Validity of aircraft noise induced awakening predictions

• Autores: M. Basner
• Localización: Noise Control Engineering Journal, ISSN 0736-2501, Vol. 57, Nº. 5, 2009, págs. 524-535
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Aircraft noise disturbs sleep and impairs recuperation. Integrative noise measures like LDN or Lnight are commonly used to implicitly describe and limit nocturnal aircraft noise effects. However, noise exposures may differ in their sleep disturbing potential but nevertheless calculate to the same LDN or Lnight. Therefore, it has recently become more popular to use exposure-response relationships for single noise events to explicitly predict the degree of sleep fragmentation for a whole night with multiple noise events. These models have been criticized for not accounting for dependence between consecutive noise events, for disregarding the placement of noise events within the night, for disregarding other important acoustic parameters beside maximum sound pressure level (SPL) or single event level (SEL), and for not considering inter-individual differences in noise sensitivity. In this analysis, a Basic Model containing maximum SPL as the only explanatory variable and an Extended Model containing elapsed sleep time, time between maxima, and noise duration as additional variables are compared according to bias and precision in predicting the number of noise induced awakenings in single nights. Random subject effect logistic regression based on field study data, where the reaction to 16279 aircraft noise events was monitored in 64 subjects and 479 subject nights, was used for both models. The results indicate that a variable should only be included in the prediction model if (a) the variable has a relevant impact on sleep, (b) information on the distribution of the variable in the target population is available, and (c) the distribution of the variable in the target population differs relevantly from the distribution in the population that was used to generate the prediction model.