Association between smoking and the peripheral vestibular disorder: a retrospective cohort study

Common inner ear diseases include peripheral vestibular disorder (PVD) and hearing impairment. The association between smoking and peripheral vestibular disorder (PVD) is unclear. We examined associations between smoking and new PVD events. In this retrospective study, we consecutively enrolled 393 participants aged ≥20 years [mean age 65.3 years; males 133 (33.8%)] treated for hypertension, dyslipidaemia, or diabetes mellitus at a primary care clinic between November 2011 and March 2013. Participants were categorized as ever-smokers (including current and past -smokers; divided per <30 and ≥30 pack-years), and never-smokers. New PVD events were reported over a 1-year follow-up period. Hazard ratios (HR) for new onset PVD were estimated using the Cox proportional hazard regression model. Compared to never-smokers, the adjusted HR was 2.22 for ever-smokers and 2.70 for all ever-smokers with ≥30 pack-years among all 393 participants. Among male participants, compared to never-smokers, the adjusted HR was 4.41 for ever-smokers with ≥30 pack-years. A smoking history of ≥30 pack-years was strongly associated with the risk of new onset PVD in males but not, females. This study may assist patients with smoking cessation for the prevention of new PVD events among males.


Methods
This study was approved by the clinical research Institutional Review Board of Jichi Medical University (approval number: 14-01; approval date: May 22, 2014). According to the ethical guidelines for Medical and Health Research Involving Human Subjects 16 , in this research design, written informed consent is not necessarily required. Therefore, we explained the outline of our study and provided opportunities for disagreement. All participants consented to involvements in this study.
All study procedures were carried out in accordance with the STROBE statement 17 . All the examinations and medical reviews involved in this study are regularly performed, standard-of-care procedures for participants with hypertension, dyslipidaemia, or diabetes mellitus regardless of enrollment; hence, there were no additional study-related procedures or risks to the participants. We noted participants' consent in their medical records.
Study design, participants, and setting. This was a retrospective cohort study. We consecutively enrolled 349 participants who were treated for hypertension, dyslipidaemia, or diabetes mellitus, for at least 6 months at a single primary care clinic (Oki Clinic) between November 2011 and March 2013. All the participants were ≥20 years old, and were followed for at least one year; we examined all participants for signs and symptoms of PVD. We have previously described the Oki Clinic 15 .
Smoking status was categorized as current, past, and never smoker; a current smoker was defined as a person who smoked at least one cigarette daily by the time of baseline data acquisition, a past smoker was someone who quit smoking by the time of baseline data acquisition, but who had smoked previously.
Alcohol consumption status was categorized as current, former, and never drinker; a current drinker was defined as a person who drank in the 12 months prior to baseline data acquisition, a former drinker was someone who quit drinking by the time of baseline data acquisition, but drank previously. Dyslipidaemia was defined as LDL-C ≥130 mg/dL, HDL-C <40 mg/dL in men and <50 mg/dL in women, TG ≥150 mg/dL, TC ≥200 mg/ dL, or taking medication for dyslipidaemia 18 . Hypertension was defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or the use of anti-hypertensive medication 19 . Diabetes mellitus was defined as fasting plasma glucose ≥126 mg/dL, non-fasting plasma glucose ≥200 mg/dL, HbA1c (NGSP) ≥6.5%, or the use of anti-hyperglycaemic medication 20 .

Smoking status and smoking consumption (pack-years).
We classified smoking status as non-smoker and ever-smoker. We combined past-smokers and current-smokers into an ever-smoker group, and classified smoking status as never-smoker or ever-smoker for comparison, referring to previous studies 21,22 . Ever-smokers were further divided into two subgroups, using the cut-off value of 30 pack-years, based on previous studies, because a smoking consumption of over 30 pack-years is a well-known established risk factor for lung cancer, chronic obstructive pulmonary disease (COPD), and cardiovascular diseases [23][24][25][26] . We calculated pack-years as the number of cigarettes packs per day × years of smoking (one pack-year = 20 cigarettes per day for one year). PVD events (primary outcome). The primary endpoint of our study was a new PVD event. During the 1-year follow-up period, the participants underwent a medical consultation within a few days of dizziness or vertigo onset, and a board-certified otorhinolaryngologist at the Oki Clinic diagnosed and reported new PVD events according to the relevant diagnostic guidelines 27 , considering medical history and clinical evidence. Moreover, all diagnoses of PVD were performed by a sonographer (M.W.) blinded to participants smoking status. We classified PVD according to the relevant diagnostic guidelines 27 .
PVD includes VN, BPPV, and MD. We adopted strict diagnostic criteria for these diseases. VN required fulfilment of the following three criteria: (1) a sudden onset of sustained vertigo associated with unidirectional mixed horizontal-torsional spontaneous nystagmus; (2) an absence of cochlear symptom or sign (deafness and tinnitus); and (3) an absence of associated neurological symptoms or signs 28,29 . BPPV required fulfilment of the following four criteria: (1) vertigo associated with a characteristic mixed torsional and vertical nystagmus provoked by the vestibular provocative test including Dix-Hallpike test and roll test; (2) a latency (typically of 1-2 seconds) between the completion of the vestibular provocative test and the onset of vertigo and nystagmus; (3) the paroxysmal nature of the provoked vertigo and nystagmus (i.e. an increase followed by a decline over a period of 10-20 seconds); and (4) fatigue (i.e. a reduction in vertigo and nystagmus when the vestibular provocative test was repeated) 30 . MD required fulfilment of the following three criteria: (1) 2 or more definitive spontaneous episodes of vertigo lasting ≥20 minutes; (2) documented hearing loss on at least one audiological examination; and (3) tinnitus or aural fullness in the treated ear 31 .

Analysis.
We analysed all participants' data, stratified by sex, because females demonstrate higher PVD incidence than males [32][33][34] . Data are presented as means ± standard deviations (SD) and population percentages. Continuous variables were compared using t-tests or ANOVA, and categorical variables were compared using chi-square tests. Incidence rates are presented as events per 1000-person years. We calculated the incidence rate for new onset PVDs, according to smoking status (never-smoker, ever-smoker with <30 pack-years and ≥30 pack-years).
We used the Kaplan-Meier method to estimate the cumulative incidence of new PVDs. The Cox proportionalhazard regression model was used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) of new PVDs. We used never-smokers as the reference category.
Statistical analyses examined the associations between the incidence of new PVDs and smoking status, after adjusting for confounding factors (age, sex, alcohol consumption status, systolic blood pressure, HbA1c (NGSP), and LDL-C). Statistical significance was defined as p < 0.05. All analyses were performed using Stata, version 12.1 (Stata Corp, College Station, TX, USA).

Data Availability Statement.
All available data can be obtained by contacting the corresponding author.
We further divided the ever-smokers into two groups, using a cut-off value of 30 pack-years, based on previous studies. Clinical and biochemical characteristics of the 393 participants stratified by smoking status (never-smoker, ever-smoker with <30 pack-years and ≥30 pack-years) are shown in Table 1. We noted significant differences among the three groups in terms of sex, hypertension, alcohol consumption status, triglyceride levels, total cholesterol, LDL-cholesterol and HDL-cholesterol. Clinical and biochemical characteristics of male and female participants stratified by smoking status (never-smoker, ever-smoker with <30 pack-years and ≥30 pack-years) are shown in Table 2. Among male participants, significant differences in alcohol consumption status and total cholesterol levels were observed among the three groups. Among female participants, significant differences in triglyceride levels, and alcohol consumption status were observed among the three groups.
Overall, there were 76 new PVD events (61 VN, 12 BPPV, and 3 MD) during 663.8 person-years of follow-up. Among male participants, there were 219.6 person-years of follow-up, and 19 new PVD events (15 VN, 3 BPPV, and 1 MD). Among female participants, there were 444.3 person-years of follow-up, and 57 new PVD events (46 VN, 9 BPPV, and 2 MD). The incidence rate of the PVD endpoint was 86.5 (95% confidence interval: 52.8-131.6) per 1,000 person-years among male participants and 128.3 (95% CI confidence interval: 98.7-163.1) among female participants. There was no significant difference in PVD incidence between male and female participants (p = 0.070).

Discussion
Following a review of the medical literature, to the best of our knowledge, this is the first study to examine associations between new onset PVD events and smoking. There was a statistically significant association between smoking and new PVD events. Specifically, among male participants, there was a strong and significant association between a ≥30 pack-year smoking history, and PVD outcomes. The prevalence of ever-smoking (current or past smoking) was higher among male than female participants. The 2015 Japan National Health and Nutrition Survey 35 , reported that 40.4% of 60-69 year-old males, and 9.5% of 60-69 year-old females were ever-smokers. The proportion of ever-smokers and never-smokers in our study, stratified by sex, was comparable to the results of prior studies 35 .
The incidence of PVD was 86.5 (95% confidence interval: 52.8-131.6) per 1,000 person-years among male participants, and 128.3 (95% confidence interval: 98.7-163.1) per 1,000 person-years among female participants. The incidence of PVD among female participants was approximately two-fold higher than that among male participants, although this difference was not significant (p = 0.070). This finding is comparable to the results of many previous studies, showing that the PVD is more common in female [32][33][34] . Several recent studies reported an association between BPPV, a major cause of PVD, and osteoporosis 36,37 . It is well known that osteoporosis is more common in elderly females than in elderly males 38,39 . There may be a relevant connection between PVD, including BPPV, and osteoporosis; this may be reason that this gender disparity exists.
Previous studies have reported an increased risk for the following conditions in participants with a smoking consumption of over 30 pack-years: lung cancer 23 , chronic obstructive pulmonary disease 25 , myocardial infarction, 25 , and stroke 26 . In this study, we showed that ever-smokers, particularly male smokers with over 30 pack-years, have an increased risk of developing new PVD events. Previous studies found an association between sensorineural hearing impairment and smoking [5][6][7][8] . Associations between sensorineural hearing impairment and endothelial dysfunction and vasospasm have also been reported [40][41][42] . PVD and hearing impairment are major inner ear diseases. Impaired vasodilatation, due to endothelial dysfunction and vasospasm that result from smoking, may cause hearing impairment and new PVD events [40][41][42] .
Sensorineural hearing impairment is associated with endothelial dysfunction, reduced flow-mediated dilation (FMD), and reduced number of endothelial progenitor cells (EPCs) 43,44 . Endothelial dysfunction is a well-known and established primary smoking effect. Cigarette smoking induces endothelial, inflammatory, and haemostatic markers, such as elevated white blood cell counts [45][46][47] , cytokines [46][47][48] , reactive oxygen species (ROS), cyclooxygenase-2 (COX-2) 48 , and increased lipid peroxidation levels [49][50][51] . These changes may be associated with dose-related, and potentially reversible, impairment of endothelium-dependent dilation 52,53 . Arterial spasms are a reversible form of arterial dysfunction, induced by cigarette smoking. Cigarette smoking increases the risk of coronary spastic angina and acute coronary syndrome 54,55 . A previous study, using multiple regression analysis, revealed that smoking predicted coronary spastic angina (p = 0.009) 54 . Another study reported that the highest prevalence of cigarette smoking was found in participants with spastic acute coronary syndrome (p = 0.031) 55 .
Smoking-induced changes cause transient bloodstream disruption to the labyrinthine artery, a feeding artery to the inner ear, potentially leading to new PVD events. Future studies are required to determine the underlying pathological mechanisms of new PVD onset. This study has 3 notable strengths. First, the follow-up rate was 100%. This enabled accurate detection of new PVD events. Second, a board-certified otorhinolaryngologist examined participants who complained of dizziness/vertigo in our clinic, allowing accurate detection of PVD. Third, there were no missing data or confounding factors. After adjusting for assumable confounding factors, there was little statistical deflection. Therefore, we estimate that minimal patient selection bias affected the results of this study. Conversely, the present study has several limitations. First, an underreporting bias and a recall bias exist because smoking behaviour were measured by self-report (assessment) 56,57 . Second, our statistical analyses were adjusted for several known confounding factors, although we were unable to account for other, unknown, factors. Third, our sample size was small, particularly for participants with <30 pack-years. There was no statistically significant difference between male ever-smoker with <30 pack-years and the outcome. It is possible that the statistical results may have differed if sample size was larger. Forth, this study involved only one primary care clinic in Japan, and all participants were Japanese. Therefore, these results may not be generalizable to other primary care clinics in Japan or to other countries. Future studies are required to assess the external validity of our findings, by evaluating participants in multicentre and multinational studies.

Conclusion
In the present study, we revealed that smoking is associated with new PVD events. Specifically, among male participants, smoking consumption of >30 pack-years was strongly associated with an increased risk of new PVD events. This study suggests that smoking habits contribute to cancer risk, cardiovascular disease risk, and PVD (dizziness) in patients with hypertension, dyslipidaemia, and diabetes mellitus. We expect these findings may lead patients with PVD to pursue more aggressive smoking cessation.  Table 3. Cox proportional-hazard analyses for the endpoints of peripheral vestibular disorder according to smoking status (never-smoker, ever-smoker with <30 and ≥30 pack-years). * Adjusted by age, alcohol consumption status, systolic blood pressure, LDL-C, and HbA1c (NGSP) among males and females. (Adjusted by age, sex, alcohol consumption status, systolic blood pressure, LDL-C, and HbA1c (NGSP) among all 393 participants) Abbreviations: H. R., hazard ratio; CI, confidence interval; LDL-C, low-density lipoprotein cholesterol; HbA1c, glycosylated hemoglobin; NGSP, National Glycohemoglobin Standardization Program.