Astrid L Martens1, Pauline Slottje2, Marie Y Meima3, Johan Beekhuizen3, Danielle Timmermans4, Hans Kromhout3, Tjabe Smid5, Roel C H Vermeulen6. 1. Institute for Risk Assessment Sciences (IRAS), Division Environmental Epidemiology, Utrecht University, Yalelaan 2, 3584 CM Utrecht, The Netherlands; Department of Public and Occupational Health, EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands. Electronic address: a.l.martens@uu.nl. 2. Institute for Risk Assessment Sciences (IRAS), Division Environmental Epidemiology, Utrecht University, Yalelaan 2, 3584 CM Utrecht, The Netherlands; Department of General Practice and Elderly Care, EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands. 3. Institute for Risk Assessment Sciences (IRAS), Division Environmental Epidemiology, Utrecht University, Yalelaan 2, 3584 CM Utrecht, The Netherlands. 4. Department of Public and Occupational Health, EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands; National Institute for Public Health and the Environment (RIVM), PO Box 1, 3720 BA Bilthoven, The Netherlands. 5. Department of Public and Occupational Health, EMGO+ Institute for Health and Care Research, VU University Medical Center, Amsterdam, The Netherlands; KLM Health Services, Schiphol, The Netherlands. 6. Institute for Risk Assessment Sciences (IRAS), Division Environmental Epidemiology, Utrecht University, Yalelaan 2, 3584 CM Utrecht, The Netherlands; Imperial College, Department of Epidemiology and Public Health, London, United Kingdom; Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht (UMCU), Utrecht, The Netherlands.
Abstract
INTRODUCTION: Geospatial models have been demonstrated to reliably and efficiently estimate RF-EMF exposure from mobile phone base stations (downlink) at stationary locations with the implicit assumption that this reflects personal exposure. In this study we evaluated whether RF-EMF model predictions at the home address are a good proxy of personal 48h exposure. We furthermore studied potential modification of this association by degree of urbanisation. METHOD: We first used an initial NISMap estimation (at an assumed height of 4.5m) for 9563 randomly selected addresses in order to oversample addresses with higher exposure levels and achieve exposure contrast. We included 47 individuals across the range of potential RF-EMF exposure and used NISMap to re-assess downlink exposure at the home address (at bedroom height). We computed several indicators to determine the accuracy of the NISMap model predictions. We compared residential RF-EMF model predictions with personal 48h, at home, and night-time (0:00-8:00AM) ExpoM3 measurements, and with EME-SPY 140 spot measurements in the bedroom. We obtained information about urbanisation degree and compared the accuracy of model predictions in high and low urbanised areas. RESULTS: We found a moderate Spearman correlation between model predictions and personal 48h (rSp=0.47), at home (rSp=0.49), at night (rSp=0.51) and spot measurements (rSp=0.54). We found no clear differences between high and low urbanised areas (48h: high rSp=0.38, low rSp=0.55, bedroom spot measurements: high rSp=0.55, low rSp=0.50). DISCUSSION: We achieved a meaningful ranking of personal downlink exposure irrespective of degree of urbanisation, indicating that these models can provide a good proxy of personal exposure in areas with varying build-up.
INTRODUCTION: Geospatial models have been demonstrated to reliably and efficiently estimate RF-EMF exposure from mobile phone base stations (downlink) at stationary locations with the implicit assumption that this reflects personal exposure. In this study we evaluated whether RF-EMF model predictions at the home address are a good proxy of personal 48h exposure. We furthermore studied potential modification of this association by degree of urbanisation. METHOD: We first used an initial NISMap estimation (at an assumed height of 4.5m) for 9563 randomly selected addresses in order to oversample addresses with higher exposure levels and achieve exposure contrast. We included 47 individuals across the range of potential RF-EMF exposure and used NISMap to re-assess downlink exposure at the home address (at bedroom height). We computed several indicators to determine the accuracy of the NISMap model predictions. We compared residential RF-EMF model predictions with personal 48h, at home, and night-time (0:00-8:00AM) ExpoM3 measurements, and with EME-SPY 140 spot measurements in the bedroom. We obtained information about urbanisation degree and compared the accuracy of model predictions in high and low urbanised areas. RESULTS: We found a moderate Spearman correlation between model predictions and personal 48h (rSp=0.47), at home (rSp=0.49), at night (rSp=0.51) and spot measurements (rSp=0.54). We found no clear differences between high and low urbanised areas (48h: high rSp=0.38, low rSp=0.55, bedroom spot measurements: high rSp=0.55, low rSp=0.50). DISCUSSION: We achieved a meaningful ranking of personal downlink exposure irrespective of degree of urbanisation, indicating that these models can provide a good proxy of personal exposure in areas with varying build-up.
Authors: Berihun M Zeleke; Christopher Brzozek; Chhavi Raj Bhatt; Michael J Abramson; Rodney J Croft; Frederik Freudenstein; Peter Wiedemann; Geza Benke Journal: Int J Environ Res Public Health Date: 2018-10-12 Impact factor: 3.390
Authors: Raquel Ramirez-Vazquez; Jesus Gonzalez-Rubio; Isabel Escobar; Carmen Del Pilar Suarez Rodriguez; Enrique Arribas Journal: Int J Environ Res Public Health Date: 2021-02-14 Impact factor: 3.390