P16-7 Reducing health inequalities in endoscopy attendance rates through non-attendance prediction and healthcare navigation: a mixed methods study

Abstract

<h3>Background</h3> Targeting 20% most deprived communities with clinical early-stage cancer diagnosis focus can reduce patient harm (Core20PLUS5 Framework). DNA (Did Not Attend) appointments impact patient health outcomes in cancer diagnostic tests; treatment delays, additional re-arranging resources and potentially more severe diagnosis/prognosis. A multi-disciplinary team developed a predictive risk model, identifying DNA high-risk patients due to health inequalities and piloted patient healthcare navigation. <h3>Methods</h3> Machine learning model trained on 2-years of 39,447 endoscopy appointments data identified endoscopy high-risk DNA patients based on risk factors including health inequalities. Probability factors contributing, positively or negatively, include; attendance behaviours, demographic factors; deprivation, age, gender, and procedure type. Variables such as prior DNAs, rurality, and socio-economic deprivation were among the most predictive. Over 12-weeks patients with 10% or greater probability of DNA received pre-appointment healthcare navigation calls; understanding and helping mitigate attendance barriers. <h3>Results</h3> Pre-appointment calls were made between 7 October 2024 and 17 January 2025. 540 predicted high-risk DNA patients were contacted (78% engagement rate). Themes identified: forgetting, anxiety and feeling un-informed. Baseline DNA rate, (same time period, previous year) estimated 17% for ‘high-risk’ patients. Pilot results: ‘high-risk’ patient DNA rate fell to 5% for patients contacted (11% including patients unable to contact). Endoscopy patient text reminders paused March 2024 resuming November 2024, 3.5 weeks into pilot. Pilot observed 52 DNAs from 650 appointments. Scaling data up over a year, assuming 10.8% DNA rate consistent, estimated preventing 150 DNAs; calls alone generate 10% DNA reduction for ‘high-risk’ patients and 20% reduction with reminder text. <h3>Conclusion</h3> Targeted pre-appointment patient calls for those predicted DNA high-risk are associated with statistically significant reduction in the proportion of DNAs. The work has scaled up; further 5 service areas developing predictor models. This small-scale study’s findings are limited to NHS trust endoscopy service in the Northeast of England; further formalised research is necessary.

Ähnliche Arbeiten