Abstract
Background
Results of previous studies examining associations between cigarette smoking and sleep-disordered breathing (SDB) are inconsistent. We therefore investigated this association in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL).
Methods
A total of 13,863 US Hispanic/Latino subjects, 18 to 76 years old, provided smoking histories and underwent home SDB testing. Logistic regression analyses were conducted to assess the independent association of smoking and SDB with covariate adjustment. Sex- and age-stratified analyses were performed.
Results
The weighted prevalence of moderate to severe SDB was 9.7% (95% CI, 9.0-10.5). No independent and statistically significant association was observed between ever smoking (defined as minimum lifetime cigarette use of 100) and moderate to severe SDB (defined as an apnea-hypopnea index ≥ 15 events per hour) (OR, 1.02; 95% CI, 0.85-1.22; P = .85). Sex and age were effect modifiers of the aforementioned association. Stratification according to age and sex revealed that younger (aged 35-54 years) female smokers had 83% higher odds of SDB compared with younger female never smokers (OR, 1.83; 95% CI, 1.19-2.81; P = .01). A significant dose-response relation was noted between smoking intensity and SDB in younger female smokers (P < .01). Lastly, use of ≥ 10 cigarettes per day was associated with a nearly threefold increase in SDB odds in younger female ever smokers. These associations were not observed in younger male subjects.
Conclusions
In the HCHS/SOL, no independent and statistically significant association was found between smoking and SDB. Sex and age stratification revealed a novel statistically significant association between smoking and SDB in younger (35-54 years old) female smokers. Our findings highlight the importance of investigating sex- and age-specific associations of SDB risk factors.
Key Words: Hispanic/Latino, sleep-disordered breathing, smoking
Abbreviations: AHI, apnea-hypopnea index; OLD, obstructive lung disease; SDB, sleep-disordered breathing
Sleep-disordered breathing (SDB) affects millions of American adults,1, 2, 3, 4 with more than one-quarter of the population at high risk for SDB.5 As the population ages and the obesity epidemic grows, the prevalence of SDB continues to increase.6 Obstructive SDB is characterized by repetitive upper airway collapse during sleep resulting in intermittent hypoxemia and sleep fragmentation.7 Decades of investigation into SDB have revealed associations with various risk factors, including age,6 obesity,8, 9 family history,10 anatomic abnormalities,11, 12 male sex,13 and menopause.4, 14 One association that has been less well understood is between cigarette smoking and SDB, with data suggesting both positive15, 16 and inverse17 associations. Mechanistically, smoking may increase the risk for SDB through upper airway inflammation18 and impaired neuromuscular reflexes.19 However, smoking is also associated with decreased BMI and obstructive lung disease (OLD), both of which may attenuate risk of SDB.20 The direction and magnitude of the association between smoking and SDB are therefore unclear. Furthermore, sex differences of this association have not been assessed. The latter is especially important as accumulating evidence suggests age- and sex-specific differences in SDB risk factors.21, 22, 23
The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) has robustly measured both smoking24 and SDB25 among 16,415 participants aged 18 to 76 years, 60% of whom are female. Using HCHS/SOL data, the current study investigated sex and age differences in the association between smoking and SDB.
Patients and Methods
Population
The HCHS/SOL is a large, multicenter, population-based cohort study of 16,415 self-identified US Hispanic/Latino subjects aged 18 to 76 years. Participants were recruited in four major US centers (Bronx, NY; Chicago, IL; Miami, FL; and San Diego, CA) between 2008 and 2011. Sampling occurred in two stages to achieve a population with diverse socioeconomic status and national background.26, 27 The study was approved by the institutional review boards at each participating institution, and all subjects provided written informed consent.
Participants who did not complete study protocols for objective monitoring of sleep (n = 1,180) were excluded from all analyses. Further excluded were those with < 30 min of valid sleep data (n = 766), incomplete smoking data (n = 336), or missing covariates data (n = 270), yielding an analytic sample of 13,863 participants.
Data Collection and Variable Definitions
All HCHS/SOL participants underwent in-person clinical examinations and overnight home sleep studies using an ARES Unicorder 5.2.28 Measurements of airflow, hemoglobin oxygen saturation, and movement were obtained. Sleep data were evaluated and scored at a central sleep reading center. SDB was defined according to an apnea-hypopnea index (AHI) ≥ 15 events per hour (moderate to severe SDB). Respiratory events were defined by a ≥ 50% reduction in airflow for ≥ 10 s with an associated ≥ 3% oxygen desaturation. Apneas and hypopneas were not distinguished due to the lack of thermistor data. Sleep time was identified during the sleep period by using a previously described approach.29 AHI was computed as the ratio of respiratory events to sleep time. Other parameters derived from sleep studies included baseline oxygen saturation, mean oxygen saturation, and percent time with oxygen saturation < 90%. Questionnaires provided subjective sleep duration data, and the Epworth Sleepiness Scale scores provided objective quantification of excessive daytime sleepiness30 with a score ≥ 10 suggestive of excessive daytime sleepiness.
Smoking information was collected by using questionnaires. Smoking status was defined as three groups of former, current, and never smokers (< 100 lifetime cigarettes). In addition, we defined ever smokers as a group that comprised former or current smokers (minimum lifetime cigarette use of 100). Pack-years were calculated by multiplying number of cigarettes per day with number of active smoking years. Categories for smoking intensity among daily smokers (0-9 vs ≥ 10 cigarettes per day) were determined by using a cut-point at the median and, for cumulative smoking exposure (< 20 vs ≥ 20 pack-years), using clinically relevant levels.
During the baseline examination at local field centers, information on sociodemographic and medical history variables were obtained via questionnaire, and anthropometry measurements were performed. BMI was defined as weight in kilograms divided by the square of height in meters. Alcohol use was classified as nondrinkers, moderate alcohol use (< 7 drinks per week for female subjects or < 14 drinks per week for male subjects), and high alcohol use (≥ 7 drinks per week for female subjects or ≥ 14 drinks per week for male subjects). Menopausal status and history of hay fever were self-reported. Spirometry was performed to obtain FEV1 and FVC, with OLD defined as an FEV1/FVC ratio < 0.7, as previously described.31
Statistical Methods
Analyses accounted for the complex survey design and sampling weights of HCHS/SOL and were performed by using SAS version 9.3 (SAS Institute, Inc.) and SUDAAN release 11.0.1 (RTI International). Baseline characteristics according to smoking status were computed from age- and sex-adjusted models as predicted marginal means for continuous variables from survey-weighted linear regression. Similarly, for categorical variables, we computed age- and sex-adjusted predicted marginals of the prevalence from survey-weighted logistic regression models. The age-adjusted prevalence of SDB was derived according to smoking status, intensity, and pack-years for the overall sample and according to sex.
Survey-weighted logistic regression analyses were used to calculate sex-stratified and overall ORs and 95% CIs. Never smokers were treated as the reference group for all analyses. Analyses of cigarettes per day were conducted after excluding former smokers, while creating separate exposure groups for current smokers who smoked intermittently (nondaily) and for current daily smokers who had consumption in the ranges of 1 to 10, 10 to 19, and ≥ 20 cigarettes per day. Initial models were adjusted for age, sex, BMI, waist-to-hip ratio, and field center. Additional models were adjusted for potential confounders determined a priori, including Hispanic/Latino background, place of birth, health insurance status, education status, annual household income, alcohol use, and hay fever. Menopausal status was included in female-specific models. In addition, the FEV1/FVC ratio was incorporated into the final models to evaluate OLD as a potential mediator. For pack-years and smoking intensity analyses, separate models were created to assess linear trends across multiple categories by entering ordinal values into regression models as continuous variables.
We examined effect modification of the association between smoking status and SDB according to sex, age categories (18-34, 35-54, and 55-76 years old), obesity status (BMI < 30 kg/m2 or BMI ≥ 30 kg/m2), or Hispanic/Latino background by incorporating interaction terms into multivariable models. Because heterogeneity in ORs might have been solely due to differences in outcome prevalence between subgroups, Poisson regression with robust variance estimation was used to test interactions on a multiplicative scale. For significant interaction terms, separate models were constructed for each subgroup to derive stratified estimates.
Results
Former smokers were, on average, older than current and never smokers (48.3 vs 40.8 vs 39.4 years of age; P < .01) (Table 1). There were more female subjects in the never smoker group compared with the former or current smoker groups. In addition, never smokers were more educated and had higher income than ever smokers. Lastly, former and current smokers were more likely than never smokers to have OLD.
Table 1.
Age- and Sex-Adjusted Levels of Characteristics at Baseline
| Characteristic | Smoking Status |
P Value | ||
|---|---|---|---|---|
| Never (n = 8,733) | Former (n = 2,649) | Current (n = 2,481) | ||
| Demographic | ||||
| Age, mean, y | 39.4 (38.9-39.9) | 48.3 (47.3-49.3) | 40.8 (39.9-41.7) | < .0001 |
| Male, % | 39.3 (37.9-40.7) | 63.3 (60.4-66.1) | 61.0 (58.0-64.0) | < .0001 |
| Hispanic/Latino background | < .0001 | |||
| Dominican | 11.7 (10.1-13.5) | 7.1 (5.4-9.2) | 5.4 (3.8-7.6) | |
| Central American | 8.5 (7.3-9.8) | 6.5 (5.2-8.2) | 4.6 (3.6-5.8) | |
| Cuban | 17.4 (14.4-20.8) | 16.8 (13.6-20.5) | 25.5 (21.3-30.3) | |
| Mexican | 41.1 (37.7-44.6) | 44.8 (40.2-49.5) | 33.2 (29.0-37.6) | |
| Puerto Rican | 12.9 (11.5-14.5) | 13.7 (11.4-16.4) | 24.0 (20.8-27.5) | |
| South American | 5.2 (4.5-6.0) | 6.1 (4.9-7.5) | 3.1 (2.4-4.0) | |
| Other/> 1 | 3.3 (2.7-4.1) | 5.1 (3.6-7.2) | 4.3 (3.3-5.7) | |
| Field center | < .0001 | |||
| Bronx, NY | 28.2 (25.3-31.2) | 23.1 (19.8-26.9) | 28.3 (24.5-32.5) | |
| Chicago, IL | 17.5 (15.3-19.9) | 16.5 (14.1-19.3) | 15.3 (13.0-17.9) | |
| Miami, FL | 26.2 (22.3-30.4) | 27.3 (23.0-32.0) | 33.3 (28.2-39.0) | |
| San Diego, CA | 28.2 (24.7-32.0) | 33.0 (28.1-38.4) | 23.1 (19.4-27.3) | |
| Born in US 50 states | 19.9 (18.2-21.7) | 20.6 (17.8-23.8) | 28.9 (25.7-32.3) | < .0001 |
| Educational attainment | < .0001 | |||
| < High school | 17.8 (16.5-19.1) | 16.4 (14.8-18.2) | 16.2 (14.3-18.3) | |
| Some high school | 12.8 (11.7-13.9) | 17.6 (15.2-20.3) | 20.0 (17.5-22.7) | |
| HS graduate/equivalent | 28.4 (27.0-29.9) | 26.3 (23.6-29.2) | 29.9 (27.3-32.8) | |
| > High school | 41.1 (39.1-43.0) | 39.7 (36.3-43.3) | 33.9 (31.1-36.9) | |
| Annual household income | < .0001 | |||
| Less than $10,000 | 12.0 (10.8-13.2) | 14.4 (12.4-16.7) | 20.5 (18.2-23.0) | |
| $10,001-$20,000 | 30.5 (28.7-32.4) | 29.3 (26.4-32.5) | 34.3 (31.1-37.7) | |
| $20,001-$40,000 | 35.2 (33.4-36.9) | 35.3 (32.1-38.7) | 29.4 (26.8-32.1) | |
| $40,001-$75,000 | 15.4 (14.0-16.8) | 15.0 (12.7-17.6) | 12.3 (10.0-15.1) | |
| More than $75,000 | 7.0 (5.5-8.8) | 6.0 (4.2-8.4) | 3.5 (2.5-5.0) | |
| Has health insurance, % | 50.7 (48.4-53.0) | 49.2 (46.2-52.1) | 48.3 (45.1-51.5) | .3772 |
| Clinical characteristics | ||||
| BMI, mean, kg/m2 | 29.2 (29.0-29.4) | 30.1 (29.7-30.6) | 29.0 (28.6-29.5) | < .0001 |
| Obese, % | 38.1 (36.5-39.8) | 45.3 (41.9-48.7) | 37.7 (34.8-40.7) | .0001 |
| Waist-to-hip ratio, mean | 0.91 (0.91-0.92) | 0.93 (0.92-0.93) | 0.92 (0.92-0.92) | < .0001 |
| Pack-years ≥ 20, % | NA | 14.7 (13.1-16.4) | 23.6 (21.4-25.9) | < .0001 |
| Obstructive lung disease | 6.8 (5.9-7.7) | 12.5 (10.3-15.1) | 15.3 (13.4-17.5) | < .0001 |
| Menopausal, % (among female subjects) | 39.0 (37.3-40.7) | 41.3 (38.5-44.2) | 39.4 (36.9-42.0) | .0632 |
| Sleep-related variables | ||||
| Average sleep duration (self-report), mean | 8.0 (8.0-8.1) | 8.0 (7.9-8.1) | 8.0 (7.9-8.1) | .7695 |
| Epworth Sleepiness Scale score, mean | 5.5 (5.4-5.7) | 5.9 (5.6-6.2) | 5.7 (5.4-5.9) | .0539 |
| Epworth Sleepiness Scale score ≥ 10, % | 17.7 (16.6-18.9) | 20.8 (18.4-23.4) | 17.9 (15.9-20.2) | .0892 |
| Baseline Spo2, mean | 97.1 (97.0-97.1) | 97.0 (96.9-97.1) | 97.1 (97.0-97.1) | .6089 |
| Mean Spo2 | 96.6 (96.5-96.6) | 96.4 (96.2-96.6) | 96.5 (96.4-96.5) | .0053 |
| Percent time Spo2 < 90, mean | 0.7 (0.6-0.8) | 1.8 (0.4-3.3) | 0.7 (0.6-0.8) | .1222 |
| Apnea-hypopnea index (3% desaturation), mean | 5.3 (4.9-5.7) | 6.5 (5.7-7.2) | 5.5 (4.9-6.1) | .0214 |
Values are least squared means (95% CIs) derived from survey linear regression for continuous variables or predicted marginals of the prevalence (95% CIs) derived from survey logistic regression for classification variables, adjusted to the age and sex distribution of the target population (47.6% male; 16.2% aged 18-24 years, 21.9% aged 25-34 years, 21.5% aged 35-44 years, 19.3% aged 45-54 years, 12.8% aged 55-64 years, and 8.3% aged ≥ 65 years). P values were derived from Wald F-statistics and test for any difference across groups. NA = not applicable; Spo2 = arterial oxygen saturation by pulse oximetry.
Former smokers had higher mean AHI (6.5 events per hour) compared with never and current smokers (5.3 and 5.5, respectively). The mean Epworth Sleepiness Scale score was 5.6 (95% CI, 5.5-5.7), with no significant differences across smoking groups (P > .05). Mean sleep duration (8 h) and average nocturnal oxygen saturation (96%-97%) were also similar across smoking groups.
The overall weighted prevalence of SDB in the study sample was 9.7% (95% CI, 9.0-10.5), representing 1,543 participants, and was higher in male subjects (13.4% prevalence; 95% CI, 12.3-14.7) than in female subjects (6.4% prevalence; 95% CI, 5.6-7.2). The age- and sex-adjusted prevalence of SDB according to smoking status was 9.5%, 10.8%, and 9.1%, in never, former, and current smokers, respectively (e-Table 1).
Multivariable adjusted logistic regression models revealed no statistically significant association between smoking and SDB (OR, 1.02; 95% CI, 0.85-1.22; P = .85). Similarly, there was no statistically significant association between smoking intensity or pack-years and SDB (e-Table 2).
Sex was a significant (defined as a P value < .1) effect modifier of the aforementioned association (P = .07). Sex-stratified analysis revealed that female ever smokers had 30% higher odds of SDB compared with female never smokers (OR, 1.30; 95% CI, 0.97-1.73; P = .07) (Table 2). After adjustment for FEV1/FVC ratio, female ever smokers had 41% higher odds of SDB (OR, 1.41; 95% CI, 1.04-1.90; P = .03) compared with female never smokers. This latter association seems to be driven by former smokers, as seen by the separate ORs for former (OR, 1.45; 95% CI, 1.05-2.00; P = .02) and current (OR, 1.35; 95% CI, 0.86-2.10; P = .19) smokers compared with never smokers. Furthermore, current female smokers who used ≥ 10 cigarettes per day had 80% higher odds of SDB than never smokers (OR, 1.80; 95% CI, 1.05-3.09; P = .03). This finding is suggestive that in either category (ie, former or current smokers), a threshold of smoking must be reached before an association between smoking and SDB becomes evident.
Table 2.
Multivariable Analysis of Moderate to Severe Sleep-Disordered Breathing (AHI3 ≥ 15 Events per Hour) Across Smoking Categories Among US Hispanic/Latino Subjects Grouped According to Sex
| Variable | Among Female Subjects (n = 8,387) |
Among Male Subjects (n = 5,476) |
||||||
|---|---|---|---|---|---|---|---|---|
| Model 1a |
Model 2b |
Model 1a |
Model 2b |
|||||
| OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | |
| Smoking status | ||||||||
| Never smoked | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | ||||
| Former smoker | 1.30 (0.95-1.77) | .10 | 1.45 (1.05-2.00) | .02 | 0.85 (0.65-1.11) | .24 | 0.91 (0.70-1.17) | .46 |
| Current smoker | 1.30 (0.85-1.99) | .23 | 1.35 (0.86-2.10) | .19 | 1.01 (0.76-1.32) | .97 | 1.08 (0.82-1.43) | .57 |
| Former/current combined (Ref = never smoked) | 1.30 (0.97-1.73) | .07 | 1.41 (1.04-1.90) | .03 | 0.91 (0.74-1.13) | .41 | 0.98 (0.79-1.21) | .85 |
| Pack-years (Ref = never smoked) | ||||||||
| < 20 pack-years | 1.36 (0.98-1.89) | .07 | 1.41 (1.00-1.99) | .05 | 1.01 (0.80-1.29) | .91 | 1.05 (0.82-1.34) | .70 |
| ≥ 20 pack-years | 1.16 (0.76-1.78) | .48 | 1.40 (0.91-2.16) | .13 | 0.73 (0.52-1.00) | .05 | 0.82 (0.59-1.13) | .22 |
| P-for-trend | .1400 | .0312 | .1035 | .3812 | ||||
| Smoking intensity (among current smokers; Ref = never smokers) | ||||||||
| Nondaily smoker | 1.27 (0.59-2.73) | .53 | 1.30 (0.60-2.82) | .51 | 1.30 (0.86-1.97) | .21 | 1.35 (0.89-2.06) | .16 |
| 0-9 cigarettes/d | 1.11 (0.60-2.03) | .74 | 1.08 (0.58-2.01) | .80 | 0.88 (0.53-1.44) | .60 | 0.88 (0.53-1.46) | .61 |
| ≥ 10 cigarettes/d | 1.80 (1.05-3.09) | .03 | 1.95 (1.10-3.45) | 0.02 | 0.93 (0.59-1.45) | .74 | 0.99 (0.62-1.57) | .96 |
| P-for-trendc | .3742 | .2878 | .5109 | .4853 | ||||
Model 1 adjusted for age, sex, BMI, waist-to-hip ratio, field center, Hispanic/Latino national background, place of birth (within vs outside US mainland), health insurance status, educational attainment, annual household income, alcohol use level, and hay fever.
Model 2 adjusted for variables in model 1, plus the FEV1/FVC ratio.
P-for-trend calculation for smoking intensity performed while restricting the analysis to current smokers.
In addition to sex, age was a significant effect modifier (P = .02). We therefore conducted stratified analyses according to both age and sex (Table 3). Among younger female subjects (aged 35-54 years) who were ever smokers, a higher risk of SDB was noted compared with younger female never smokers (OR, 1.83; 95% CI, 1.19-2.81; P = .01). Furthermore, smoking ≥ 10 cigarettes per day conferred an almost three times higher odds of SDB among younger female smokers vs younger female never smokers (OR, 2.72; 95% CI, 1.45-5.12; P < .01). Taken together, these findings support a dose-response relation between smoking and SDB among younger female subjects, with P values for linear trend for pack-years and smoking intensity of .01 and < .01, respectively.
Table 3.
Multivariable Analysis of Moderate-to-Severe Sleep Disordered Breathing (AHI3 ≥ 15 Events per Hour) Across Smoking Categories Among US Hispanics/Latino Subjects According to Age and Sex
| Variable | Female Subjects Aged 18-34 Years |
Female Subjects Aged 35-54 Years |
Female Subjects Aged 55-76 Years |
Male Subjects Aged 18-34 Years |
Male Subjects Aged 35-54 Years |
Male Subjects Aged 55-76 Years |
||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (n = 1,683) |
(n = 4,173) |
(n = 2,531) |
(n = 1,423) |
(n = 2,548) |
(n = 1,505) |
|||||||
| OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | OR (95% CI) | P Value | |
| Smoking status | ||||||||||||
| Never smoked | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | ||||||
| Former smoker | < 5 cases | NA | 2.11 (1.22-3.66) | .008 | 1.10 (0.77-1.59) | .595 | 1.55 (0.59-4.03) | .370 | 1.18 (0.82-1.70) | .375 | 0.54 (0.36-0.79) | .002 |
| Current smoker | < 5 cases | NA | 1.53 (0.94-2.51) | .088 | 1.42 (0.84-2.39) | .194 | 1.26 (0.52-3.02) | .610 | 1.28 (0.88-1.86) | .204 | 0.50 (0.30-0.82) | .006 |
| Former/current combined (Ref = never smoked) | < 5 cases | NA | 1.83 (1.19-2.81) | .006 | 1.20 (0.85-1.69) | .292 | 1.37 (0.64-2.92) | .414 | 1.23 (0.92-1.64) | .170 | 0.52 (0.37-0.74) | < .001 |
| Pack-years (Ref = never smoked) | ||||||||||||
| < 20 pack-years | < 5 cases | NA | 1.84 (1.15-2.95) | .011 | 1.41 (0.94-2.11) | .093 | 1.39 (0.64-3.01) | .399 | 1.22 (0.88-1.70) | .228 | 0.64 (0.44-0.95) | .027 |
| ≥ 20 pack-years | < 5 cases | NA | 1.80 (1.00-3.24) | .052 | 0.90 (0.53-1.53) | .689 | < 5 cases | NA | 1.23 (0.80-1.88) | .347 | 0.42 (0.27-0.64) | < .001 |
| P-for-trend | NA | .0053 | .739 | .3992 | .1994 | .0001 | ||||||
| Smoking intensity (among current smokers; Ref = never smoked) | ||||||||||||
| Nondaily smoker | < 5 cases | NA | 0.62 (0.26-1.46) | .271 | 2.40 (0.96-6.01) | .062 | 1.38 (0.49-3.84) | .540 | 1.45 (0.84-2.50) | .183 | 0.91 (0.41-2.06) | .828 |
| 0-9 cigarettes/d | < 5 cases | NA | 1.45 (0.61-3.45) | .401 | 1.08 (0.56-2.10) | .814 | 0.92 (0.21-4.06) | .913 | 1.19 (0.63-2.28) | .591 | 0.44 (0.20-0.97) | .043 |
| ≥ 10 cigarettes/d | < 5 cases | NA | 2.72 (1.45-5.12) | .002 | 1.47 (0.64-3.36) | .366 | 1.44 (0.39-5.30) | 0.584 | 1.16 (0.65-2.06) | .617 | 0.43 (0.21-0.88) | .021 |
| P-for-trenda | NA | .004 | .3136 | .9684 | .5454 | .1349 | ||||||
Models adjusted for age within subgroups, as well as BMI, waist-to-hip ratio, field center, Hispanic/Latino national background, place of birth (within vs outside US mainland), health insurance status, educational attainment, annual household income, alcohol use level, hay fever, and menopausal status. See Table 1 legend for expansion of abbreviation.
P-for-trend calculation for smoking intensity performed while restricting the analysis to current smokers.
We conducted additional exploratory analyses using an AHI threshold of ≥ 30 events per hour. In these analyses, we continued to observe no statistically significant association between ever smoking and SDB (OR, 1.12; 95% CI, 0.85-1.48; P = .42). Fully adjusted analyses stratified according to sex and age continue to reveal an increased odds of SDB (OR, 1.48; 95% CI, 0.81-2.72; P = .20) in younger female ever smokers (aged 35-54 years) compared with younger female never smokers; however, the effect estimate is attenuated compared with our original reported findings (OR, 1.83; 95% CI, 1.19-2.81; P = .01). This is likely due to loss of power with a 66% reduction in our new sample size when using an AHI cutoff of ≥ 30 events per hour (n = 35 compared with n = 103 in our original analysis). Moreover, using this new cutoff of AHI ≥ 30 events per hour, we noted a statistically significant association in older female ever smokers (aged 55-76 years) (OR, 2.31; 95% CI, 1.38-3.86; P = .01) compared with older female never smokers. This scenario was not observed in our original analysis with a cutoff of AHI ≥ 15 events per hour (OR, 1.20; 95% CI, 0.85-1.69; P = .29). We also repeated the analysis using a more stringent cutoff of ≥ 4 h of analyzable sleep data. This approach did not change our main conclusion; that is, there was no statistically significant association between smoking and SDB in the overall cohort (OR, 0.98; 95% CI, 0.80-1.19; P = .82). In our sex- and age-stratified analysis restricting the study sample to only those with ≥ 4 h of analyzable sleep data, we continued to observe a statistically significant independent association between smoking and SDB among younger female ever smokers compared with younger female never smokers, with a strengthening of the effect estimate (OR, 2.01; 95% CI, 1.32-3.08; P = .01) compared with our previous analysis (OR, 1.83; 95% CI, 1.19-2.81; P = .01). Due to the exploratory nature of these analyses, we do not provide the raw data.
A similar relation between smoking and SDB was not seen in younger male subjects in the 35- to 54-year-old group. In fact, older ever smoker male subjects (aged 55-76 years) had lower odds of having SDB compared with never smokers (OR, 0.52; 95% CI, 0.37-0.74; P < .01). This inverse association between smoking and SDB persisted among male subjects in the 55- to 76-year-old age category when assessing pack-years and smoking intensity. A similar inverse association was not observed between smoking and SDB in older female subjects aged 55 to 76 years (Table 3). Among older male subjects, the association was attenuated (although remained statistically significant) after adjustment for the FEV1/FVC ratio (e-Table 3).
BMI, waist-to-hip ratio, and Hispanic/Latino background were not significant effect modifiers of the association of interest. In addition, results did not differ when excluding those who reported a previous diagnosis of sleep apnea (data not shown).
Discussion
Cross-sectional analysis of a large US Hispanic/Latino cohort did not confirm an independent association between smoking and SDB. However, sex- and age-stratified analysis revealed a novel association between smoking and SDB among female ever smokers who were aged 35 to 54 years. According to our findings, these younger female subjects who smoked at least 100 cigarettes in their lifetime had 83% higher odds of SDB compared with younger female subjects who never smoked. Moreover, US Hispanic/Latino female subjects in this age group who smoked ≥ 10 cigarettes per day had a nearly threefold increase in the odds of SDB compared with younger female subjects who never smoked. This association was not observed in male smokers.
Our finding of a strong association between smoking and SDB among younger female subjects is novel and, to the best of our knowledge, has not been reported by others. This finding may be largely due to lack of adequate number of female subjects across the age spectrum in other studies of SDB. We propose several potential explanations for our finding. First, it is possible that compared with male subjects and postmenopausal female subjects, premenopausal female subjects have fewer SDB risk factors, thereby allowing smoking to emerge as a statistically significant risk factor for SDB. Evidence suggests that menopause is a very strong risk factor for SDB4, 14 and that male sex alone confers a twofold to threefold increase in the risk of SDB.13 Therefore, in the presence of menopause and male sex, smoking’s influence may be less robust, which is consistent with our data. Another mechanism for the link between smoking and SDB may be increased upper airway inflammation.18 This condition may be magnified in female subjects compared with male subjects, as female subjects have shorter upper airway length32 and therefore are more prone to irritant injury such as smoking. This theory needs to be confirmed in future studies that quantify upper airway inflammation in male subjects and female subjects with SDB and smoking history.
There are many other potential mechanisms for an association between smoking and SDB. Previous studies have found that progesterone and estrogen are protective against SDB.33, 34 Derangements in these sex hormones caused by smoking may represent one such mechanism. Previous studies have also revealed smokers to have worse sleep efficiency,35 an increased density of lighter stages of sleep,36 and a higher arousal index37 compared with nonsmokers. These factors can further result in SDB via increased loop gain. Taken together, the aforementioned additional mechanisms may contribute to increased odds of SDB among younger female smokers. In addition, because previous studies have shown an association between SDB and OLD, we controlled for the FEV1/FVC ratio in the final models. As predicted, after adjustment for OLD, even higher odds of SDB were observed among younger female smokers. These findings are in agreement with evidence supporting lower odds of SDB in patients with OLD.20 The mechanism and sex differences of this relation between SDB and OLD among smokers will need to be the subject of future investigation.
Our findings contrast with several previous studies of the general US population that have shown an independent association between smoking and SDB. Wetter et al15 examined the association between smoking and SDB. Their Wisconsin-based cohort of predominantly white individuals reported a more than fourfold increase in the odds of SDB in current smokers compared with never smokers; furthermore, the highest intensity smokers (≥ 40 cigarettes per day) had a 40-fold increase in the odds of SDB compared with never smokers. These results were affirmed in a smaller study by Kashyap et al,16 which found an almost threefold increase in the odds of SDB (defined as AHI > 10 events per hour) in current smokers vs nonsmokers. Both the aforementioned studies were limited by lack of formal assessment of interaction according to age and sex as performed in the current study. Furthermore, these studies may not be generalizable to the current study population of US Hispanic/Latino subjects. To the best of our knowledge, only one prior study has explored the association between SDB and smoking in this population specifically. A New Mexico referral-based cohort of 1,222 Hispanic/Latino adults found an association between self-reported snoring and cigarette smoking.38 However, no objective testing for SDB was conducted in this study.
The current study has several limitations. The overall prevalence of smoking in the HCHS/SOL population was low, which may have resulted in our null findings in the overall cohort. Furthermore, the home sleep studies were conducted by using a standardized protocol with a recommended Level 3 device and scored by a research staff blinded to other data. This home sleep study data may underestimate severity of SDB, affecting the prevalence of disease and thereby reducing the likelihood of finding a significant association. Nonetheless, the current study had a large sample size, and therefore we do not believe lack of power contributed to our overall null finding. Furthermore, the cross-sectional study design limited our ability to comment on the natural history of SDB progression, and a causal relation between smoking and SDB cannot be inferred. In addition, the large number of statistical tests and exploratory subgroup analyses conducted raises the possibility of chance findings. However, evidence for a dose-response relation between increasing smoking intensity and SDB in younger female subjects supports biological plausibility and therefore may suggest a causal association. Lastly, the HCHS/SOL did not include hormonal data or biomarkers to provide insight into mechanistic pathways that might explain the sex-specific effect of smoking on SDB in younger female subjects.
This study cohort represents, to the best of our knowledge, the largest population-based study to examine the relation between smoking and SDB. It is also one of the few studies to examine this relation in US Hispanic/Latino subjects. Due to the large sample size and collection of multiple risk factor data, we were able to adequately control for potential confounders and account for mediators such as OLD. We also used this large cohort recruited across the age spectrum to examine sex and age as potential modifiers of the association between smoking and SDB. Although our study did not find an overall statistically significant association between SDB and smoking, differing trends among female and male subjects in our primary analysis prompted further exploration into sex and age modifications. This exploration uncovered hypothesis-generating evidence that the association between smoking and SDB is stronger among younger female subjects compared with male subjects or postmenopausal female subjects. Additional strengths of this study are that much of the original sampled population were included in our analyses as there was robust participation in providing sleep and smoking data.
Conclusions
Across a large sample of US Hispanic/Latino subjects, we found no statistically significant association between smoking and SDB. However, sex and age stratification revealed a novel association between smoking and SDB in younger (aged 35-54 years) female subjects, such that younger female smokers had statistically significantly higher odds of moderate to severe SDB compared with younger female never smokers. Because our sex-specific findings were generated in subgroup analyses, future research should confirm our observed associations in younger female subjects. Future investigations into age- and sex-specific phenotypes of SDB are needed for targeted and tailored SDB screening and treatment.
Acknowledgments
Author contributions: As corresponding author, N. A. S. assumes full responsibility for the integrity of the submission as a whole from inception to publication. O. C. was the primary author responsible for the writing of the manuscript. G. M. S. had full access to all the study data and takes responsibility for the integrity and accuracy of the data and analysis. All authors contributed substantially to the study design, data interpretation, and drafting or revising of the manuscript.
Financial/nonfinancial disclosures: The authors have reported to CHEST the following: P. C. Z. serves on the scientific advisory board and as a consultant for Jazz Pharmaceuticals, Philips/Respironics, and Eisai; and has stock ownership in Teva Pharmaceutical. S. R. reports that her institution received grant support from Jazz Pharmaceuticals and personal consulting fees from Jazz Pharmaceuticals unrelated to the current article. None declared (O. C., G. M. S., A. R. R., K. J. R., V. M., D. M. R., R. C. K., N. A. S.).
Role of sponsors: The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript.
Other contributions: The authors thank the staff and participants of HCHS/SOL for their important contributions. The HCHS/SOL investigators’ Website is http://www.cscc.unc.edu/hchs/.
Additional information: The e-Tables can be found in the Supplemental Materials section of the online article.
Footnotes
FUNDING/SUPPORT: The Hispanic Community Health Study/Study of Latinos (HCHS/SOL) is a collaborative study supported by contracts from the National Heart, Lung, and Blood Institute to the University of North Carolina [HHSN268201300001I/N01-HC-65233], University of Miami [HHSN268201300004I/N01-HC-65234], Albert Einstein College of Medicine [HHSN268201300002I/N01-HC-65235], University of Illinois at Chicago [HHSN268201300003I/N01-HC-65236 Northwestern University], and San Diego State University [HHSN268201300005I/N01-HC-65237]. The following institutes/centers/offices have contributed to the HCHS/SOL through a transfer of funds to the National Heart, Lung, and Blood Institute: National Institute on Minority Health and Health Disparities, National Institute on Deafness and Other Communication Disorders, National Institute of Dental and Craniofacial Research, National Institute of Diabetes and Digestive and Kidney Diseases, National Institute of Neurological Disorders and Stroke, and the Office of Dietary Supplements of the National Institutes of Health. N. A. S. received funding from the National Institute of Health/National Heart, Lung, and Blood Institute during the time of our study [5K23HL125923-03, 1R03HL140273-01, and 1R01HL143221-01].
Supplementary Data
References
- 1.Berry D.T., Webb W.B., Block A.J., Switzer D.A. Sleep-disordered breathing and its concomitants in a subclinical population. Sleep. 1986;9(4):478–483. doi: 10.1093/sleep/9.4.478. [DOI] [PubMed] [Google Scholar]
- 2.Young T., Palta M., Dempsey J., Skatrud J., Weber S., Badr S. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328(17):1230–1235. doi: 10.1056/NEJM199304293281704. [DOI] [PubMed] [Google Scholar]
- 3.Bixler E.O., Vgontzas A.N., Ten Have T., Tyson K., Kales A. Effects of age on sleep apnea in men: I. Prevalence and severity. Am J Respir Crit Care Med. 1998;157(1):144–148. doi: 10.1164/ajrccm.157.1.9706079. [DOI] [PubMed] [Google Scholar]
- 4.Young T., Peppard P.E., Gottlieb D.J. Epidemiology of obstructive sleep apnea: a population health perspective. Am J Respir Crit Care Med. 2002;165(9):1217–1239. doi: 10.1164/rccm.2109080. [DOI] [PubMed] [Google Scholar]
- 5.Hiestand D.M., Britz P., Goldman M., Phillips B. Prevalence of symptoms and risk of sleep apnea in the US population: results from the National Sleep Foundation Sleep in America 2005 Poll. Chest. 2006;130(3):780–786. doi: 10.1378/chest.130.3.780. [DOI] [PubMed] [Google Scholar]
- 6.Peppard P.E., Young T., Barnet J.H., Palta M., Hagen E.W., Hla K.M. Increased prevalence of sleep-disordered breathing in adults. Am J Epidemiol. 2013;177(9):1006–1014. doi: 10.1093/aje/kws342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Hudgel D.W. Mechanisms of obstructive sleep apnea. Chest. 1992;101(2):541–549. doi: 10.1378/chest.101.2.541. [DOI] [PubMed] [Google Scholar]
- 8.Peppard P.E., Young T., Palta M., Dempsey J., Skatrud J. Longitudinal study of moderate weight change and sleep-disordered breathing. JAMA. 2000;284(23):3015–3021. doi: 10.1001/jama.284.23.3015. [DOI] [PubMed] [Google Scholar]
- 9.Newman A.B., Foster G., Givelber R., Nieto F.J., Redline S., Young T. Progression and regression of sleep-disordered breathing with changes in weight. Arch Intern Med. 2005;165(20):2408. doi: 10.1001/archinte.165.20.2408. [DOI] [PubMed] [Google Scholar]
- 10.Redline S., Tishler P.V. The genetics of sleep apnea. Sleep Med Rev. 2000;4(6):583–602. doi: 10.1053/smrv.2000.0120. [DOI] [PubMed] [Google Scholar]
- 11.Schwab R.J., Gupta K.B., Gefter W.B., Metzger L.J., Hoffman E.A., Pack A.I. Upper airway and soft tissue anatomy in normal subjects and patients with sleep-disordered breathing: significance of the lateral pharyngeal walls. Am J Respir Crit Care Med. 1995;152(5 I):1673–1689. doi: 10.1164/ajrccm.152.5.7582313. [DOI] [PubMed] [Google Scholar]
- 12.Dempsey J.A., Skatrud J.B., Jacques A.J. Anatomic determinants of sleep-disordered breathing across the spectrum of clinical and nonclinical male subjects. Chest. 2002;122(3):840–851. doi: 10.1378/chest.122.3.840. [DOI] [PubMed] [Google Scholar]
- 13.Strohl K.P., Redline S. Recognition of obstructive sleep apnea. Am J Respir Crit Care Med. 1996;154(2):279–289. doi: 10.1164/ajrccm.154.2.8756795. [DOI] [PubMed] [Google Scholar]
- 14.Mirer A.G., Young T., Palta M., Benca R.M., Rasmuson A., Peppard P.E. Sleep-disordered breathing and the menopausal transition among participants in the Sleep in Midlife Women Study. Menopause. 2017;24(2):157–162. doi: 10.1097/GME.0000000000000744. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wetter D.W., Young T.B., Bidwell T.R., Badr M.S., Palta M. Smoking as a risk factor for sleep-disordered breathing. Arch Intern Med. 1994;154(19):2219–2224. [PubMed] [Google Scholar]
- 16.Kashyap R., Hock L.M., Bowman T.J. Higher prevalence of smoking in patients diagnosed as having obstructive sleep apnea. Sleep Breath. 2001;5(4):167–172. doi: 10.1007/s11325-001-0167-5. [DOI] [PubMed] [Google Scholar]
- 17.Newman A., Nieto F., Guidry U. Relation of sleep-disordered breathing to cardiovascular disease risk factors: the Sleep Heart Health Study. Am J Epidemiol. 2001;154(1):50–59. doi: 10.1093/aje/154.1.50. [DOI] [PubMed] [Google Scholar]
- 18.Kim K.S., Kim J.H., Park S.Y. Smoking induces oropharyngeal narrowing and increases the severity of obstructive sleep apnea syndrome. J Clin Sleep Med. 2012;8(4):367–374. doi: 10.5664/jcsm.2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Haxhiu M.A., Deal E.C., Norcia M.P., Van Lunteren E., Mitra J., Cherniack N.S. Medullary effects of nicotine and GABA on tracheal smooth muscle tone. Respir Physiol. 1986;64(3):351–363. doi: 10.1016/0034-5687(86)90128-3. [DOI] [PubMed] [Google Scholar]
- 20.Zhao Y.Y., Blackwell T., Ensrud K.E., Stone K.L., Omachi T.A., Redline S. Sleep apnea and obstructive airway disease in older men: outcomes of Sleep Disorders in Older Men Study. Sleep. 2016;39(7):1343–1351. doi: 10.5665/sleep.5960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Redline S., Kump K., Tishler P.V., Browner I., Ferrette V. Gender differences in sleep disordered breathing in a community-based sample. Am J Respir Crit Care Med. 1994;149(3 pt 1):722–726. doi: 10.1164/ajrccm.149.3.8118642. [DOI] [PubMed] [Google Scholar]
- 22.Bixler E.O., Vgontzas A.N., Lin H.M. Prevalence of sleep-disordered breathing in women: effects of gender. Am J Respir Crit Care Med. 2001;163(3 I):608–613. doi: 10.1164/ajrccm.163.3.9911064. [DOI] [PubMed] [Google Scholar]
- 23.Quintana-Gallego E., Carmona-Bernal C., Capote F. Gender differences in obstructive sleep apnea syndrome: a clinical study of 1166 patients. Respir Med. 2004;98(10):984–989. doi: 10.1016/j.rmed.2004.03.002. [DOI] [PubMed] [Google Scholar]
- 24.Kaplan R.C., Bangdiwala S.I., Barnhart J.M. Smoking among US Hispanic/Latino adults. Am J Prev Med. 2014;46(5):496–506. doi: 10.1016/j.amepre.2014.01.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Redline S., Sotres-Alvarez D., Loredo J. Sleep-disordered breathing in Hispanic/Latino individuals of diverse backgrounds. The Hispanic Community Health Study/Study of Latinos. Am J Respir Crit Care Med. 2014;189(3):335–344. doi: 10.1164/rccm.201309-1735OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Sorlie P.D., Avilé S-Santa L.M., Wassertheil-Smoller S. Design and implementation of the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20:629–641. doi: 10.1016/j.annepidem.2010.03.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lavange L.M., Kalsbeek W.D., Sorlie P.D. Sample design and cohort selection in the Hispanic Community Health Study/Study of Latinos. Ann Epidemiol. 2010;20:642–649. doi: 10.1016/j.annepidem.2010.05.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Ayappa I., Norman R.G., Seelall V., Rapoport D.M. Validation of a self-applied unattended monitor for sleep disordered breathing. J Clin Sleep Med. 2008;4(1):26–37. [PMC free article] [PubMed] [Google Scholar]
- 29.Zhao Y.Y., Weng J., Mobley D.R. Effect of manual editing of total recording time: implications for home sleep apnea testing. J Clin Sleep Med. 2017;13(01):121–126. doi: 10.5664/jcsm.6404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Johns M.W. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep. 1991;14(6):540–545. doi: 10.1093/sleep/14.6.540. [DOI] [PubMed] [Google Scholar]
- 31.Barr R.G., Avilés-Santa L., Davis S.M. Pulmonary disease and age at immigration among Hispanics. Results from the Hispanic Community Health Study/Study of Latinos. Am J Respir Crit Care Med. 2016;193(4):386–395. doi: 10.1164/rccm.201506-1211OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ronen O., Malhotra A., Pillar G. Influence of gender and age on upper-airway length during development. Pediatrics. 2007;120(4):e1028–e1034. doi: 10.1542/peds.2006-3433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Pickett C.K., Regensteiner J.G., Woodard W.D., Hagerman D.D., Weil J.V., Moore L.G. Progestin and estrogen reduce sleep-disordered breathing in postmenopausal women. J Appl Physiol. 1989;66(4):1656–1661. doi: 10.1152/jappl.1989.66.4.1656. [DOI] [PubMed] [Google Scholar]
- 34.Popovic R.M., White D.P. Upper airway muscle activity in normal women: influence of hormonal status. J Appl Physiol. 1998;84(3):1055–1062. doi: 10.1152/jappl.1998.84.3.1055. [DOI] [PubMed] [Google Scholar]
- 35.Wetter D.W., Young T.B. The relation between cigarette smoking and sleep disturbance. Prev Med (Baltim) 1994;23(3):328–334. doi: 10.1006/pmed.1994.1046. [DOI] [PubMed] [Google Scholar]
- 36.Zhang L., Samet J., Caffo B., Punjabi N.M. Cigarette smoking and nocturnal sleep architecture. Am J Epidemiol. 2006;164(6):529–537. doi: 10.1093/aje/kwj231. [DOI] [PubMed] [Google Scholar]
- 37.Conway S.G., Roizenblatt S.S., Palombini L. Effect of smoking habits on sleep. Brazilian J Med Biol Res. 2008;41(8):722–727. doi: 10.1590/s0100-879x2008000800014. [DOI] [PubMed] [Google Scholar]
- 38.Schmidt-Nowara W.W., Coultas D.B., Wiggins C., Skipper B.E., Samet J.M. Snoring in a Hispanic-American population. Risk factors and association with hypertension and other morbidity. Arch Intern Med. 1990;150(3):597–601. doi: 10.1001/archinte.150.3.597. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.