Behavioural text message reminders to improve participation in breast screening: a randomised controlled trial

Abstract

BackgroundBreast screening coverage in London, UK, consistently falls below the national average (69·3% vs 75·6% in 2016). West London has among the lowest coverage rates nationally. Text message reminders (SMS-R) can improve screening rates. Evidence shows that message content can further improve the impact of SMS-R on hospital appointment attendance. This randomised controlled trial aimed to test the effect of SMS-R message manipulations on breast screening participation. MethodsWomen aged 47–73 years in the West of London Breast Screening Service (WoLBSS) catchment area, who were due for screening between Aug 10, 2016, and Nov 2, 2016, and who had provided their mobile number to the service, were eligible. Two intervention messages were tested against the current SMS-R (appendix). A survey of 1000 women informed the content, which indicated that women who prioritised or planned to attend breast screening were more likely to attend. Participants were randomised to control, planning, and priority arms (1:1:1) by the SMS provider platform iPLATO using a Mersenne twister algorithm and sent two SMS-R before their appointment. Participants and the routine care team, but not researchers, were masked to allocation. The primary outcome was uptake of screening by trial arm. Analysis was by intention to treat. Ethics approval was obtained from IRAS (IRAS project ID 185492, REC reference 16/YH/0276). This trial is registered with ClinicalTrials.gov, number NCT02872363. Findings2696, 2614, and 2634 women were allocated to the control, planning, and priority arms, respectively (mean age 58·8 years, SD 6·09), with 854, 883, and 860 women analysed after exclusions for opt-outs, self-referrals, and SMS-R cross-contamination. Uptake was 82·8% (n=707), 84·4% (745), and 82·7% (711), respectively (χ2 [df 2]=1·131, p=0·568). No adverse events were observed. InterpretationAlthough no statistical difference between trial arms was noted, the target sample size (11 496) was not achieved because of a reconfiguration of services prematurely ending the trial. A large number of participants received different SMS-R messages or more than two SMS-R in total, probably because of appointment rescheduling, which re-entered women into the randomisation process. These participants were excluded. Owing to information governance regulations, women who provided their mobile number to their general practitioner but not directly to WoLBSS could not be sent an SMS-R. Consequently, the sample is likely to represent highly motivated women, reflected in the high participation rate. Therefore, first-time attenders and non-attenders may be disadvantaged by SMS-R interventions for breast screening. FundingNational Institute for Health Research Imperial Patient Safety Translational Research Centre and the Association of Breast Surgeons and Prevent Breast Cancer.