Occupational Noise Exposure and the Risk of Hypertension

Abstract

Background: 

Noise may increase the risk of hypertension, but findings are inconsistent with respect to both community and occupational noise exposure. We used a large sample of noise-exposed industrial trades to analyze the association of occupational noise exposure and the risk of hypertension.

Methods: 

The 7-year prospective cohort study included 145,190 workers from 625 companies representing 10 industrial trades and 100 companies from the finance sector. They were followed from 2001 to 2007 by record linkage with several Danish national registries. Full-shift noise exposure levels, measured in a random subset of 710 workers at the start and the end of follow-up, ranged from 70 to 86 dB(A); based on this information, historical levels back to the 1960s were estimated. Hypertension (defined by the prescription of antihypertensive medication or a hospital discharge diagnosis of hypertension) was regressed on the trade mean sound levels (L_AEq) adjusting for a number of covariates.

Results: 

Women had increased rate ratios for hypertension when comparing blue-collar industrial workers with white-collar financial workers (adjusted rate ratio = 1.17 [95% confidence interval = 1.09–1.26]). For men, the corresponding relative risk value was 1.06 (0.98–1.14). Within blue-collar industrial workers, however, increasing noise exposure level was not associated with an increasing risk of hypertension among either men or women.

Conclusion: 

Our study shows no increased risk of hypertension with exposure to noise in the lower half of the 80–90 dB(A) range.

In Europe, 30% of the population is exposed daily to community noise exceeding 55 dB(A), and 30% of the work force report that they are exposed to noise so loud that they have to raise their voice.¹ Community and occupational noise have been related to increased risk of hypertension,^2–4 hypothetically through altered activation of the autonomic nerve system.^5,6 The results seem more consistent for community noise, although a recent study showed marked risk differences among countries.^7,8 The evidence regarding occupational noise is less consistent.⁴ Nevertheless, two recent large studies of community and occupational noise showed increased risk of hypertension.^9,10 We report the relation of occupational noise exposure to hypertension in a large 7-year follow-up study of workers in industrial trades and financial services.

METHODS

Study Population

In 2001, we identified the 10 industrial trades with the highest levels of compensation claims for occupational hearing loss according to the Danish Occupational Safety and Health Authority.¹¹ From Statistics Denmark we retrieved data from 625 companies in the industrial trades and 100 companies in financial services (the latter with expected low noise exposure levels), all from Aarhus County. All workers employed in 2001–2007 were identified in the Supplementary Pension Fund Register, which provides information on employer, trade, and percentage of employment for each year since 1964.¹² Data on occupation (1991–2007), defined by a Danish version of the International Standard Classification of Occupations (ISCO-88)¹³ and socioeconomic status (1980–2007),¹⁴ were obtained from the Integrated Database for Labour Market Research,¹⁵ and vital status information was obtained from the Danish Civil Registration System. These procedures were possible owing to the unique identification number used in all Danish national registers.¹⁶

In total, 219,550 subjects were employed in the 725 study companies in 2001–2007. The industrial trades employed 115,573 blue-collar workers (ISCO codes 7–9) and 48,838 white-collar workers (ISCO codes 1–4), whereas the financial companies employed 4039 blue-collar and 51,100 white-collar workers. We excluded white-collar workers from industrial trades and blue-collar workers from financial companies, as well as 1795 white-collar workers from financial companies who had been employed in a blue-collar job before 2001. We also excluded workers living outside Denmark (n = 164), as well as those who had redeemed antihypertensive prescriptions (n = 11,652) or been diagnosed with hypertension (n = 295) or a cardiovascular diseases other than hypertension (n = 7,577) before 1 January 2001 or in their first year of employment if after that date. The study population then consisted of 145,190 workers (108,402 men and 36,788 women).

Health Outcomes

Individual data on redeemed antihypertensive prescriptions were obtained from the Danish National Prescription Registry, which records all prescribed medications purchased at pharmacies in Denmark since 1995.¹⁷ In Denmark, the national health security system covers all residents and partially reimburses drug expenses. Because of this, pharmacies are required by law to register all prescriptions dispensed at an individual level, which makes the registry highly valid. We identified registrations for five classes of antihypertensive medications coded by the Anatomical Therapeutic Chemical classification system¹⁸: centrally acting α-2 agonists and others (Anatomical Therapeutic Chemical code C02), diuretics (C03), β-blockers (C07), calcium-channel blocker (C08), and ACE-inhibitors (C09).

Information on hospital discharge diagnoses of essential hypertension and other cardiovascular diseases was retrieved from the Danish National Patient Register, which provides diagnoses for all contacts with Danish hospitals according to the International Classification of Diseases, revision 8 (ICD-8) 1977–1993 and revision 10 (ICD-10) since 1994.¹⁹ Cases of hypertension and date of diagnosis were defined by the first entry of a relevant Anatomical Therapeutic Chemical code or by a discharge diagnosis of essential hypertension.

The population was followed from first year of employment or 1 January 2001, if later, until (1) becoming cases; (2) being censored, when recorded with a hospital discharge diagnosis of cardiovascular disease other than essential hypertension, emigration, disappearance, death; or (3) end of follow-up at 31 December 2007, whichever came first.

Noise Exposure Assessment

In 2001, we conducted a noise exposure survey of 80 randomly selected study companies representing all 11 trades.²⁰ We recorded full-shift noise exposure levels (L_AEq values) by personal dosimeters (Brüel and Kjær, 4443 and 4445) for 649 blue-collar workers from the industrial companies and 61 white-collar workers from the financial companies (eTable 1 https://links.lww.com/EDE/A604). In 2009–2010, the survey was repeated for 589 workers from 132 study companies according to the same protocol, and we analyzed whether there was a general trend in noise exposure levels during this time period. From these data, we predicted noise exposure level for each combination of trade and calendar year since 1964.

Table 2: Crude and Adjusted RRs of Hypertension Among 103,687 Blue-Collar Industrial Workers Compared with a Reference Group of 41,503 White-Collar Financial Workers

We cumulated noise exposure as the product of noise exposure level (L_AEq in dB(A)) and duration of exposed employment (T) since 1964, according to the following formula: 10 × log [Σ(10^dB(A)/10 × T)], resulting in “dB(A)-year” on a logarithmic scale. The computation took into account percentage of employment in any year and stopped accumulating during years with no exposed employment (blue-collar industrial workers or white-collar financial workers).

Because there may be a threshold below which noise exerts no effect on hypertension, we also classified noise exposure by two metrics of duration of exposure above threshold: duration of exposure above 80 dB(A) and duration of exposure above 85 dB(A).

Early year of employment was expected to be a strong proxy of high noise exposure level, as well as low prevalence of hearing protector.^21,22 For this reason, we categorized participants by first year of exposure as a sensitivity check of noise exposure level at the ear.

Statistical Analyses

We estimated rate ratios (RRs) and 95% confidence intervals (95% CIs) of hypertension by logistic regression using STATA version 11 (STATA Corp., College Station, TX). The analyses were performed as a discrete survival function, because person-year was the unit of analysis, thereby yielding RRs.²³ All models either stratified by sex or included an interaction term between sex and occupation (blue-collar industrial worker vs. white-collar financial worker) and furthermore adjusted for age (six categories), socioeconomic status (five categories), calendar year (seven categories, 2001–2007), employment status that for each year indicated if the participant was gainfully employed, and duration of employment to account for a possible healthy-worker survivor effect.^24,25 In addition, we analyzed age as a continuous and squared variable, and we adjusted for the prescription of any medications other than antihypertensives. Because of possible lifestyle and other differences not captured in the adjusted analyses, we also conducted internal trend analyses restricted to the 10 industrial trades.

In addition, we analyzed noise exposure variance components between and within the 11 trades. We had information only on blue-collar versus white-collar status for the noise-measured workers, and inclusion of occupation in this analysis would not add further information.

RESULTS

The 108,402 male workers cumulated 577,200 person-years (mean = 5.3 years; maximum = 7 years) and 7587 cases of hypertension (88% defined by redemption of antihypertensives and 12% defined by a hospital discharge diagnosis of hypertension). The 36,788 female workers cumulated 191,086 person-years (mean = 5.2 years; maximum = 7 years) and 3808 cases (93% defined by redemption and 7% defined by a hospital discharge diagnosis).

Among both men and women, industrial workers were slightly younger, had lower socioeconomic status, shorter duration of employment, and completely different occupational titles than financial workers (the latter as expected owing to the definition of the study population) (Table 1). Redemption rates for medication other than antihypertensives were comparable for industrial and financial workers, for men as well as for women—except for nervous system medications, which showed higher rates among industrial workers.

Table 1: Characteristics of 103,687 Blue-Collar Industrial Workers and 41,503 White-Collar Financial Workers

Noise exposure levels from the 2001–2002 survey that classifies participants according to trade are presented in eTable 1 (https://links.lww.com/EDE/A604). Mean noise level was below 70 dB(A) in the financial services and above 80 dB(A) for all industrial trades. Manufacture of wood products and manufacture of fabricated metal had mean levels above 85 dB(A). Analyses of time trend showed a 0.1 dB(A) decline annually during the 8-year period from 2001–2002 to 2009–2010.²⁶ Analyses of noise exposure levels showed variance components of 53% between and 48% within the 11 trades.

Among women, the adjusted RR of hypertension was 1.17 among industrial, blue-collar workers compared with white-collar workers of financial services (95% CI = 1.09–1.26) (Table 2). Among men, the corresponding value was 1.06 (0.98–1.14). For the interaction term between sex and occupation (blue-collar industrial worker vs. white-collar financial worker), P was < 0.001.

Table 3 presents RRs of hypertension by cumulative noise exposure. Crude analyses showed increasing risk by increasing noise exposure level for both men (RR = 1.02 [95% CI = 1.02–1.03]) and women (1.02 [1.02–1.03]). When analyses were restricted to internal comparisons within the industrial trades, stronger trends were seen (for men, 1.04 [1.03–1.04] and for women, 1.03 [1.02–1.04]). Analyses that adjusted for age, socioeconomic status, calendar year, and employment status showed a weak declining trend for male workers when the financial sector was included, but no trend when restricted to the industrial trades (0.99 [0.99–1.00]). For female workers, the adjusted analyses showed an increasing trend by cumulative exposure that vanished when we excluded the financial sector.

Table 3: Association of Hypertension with Cumulative Noise Exposure for Male and Female Blue-Collar Industrial and White-Collar Financial Workers

Table 4 presents RRs for hypertension by duration of noise exposure above 80 dB(A) and above 85 dB(A). For male workers, the adjusted RRs showed no increase by duration of employment above either 80 dB(A) or 85 dB(A). For female workers, an increasing adjusted RR was seen by duration of exposure above 80 dB(A). The opposite trend was seen in analyses restricted to blue-collar industrial workers. Exposure above 85 dB(A) showed no positive trend by duration.

Table 4: Association of Hypertension with Duration of Exposure Above 80 dB(A) and Above 85 dB(A), Respectively, for Male and Female Blue-Collar Industrial and White-Collar Financial Workers

Analyses considering first year of exposure as a proxy measure of noise exposure at the ear showed declining RRs of hypertension by increasing calendar year (Table 5). The RR for a linear trend was 0.99 (95% CI = 0.98–0.99) for women.

Table 5: Association of Hypertension with First Year of Exposure for Male and Female Blue-Collar Industrial Workers

Additional analyses were performed with a continuous and squared age variable, without any effect on the results (eTables 2–4, https://links.lww.com/EDE/A604). When we included medication other than antihypertensives in the analyses, point estimates were not substantially altered (eTables 2–4, https://links.lww.com/EDE/A604).

DISCUSSION

We found higher risk of hypertension in female noise-exposed industrial blue-collar workers compared with white-collar workers from the financial sector. Among women, the risk increases with cumulative noise exposure, but this trend was not seen in the subsample of industrial workers. When women’s person-years were classified by duration of exposure above 80 dB(A) and 85dB(A) and restricted to industrial workers, no increasing risks of hypertension were apparent, and thus no indication of a threshold effect. Among men, we found no effect of noise.

There are several possible interpretations of this risk pattern. First, there may be a sex-dependent effect of noise on hypertension, supported by the fact that the risk pattern for hypertension is different for men and women.²⁷ However, the effect of noise on hearing handicap does not differ between men and women.²²

Second, the hypertension difference seen in women may not be related to noise exposure but to individual or lifestyle factors. We know from an earlier study of a subsample of this population that the blue-collar industrial workers smoke more than twice as much as the white-collar financial workers.²² Furthermore, it is well documented that obesity, physical inactivity, and smoking show a strong socioeconomic gradient in Denmark.²⁸ These factors are all risk factors for hypertension²⁷ and factors for which we had no information.

Third, noise exposure is assessed by the mean values for each trade, and nondifferential misclassification could explain our lack of an effect. However, mean values are expected to provide better estimates of the individual, long-term, average noise exposure, given that higher day-to-day variability in noise exposure level is assumed within, rather than between, workers of each trade.^29–32 Furthermore, this misclassification is expected to exert Berkson-type error, which does not underestimate a true exposure relation.³³ We did not have access to repeated noise recording for individual workers, which would allow us to assess this assumption.

A recent meta-analysis³ presented an increased average blood pressure level among industrial workers exposed to noise compared with those not exposed. Another meta-analysis⁴ concluded that the epidemiologic evidence of a causal association between occupational noise exposure and increased blood pressure is inconsistent. The majority of the studies included in both meta-analyses investigated work-site blood pressure measurements, rather than hypertension.^3,4 One recent, large cross-sectional study³⁴ found no association between occupational noise exposure and hypertension defined as systolic blood pressure ≥140 mmHg, diastolic blood pressure ≥90 mmHg, or self-reported hypertension. Moreover, few studies show positive associations between exposure to industrial noise and change in blood pressure over time.^6,35 To our knowledge, only one previous study¹⁰ has investigated the incidence of hypertension, and it showed a positive association with noise exposure.

Our study has several strengths compared with earlier studies. It has a follow-up design, thus minimizing any potential problems related to the temporal relation between noise exposure and hypertension. Our study is large owing to the availability of registry data, and thus has the statistical power to detect small effects.

We have access to full-shift noise recordings from a sample of companies and workers of the 11 study trades, as well as full work histories since 1964. Moreover, we measured noise exposure levels at two points in time and thereby were able to address changes in exposure over time.

We analyze antihypertensive medication and hospital discharge diagnoses as a proxy for hypertension. This measure is expected to be the end point of repeated blood pressure measurements by a physician, at home, and probably also ambulatory recordings, as well as clinical examination in accordance with contemporary clinical guidelines.³⁶ Thus, the specificity is expected to be higher than single measurements that may be biased by transient increases related to noise exposure, as well as physical and psychological strain at work.^37,38 Self-reported information on hypertension may be influenced by recall bias.³⁹

Antihypertensives may be prescribed for other diseases than hypertension, especially for heart diseases, but few patients receive these drugs because of hypertension without a verified diagnosis.⁴⁰ To reduce misclassification of outcome, all workers diagnosed with a cardiovascular disease were censored.

All Danes have free access to general practitioner and hospital care, and thus selection bias is probably not an issue. If blue-collar industrial workers contact the health care system less often than white-collar financial workers, this could, however, bias a true effect of noise toward null. The prevalence of redemptions for medications other than antihypertensives was similar for the industrial and financial workers, and thus we think this bias is unlikely.

We noticed a 50% increase in the redemption of antihypertensive medications during the 7 years of follow-up not explained by increasing age or other variables included in our models. This should not have biased our findings, however, because all analyses were adjusted by calendar year.

The healthy-worker survivor effect could explain the lack of an exposure-outcome relation by duration of exposure if hypertensive noise-exposed workers hasten the termination of employment compared with normotensive workers. The adjusted analyses took account of this by including employment status and duration of employment.^24,25

Our study also has limitations that should be considered. Individual information on the use of hearing protection devices was not available. Because the use of hearing protection is associated with noise exposure level, this will attenuate the exposure contrast and thus reduce the power to detect an effect. Later year of first exposure shows lower RR of hypertension. First year of exposure may be an indirect measure of noise exposure level at the ear because noise exposure level is expected to be higher and prevalence of hearing protection lower during early years. Information on the use of hearing protection devices would have been useful but is not feasible in a register-linked study.^21,22

To conclude, our study showed no increased risk of hypertension at noise exposure levels within the lower half of the 80–90 dB(A) range.

ACKNOWLEDGMENTS

Camilla Skovbjerg Jensen, Henriette Lund Christensen, Malene Amondin Tousgaard, and Tina Bahn Larsen collected the noise exposure data, and Matias Brødsgaard Grynderup gave invaluable assistance with the data handling.