Towards a representative and affordable national survey on radon in dwellings: a new methodological approach with findings from Italy

National surveys on radon concentration in dwellings that are representative of population exposure are recommended by international regulations but are often difficult and costly to carry out. Furthermore, evaluation of sample representativeness is usually not reported, making it impossible to quantify or correct potential biases. This study aimed to develop and test, in a real-world setting, methods to plan and implement a cost-effective national survey on radon concentration in dwellings that includes a large proportion of municipalities, allows for the assessment of representativeness, and enables bias correction when the sample is not fully representative. A comprehensive national survey was conducted efficiently and at low cost in collaboration with a national company with numerous worksites. The company provided: i) a list of all employee dwelling addresses, enabling random selection across all 110 Italian provinces; ii) internal email and web platforms, for communication with workers and completion of questionnaire data, iii) internal mail service, for distributing and retrieving radon detectors which were centrally prepared and read out. Moreover, questionnaire data on dwelling characteristics were collected and compared with census data to assess representativeness and adjust for potential biases. Valid radon measurements were obtained for 5,354 dwellings. After correcting for two major identified biases, the national Italian arithmetic mean was 79 Bq/m 3 with values ranging from 34 to 132 Bq/m 3 across the 21 Regions and from 25 to 343 Bq/m 3 across the 110 Provinces. Without corrections, the national mean would have been underestimated by about 10%, regional means by up to 30%, and even more at the provincial level. The results demonstrate that a national survey can be conducted effectively and affordably, using methods that can be adopted by other countries. Furthermore, the study highlights the importance of assessing representativeness to avoid significant biases.