OP XI – 4 A comparison of multiple noise indicators for the assessment of short-term associations between personal exposure to noise and heart rate variability: a sensor-based study in a real-life setting

Abstract

Background/aim Part of the complexity of noise assessment resides in its temporal variations as so, it is still unclear which indicators of exposure are appropriate. This study aims at (1)describing the concomitant associations between personal noise exposure summarised using various noise indicators and heart rate variability (HRV) parameters (2)identifying the noise indicators that best predict HRV parameters. Methods The RECORD MultiSensor Study collected between July 2014 and June 2015 noise and heart rate (HR) data for 75 participants, aged 34–74 years, in their living environments for 7 days using an electrocardiography sensor on the chest and a personal dosimeter recording 1 s A-weighted equivalent sound pressure levels (L Aeq,1s ). HRV parameters and noise levels were calculated for 5 min windows. Noise was summarised as L Aeq , L AX (noise level exceeded X% of the time) with L A90 , L A50 , L A10 , L A01 , L A10 -L A90 and the standard deviation of L Aeq,1s . Short-term associations of noise level and HRV parameters were assessed using mixed effects models with a random intercept for participants adjusted for HR, accelerometry, context and short-term trends. The models’ goodness of fit was assessed using the BIC. Results The classically used indicator, the LAeq, was highly correlated with L A10 (r=0.94) and L A01 (r=0.97) and moderately with L A90 (r=0.66) and L A10 -L A90 (r=0.52). All of the noise indicators, when examined in separate models, were positively associated with the Standard deviation of N-N intervals (SDNN) and with the Low frequency on High frequency HRV ratio (LF/HF), with the exception of the association between the L A90 and SDNN which decreased by −0.17% (95% CI: −0.21 to −0.13) per increase of one dB(A). Based on the BIC, the noise indicators that best predicted HRV parameters were the standard deviation of L Aeq,1s for the SDNN followed by L A10 -L A90 (ΔBIC=221.66) and L A01 for the LF/HF ratio followed by L A10 (ΔBIC=190.36). Conclusion The results suggest that short-term effects of noise on overall heart rate variability (SDNN) are better explained by the amplitude of noise level variations (σ LAeq,1s ), while those on the balance of the autonomic nervous system (LF/HF) are better explained by sporadic noise events (L A01 ). In addition a negative association between the background noise level (L A90 ) and SDNN was found.