Introduction

Obstructive sleep apnea (OSA) is characterized by the repeated cessation of breathing during sleep due to obstruction or collapse of the upper airway. When this symptom occurs, the oxygen saturation of the blood decreases, and sleep is interrupted to breathe. This is also known as the “apnea event.” OSA occurs in 14% of men and 5% of women in the general adult population1. The prevalence of OSA increases with obesity1, and the five-year incidence in middle-aged adults is 7–11%2. Nevertheless, only 1 in 50 patients with symptoms suggesting OSA syndrome is diagnosed and treated3. OSA is a common comorbidity in patients with lung diseases, and recognizing its prevalence and clinical significance may improve patients’ quality of life4. Particularly, the appropriate and timely circadian rhythm treatment of OSA leads to increased positive outcomes for patients with sepsis, chronic obstructive pulmonary disease, and cancer5. The best standard to diagnose OSA is polysomnography. However, it is challenging for use in epidemiological studies because it requires a lot of money and time. Therefore, there was a lot of interest in OSA’s screening tools around the world. OSA’s screening tools include the Berlin questionnaire6 and STOP-Bang7.

The risks of respiratory diseases in smokers are known worldwide8,9. A study that investigated the health outcomes of ex-smokers observed that the risk of lung cancer was reduced10. Other studies revealed that smoking cessation for a prolonged duration was associated with a decreased tendency to have cancer and obstructive spirometry pattern11. These respiratory diseases are associated with sleep apnea, and there is an association between smoking and sleep disorders, including sleep patterns, sleep maintenance, and daytime sleepiness12.

According to a cohort study, smokers have reduced sleep quality, less restorative sleep, and take longer to fall asleep than non-smokers13,14,15. Sleep disorders can affect well-being and mood during waking hours and frequently occur among smokers. A study has detected a decrease in sleep quality and changes in sleep patterns in young, healthy smokers16. Some of these changes are based on daily tobacco consumption and serum nicotine levels, and this supports the effect of smoking on sleep patterns16.

Previous OSA studies have observed associations with asthma, metabolic syndrome, insulin resistance, hypothyroidism, and depression17,18,19,20,21,22. Furthermore, based on previous studies, a biological association between obstructive sleep apnea and smoking can be found. Smoking is widely known to negatively affect respiratory function, causing inflammation and swelling of the upper airway23. This phenomenon can also lead to obstruction and difficulty breathing during sleep, resulting in symptoms such as snoring, panting, and interrupted sleep. We also assume that smoking reduces muscle tension in people and causes structural changes in the upper airway by accumulating pharyngeal fat, which can contribute to the severity of OSA symptoms24. Finally, it increases oxidative stress and inflammation throughout the body, leading to sleep disorders and the development of an arousal threshold that makes it easier to wake up25.

There is biological plausibility of disorders between smoking and OSA; however, a significant relationship has not been conclusively established within more definitive physiologic studies26. Therefore, this study aimed to investigate the association between smoking and OSA using the STOP-Bang index in a representative population of Korean adults27.

Methods

Data

Data used in this study were obtained from the Korea National Health and Nutrition Examination Survey28, a cross-sectional survey. This survey is conducted annually by the Korea Disease Control and Prevention Agency (KDCA). It uses a stratified, multi-stage cluster sampling design based on age, gender, and geographic area28. The KNAHNES is reliable for Korean health policies and programs, and it is publicly accessible data. This study did not require ethical approval as the KNHANES complies with the Helsinki Declaration29.

Study population

In 2020, the total number of respondents was 7359. Individuals aged between 1 and 18 years were excluded due to the lack of smoking information (N = 1,820). Additionally, respondents under age 40 were excluded because they were not measured using the STOP-Bang index (N = 1804). Participants with missing data were also excluded (N = 293). Finally, a sample of 3442 participants (1465 males and 1977 females) was analyzed in this study (Fig. 1).

Figure 1
figure 1

Flowchart of study participants displaying the inclusion and exclusion.

Full size image

Variables

A dependent variable was defined by the STOP-Bang score, which reflects OSA risk using an 8-item questionnaire7,30. Polysomnography (PSG), a time and money-intensive complex procedure, is performed to diagnose OSA. Therefore, the STOP-Bang indicator is generally used27,31. The STOP-Bang questionnaire has high sensitivity and is a valid screening tool for OSA worldwide32,33. A relatively easy screening tool, STOP-Bang, has been validated with higher sensitivity than the Berlin questionnaire in South Korean patients34. KNHANES, which represents Koreans, introduced newly a STOP-Bang as a simple and user-friendly screening tool in 2019, and many research using it is being actively conducted. STOP-Bang is an acronym for the first letter of each symptom or physical attribute often associated with OSA, S, T, O, P, and BANG. The questionnaire included sleep-related symptoms, blood pressure, body mass index (BMI), neck circumference, age, and sex. Each option was rated as 0 (“No”) or 1(“Yes”), resulting in a maximum score of 8 points. In addition, sleep-related symptoms (snoring, tiredness, and observed apnea) were self-reported: (1) Is your snoring louder than the conversation or loud enough to be heard in the next room? (2) Do you often feel tired or sleepy during the day? and (3) Has anyone seen you stop breathing when you sleep? Blood pressure was measured using a non-mercury automatic sphygmomanometer (Microlife WatchBP Office AFIB), an internationally certified device. BMI was calculated as weight (kg) divided by height squared (m2). Neck circumference was measured using a Lufkin W606PM device below the thyroid cartilage (Adam’s apple). The risk of OSA was based on a global cutoff32,33, with a low risk of OSA if the score was ≤ 2 and a high risk if it was ≥ 3. In additional subgroup analysis, a cutoff was used by dividing the degree of OSA risk using the STOP-Bang score into three categories, with a mild risk if the score was from 0 to 2, moderate risk if the score was from 3 to 4, and severe risk if the score was from 5 to 8. Also, a high STOP-Bang score means a high risk of OSA, which does not confirm a full OSA diagnosis.

The main independent variable was “smoking,” which was classified into three groups: (1) non-smokers, (2) ex-smokers who had been using conventional or e-cigarettes previously, and (3) current smokers who use conventional or e-cigarettes. This classification was similar to that of previous studies investigating smoking behavior using the same tools10.

The covariates included demographic factors, socioeconomic factors, health conditions, and health behaviors of the participants. Demographic factors included age (40–49 years/50–59 years/60–69 years/ ≥ 70 years), marital status (married/single and widow/divorced and separated), and educational level (middle school or below/high school/college or over). Socioeconomic factors included household income (low/mid-low/mid-high/high), region (urban/rural), and occupational categories (white/pink/blue/inoccupation). Health conditions and health behaviors included high-risk drinking (no-drinker/low-risk drinker/high-risk drinker), physical activity (active/inactive), BMI (underweight and normal/overweight/obesity of stage 1/obesity of stage 2 and 3), status of hypertension (normal/warning/pre-/stage 1/stage 2), status of diabetes (normal/pre-/diabetes), allergic rhinitis history (yes/no), and life disturbance due to rhinitis (yes/no).

Statistical analysis

All estimates were calculated using sample weight procedures, clusters, and strata assigned to the study participants. In addition, we analyzed each variable stratified by sex. Descriptive analysis was performed to examine the general characteristics of the study population. To assess the association between smoking and OSA (low risk/ high risk), we used multi-logistic regression analysis. Furthermore, a multinomial regression analysis stratified by dependent variables as OSA (mild/moderate/severe). Subgroup analyses stratified by interesting variable were performed to investigate the combined effects of OSA and indicators of severity of smoking through the smoking cessation status and pack-years among ex- and current smokers. In addition, we analyzed the association between moderate risk and severe risk compared to the mild risk of OSA, respectively (supplementary 1 and 2). The results were reported as odds ratios (ORs) and 95% confidence intervals (CIs). SAS version 9.4 (SAS Institute Inc.; Cary, NC, USA) was used for all the analyses, and a p-value of 0.05 was considered significant.

Results

Table 1 presents the general characteristics of the study population. Of the 3442 participants, 1465 were males (42.6%) and 1977 were females (57.4%). Among the males, 435 (29.7%) were current smokers, 725 (49.5%) were ex-smokers, and 305 (20.8%) were non-smokers. Among the females, 61 (3.1%) were current smokers, 86 (4.4%) were ex-smokers, and 1830 (92.6%) were non-smokers. The relationship between smoking and OSA was significant in males (p-value = 0.0270).

Table 1 General characteristics of the study population.
Full size table

Table 2 presents the association between smoking and OSA in males and females after adjusting for all covariates. Among the males, the ex-smokers (OR: 1.53, 95% CI 1.01–2.32) and current smokers (OR: 1.79, 95% CI 1.10–2.89) had a significantly higher association with OSA than the non-smokers. Among the females, the ex-smokers and current smokers exhibited an increasing trend of ORs for a high risk of OSA, although the association with OSA was not significant.

Table 2 Results of factors associated between smoking and obstructive sleep apnea.
Full size table

Table 3 presents the results of a multinomial regression analysis of the three groups in the STOP-Bang index based on sex. It shows the results of a stratified analysis of the association between smoking and the degree of OSA risk. The degree of OSA risk was based on a global cutoff using the STOP-Bang score32,33: mild risk (0–2), moderate risk (3–4), and severe risk (5–8). Considering the mild risk of OSA as the reference category, the ORs of smoking status were linearly higher in ex-smokers and current smokers. In males, ex-smokers (OR: 1.61, 95% CI 1.05–2.48) were significantly associated with a moderate risk of OSA, and current smokers (OR: 1.88, 95% CI 1.07–3.29) were significantly associated with a severe risk of OSA.

Table 3 Multinomial regression of three-groups in STOP-BANG index.
Full size table

Figure 2 presents the results of the subgroup analysis of changes in ORs based on smoking cessation status (SCS) and pack-years (the number of cigarettes smoked and the smoking period). The ORs increased linearly as the SCS and pack-years increased in males. Specifically, ex-smokers with < 15 years of smoking cessation (OR: 1.69, 95% CI 1.01–2.84), current smokers (OR: 1.80, 95% CI 1.11–2.91), and over 20 pack-years (OR: 1.88, 95% CI 1.15–3.05) were more likely to have high ORs for OSA compared to non-smokers.

Figure 2
figure 2

Results of subgroup analysis stratified by the smoking cessation status and pack-year.

Full size image

Discussion

In this study, we observed that OSA risk was higher in ex-smokers and current smokers than in non-smokers in males. In addition, the risk had a positive dose-dependent relationship with pack-years and negative with the duration of smoking cessation. According to the STOP-Bang, there was a significant association between a higher OSA risk and ex-smokers and the highest OSA risk and current smokers. In other words, current smokers quitting will contribute to a reduced OSA risk. However, we did not observe a significant association between smoking and OSA due to the few female smokers. Nevertheless, the increasing ORs based on smoking behavior were similar in men. This can be attributed to the low awareness of smoking among Korean women, regarded as a recall bias in self-reported data35,36,37,38. This underreporting of women’s smoking is also linked to the critical view of society. That’s because the social stigma of women smokers tends to hide and cover up smoking more than men39.

According to previous studies, the association between smoking and OSA has many conflicting opinions, although many research groups have not studied it. A study observed that patients with OSA had more current smokers than those without OSA40. This finding suggests that smoking may be an independent risk factor for OSA40. Also, it is an independent risk factor for intermittent hypoxemia, and smokers with OSA have been associated with the most severe endothelial function impairment41. It occurs more frequently in children exposed to secondary smoking, and adults smokers42,43, and these children exhibit more concentration challenges, fatigue, irritability, and hyperactivity due to OSA42. Conversely, a study analyzed medical records and polysomnography results and observed no significant correlation between smoking and OSA44,45,46. Additionally, a relationship between smoking and OSA is plausible; however, there is no conclusive evidence suggesting that untreated OSA is associated with smoking addiction26. However, categorizing the medical records and polysomnography results revealed that smokers with a low BMI developed OSA, and heavy smokers had a light sleep stage and a high OSA risk44. The most likely explanation for results that differ from this study is the measurement methods of OSA and study population and year.

Several mechanisms have been proposed to explain the association between smoking and OSA. First, we described some biological mechanisms. Smoking alters the uveal mucosa in patients with OSA, resulting in thickening and edema through calcitonin gene-related peptide (CGRP)-induced neurogenic inflammation47. Nasal obstruction due to chronic mucosal inflammation associated with smoking, such as impaired ciliary function, mucosal edema, cell proliferation, and thickened epithelium, is also considered a potential mechanism48,49,50. This supports our findings that prolonged smoking increases the risk of moderate or severe OSA. Mild risk of OSA are common, and typically untreated51. Moderate and severe risk of OSA have been well established to increase risk in the presence of incident cardiometabolic comorbidities such as heart failure, ischemic heart disease, hypertension, diabetes, and so on51. If people with moderate or severe risk of OSA are not treated, the risk of death from cardiovascular and all causes increases significantly52,53,54, so they should always be prevented and monitored.

Second, other potential associations between smoking and OSA have been suggested as nicotine-induced impairment from neuromuscular protection reflexes in the upper airway55. Finally, smoking can lead to OSA due to its respiratory effects, nicotine withdrawal, and the stimulant effects of nicotine. According to previous studies, smokers are more likely to experience daytime sleepiness56,57 and respiratory sleep disorders than non-smokers58.

On the other hand, Gothe et al.59 demonstrated that chewing nicotine gum before sleep reduces the frequency of obstructive apnea episodes during the first two hours of sleep. This suggests that nicotine can reduce sleep breathing disorders59. However, during the night, as nicotine blood levels decreases and upper airway resistance increases, the apnea–hypopnea index (AHI) increases with nicotine withdrawal symptoms or respiratory effects associated with smoking59.

This study had certain limitations. First, it was a cross-sectional study. It may be an inverse causal relationship; therefore, caution should be exercised in the interpretation. Future studies are required to clarify the relationship between smoking and OSA, and longitudinal studies are required to establish a causal relationship. Second, KNHANES data were self-reported. Data on smoking status, the STOP-Bang index, and health-related and socioeconomic variables might be reliable but not accurately measured. In addition, the STOP-Bang index indicates OSA risk, not its prevalence. It may result in recall bias, which has been underestimated in smoking. However, we minimized this bias by using a globally reliable STOP-Bang index. Third, the available results regarding the causes of the STOP-Bang index were only available for 2020, and the KNHANES data for other years could not be obtained. If STOP-Bang data continue to emerge in future KNHANES, studies using a larger study population would be required. Fourth, STOP-Bang index has not been investigated for people under the age of 40, so our participants are adults over the age of 40. This is the limitation of the data, and future studies need to be conducted for all age groups. The results of this study cannot be generalized to young people at OSA risk. Fifth, the STOP-Bang index is not an objective measurement method to define OSA. However, it is mostly used as a screening tool in place of polysomnography. Therefore, further research using the test result data for polysomnography is needed. Finally, we could not identify the type and smoking behaviors, such as conventional cigarette smoking, e-cigarette smoking, or both. Furthermore, the pack-years of liquid e-cigarettes could not be calculated because the KNHANES did not contain this information.

Despite these limitations, our study had notable strengths. First, this study was based on the KNHANES, a nationally representative data collected by the KDCA. It is conducted using random cluster sampling, which is reliable and representative. Therefore, the study results can be generalized to adults over the age of 40 in Korea. Second, to the best of our knowledge, this is the first study to investigate the association between smoking and OSA using the STOP-Bang index, an appropriate tool for measuring OSA in the general population. Our research suggests appropriate interventions to improve sleep quality in smokers at risk of OSA based on the STOP-Bang index should be developed and applied.

Conclusion

In conclusion, our study suggests that smoking may be associated with OSA and may further affect sleep quality. Given these results, current smokers are at risk of OSA; however, it is also related to smoking cessation and pack-years, which are calculated by the amount and duration of smoking. These findings provide direction for future studies on smoking disadvantages and OSA and educate patients with OSA who are unaware of it. Knowing our health status and quitting smoking are the best ways to avoid life-threatening diseases.