Blood pressure (BP) has been well documented to be associated with hearing loss previously. However, the role of blood pressure variability (BPV, representing BP fluctuation over a time period) on hearing remains unknown. We aimed to evaluate the relationship between BPV and hearing in Chinese population. We included 8646 male subjects from a population-based study (the Kailuan study). BP was measured every two years at routine physical examinations from 2006 to 2015. Based on five annual BP measurements, BPV was estimated by standard deviation of BP (SD), coefficient of the variation of BP (CV), and variation independent of mean of BP (VIM). Hearing was estimated by pure-tone average threshold (PTA) at low, intermediate, and high frequencies in the year of 2014. Regression models were used to evaluate the relationship between BPV and hearing. The results showed that PTAs and percentages of hearing loss at low, intermediate, and high frequencies grew gradually with increasing systolic SD (SSD) (p<0.05). After adjusting for multiple covariates, multivariate regression analyses demonstrated that variations of SBP (SSD, SCV, and
Hearing loss ranks as the fifth leading cause of years lived with disability, affecting 360 million people worldwide [
Hypertension is a major global health burden and a leading risk factor for cardiovascular diseases and premature death [
Clinical and experimental studies have demonstrated that arterial hypertension was an independent risk factor for the hearing loss. Patients with higher BP had worse pure-tone thresholds [
This study was performed based on the Kailuan Community in Tangshan. Physical examinations were conducted every two years on both in-service and retired workers of Kailuan Community. Eleven hospitals participated in the physical examination. A total of five physical examinations were performed during 2006-2007, 2008-2009, 2010-2011, 2012-2013, and 2014-2015, respectively. The measurement of PTAs was performed in 2014-2016.
The inclusion criteria were as follows: giving signed informed consent to participate in the current study, providing complete information from at least three of the five physical examinations, and providing complete information of PTA measurements. The exclusion criteria were as follows: history of stroke, history of head injury, history of myocardial infarction, history of atrial fibrillation, missing BP data from more than two of the five physical examinations, missing data of the measurements of PTA, and female individuals (female individuals were ruled out because the sample size of female was too small compared to male subjects and the unbalanced gender distribution may result in gender bias).
The questionnaire was completed by individuals and then verified by research doctors. The questionnaire items were consisted of demographic information, occupation situation (the questionnaire items about occupation situation were consisted of type of occupation and nature of work (mental work or physical work), employing conditions (serving or retired), labor intensity (extremely light, light, intermediate, and heavy), occupational hazard (e.g., noise, high temperature, and microwave, harmful chemicals, and dust exposure)), lifestyle (e.g., cigarette smoking, exercise, and diet), disease history and family history, and physical examination profiles (e.g., blood pressure, height, weight, waist circumference, etc.). Smoking was defined as ≥1 cigarette/day, continuous smoking ≥1 year, or giving up smoking ≤1 year. Alcohol consumption is defined as continuous drinking≥1 year (alcohol content >50%, amount >100ml). Physical exercise is defined as aerobic exercise (such as walking, jogging, ball games, and swimming) ≥3 times/weeks and ≥ 30min/times. Occupational noise exposure is defined as working places equivalent environmental noise level ≥85 decibel (dB) at least 8 hours per day.
Standard protocols were used for all of the measurements as described earlier by our group [
BP was measured between 7:00 and 9:00. Individuals were asked to refrain from smoking and drinking tea or coffee for more than 30 min and to sit and rest for 15 min prior to measurement. During BP measurement, individuals sat with their arms and feet flat and their upper arms at the height of their heart. Right brachial artery BP was measured by a corrected mercury sphygmomanometer with an appropriate sized cuff. Systolic blood pressure (SBP) was recorded on hearing the phase I Korotkoff sound. Diastolic blood pressure (DBP) was recorded on hearing the phase V Korotkoff sound. Sitting BP was measured two times first with a 30s interval. If two measurements differed by <5 mmHg, BP was recorded as the mean of the two measurements. If two measurements differed by >5 mmHg, BP was remeasured and the final BP in each examination was calculated as the mean of three measurements. Hypertension in each examination was defined as SBP≥140 mmHg and/or DBP≥90 mmHg or BP<140 mmHg and DBP<90 mmHg with regular antihypertensive drugs usage. Individuals were regarded as having hypertension if they were recorded as having hypertension in at least two examinations.
BPV was calculated by three methods. (1) Standard deviation (SD) of the BP levels is obtained from the physical examinations. The SD of SBP was recorded as SSD, and the SD of DBP was recorded as DSD. (2) The coefficient of the variation of BP (CV) was calculated as SD/mean of BP levels obtained from the physical examinations
BPV levels in the current study were calculated based on the BP levels measured every two years. Several previous studies have demonstrated that the two-year BP measurement interval can sufficiently reflect the BP fluctuation over time [
Trained professional staff performed audiometric testing in a sound-isolating room using the Otometrics MADSEN Xeta audiometer (GN Group Co., Ltd., Ballerup, Denmark). Air-conduction hearing thresholds were measured for each ear using pure tone at six frequencies (0.5, 1, 2, 3, 4, and 6 kHz). 1KHz was used as the first pure-tone frequency from an intensity of -20 dB. If no response was observed, 5dB was added each time until response is observed. Then pure-tone frequencies at 0.5, 2, 3, 4, and 6 KHz were measured. PTAs were measured at low, intermediate, and high frequencies, respectively. PTA of low frequency was calculated by the mean of PTA at 0.5 and 1 KHz. PTA of intermediate frequency was calculated by the mean of PTA at 1 and 2 KHz. PTA of high frequency was calculated by the mean of PTA at 3, 4, and 6 KHz [
The testing should begin at relatively low frequencies ranging from 0.5 to 1 KHz, because this frequency is easily heard by most patients and has the greatest test-retest reliability. After that, the hearing test is performed at frequencies ranging from 1 to 2 KHz, which represent the intermediate frequencies of speech range. Then, the hearing test is performed at high frequencies ranging from 3 to 6 KHz. In clinical settings, many factors, such as sound injury, ototoxic drugs, and senile auditory system degeneration, may firstly affect the function of basal gyrus of cochlea. Consequently, early manifestation is the change of high-frequency hearing threshold [
The design of the current study is cross-sectional rather than a cohort one and the aim of this study is to reflect the distribution of BPV at the specific time-point (the year of 2014) and uncover the correlation between BPV and hearing ability. Therefore, we only focus on the relationship between BPV and hearing ability instead of their causal relationship. Consequently, only one hearing measurement is sufficient to identify the correlation between BPV and hearing ability.
Data were entered in the terminal of each hospital and then uploaded to the computer room of the Kailuan General Hospital for storage in an Oracle 10.2g database. SPSS 13.0 statistical software was utilized for statistical analysis. Normally distributed measurement data were recorded as mean±SD. Trend test was used to compare differences of multiple groups. If the variance is homogeneous, the LSD test is used. If the variance is not homogeneous, Dunnett’s T3 test is used. Categorical variables were described as percentages and compared by the chi-square test. Multivariate linear regression analysis was used to investigate the impacts of BPV on PTAs and hearing loss. The collinearity was analyzed using variance inflation factor (VIF). Multivariate logistic regression analysis was used to analyze the effect of each SD increase in different BPV measurements on hearing loss. Sensitivity analyses were performed by removing individuals with occupational noise exposure and individuals with hypertension, respectively. P<0.05 (bilateral) was considered as statistically significant.
Among the 101510 workers who participated in the 2006-2007 health examination, a total number of 8875 subjects participated in at least three physical examinations and had complete pure-tone threshold measurement data. In the 8875 individuals, 229 were excluded for female gender (n=138), the history of head injury (n=36), the history of stroke (n=24), myocardial infarction, or history of atrial fibrillation (n=31). As a result, a total of 8646 participants were included in the final statistical analysis.
As the Kailuan Group Corporation is a highly industrialized enterprise, the vast majority of the employees of the Kailuan Group Corporation are men (more than 80%). Moreover, the measurement of PTAs was performed on employees who work in coal mines. Since female employees rarely work in mines, the number of women is significantly less than that of men. Consequently, we excluded the 138 female subjects from further analyses. See Figure
A flow chart of the current study.
The 8646 participants were divided into four groups according to the quartiles of their SSD levels: (1) quartile 1 (n=2165): SSD<6.38; (2) quartile 2 (n=2156): 6.38≤SSD<9.07; (3) quartile 3 (n=2166): 9.07≤SSD<12.31; and (4) quartile 4 (n=2159): SSD≥12.31. Table
Clinical characteristics of participants in different BPV groups.
Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | p for trend | |
---|---|---|---|---|---|
(n=2165) | (n=2156) | (n=2166) | (n=2159) | ||
Age, year | 45.2±8.6 | 46.6±8. | 47.0±8. | 47.9±8. | <0.001 |
| |||||
SBP (mmHg) | 126±11 | 129±1 | 131±1 | 137±2 | <0.001 |
| |||||
DBP (mmHg) | 81±9 | 82± | 83±1 | 86±1 | <0.001 |
| |||||
Mean of times (BP) | 4.3±0.8 | 4.5±0. | 4.5±0. | 4.4±0. | <0.001 |
| |||||
SSD (mmHg) | 5±1 | 8± | 11± | 16± | <0.001 |
| |||||
DSD (mmHg) | 5±3 | 6± | 7± | 9± | <0.001 |
| |||||
SCV | 3.8±1.0 | 6.3±0. | 8.5±0. | 12.6±2. | <0.001 |
| |||||
DCV | 6.6±3.1 | 7.2±3. | 8.2±3. | 10.7±4. | <0.001 |
| |||||
| 4.9±1.5 | 8.0±1. | 10.8±1. | 15.7±3. | <0.001 |
| |||||
| 5.5±2.7 | 6.0±2. | 6.9±2. | 8.9±3. | <0.001 |
| |||||
BMI (kg/m2) | 25.2±3.2 | 25.3±3.2 | 25.2±3.3 | 25.0±3.3 | 0.034 |
| |||||
FBG (mmol/L) | 5.5±1.5 | 5.6±1.5 | 5.6±1. | 5.7±1. | <0.001 |
| |||||
TC (mmol/L) | 4.8±1.4 | 5.0±1. | 5.0±1. | 5.0±1. | <0.001 |
| |||||
Occupational noise exposure, n (%) | 560 (25.9) | 502 (23.6) | 522 (24.3) | 493 (23.1) | 0.066 |
| |||||
Cigarette smoking, n (%) | 1094 (60.2) | 1157 (62.9) | 1138 (61.3) | 1144 (62.6) | 0.272 |
| |||||
Alcohol consumption, n (%) | 217 (10.7) | 240 (11.7) | 221 (10.7) | 270 (13.2) | 0.037 |
| |||||
Physical exercise, n (%) | 170 (8.4) | 160 (7.8) | 164 (8.0) | 129 (6.3) | 0.022 |
| |||||
Hypertension, n (%) | 521 (23.9) | 747 (34.9) | 878 (40.6) | 1345 (62.1) | <0.001 |
| |||||
Antihypertensive drug usage, n (%) | 57 (2.6) | 73 (3.4) | 112 (5.2) | 240 (11.1) | <0.001 |
| |||||
Diabetes mellitus, n (%) | 140 (6.6) | 159 (7.5) | 182 (8.6) | 268 (12.6) | <0.001 |
| |||||
Dyslipidemia, n (%) | 1524 (70.0) | 1549 (72.3) | 1588 (73.6) | 1651 (76.2) | <0.001 |
SBP, systolic blood pressure; DBP, diastolic blood pressure; mean of times (BP), mean times of blood pressure measurement; SSD, standard deviation of systolic blood pressure; DSD, standard deviation of diastolic blood pressure; SCV, coefficient of the variation of systolic blood pressure; DCV, coefficient of the variation of diastolic blood pressure;
Participants’ PTA values and hearing loss percentages in different BPV groups are shown in Table
Pure-tone average thresholds (PTAs) and hearing loss distribution in different BPV groups.
Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | p for trend | ||
---|---|---|---|---|---|---|
(n=2165) | (n=2156) | (n=2166) | (n=2159) | |||
Pure-tone average threshold (PTA, dB) | Low frequency | 20.0±9.3 | 20.3±9.6 | 20.6±10.5 | 21.1±10. | <0.001 |
Intermediate frequency | 20.8±10.7 | 21.3±11.0 | 21.6±11.5 | 22.1±12. | <0.001 | |
High frequency | 26.1±19.2 | 26.9±19.5 | 27.1±19.9 | 28.5±21. | <0.001 | |
| ||||||
Hearing loss, n (%) | Low frequency | 267 (12.3) | 334 (15.6) | 302 (14.0) | 354 (16.3) | 0.002 |
Intermediate frequency | 334 (15.2) | 400 (18.3) | 386 (17.6) | 470 (21.4) | <0.001 | |
High frequency | 532 (24.1) | 592 (27.1) | 582 (26.5) | 685 (31.2) | <0.001 |
a, p<0.05 compared with quartile 1.
To identify the factors associated with PTA values, we performed a multivariate linear regression analysis with PTA as dependent variable. Independent variables included SSD, DSD, SCV, DCV,
Multivariate linear regression analysis between BPV and PTA.
BPV | PTA at low frequency | PTA at intermediate frequency | PTA at high frequency | |||
---|---|---|---|---|---|---|
B value (95% CI) | p value | B value (95% CI) | p value | B value (95% CI) | p value | |
SSD | 0.05 (0.00-0.11) | 0.050 | 0.07 (0.09-0.13) | 0.024 | 0.16 (0.05-0.27) | 0.003 |
| ||||||
DSD | 0.05 (-0.03-0.12) | 0.195 | 0.05 (-0.03-0.14) | 0.195 | 0.07 (-0.08-0.21) | 0.356 |
| ||||||
SCV | 0.08 (0.01-0.15) | 0.035 | 0.10 (0.02-0.18) | 0.017 | 0.19 (0.05-0.33) | 0.009 |
| ||||||
DCV | 0.03 (-0.03-0.10) | 0.295 | 0.04 (-0.03-0.11) | 0.298 | 0.04 (-0.08-0.17) | 0.518 |
| ||||||
| 0.04 (-0.02-0.09) | 0.169 | 0.05 (-0.01-0.11) | 0.115 | 0.12 (0.01-0.23) | 0.029 |
| ||||||
| 0.11 (-0.10-0.32) | 0.319 | 0.08 (-0.16-0.32) | 0.528 | 0.10 (-0.32-0.52) | 0.642 |
BPV, blood pressure variation; PTA, pure-tone average threshold; SSD, standard deviation of systolic blood pressure; DSD, standard deviation of diastolic blood pressure; SCV, coefficient of the variation of systolic blood pressure; DCV, coefficient of the variation of diastolic blood pressure;
The results showed that variations of SBP (SSD, SCV, and
Because hypertension and antihypertensive drugs affect BPV levels, we further divided the individuals into nonhypertension group (n=5111) and hypertension group (n=3491) and reanalyzed the relationship between BPV and PTA by multivariate linear regression model. In nonhypertension group, the results showed that variations of SBP (SSD and SCV) were positively correlated with PTAs at intermediate (p=0.017 for SSD) and high frequency (p=0.004 for SSD; p=0.017 for SCV). However, no significant relationship was revealed in hypertension group. See Supplementary Table
To further demonstrate the relationship between BPV and hearing loss, multivariate logistic regression analysis was performed with the existence of hearing loss as dependent variable (0 = without hearing loss; 1 = with hearing loss). Independent variables included each SD increase of SSD, DSD, SCV, DCV,
Multivariate logistic regression analysis between BPV and hearing loss.
BPV groups | PTA at low frequency | PTA at intermediate frequency | PTA at high frequency | |
---|---|---|---|---|
OR value (95% CI) | OR value (95% CI) | OR value (95% CI) | ||
SD | SSD (+SD) | 1.07 (0.99-1.15) | 1.09 (1.02-1.17) | 1.07 (1.01-1.14) |
DSD (+SD) | 1.04 (0.97-1.12) | 1.04 (0.98-1.11) | 1.02 (0.96-1.08) | |
| ||||
CV | SCV (+SD) | 1.06 (0.99-1.14) | 1.08 (1.01-1.15) | 1.06 (1.003-1.12) |
DCV (+SD) | 1.03 (0.96-1.11) | 1.03 (0.97-1.10) | 1.01 (0.95-1.07) | |
| ||||
VIM | | 1.05 (0.98-1.12) | 1. 07(1.004-1.14) | 1.05 (0.997-1.11) |
| 1.02 (0.96-1.10) | 1.03 (0.96-1.09) | 1.00 (0.95-1.06) |
BPV, blood pressure variation; PTA, pure-tone average threshold; SD, standard deviation; CV, coefficient of the variation; VIM, variation independent of mean. Multivariate logistic regression analysis was performed with the existence of hearing loss as dependent variable (0 = without hearing loss; 1 = with hearing loss). Independent variables included each SD increase of SSD, DSD, SCV, DCV,
The results indicated that each SD increase in SSD was positively associated with hearing loss at intermediate and high frequencies (OR (95% CI)=1.09 (1.02-1.17) for intermediate frequency; OR (95% CI)=1.07 (1.01-1.14) for high frequency). For CV, the results showed that each SD increase in SCV was positively associated with hearing loss at intermediate and high frequencies (OR (95% CI)=1.08 (1.01-1.15) for intermediate frequency; OR (95% CI)=1.06 (1.003-1.12) for high frequency). In terms of VIM, the results suggested that each SD increase in
We also divided the individuals into nonhypertension group (n=5111) and hypertension group (n=3491) and reanalyzed the relationship between BPV and hearing loss by multivariate logistic regression model. In nonhypertension group, the results showed that variations of SBP (SSD and SCV) were positively correlated with hearing loss at intermediate (OR (95% CI)=1.12 (1.02-1.24) for SSD; OR (95% CI)=1.09 (1.00-1.19) for SCV) and high frequency (OR (95% CI)=1.11 (1.01-1.21) for SSD; OR (95% CI)=1.08 (1.001-1.17) for SCV). However, no significant relationship was observed in hypertension group. See Supplementary Table
As occupational noise exposure is one of the predominant risk factors for hearing loss [
In this large-scale population-based cross-sectional study, we firstly investigated the relationship between BPV and hearing and found that PTAs and percentages of hearing loss at low, intermediate, and high frequencies grew gradually with increasing BPV levels. After adjusting for multiple covariates, multivariate linear regression analysis demonstrated that variations of SBP (SSD, SCV, and
Hypertension has long been regarded as one of the essential risk factors underlying pathophysiological processes of the cochlea from early in the twentieth century [
As occupational noise exposure is an important risk factor of hearing loss, we remove subjects with noise exposure and reanalyzed the relationship between BPV and hearing loss. We found that the correlation between each SD increase in SSD, SCV, and
It has been reported by previous studies that greater BPV is associated with higher risk of target organ damage [
Cochlea, as the main hearing organ, is supplied by the labyrinthine artery and they are terminal arteries without collateral vessels. The hair cells of cochlea are movable cells and play an important role in the process of acoustic amplification. Their function requires a lot of energy. Therefore, the cochlear hair cells are extremely sensitive to ischemia [
This study has some limitations. First, the participants of the current study are all male subjects and thus the relationship between BPV and hearing among female individuals remains unknown. Second, we did not validate the damage of BPV on cochlea by cellular or animal models. Third, this observation was performed on Chinese population. Whether the results can be generalized to individuals of other ancestries warrants further investigations.
This is the first and largest-scale population-based study to analyze the relationship between long-term BPV and hearing. After adjusting for multiple factors, we found that variations of SBP (SSD, SCV, and
The population-based data used to support the findings of this study may be released upon application to the Kailuan General Hospital that can be contacted at drwusl@163.com.
The authors declare that there are no conflicts of interest.
Minghui Bao and Yongjian Song contributed equally to this work. Xinchun Yang and Shouling Wu contributed equally as co-corresponding authors.
This work was supported by the National Natural Science Foundation of China (81670214).
Supplementary materials description: this file included a total number of six supplementary tables. In Supplementary Tables 1-2, we divided the individuals in the current study into nonhypertension group (n=5111) and hypertension group (n=3491) and the multivariate linear regression analysis between BPV and PTA and the multivariate logistic regression analysis between BPV and hearing loss were reanalyzed in Supplementary Table