Author's reply

Sanford Fidell 
 Fidell Associates, Inc., 23139 Erwin Street, Woodland Hills, California, 91367, USA

Correspondence Address:
Sanford Fidell
Fidell Associates, Inc., 23139 Erwin Street, Woodland Hills, California, 91367
USA

Full Text

Sir,

Basner et al.[1] indicate that they disagree more strongly with matters of interpretation than with the substantive findings of the Fidell et al.[2] review of methods for predicting noise-induced sleep disturbance. One major interpretive disagreement may, as Basner et al. hint, concern the definition of health. The World Health Organization's absolutist definition of health (not merely the absence of disease, but complete mental and physical well-being) is one that implicitly categorizes large segments of the world's urban population as less-than-healthy. After all, whose mental well-being (and by implication, health) could not be made more complete by living in a world freed by regulation from all supposed noise hazards? All regulation comes at a cost, however. Those who recommend regulation to reduce putative health risks of noise-induced sleep disturbance by unknown amounts must make the case that the cost of regulation is commensurate with the risk.

Additional interpretive disagreements may be due at root to differences in societal rather than technical perspectives on risk and the role of regulation in public health. A Eurocentric view [as typified in Ising and Kruppa's [3] assertion that "In the case of preventive health protection…any reasonable assumption of a possible health hazard justifies protective measures"] may err in the direction of a lesser standard of proof of the value of regulation, while the North American view may err in the opposite direction. Similarly, the conceit that no citizen should suffer a health risk of any magnitude - however different his or her ability to adapt to noise-induced sleep disturbance in urban settings may be from that of most other citizens - may also be a product of societal preferences rather than of technical considerations.

The health risks of noise-induced sleep disturbance are not usefully analogized to those of smoking, as Basner et al. insinuate ["the fact that there are other reasons for lung cancer does not mean smoking does not substantially increase its risk"]. This is not a matter of quibbling over the meaning of morbidity or mortality statistics, but rather a matter of the near-complete absence of any quantitative information on the health risks of noise-induced sleep disturbance. No evidence exists that nighttime noise intrusions from noise sources potentially amenable to regulation substantially increase the health risks (if any) of sleep disturbance over the risks (if any) of disturbance by indoor noise sources, or by the myriad other factors that affect sleep quality.

Basner et al. agree that no evidence exists of a direct link between cardiovascular disease and noise-induced awakening. Nonetheless, they cite a study that suggests a link between obstructive sleep apnea and "cardiovascular endpoints", and note that sleep fragmentation is believed to contribute to an elevated risk of such endpoints. This interpretation of a potential health risk of noise-induced sleep disturbance is a tenuous one that does not address the relative contributions of noise-induced sleep disturbance and other causes of sleep fragmentation to "cardiovascular endpoints". The interpretation also does not address whether such endpoints, by themselves, constitute meaningful health risks, in the context of all of the other risks which people routinely and voluntarily incur, but for which regulation is not a plausible palliative.

The observation that detectable cardiovascular responses to noise in sleeping quarters do not fully adapt likewise has few clear health implications. Transient changes in pulse rate and blood pressure occur normally and routinely throughout the day and night. They may be no more than harmless indications that the auditory system does not, for evolutionarily useful reasons, shut down during sleep.

Although Basner et al. suggest "nature … seems to favor continuous sigmoidal functions" over unambiguous inflection points in dosage-response relationships, it may be analysts who favor them more (cf. Fidell and Silvati [4] for discussion of the arbitrariness of statistical assumptions underlying development of dosage-effect relationships for population-level noise effects.) A relationship with a shallow slope, whether linear or sigmoidal, is in any event of little practical use for regulatory purposes. Much of the remaining statistical observations made by Basner et al. support the view that behaviorally confirmed awakening is too crude a measure to serve as a useful index of sleep disturbance. This could well be the case from some perspectives. The opposite argument could equally well be made, however: that covert, short-term physiological activity is an excessively sensitive, costly and inconvenient to measure, poorly interpretable, and non-persuasive measure for justifying societal action such as regulation.

The notion that "experts in the field can define a decision set … for the definition of future limit values" for political consideration may also reflect differing societal perspectives. When uncertainty about appropriate measures, dosage-effect relationships, quantitative health risks, and epidemiologic evidence is great, even experts may not be able to offer much useful political guidance. Consensus recommendations reached by committees of researchers in such circumstances are often of the least-common-denominator variety; or worse yet, strongly influenced by the views of individual researchers. Thus, for example, Lnight values that may seem appropriate for regulatory consideration in one political jurisdiction may provide meager pragmatic guidance in another.

References