Variation in the use of primary care-led investigations prior to a cancer diagnosis: analysis of the National Cancer Diagnosis Audit.

Introduction: Use of investigations can help support the diagnostic process of patients with cancer in primary care, but the size of variation between patient group and between practices is unclear.

Methods: We analysed data on 53 252 patients from 1868 general practices included in the National Cancer Diagnosis Audit 2018 using a sequence of logistic regression models to quantify and explain practice-level variation in investigation use, accounting for patient-level case-mix and practice characteristics. Four types of investigations were considered: any investigation, blood tests, imaging and endoscopy.

Results: Large variation in practice use was observed (OR for 97.5th to 2.5th centile being 4.02, 4.33 and 3.12, respectively for any investigation, blood test and imaging). After accounting for patient case-mix, the spread of practice variation increased further to 5.61, 6.30 and 3.60 denoting that patients with characteristics associated with higher use (ie, certain cancer sites) are over-represented among practices with lower than the national average use of such investigation. Practice characteristics explained very little of observed variation, except for rurality (rural practices having lower use of any investigation) and concentration of older age patients (practices with older patients being more likely to use all types of investigations).

Conclusion: There is very large variation between practices in use of investigation in patients with cancer as part of the diagnostic process. It is conceivable that the diagnostic process can be improved if investigation use was to be increased in lower use practices, although it is also possible that there is overtesting in practices with very high use of investigations, and in fact both undertesting and overtesting may co-exist.