ABSTRACT
Background: Germ Defence (www.germdefence.org) is an evidence-based interactive website that promotes behaviour change for infection control within households. To maximise the potential of Germ Defence to effectively reduce the spread of COVID-19 the intervention needed to be implemented at scale rapidly. Methods: With the approval of NHS England, we implemented an efficient two-arm (1:1 ratio) cluster randomised controlled trial (RCT) to examine the effectiveness of randomising implementation of Germ Defence via GP practices across England, UK, compared with usual care. GP practices randomised to the intervention arm (n=3292) were emailed and asked to disseminate the Germ Defence intervention to all adult patients via mobile phone text, email or social media. GP practices randomised to the usual care arm (n=3287) maintained standard management for the 4-month trial period after and then asked to share Germ Defence with their adult patients. The primary outcome was the rate of GP presentations for respiratory tract infections (RTI) per patient. Secondary outcomes comprised rates of acute RTIs, confirmed COVID-19 diagnoses, suspected COVID-19 diagnoses, COVID-19 symptoms, gastrointestinal infection diagnoses, antibiotic usage, hospital admissions. The impact of the intervention on outcome rates was assessed using negative binomial regression modelling within the OpenSAFELY platform. The uptake of intervention by GP practice, and by patients were measured via website analytics. Results: Germ Defence was used 310,731 times. The average satisfaction score after using the website was 7.52 (0-10 not at all to very satisfied, N = 9933). There was no evidence of a difference in the rate of RTIs between intervention and control practices (rate ratio (RR) 1.01, 95%CI 0.96, 1.06, p=0.70). This was similar to all other eight health outcomes. Patient engagement within intervention arm practices ranged from 0- 48% of a practice list. Practices with high levels of intervention uptake (>11%) had a lower proportion of minority ethnic groups. Conclusions: We demonstrated that rapid large-scale implementation of a digital behavioural intervention can be evaluated with a novel efficient prospective RCT methodology analysing routinely collected patient data entirely within a trusted research environment. Trial registration: This trial was registered in the ISRCTN registry (14602359) on 12 August 2020.