A real-world quality assessment study in six ExpoM-RF measurement devices.

Exposimeters measuring radiofrequency electromagnetic fields (RF-EMF) are commonly used to assess personal exposure to RF-EMF in real-life environments. They are usually calibrated in an anechoic chamber using single, well-defined signals such as the center frequency of each band, and standardized orientations, but it is not clear how different devices compare in the real environment where complex mixtures of signals from all directions are present. We thus tested the comparability of six ExpoM-RF exposimeters before and after calibration in an anechoic chamber by varying their position and orientation while repeatedly measuring 15 microenvironments (9 walking routes, 4 tram routes and 2 bus routes) on 6 different days. We modelled the geometric mean levels of RF-EMF as a function of orientation, position, device ID, whether the device was recently calibrated, correcting for the microenvironment in which each measurement took place. We found that systematic differences introduced by device ID, calibration, day of the week, orientation and position are relatively small compared to exposure differences between microenvironments. Any corrections (if desired) should include both device ID and calibration session, but would have a small impact considering the negligible differences between devices. This supports the validity of previous exposure measurement studies relying on ExpoM-RF devices, which did not correct for device ID. We further found that summarizing the exposure per microenvironment as geometric means results in better models than arithmetic means and medians, and recommend that further exposure assessment studies report observed levels as geometric means.