Missed/delayed referrals for investigation of suspected cancer: a candidate indicator of practice-level care quality

Problem

Assessing the quality of cancer diagnostic activity in general practice is difficult, with outcome indicators dependant on small numbers of patients. We investigated the utility of a novel indicator tracking expidited referrals following ‘red flag’ cancer symptoms in accordance with National Institute for Health and Care Excellence (NICE) guidance.

Approach

Clinical Practice Research Datalink electronic primary care records from 2014-2015 for patients reporting one of six cancer symptoms for which expedited referral is suggested under NICE guidance on for suspected cancer (dysphagia, breast lump, rectal bleeding, post-menopausal bleeding, haematuria, iron-deficiency anaemia) were linked to Hospital Episode Statistics referrals data. Practices’ proportion of patients receiving an expedited referral request within 14-days of presentation was assessed for adequate variation (odds ratio ≥1.50 between 10th and 90th centiles) and reliability (median ≥0.70). The effect of case-mix adjustment on performance was explored using Kendall's tau. The indicator was also investigated at the level of individual practitioners.

Findings

Altogether, 19,787/48,847 (40.5%; 95%CI 40.1%-40.9%) patients from 279 practices received an expedited referral; median practice performance was 40.1% (IQR 33.5%-46.5%). The odds ratio for the difference in performance between the 10th and 90th centiles was 2.55, indicating considerable variation in performance. Very good reliability was also demonstrated, with a median of 0.82 (IQR 0.70 to 0.88), and was sufficient when using a single year of data. Single symptom indicators for dysphagia, breast lump, post-menopausal bleeding, and haematuria individually showed sufficient variability and reliability. Adjustment for age, gender, and symptom led to modest reordering of practices and is recommended (Kendall’s tau=0.74). At the practitioner-level, the indicator had sufficient variability, but not reliability, due to the lower counts at this level of aggregation.

Consequences

Practice-level proportion of patients receiving an expedited referral following six red flag cancer symptoms has the potential to be an effective care quality indicator, with one year of data sufficient for adequate reliability.