Obstructive sleep apnea (OSA) is a common sleep-disordered breathing (SDB) affecting as much as 6%−13% of the common population (1). It is caused by repeated upper airway collapse during sleep, resulting in intermittent hypoxia, and manifests as sleep fragmentation, snoring, and daytime sleepiness. Current common OSA therapies, e.g., positive airway pressure and surgery (2), could relieve sleep apnea symptoms. However, these therapies also come with poor compliance (3) and financial burdens, compromising the quality of life for OSA patients. Lifestyle modifications (4), e.g., weight loss and smoking cessation, have been recommended for OSA patients. However, there is still a gap in offering scientific, evidence-based health advice for which kind of lifestyle is beneficial for the prevention and clinical management of OSA.

Previous clinical research has linked several common lifestyles to OSA pathogenesis. Smoking has been linked to a direct contribution to upper airway inflammation, a factor known to heighten the susceptibility to episodes of sleep apnea (5). Alcohol relaxes muscles, including the muscles in the throat, which can lead to increased airway collapse during sleep (6, 7). This relaxation contributes to the development or worsening of OSA. While stimulant drinks are often associated with wakefulness, they can also disrupt sleep patterns and exacerbate OSA symptoms. Multiple observational studies have described a positive association between cigarette smoking and an increased risk of OSA. On the other hand, alcohol consumption was reported as an independent risk factor for OSA (8), while OSA may increase the incidence of alcohol use disorder in turn (9). Furthermore, smoking, alcohol consumption, and OSA are linked to cardiovascular impairments, several other sleep traits, and morbidity (10–13). Common stimulant drinks such as coffee and tea (14) were also linked to OSA risks, while there is only a limited number of observational studies. Furthermore, leisurely sedentary behaviors such as TV watching, computer use, driving, etc., have been linked to health concerns (15). OSA has been widely reported to negatively affect driving performance through cognitive impairment (16).

Most of the previous observational studies are prone to biases from reverse causation and residual confounding. There is also a possible inverse association that OSA patients may be predisposed to smoking to cope with daytime sleepiness (17) or mental health comorbidities (18). Meanwhile, there are practical difficulties in conducting randomized controlled trials (RCTs) for any of these common lifestyle risk factors, or more specifically taking behavioral details into consideration, e.g., smoking initiation or cessation, and frequency and duration of drinking. For example, chronic smoking over an extended period is more likely to increase the frequency and severity of OSA episodes (19). Adding more to it, heterogeneity in current clinical evidence, selective bias, and potential inverse association may cloud the understanding of the causal relationship between these common lifestyles and OSA. Mendelian randomization (MR) estimates the potential causal relationship between exposure and outcome by using genetic variants as instrumental variables (IV) for the exposure factor of interest. As genetic variants are assigned at random during conception, the MR approach is not susceptible to confounding factors. It can also avoid reverse causality bias, with genetic variants being assigned prior to disease development. Similar methods were applied to study the associations among sleep traits, inflammation protein levels, and glycemic traits (20–22).

Lifestyle factors, such as smoking, alcohol, stimulant drinks, and sedentary behaviors, are implicated in obstructive sleep apnea (OSA) pathogenesis. Utilizing an MR can overcome biases in observational studies, offering a more robust understanding of their causal relationships. In this study, we aimed to evaluate a possible causal association between common modifiable lifestyles (i.e., smoking, alcohol, coffee and tea consumption, TV watching, computer use, and driving) and OSA via an MR approach. Our findings will offer valuable insights for understanding OSA pathogenesis and clinical management of OSA.

This study comprehensively evaluated the casual association between common modifiable lifestyles (including smoking, alcohol, coffee, tea consumption, TV watching, computer use, and driving) and OSA via an MR approach. Our results showed that ever regularly smoking showed a robust causal role in increasing OSA risks. Frequent alcohol intake, lifetime smoking index, and coffee intake are risk factors for OSA. Although the clinical association between these lifestyles and OSA has been widely reported by observational studies (8, 9, 32), our research confirmed the causal role of these lifestyles in OSA risk. Hopefully, our results will provide some insights for evidence-based precaution and clinical care for OSA.

First, we found that both ever-regular smoking and the lifetime smoking index play causal roles in increasing the risk of OSA. Although the clinical association between cigarette smoking and OSA has been widely reported by observational studies, our research first confirmed the causal role of smoking in OSA pathogenesis. Although this causal relationship is still not well understood, there are certain theories. One possible explanation for smoking's effect on sleep apnea is nicotine's impairment of upper airway neuromuscular protective reflexes (33). Additionally, smoking could change sleep structure into a higher proportion of light sleep (namely, the N1 and N2 periods) and correspondingly a lower proportion of slow-wave sleep (the N3 period), while sleep apnea episodes are also more likely to occur in the N1 period. Meanwhile, smoking could also contribute to a hyperarousal state (34), which could also increase the risk for sleep apnea episodes. Aside from enhancing OSA itself, smoking could also increase the chances of common complications, especially cardiovascular diseases and hypertension (35).

However, neither smoking cessation nor cigarettes per day showed any effects on OSA risk. This is inconsistent with previous clinical research reporting that increased pack-years could increase OSAS severity (32). Although smoking cessation should theoretically enhance sleep health by withdrawing from this harmful lifestyle, both our results and several previous clinical studies reported otherwise. In a large epidemiologic study, former smokers were found not to have a significantly increased prevalence of sleep-disordered breathing, as compared with non-smokers (36). To our knowledge, there have been no previous clinical trials on the impact of smoking cessation on OSA patients. Nicotine is a highly addictive substance, while during smoking cessation, increased insomnia and irritability, even mimicking OSA symptoms, were observed, especially in the first 1–2 days of smoking cessation (33). However, after fighting through this tough acute withdrawal phase, smoking cessation should still benefit sleep health in the long term. One previous clinical study observed better sleep quality in former smokers than in current smokers (37).

In addition, we observed a probable causal association between frequent alcohol intake and increased OSA risks. This finding is consistent with clinical research reporting that alcohol consumption may increase OSA risks (8, 9) and severity (38). We also offer some possible explanations for this causal role of frequent alcohol use in increasing OSA. Alcohol has a depressant effect on respiratory centers in the central nervous system, which controls the tone of pharyngeal muscles (39). Therefore, alcohol may increase the susceptibility to pharyngeal closure and subsequent upper airway obstruction during sleep. Additionally, there may be a potential dose–effect relationship with moderate drinking decreases the risk of OSA (40), whereas problematic drinking increases the OSA risk. For instance, the Mediterranean diet encourages moderate use of wine, which has been widely recognized to protect against risks of OSA and cardiovascular diseases (41, 42).

For common stimulant drinks in daily life (coffee, tea), we found that coffee intake has a probable causal role in OSA risk. For leisure behaviors, although we observed no robust causal impact on OSA, computer use showed a trend for increasing OSA risk after outlier removal (OR = 1.34, p = 0.055). Only a few clinical studies have reported that coffee and tea may increase OSA risks, while there has not been a previous MR analysis focusing on their causal association. Our findings, while highlighting the causal role of coffee consumption on OSA risks, also provide practical lifestyle advice for OSA patients. Daytime sleepiness or daytime function impairments are common symptoms of OSA (43), while coffee, as the most common stimulant drink (44), may seem to be the perfect fix. However, considering that coffee may also increase OSA risk, it may be better in other alternative ways (i.e., moderate outdoor exercise) to cope with daytime symptoms of OSA.

The definition of ever smoked regularly, age at initiation of regular smoking, cigarettes per day, and the lifetime smoking index encompasses various dimensions of smoking behavior. However, nuances in self-reported smoking history and potential recall biases should be acknowledged. Similarly, for alcohol consumption, the categorization into ever regularly drinking or not simplifies a complex behavior, potentially oversimplifying the spectrum of alcohol consumption patterns. Careful consideration of these definitional nuances is crucial to interpreting the associations between smoking, alcohol, and OSA accurately. Regarding the duration of measurement, our observations on smoking, alcohol consumption, and lifestyle span multiple years; however, the dynamic nature of individual behaviors might not be fully captured over extended periods. Despite our efforts in mitigating confounding factor BMI, the possibility of unaccounted confounders remains. Therefore, further studies should consider the dose effect and develop more specific health advice for drinking alcohol in OSA patients.

Additionally, in our study, there is none of the sample overlap between the exposure and outcome datasets, as they are derived from distinct cohorts. The genetic variant–exposure associations are independently evaluated in each dataset, ensuring that the assessments are specific to their respective sources. Furthermore, the exposure and outcome samples are sourced from different cohorts, emphasizing the robustness of our study design in mitigating confounding factors and bolstering the credibility of the observed associations.

Our study still has certain limitations. First, the definition of OSA from the FinnGen database in our study relies on ICD codes. In addition, it is acknowledged that without individual screening or a sleep study in the control group, subtle variations in OSA status could exist. In addition, while efforts were made to comprehensively define exact behavior, the retrospective nature of data collection may introduce recall bias, impacting the accuracy of self-reported information. Additionally, there is considerable heterogeneity in this study, which may reduce the statistical significance of our findings. Meanwhile, we extensively explored reverse causality in various facets of our analysis and observed OSA exhibited a weak causal impact on increased coffee intake [beta = 0.09, 95% CI = (4.98 × 10⁻³, 0.17), p = 0.04] and alcohol intake frequency [outliers removed beta = 0.08, 95% CI = (0.01, 0.14), p = 0.02]. The potential for bias in our study demands careful consideration, with attention to both direction and magnitude. Last but not least, the GWAS used in this study mainly included populations from Europe, which may make our findings less applicable in other regions. In addition, it is essential to acknowledge that the validity of MR assumptions, including relevance, no confounding, and no pleiotropy. While we have endeavored to adhere to established protocols and methodologies, the potential for deviations in MR assumptions remains an intrinsic challenge. These limitations collectively emphasize the need for a careful interpretation of our findings and underscore the importance of ongoing methodological advancements in MR research.

In conclusion, ever regularly smoking showed a robust causal role in OSA risks. Frequent alcohol intake, lifetime smoking index, and coffee intake are probable causalities for OSA. These lifestyles should be discouraged as precautions from developing OSA. More advanced studies are still needed to further authenticate these associations.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.

KL: Data curation, Methodology, Writing – original draft, Formal analysis, Resources, Visualization. CZ: Data curation, Formal analysis, Methodology, Resources, Writing – original draft, Conceptualization. JW: Writing – review & editing, Data curation, Methodology, Software. JL: Conceptualization, Data curation, Methodology, Software, Writing – original draft, Writing – review & editing. ZC: Writing – review & editing, Data curation, Methodology, Software. MH: Writing – review & editing, Data curation, Methodology, Software. BL: Writing – review & editing, Data curation, Methodology, Software. XS: Writing – review & editing, Data curation, Methodology, Software. YZ: Writing – original draft, Data curation, Methodology, Software. MY: Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Software, Supervision, Writing – original draft, Writing – review & editing.