IEA_Radon_Reporter_January_2024r

20 | January 2024 RESEARCH RESEARCH Behaviourally-Informed Household Communications Increase Uptake of Radon Tests in a Randomised Controlled Trial Shane Timmons & Peter D. Lunn (Excerpt as published in Nature.com / Scientific Reports / November 21, 2023 For the full paper including letters used in the trial, please visit https://rdcu.be/du5FN Sci Rep 13, 20401 (2023). Received 17 August 2023. Accepted 19 November 2023. Published 21 November 2023.) We report a pre-registered randomised controlled trial of communications to encourage test uptake. Households (N= 3500) in areas at high risk of radon exposure were randomly assigned to receive (i) a the control letter from the national Environmental Protection Agency; (ii) a behaviourally-informed version of the control letter that incorporated multiple nudges, including reciprocity messages and numeric frequencies of risk; (iii) this same behaviourally-informed letter in a re-designed envelope; (iv) the behaviourally-informed letter in the re-designed enveloped with a radon risk map of the household’s county. The behaviourally-informed letter led to a large increase in test uptake, from 22% in the control condition to 33% (a 50% increase). There was no additional benefit of the re-designed envelope, which generated uptake of 30%. Including the map led some households to respond faster, but the overall uptake (26%) was weaker. The results have implications for public health communications with households and show the potential for techniques from behavioural science to help mitigate environmental risks. As a result, most previous attempts to increase radon testing rates have had limited success. Taking randomised controlled trials (RCTs) as the best evidence, few tests of ways to communicate with householders have observed significant increases in testing rates and none have generated rates above 20% among those who hadn’t already decided to order a test. Weinstein, Sandman and Roberts (1991) issued letters to householders and varied whether they contained general advice to test or specific information about high levels of radon in their area. Results showed no difference in test kit orders. Weinstein, Lyon, Sandman and Cuite (1998) issued videos to householders that either emphasised the risks of radon, how easy it is to test for radon, or both. Compared to a control group order rate of 5.1% among those who had not previously decided to test, 18.7% of those who received the combination video ordered a test kit. However, the authors note the length of the video and costs of production limit the scalability of the intervention. Another (non-randomised) evaluation of costly awareness campaigns showed positive effects, with approximately one quarter of households in participating areas having tested for radon versus less than 15% of those in non-participating areas. Other trials that have shown positive effects have relied on rating-scale intentions rather than measuring behaviour or on small, non-representative samples. Thus no trials to date have observed testing rates above 20% using low-cost, scalable interventions. Our approach differs from previous trials and follows recent calls from experts to apply rigorous methods from social science to radon communications. The Potsdam Radon Communication Manifesto emphasises the need for science-based communication strategies that rely on theory and evidence in place of gut-feelings on what may work, for collaboration between policymakers and social scientists and the use of interactive risk maps to bolster personalised communications. McLaughlin, Gutierrez-Villanueva and Perko (2022) stress the need to address deficits in radon risk communication as a priority for National Radon Control Strategies. A recent systematic review of radon communication campaigns highlights substantial deficits in the methods used to evaluate campaigns, the need to pre- test communications and the absence of behaviourally-informed interventions (or “nudges”). We worked collaboratively with Ireland’s Environmental Protection Agency (EPA) to inform postal communications issued to householders using behavioural science and test these new communications via an RCT. Read the full paper to better understand the author’s approach which differs from previous trials and follows recent calls from experts to apply rigorous methods from social science to radon communications. Exposure to radon – and the associated risk of lung cancer – is increasing in multiple countries. Household testing is hence a public health priority, but the lack of sensory cues from radon means most people don’t consider radon to be a risk. Indeed, previous RCTs of direct communications with householders have failed to generate testing rates above 20%. We therefore combined multiple behavioural interventions into one BI letter issued to householders in high radon areas. The effect of combining interventions in this way was large. We observed an almost 50% (11%-point) increase in test uptake among those who received a BI letter, bringing the uptake rate to roughly 1-in-3. This effect is much larger than previous radon testing trials16 and the typical effect from interventions run by “Nudge Units”. Although we can merely speculate, the size of the effect is perhaps driven by the combination of multiple behavioural levels into one intervention (e.g., reciprocity, numeric frequencies of risk, etc.) in a simple-to-read format, rather than relying on lengthy videos or singular levers. If all households in our 3500-home trial were issued our BI letter, 300 more homes in high radon areas would have ordered radon test kits over the trial period. The simplicity of the intervention – mere changes to a standard letter – means it is cheap to scale and is readily adaptable for testing in other countries. While the strength of the effect for policymakers is encouraging, identifying the mechanism underlying it is not straightforward. The effect could have been driven by one feature (e.g., merely simplifying the text), two features (e.g., simplification and using numeric frequencies), three features, or all features (simplification, numeric frequencies, reciprocity, endowment and urgency). It’s also possible that one or more features had a negative effect, and the effect size of combining fewer features would have been even larger.