Epidemiologist Andrei Pyko and colleagues say so, and their new study has garnered the usual press attention. But the claim is false. Their study wasn’t even designed to test this hypothesis, so its publication indicates a failure of scholarly peer review
The study by Pyko et al., published in BMJ Occupational & Environmental Medicine, claims that every increase of 5 dB(A) increases waist circumference by 0.21 cm. The strength of this association is weak (odds ratio: 1.18 for those exposed to > 45 dB(A)), as is its calculated statistical significance (the lower 95th percentile confidence interval is 0.01). And there is no association between noise and BMI, the more commonly used index of obesity. That alone should have tipped off the authors that the association they found not biologically meaningful.
Nevertheless, they persisted, concluding:
Our results suggest that traffic noise exposure can increase the risk of central obesity. Combined exposure to different sources of traffic noise may convey a particularly high risk.
Even if the authors’ results were stipulated to be correct, this conclusion is impossible given the study’s cross-sectional research design. The most that a cross-sectional design could ever show is that people who live in noisy locations are more likely to be obese. No matter how strong the association, this design cannot show that noise causes anything.
Any competent peer reviewer should have detected this fundamental error and advised that the manuscript be rejected as written. Meanwhile, many press reports have been published, including by The Independent, Daily Mail, CBS News, and The Atlantic, appealing to the authority of a scholarly journal whose peer review failed. Some of these reports seem to welcome the results of the study because it supports a preferred policy agenda.