BT08 Real world variability in teledermatologist discharge rates in AI-enabled urgent suspected cancer pathways

Abstract

Teledermatology is increasingly used in urgent suspected cancer (USC) pathways. However, there remains no standardized approach to monitoring and little is known about practice variation in real-world settings. The aim of this study was to describe variation in teledermatologist discharge rates within USC pathways and to explore factors associated with this variability. We conducted a retrospective service evaluation of 11 AI-enabled USC pathways. Teledermatologists included in the analysis reviewed all cases that had not been discharged after AI assessment with a clinician second-read review. Teledermatologists who assessed ≥ 50 cases between September 2020 and August 2024 were included. Statistical analyses included Spearman correlation and variance analysis to distinguish between interclinician and interorganizational variability. Additional sensitivity analyses were conducted across different case volume thresholds. The primary analysis was based on 64 827 case assessments made by 124 teledermatologists. Discharge rates varied significantly, ranging from 0.0% to 54.1% (mean 19.5%, SD 13.0). Substantial variability remained within individual organizations, where clinicians review the same referral pool: the median within-­organization range was 19.2 percentage points (maximum 52.9%), with a mean within-organization SD of 7.8%. The number of cases assessed did not correlate with discharge behaviour (Spearman’s ρ = −0.01, P = 0.89), which remained consistent across sensitivity analysis thresholds. This multicentre service evaluation demonstrates significant and previously unreported variability in teledermatologist discharge rates. Substantial within-organization differences highlight interclinician variability that cannot be attributed to referral populations. The absence of correlation with case volume suggests that factors beyond experience drive this variation. These findings support the need for standardized audit methodologies and regular feedback to reduce unwarranted variation in teledermatology services. This study was conducted by an AI provider as part of their clinical governance framework. All authors are employed by the AI provider.