The signal and the noise in colorectal cancer diagnosis: exploring and explaining the relationship between diagnostic delays and stage at diagnosis using the Ca-PRI Colorectal Cancer Collaboration dataset

Objective: To explain the relationship between time to diagnosis and stage at diagnosis of colorectal cancer using data collected from primary care in six countries. Study design and setting: We identified eight population-based studies in Scotland, England, Canada, Denmark, Spain and the Netherlands using data from records, registries, audit sand questionnaires. Data allowed us to (1) explore the length of diagnostic interval and its primary and secondary care components for more than 10 000 colorectal cancer patients for whom the general practitioner had been involved in the diagnosis – and (2) analyse the impact of delays on the stage at diagnosis while taking account of differences in gender, age at diagnosis, alarm symptoms at presentation and emergency admissions from primary care.Results: Despite large variation in the distribution of diagnostic delays and tumour staging, the studies had consistent results: a complex two-sided ‘waiting time paradox’. For the primary care component of the diagnostic interval, we saw a concave, ∩-shaped association with increasing and subsequently decreasing odds of advanced disease with longer primary care intervals. For the secondary care component, we saw a convex, ∪-shaped association with decreasing and subsequently increasing odds of advanced disease with longer secondary care intervals.Conclusion: Clinical triage and hence confounding by indication is likely to explain counter-intuitive findings in studies of diagnostic delay and cancer outcomes. In primary care, GPs are trained to prevent the healthiest looking people from going to hospital, whereas in secondary care, the sickest looking people are expedited for investigation and diagnosed more quickly. Yet, along the pathway to cancer diagnosis, many patients do not fall squarely into set categories of healthy and sick. Based on the premise that delays are more randomly assigned for a significant part of intermediate patients in both sectors, we conclude that longer diagnostic intervals do increase the odds of being diagnosed with an advanced stage colorectal cancer.