associated with history of previous stroke. There was a dose-dependent relationship between exposure to SHS and all-cause mortality after stroke, but this relationship was not observed in people without stroke. Although prospective studies are needed to assess causality, this study highlighted the importance of obtaining exposure to SHS history and counseling patients and their families on the potential impact of SHS on poststroke outcomes.
Sources of Funding
B. Ovbiagele is supported by award number U01 NS079179 from the National Institute of Neurological Disorders and Stroke.
A. Towfighi is supported by 1U54NS081764-01 from the National Institute of Neurological Disorders and Stroke and 11SDG7590160 from the American Heart Association.
Guest Editor for this article was Stephen M. Davis, MD, FRACP.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.116.014099/-/DC1.
- Received May 16, 2016.
- Revision received August 18, 2016.
- Accepted September 12, 2016.
- © 2016 American Heart Association, Inc.
- Kernan WN,
- Ovbiagele B,
- Black HR,
- Bravata DM,
- Chimowitz MI,
- Ezekowitz MD,
- et al
- 5.↵Office of the Surgeon General, US. The Health Consequences of Involuntary Exposure to Tobacco Smoke: A Report of the Surgeon General. Publications and Reports of the Surgeon General. Atlanta, GA: Centers for Disease Control and Prevention (US); 2006.
- Oono IP,
- Mackay DF,
- Pell JP
- 7.↵US Department of Health and Human Services. The Health Consequences of Smoking—50 Years of Progress: A Report of the Surgeon General. Reports of the Surgeon General. Atlanta, GA: Centers for Disease Control and Prevention (US); 2014.
- 8.↵National Center for Statistics. National Health and Nutrition Examination Survey (NHANES): questionnaires, datasets, and related documentation. http://www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm. Accessed April 1, 2016.
- He Y,
- Lam TH,
- Jiang B,
- Wang J,
- Sai X,
- Fan L,
- et al
- Callinan JE,
- Clarke A,
- Doherty K,
- Kelleher C
- Tan CE,
- Glantz SA
- Zhang X,
- Shu XO,
- Yang G,
- Li HL,
- Xiang YB,
- Gao YT,
- et al
- Glantz SA,
- Parmley WW
- Talukder MA,
- Johnson WM,
- Varadharaj S,
- Lian J,
- Kearns PN,
- El-Mahdy MA,
- et al