Gan et al, 2011 [1] concluded that long-term, occupational noise exposure was associated with increased prevalence of coronary heart disease (CHD), for which the authors report a clear exposure-response relationship that was particularly strong for participants aged < 50 years, men and current smokers. We do not believe the results support these conclusions, particularly in light of notable study limitations.

The cross-sectional design is a significant drawback. Self-reported exposure and outcome data created uncertainty, regarding time ordering of events. Additionally, exposed subjects had many well-known risk factors - representing strong confounders - for cardiovascular disease (CVD)/CHD, the effects of which cannot be entirely removed by statistical adjustment. Moreover, unrecognized bias was likely introduced by including potential causal intermediates and over-adjusting logistic models,[2] for which a mechanistic basis is not well established. Although the authors dismissed the presence of misclassification as non-differential, simulations have shown this argument to be less persuasive.[3]

We raise additional concerns. First, heterogeneity of effect (Figure 2) precluded reporting summary estimates. It was incorrect to state, for example, that an increased odds of CHD were observed particularly for men when they were observed only for them. Second, the authors misapplied the test for trend of a monotonic exposure-response relationship.[4] Finally, potentially important effect patterns were obscured by aggregating outcomes to ascertain CVD/CHD prevalence. CVD, the primary endpoint of interest and the aggregate of six variables, was not discussed. The significant association reported for CHD was largely driven by the relatively strong association observed only for angina pectoris.

Given the prevalence of occupational noise exposure, even a modest association would represent an important contributor to the development of CVD/CHD. Unfortunately, Gan et al [1] does little to advance our understanding of whether such an association exists, particularly because the mechanism of action is so poorly understood. We agree with the authors that cohort studies are warranted. These studies should use objective exposure and outcome measurements.

References

1 Gan WQ, Davies HW, Demers PA. Exposure to occupational noise exposure and cardiovascular disease in the United States: the National Health and Nutrition Examination Survey 1999-2004. Occup Environ Med 2010;68:183-190

2 Schisterman EF, Cole SR, Platt RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology 2009;20:488 -495

3 Jurek AM, Greenland S, Maldonado G. How far from non-differential does exposure or disease misclassification have to be to bias measures of association away from the null? Int J Epidemiol 2008; 37:382-5

4 Maclure M, Greenland S. Tests for trend and dose response: misinterpretations and alternatives. Am J Epidemiol 1992;135:96-104

Conflict of Interest:

None declared