Format

Send to

Choose Destination
J Acoust Soc Am. 2017 May;141(5):3175. doi: 10.1121/1.4982922.

Noise exposure-response relationships established from repeated binary observations: Modeling approaches and applications.

Author information

1
Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory for Acoustics/Noise Control, Überlandstrasse 129, 8600 Dübendorf, Switzerland.
2
German Aerospace Center (DLR), Institute of Aerospace Medicine, 51147 Cologne, Germany.
3
Department of Psychiatry, Unit for Experimental Psychiatry, Division of Sleep and Chronobiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania 19104-6021, USA.
4
Federal Office for the Environment, Noise and NIR Division, 3003 Bern, Switzerland.

Abstract

Noise exposure-response relationships are used to estimate the effects of noise on individuals or a population. Such relationships may be derived from independent or repeated binary observations, and modeled by different statistical methods. Depending on the method by which they were established, their application in population risk assessment or estimation of individual responses may yield different results, i.e., predict "weaker" or "stronger" effects. As far as the present body of literature on noise effect studies is concerned, however, the underlying statistical methodology to establish exposure-response relationships has not always been paid sufficient attention. This paper gives an overview on two statistical approaches (subject-specific and population-averaged logistic regression analysis) to establish noise exposure-response relationships from repeated binary observations, and their appropriate applications. The considerations are illustrated with data from three noise effect studies, estimating also the magnitude of differences in results when applying exposure-response relationships derived from the two statistical approaches. Depending on the underlying data set and the probability range of the binary variable it covers, the two approaches yield similar to very different results. The adequate choice of a specific statistical approach and its application in subsequent studies, both depending on the research question, are therefore crucial.

PMID:
28599544
DOI:
10.1121/1.4982922
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for American Institute of Physics
Loading ...
Support Center