Abstract
BACKGROUND: In a previous study, we demonstrated that military aircraft noise increases the short-term probability of on-demand drug administrations in patients in a psychiatric clinic located in close proximity to a military airfield. The aim of this article was to: (1) apply refined exposure assessment methods; (2) compare results for various noise exposure measures reflecting different assumptions about outdoor-indoor noise attenuation and explore possible effect thresholds; and (3) conduct an independent validation study with 1 year of more recent follow-up. METHODS: We conducted noise measurements during 1 year at the clinic to validate the modeled noise exposure of hourly military aircraft noise (equivalent continuous sound pressure level, L (Aeq) [dB]). We then compared results from case-time series analyses between on-demand drug administration and military aircraft noise using modeled versus measured aircraft noise exposure, different censoring approaches for background noise, and explored a potential threshold using nonlinear exposure-response modeling. RESULTS: There was a moderate (Pearson's r = 0.54) to strong (r = 0.65) correlation between modeled and measured aircraft noise levels, depending on the wind situation. Applying censoring at levels reflecting background noise in the range of 40-50 dB instead of censoring exposure at 20 dB resulted in higher effect estimates, but also larger uncertainty compared with the previous analysis (original analysis, censoring at 20 dB: odds ratio [OR] for sedative administration = 1.016 per 10 dB; realistic background analysis: OR = 1.036, 95% confidence interval [CI]: 0.994, 1.079). Additionally, we found indications of a nonlinear exposure-response relationship. Using measured instead of modeled aircraft noise exposure resulted in smaller effect estimates (modeled OR = 1.056, 95% CI: 0.970, 1.149; measured: 1.038, 95% CI: 0.983, 1.096). Analyses exclusively using new data from the year 2022 revealed almost identical results compared with the original analysis (2016-2021). CONCLUSION: Our results suggest that predicting indoor noise levels and the addition of informed background noise estimates improves specificity in the exposure assessment and likely provides more reliable exposure-response associations. With additional follow-up data, we could confirm our previous findings that loud noise events can have acute effects on psychiatric patients' sedative drug consumption.