Occupational noise and myocardial infarction: considerations on the interrelation of noise with job demands.

The present analysis aims to differentiate the association of noise on myocardial infarction (MI) by job specific demands using International Standard Classification of Occupations (ISCO)-88 codes as a proxy. Data of a German case-control study were supplemented by job descriptions (indicated by ISCO-88). It was examined whether the demands in the various occupational groups modify the effect of noise. Noise and occupational groups are combined to form new exposure categories. Conditional logistic regression models were fitted to identify effects of combined job-noise categories. For the highest noise range (95-124 dB(A)) we found a significant odds-ratio (OR) of 2.18 (confidence interval [CI] 0.95 = 1.17-4.05) independent of the profession. Some interesting results were found indicating ISCO groups with possible risk. In men, noticeable effects for the exposure category between 62 dB(A) and 84 dB(A) are calculated in the group of legislators and senior officials (ISCO-group 11; OR=1.93; CI (0.95) = 0.50-7.42), the group consisting of life science and health professionals (ISCO-group 22; OR=2.18; CI 0.95 = 0.36-13.1), the group of life science and health associate professionals (ISCO-group 32; OR = 2.03; CI 0.95 = 0.50-8.24), and the group of "precision, handicraft, printing, and related trades workers" (ISCO-group 73; OR = 2.67; CI(0.95) = 0.54-13.0). In the exposure range of 85-94 dB(A), high ORs are calculated for "skilled agricultural, fishery, and forestry workers" (ISCO-group 6; OR = 4.31; CI(0.95) = 0.56-33.3). In women, there are high (nonsignificant) ORs in ISCO-group 1 (OR = 2.43; CI(0.95) = 0.12-50.0), ISCO-group 2 (OR = 1.80; CI 0.95 = 0.31-10.5), and ISCO-group 9 (OR = 2.45; CI(0.95) = 0.63-9.51) for a noise exposure between 62 dB(A) and 84 dB(A). When investigating noise at the workplace in relation to cardiovascular diseases it is important to take the specific requirements of a job into account. Thus, work tasks with high health risks can be identified that helps to develop appropriate prevention strategies.

Introduction

Occupational noise is a frequent exposure in the work environment. Results from surveys[1] show that in Germany 23.9% of the working population are often exposed to noise, and 54% of them feel burdened by noise. Besides the effect on hearing, nonauditory effects such as those on the cardiovascular system are discussed.

Within the scope of the discussion about aging workforce and the expected increase of workers with chronic diseases, such as cardiovascular diseases, prevention becomes more important. Besides individual changes in lifestyle factors, the consideration of occupational risk factors such as noise or stress is necessary. Definition of high-risk groups is important to address preventive measures.

There is physiological evidence from experimental as well as field studies showing that the underlying mechanisms of the cardiovascular effects of noise are unspecified biological stress responses in terms of activation of the autonomic system and neuroendocrine pathways.[23] Noise-induced dysregulation may result in the promotion of the atherosclerotic process, which in the long run causes hypertension and cardiovascular events.

Epidemiological studies have shown contradictory results on the association between noise and cardiovascular disease. Data from the Copenhagen male study could not detect an association between noise and death from ischemic heart disease (IHD).[4] A Finnish census study reported fairly weak associations.[5] Within the Helsinki heart study, noise was associated with a moderate but statistically significant increase in coronary heart disease (CHD) risk that persisted even after the workers had passed the age of retirement.[6] A Canadian follow-up study[7] investigating lumber mill workers found a relative risk (RR) of 1.5 in the highest exposure category. A recent study investigating occupational noise in combination with job strain found a nonsignificant increase of disease risk, when only occupational noise was considered in relation to myocardial infarction (MI).[8]

Usually the magnitude of noise exposure is described by the sound energy level (decibel). The risk assessment is based on the exposure data. But as Babisch discussed, the nonauditory noise effects, such as those on the cardiovascular system, do not follow the “toxicological principle of dosage.”[9] Thus, not only the levels of sound energy should be taken into account when describing the magnitude of exposure, but also the individual situation or the degree of disturbance of the activity or the work task of the noise exposed needs to be considered.

Laboratory studies looking at the effect of mental tasks in combination with noise on blood pressure,[10] heart rate, and norepinephrine and cortisol levels[11] suggest that work characteristics such as task complexity, task difficulty, and required effort can modulate the adverse effect of noise. The Cardiovascular Occupational Risk Factors Determination in Israel (CORDIS) study also showed an association of noise exposure with excess mortality in workers performing complex jobs.[12] A population-based case-control study from Sweden found that the combination of noise and job strain increases the risk of MI substantially.[8]

It seems that different types of “job stress” may act as a modifier of noise health effects. Thus, we assume that the effect of noise may be different in subjects with different levels of job complexity and skill specialization. Additionally, it may make a difference whether workers perceive noise as caused by themselves and thus as a normal circumstance of their work, or whether noise comes from a foreign source. Different levels of job complexity and skill specialization are grouped in the International Standard Classification of Occupation (ISCO). Thus, data of a German case-control study[1314] looking at the association of noise on MI were reconsidered taking job description (indicated by ISCO-88) as well as noise levels (decibels/dB) into account.

Methods

Study population

This study comprised 353 female cases and 706 female controls as well as 1,527 male cases and 1,527 male controls. The cases and controls, aged 20-69 years, were recruited from 32 major hospitals in Berlin. Cases of acute MI or survivors of sudden cardiac arrest were classified by diagnostic criteria following the World Health Organization definitions, including ischemic changes in the electrocardiogram, clinical symptoms, and enzymatic changes. Controls and patients from the same hospitals for diagnoses that were presumably not related to noise were individually matched according to sex, age (5-year categories), and hospital. The study design was approved by the Ethical Commission of the medical faculty of the Humboldt University of Berlin.[13]

Data elicitation

After being informed about the objective of the study, those who consented to study participation were invited to a computer-assisted standardized interview. During the interview, medical, occupational, and sociodemographic histories were recorded. The occupational history contains information about the type of job, company and machines/devices used by the study subjects at work during the preceding 10 years (for up to three jobs).

Exposure assessment

The daily sound exposure (in dB) level (ISO 1999), which is supposed to be an “objective” exposure measure, was estimated for all workplaces retrospectively for 10 years from information about workplaces and machines with the help of catalog specifications (LEX,8h,obj). To validate these estimates, measurements were performed on 146 existing workplaces. Information about hearing protection (type and duration of wear) was considered in the assessments of “objective” sound pressure level (LEX,8h,obj,hp). Using these values, the “10-year sound exposure level” was calculated as an equivalent continuous A-weighted sound pressure level averaged over the past 10 years and normalized to 10 years (LEX,10y,obj / LEX,10y,obj,hp). For periods of unemployment, 55 dB(A) were presumed. The final analyses included the objective 10-year rating level corrected for the use of ear protection. The metric sound pressure values in the study data have a bimodal distribution with a hollow at 61 dB(A). The German Standard VDI 2058/3 cuts the dB(A) scale at 85 dB(A); the values above >85 dB(A) are subsumed in one category. A noise level of about 95 dB(A) or above is associated with an increased activation of the autonomic nervous system and increased secretion of stress hormones.[15] Thus, noise levels >85 dB(A) will be split at 95 dB(A). Finally, the estimated sound values were categorized into four categories [46-61 dB(A), 62-84 dB(A), 85-94 dB(A), and 95-124 dB(A)].

Job classification

Classification of occupations according to ISCO-88 is used here as a proxy for a specific situation at the workplace, including other workplace stressors. The ISCO groups job activities in terms of tasks and duties specific for a job. These groups are structured according to their similarity based on two dimensions, skill level und skill specialisation. The skill level reflects the complexity of the corresponding job; the skill specialization is a criterion of professionalism and competence, which reflects the type of performed work. Characteristics of a job — e.g., decision latitude, quantitative, and qualitative work demands — differ between different ISCO groups. According to our hypothesis the effect of noise may be different in subjects with different levels of job complexity and skill specialization reflected by the ISCO levels, e.g., in jobs with complex tasks (ISCO major groups 1, 2, and 3) an interference due to noise might be experienced as a more adverse stressor than in jobs with less complex tasks.

Information on the type of job was collected using the job characteristics, not according to the trained qualification. Thus, it was only possible to use a 1- and 2-digit ISCO grouping. For an accurate classification, the level of graduation, the net income, and the type of health insurance were used as additional information.

Statistical methods

Separate models were calculated for men and women, male cases and controls were matched 1:1 and female cases and controls were matched 1:2. Conditional logistic regression analyses were performed to calculate odds-ratios (ORs) and confidence intervals (CI). In a former analysis,[1314] a large set of potentially confounding factors was used to adjust the effects. We revised this set by creating a directed acyclic graph (DAG)[16] and deleted some variables as intermediate factors (diabetes mellitus, hypertension, hyperlipidaemia) or as factors without effect on the exposure (family history of MI, smoking status, body mass index). The minimal sufficient adjustment set to adjust the estimated total effect of occupational noise within an ISCO group on MI are current employment status (CEM), shift work (SW), work >40 h per week (WHW), <12 years at school (YS).

All analyses were performed using the STATA package release 12.[17] Our model has the following form for the expected probability pi of MI:

Logit(pi) = b0 + b1 *EXPi + b2 *CEMi + b3 *SWi + b4 *WHWi + b5 *YSi

whereas EXPi is the combined exposure group from ISCO and noise category. For each ISCO group, there is one regression model and ISCOj *[46 – 61 dB(A)] (j=1,…,9) is the reference group to estimate the OR for the groups with higher noise exposure. For each regression model, all subjects from other ISCO groups are summarized in one exposure group, “remainders.” Subjects without information on noise or profession were also assigned to the group “remainders.”

Since there was a low number of cases and controls in some ISCO groups, sample sizes needed with a given exposure were calculated for further study protocols.

Results

Our data set contains 3,054 male subjects. For 282 men, there was no job information and for 322 men, no information on occupational noise was available. The intersection is 280 male subjects without both these information. A total of 1,059 female subjects were included: 253 without job information, 275 without information on occupational noise, and the intersection is 251 subjects without both these information.

There was a limited number of cases and controls in some of the ISCO groups and noise exposure categories under investigation. Table 1a and 1b show the numbers of male and female cases and controls grouped for ISCO-88 and noise categories.

Table 1a

Number of male cases and controls (1: 1-matching) for ISCO-88 groups and noise categories

ISCO-88/dB(A)	46-61	62-84	85-94	95-124
All	436/382	789/771	123/177	39/15
1 Legislators, senior officials and managers	67/64	34/24	2/4	0/0
11 Legislators and senior officials	18/21	8/4	0/1	0/0
12 Corporate managers	19/31	5/6	0/1	0/0
13 General managers	30/12	21/14	2/2	0/0
2 Professionals	60/72	46/75	3/3	0/1
21 Physical, mathematical and engineering science professionals	28/32	11/28	0/0	0/0
22 Life science and health professionals	6/8	5/3	0/0	0/0
23 Teaching professionals	1/1	22/35	0/0	0/0
24 Other professionals	25/31	8/9	3/3	0/1
3 Technicians and associate professionals	164/130	135/131	7/12	5/1
31 Physical and engineering science associate professionals	48/49	59/73	5/10	3/0
32 Life science and health associate professionals	5/11	9/10	0/0	0/0
33 Teaching associate professionals	2/0	21/12	0/0	0/1
34 Other associate professionals	109/70	46/36	2/2	2/0
4 Clerks	71/63	40/40	0/0	0/0
41 Office clerks	63/53	18/19	0/0	0/0
42 Customer services clerks	8/10	22/21	0/0	0/0
5 Service workers and shop and market sales workers	18/6	99/70	3/1	4/1
51 Personal and protective services workers	17/6	67/60	2/1	4/1
52 Models, salespersons and demonstrators	1/0	32/10	1/0	0/0
6 Gardeners (skilled agricultural and fishery workers)	1/0	9/17	5/2	0/0
7 Craft and related trades workers	31/28	216/222	85/123	29/12
71 Extraction and building trades workers	9/9	101/110	55/77	11/5
72 Metal, machinery and related trades workers	14/12	82/88	21/31	17/7
73 Precision, handicraft, printing and related trades workers	5/5	13/6	4/6	1/0
74 Other craft and related trades workers	3/2	20/18	5/9	0/0
8 Plant and machine operators and assemblers	8/5	161/146	14/24	0/0
81 Stationary-plant and related operators	3/1	12/12	6/7	0/0
82 Machine operators and assemblers	2/1	14/19	3/9	0/0
83 Drivers and mobile-plant operators	3/3	135/115	5/8	0/0
9 Elementary occupations	15/14	49/45	4/8	1/0
91 Sales and services elementary occupations	10/11	22/23	0/0	0/0
92 Agricultural, fishery and related labourers	1/0	2/3	1/2	0/0
93 Labourers in mining, construction, manufacturing and transport	4/3	25/19	3/6	1/0

Table 1b

Number of female cases and controls (1: 2-matching) for ISCO-88 groups and noise categories

ISCO-88/dB(A)	46-61	62-84	85-94
All	110/251	158/254	4/7
1 Legislators, senior officials and managers	4/12	1/1	0/0
2 Professionals	2/17	5/21	0/2
3 Technicians and associate professionals	40/78	25/43	2/3
4 Clerks	54/125	12/22	0/0
5 Service workers and shop and market sales workers	5/6	55/77	0/0
6 Gardeners (skilled agricultural and fishery workers)	0/1	3/4	0/1
7 Craft and related trades workers	1/1	16/29	2/0
8 Plant and machine operators and assemblers	1/1	8/15	0/0
9 Elementary occupations	3/10	33/40	0/1

In the 54 subjects with very high noise exposure [95-124 dB(A)], we have found a significant OR = 2.18 (CI0.95 = 1.17-4.05) compared to the low noise category [Table 2]. The group with the highest exposure, which means at least 100 dB(A) are subjects working in the country police department (n = 10, 8 of them with MI), e.g., armorers and shooting coaches. Other job titles with an exposure of 95-106 dB(A) are engine fitters, engine drivers, and metal workers in boiler plants, turbine hangars, the ship building industry, steel construction, etc.

Table 2

Adjusted odds-ratios and asymptotic 0.95-confidence intervals for all subjects (males and females): One conditional logit model per row, basis of comparison: 46-61 dB(A) noise category

ISCO-88	62-84 dB(A)	85-94 dB(A)	95-124 dB(A)
Male	0.89 (0.74-1.06)	0.61 (0.46-0.80)	2.18 (1.17-4.05)
Female	1.24 (0.91-1.69)	1.22 (0.34-4.32)	—

In the medium noise category [85-94dB(A)], male subjects showed decreased ORs compared to the low noise category [46-62 dB(A)]. This is an effect, which is probably due to the subjects of ISCO groups 7-9 and will be discussed later.

There are only few women (n = 11) who are exposed to noise above 85 dB(A). These women are mainly active in the education of children, some are employed at the opera, and some are working with machines. Female subjects showed slightly increased OR compared to the low noise category. With view on the frequencies in the ISCO groups [Table 1b], the noise effect cannot be assigned to any activity or occupational group.

Table 3a summarizes the ORs and CIs for male subjects in 1- and 2-digit ISCO groups. Because of the high number of ISCO groups, which also implies low case numbers per group, we have found no significant results. However, some distinctive features may be noticed. ORs greater than or nearly 2 are calculated for an exposure between 62 dB(A) and 84 dB(A) in the group of legislators and senior officials (ISCO-group 11; OR = 1.93; CI0.95 = 0.50-7.42), the group consisting of life science and health professionals (ISCO-group 22; OR = 2.17; CI0.95 = 0.36-13.1), and the group of life science and health associate professionals (ISCO-group 32; OR = 2.03; CI0.95 = 0.50-8.24). In the latter two groups physicians and medical professionals are mainly enclosed. Within a further group with an OR >2 are the “precision, handicraft, printing, and related trades workers” (ISCO-group 73; OR = 2.67; CI0.95 = 0.54-13.0). In this group 31 of the 40 subjects are printing workers.

Table 3a

Adjusted odds-ratios and asymptotic 0.95-confidence intervals for ISCO-88 groups and noise categories (males): One conditional logit model per row (within each ISCO class), basis of comparison: 46-61 dB(A) noise category of the same ISCO class

ISCO-88	62-84 dB(A)	85-94 dB(A)	Remainders
1	1.34 (0.70-2.53)	0.52 (0.09-2.96)	1.02 (0.71-1.47)
11	1.93 (0.50-7.42)		1.13 (0.60-2.14)
12	1.57 (0.42-5.93)		1.91 (1.06-3.42)
13	0.60 (0.22-1.58)	1.17 (0.16-8.60)	0.41 (0.21-0.81)
2	0.75 (0.46-1.24)	1.56 (0.29-8.30)	1.40 (0.95-2.05)
21	0.46 (0.20-1.07)		1.22 (0.72-2.06)
22	2.18 (0.36-13.1)		1.42 (0.48-4.18)
23	0.78 (0.046-13.2)		1.25 (0.077-20.2)
24	1.04 (0.34-3.13)	1.41 (0.25-7.86)	1.28 (0.73-2.25)
3	0.80 (0.57-1.12)	0.42 (0.16-1.12)	0.75 (0.58-0.97)
31	0.79 (0.46-1.34)	0.47 (0.15-1.48)	1.01 (0.68-1.52)
32	2.03 (0.50-8.24)		2.08 (0.72-6.06)
33	—		—
34	0.78 (0.46-1.34)	0.61 (0.082-4.55)	0.61 (0.45-0.84)
4	0.87 (0.50-1.52)		0.89 (0.62-1.27)
41	0.76 (0.36-1.59)		0.85 (0.58-1.23)
42	1.30 (0.43-3.89)		1.24 (0.48-3.16)
5	0.45 (0.17-1.19)	1.12 (0.10-13.1)	0.33 (0.13-0.84)
51	0.39 (0.14-1.05)	0.83 (0.062-11.2)	0.37 (0.14-0.95)
52	—		—
6	1.001	3.98 (0.63-25.0)	1.75 (0.78-3.96)
7	0.90 (0.52-1.56)	0.64 (0.36-1.14)	0.94 (0.56-1.59)
71	0.90 (0.34-2.38)	0.72 (0.27-1.92)	1.01 (0.40-2.60)
72	0.78 (0.34-1.81)	0.57 (0.22-1.49)	0.84 (0.38-1.83)
73	2.67 (0.54-13.0)	0.64 (0.11-3.83)	1.10 (0.32-3.84)
74	0.72 (0.11-4.75)	0.30 (0.036-2.53)	0.63 (0.10-3.85)
8	0.72 (0.23-2.29)	0.40 (0.11-1.51)	0.69 (0.22-2.14)
81	0.30 (0.027-3.40)	0.29 (0.023-3.65)	0.31 (0.031-3.03)
82	0.48 (0.039-5.90)	0.19 (0.012-3.02)	0.65 (0.058-7.26)
83	1.31 (0.25-6.77)	0.76 (0.11-5.48)	1.17 (0.23-5.93)
9	0.96 (0.41-2.22)	0.39 (0.09-1.63)	0.84 (0.40-1.76)
91	1.04 (0.36-2.97)		1.04 (0.43-2.49)
92	—		—
93	0.88 (0.17-4.46)	0.29 (0.037-2.28)	0.65 (0.14-2.94)

Basis of comparison 62-84 dB(A), because no controls exist with 46-61 dB(A)

Also, in the ISCO-group 6, a high OR has been calculated for the exposure range between 85 dB(A) and 94 dB(A). This ISCO group normally contains skilled agricultural and fishery workers, but in this data set there are almost exclusively gardeners. The OR of 4.31 (CI0.95 = 0.56-33.3) is quite high, but the result is limited by the low case number (n = 7). The high noise level probably results from machines they need to run or to drive but at the same time, this work is only slightly dependent on physical fitness.

In the ISCO-groups 7-9, the ORs for those with an exposure between 85 dB(A) and 94 dB(A) are lower than for the lower exposure category [62-84 dB(A)]. Possible explanations may be a correlation of noise and heavy work in these jobs, which might be protective and also an expression of good physical fitness combined with lower risk for MI. These findings may also be the result of a healthy worker effect.

Table 3b summarizes the ORs and CIs for female subjects in 1-digit ISCO groups. In ISCO-group 1 (legislators and managers) and a noise exposure between 62 dB(A) and 84 dB(A), there is a high (nonsignificant) OR = 2.43 (CI0.95 = 0.12-50.0). This result is based on a low frequency of subjects in this group, but is in agreement with the results observed for men.

Table 3b

Adjusted odds-ratios and asymptotic 0.95-confidence intervals for ISCO-88 groups and noise categories (females): One conditional logit model per row (within each ISCO class), basis of comparison: 46-61 dB(A) noise category of the same ISCO class

ISCO-88	62-84 dB	85-94 dB	Remainders
1	2.43 (0.12-50.0)	—	1.04 (0.32-3.40)
2	1.80 (0.31-10.5)	—	2.82 (0.62-12.8)
3	1.09 (0.58-2.03)	1.17 (0.18-7.44)	0.82 (0.54-1.25)
4	1.02 (0.46-2.26)	—	1.26 (0.87-1.83)
5	0.84 (0.25-2.90)	—	0.66 (0.20-2.20)
6	—	—	—
7	0.60 (0.03-10.4)	—	0.61 (0.04-9.95)
8	0.53 (0.03-9.28)	—	0.55 (0.03-8.99)
9	2.45 (0.63-9.51)	—	1.70 (0.46-6.30)

In ISCO-group 2 and a noise exposure between 62 dB(A) and 84 dB(A), there is a (nonsignificant) heightened OR = 1.80 (CI0.95 = 0.31-10.5). Likewise, this is in agreement with the results analyzed for men in ISCO-group 2. The distribution of occupations is similar to that of the men; more than half of the women in this ISCO group are teachers, physicians, or work in science. Most of the subjects with MI were women who were teaching.

Another noticeable result has been found for the group of women in the ISCO-group 9 and a noise exposure between 62 dB(A) and 84 dB(A). In contrast to men, a high OR = 2.52 (CI0.95 = 0.51-12.4) is calculated. Unlike the men in ISCO-group 9, these women work in professions that are low paid, have low social prestige, and provide less physical requirements, such as cleaning personnel, kitchen helpers, and assistants in health care. The men of ISCO-group 9, however, are mainly engaged in jobs with physical demands typically found in construction industry or logistics.

Discussion

Very high occupational noise exposure is, independent of the ISCO group, a risk factor for MI, which is demonstrated by the significant results in the high exposure category [>95 dB(A)]. Other studies have also demonstrated health effects of noise in industry jobs[7181920] but there is little information about the risk group working in the police department with the need to handle firearms during their work. Lwow et al. describe that in the exposure during shooting training, they observed C-weighted peak sound pressure levels of 138.2-165.2 dB.[21] Wu and Young 2009 also point to the hearing damage effect of shooting training for police officers although they are wearing hearing protection.[22] Thus, the extraaural effects of impulse noise in individuals with frequent handling of firearms should be given more attention, especially because wearing of ear protection during these tasks does not seem to have a sufficient protective effect.

Although most other results were not significant, because of the low case numbers in the particular ISCO groups, we think that those results with an OR greater than or nearly 2 for men or 1.8 for women indicate vulnerable groups and need to be discussed.

The most interesting results, found for men and women, are the higher noise associated risks (even at moderate noise exposure [62-84 dB(A)] for MI in professions that are known to have several other stressors and need to deal with several high demands and responsibilities. This concerns the ISCO groups 11, 22, 32, and 73 for men and ISCO groups 1, 2, and 9 for women. In the groups 1-3 legislators and senior officials, physicians, nurses, and teachers are included.

Also, Malinauskiene et al. found an increased risk of MI for male legislators, senior officials, and managers (ISCO-group 1; OR = 2.18; CI0.95 = 1.59-2.99)[23] in relation to ISCO-group 7, and for female legislators, senior officials, and managers (OR = 2.64)[24] compared to ISCO-group 3. We could not confirm both results with our data; neither men nor women of ISCO-group 1 have a risk difference to the aforementioned reference groups.

The increased risk for MI in health professionals and health associate professionals may be associated with the particular background noise in hospitals[252627] during special tasks, such as surgery.[282930] This is documented in numerous publications. Siegmann and Notbohm[31] summarize: “During operation sessions lasting from 30 min to several hours, reported average Leq values ranging from 58 dB(A) to 72 dB(A) with maximum levels above 105 dB(A).” Similar noise levels are reported from emergency departments. Since concentration, precise communication, and fast decisions are necessary in these situations, this has to be considered an enormous strain for the staff and a potential risk with regard to faults at work. An association of stressors during these tasks and cardiovascular reactivity are described.[3233]

Teachers also seem to be a vulnerable group, since they feel very annoyed by noise in the classroom,[34] which may intensify effects by other stressors.

A lower risk for MI in association with noise has been calculated for men, who work in jobs with high physical demands (ISCO-groups 7-9). On the one hand, high physical demands are associated with a higher risk for IHD[353637] but on the other hand, high physical fitness was found to reduce the risk for IHD mortality among men with high physical demands.[38] There may perhaps be selection processes, leaving those with high physical fitness and/or high ability to cope with different stressors in these types of noisy jobs. An exception are workers of the ISCO-group 73, especially those working in the printing industry.

Women in the ISCO-group 9 show a higher noise associated risk for MI, which is in contrast to the results for men in this group. But other types of professions are included here, e.g., the cleaners or staff helping in the kitchen and in health care. For example, cleaners seem to have a high cardiovascular load at work and low cardiorespiratory fitness;[39] this load may be increased by noise.

With the proxy of the occupational group (described by the ISCO classification) as one approach to look more precisely on noise-exposed subjects and their specific work tasks, it is possible to identify groups with high risk. This provides a basis for new research projects.

As a precondition for possible further studies we calculated the numbers of cases and controls that need to be investigated within a specific group. As shown in Table 4 for a basic study design,[40] such studies can be performed with a reasonable effort. With a 1:2 matching, studies may be performed with 100-150 cases for a proportion of 15-50% of the higher noise category. A better power can be achieved using metrical exposure data.

Table 4

Approximate sample sizes of cases, 1: M matched case-control study with dichotomous exposure (ORalternative = 2.0; α = 0.05, two-sided test; β = 0.20)

Proportion of higher noise in the control group [%]
M	5	10	15	20	25	30	35	40	45	50
1	516	283	208	172	152	141	135	133	133	137
2	371	205	151	126	112	105	101	99	100	103
3	322	179	132	111	99	92	89	88	89	92
4	298	165	123	103	92	86	83	82	83	86

With the concentration of the analysis on subjects working in groups with similar job characteristics, it is possible to avoid the effects of different noise levels in different job groups from being mixed up. In summary, by investigating and analyzing the association between noise and cardiovascular disease in specific groups with their work tasks and special challenges, future research may come to more consistent and reliable results.