Comparison of the urgent referral for suspected breast cancer process with patient age and a predictive multivariable model

Abstract

Abstract Background The urgent 2-week wait referral for suspected breast cancer system (U2WW) in the UK prioritizes primary care referrals to one-stop breast clinics as ‘urgent’ or ‘choose and book’ (C&B). The aim of this study was to evaluate the accuracy of U2WW in discriminating cancer versus no cancer, and to consider alternative criteria. Methods Clinical features elicited in primary care and demographics of consecutive female patients in a specialist breast clinic were collated at the time of consultation from May 2008 to July 2017. U2WW was compared with patient age alone and a multivariable model in terms of accuracy and net cost for eight underlying cost–benefit assumptions. Results There were 7915 eligible referrals: 4877 urgent (61.6 per cent) and 3038 C&B (38.4 per cent) referrals. Breast cancer was diagnosed in 546 patients (6.9 per cent): 491 (10.1 per cent) in urgent and 55 (1.8 per cent) in C&B referrals (P < 0.001). The multivariable model summated the significant variables: age (odds ratio (OR) 1.07, 95 per cent c.i. 1.07 to 1.08), tumour (OR 4.85, 3.62 to 6.52), observed change (OR 1.73, 1.34 to 2.23), pain (OR 0.46, 0.35 to 0.61) and gravidity (OR 0.72, 0.54 to 0.95). The area under the curve was 0.651 for U2WW, 0.784 for age alone, and 0.824 for the multivariable model (P <0.001 for all comparisons). Considering the cost assumptions, age alone and the multivariable model were either more accurate than U2WW, or as accurate but less costly. Conclusion The U2WW is surpassed by patient age as a single referral criterion. A multivariable model based on demographics and simple clinical features outperformed both. The continued use of the U2WW needs to be reconsidered.