Association between use of the urgent suspected cancer referral (two week wait) pathway and cancer survival: a 5 year national cohort study

Problem

Urgent referral pathways for suspected cancer have been available to GPs in England since the early 2000s, with considerable variation in use between practices. A previous study found that higher practice propensity to use the urgent referral pathway was associated with lower mortality of patients from all cancers, based on a single year cohort (2009). We aimed to determine if this association was replicated over a longer time period , and for the main cancer types.

Approach

National cohort study of over 1.4m patients diagnosed with cancer in England 2011-2015, from Public Health England’s (PHE) Cancer Analysis System, linked to practice-level referral data. Cox proportional hazards regression was used to quantify the risk of death over five years from any cause. Risk was stratified according to quintiles of practice referral metrics (referral ratio, detection rate and conversion rate), stratified for cancer sub-types and for all cancers combined , with adjustment for potential confounders.

Findings

During five years of follow up over 664,423 deaths occurred (45.2%). There was reduced hazard of death for patients from practices within the highest quintiles of referral ratio (4% reduction, hazard ratio (HR) 0.96; 95% CI 0.95-0.97) and detection rate (6%, HR 0.94: 95% CI 0.94-0.95). Similar patterns were found for colorectal (5% and 5%), lung (5% and 2%) and prostate (12% and 5%) cancers respectively. There was a 4% reduced hazard of death with highest referral ratio quintile for breast cancer but no association with detection rate. Conversion rate was not associated with survival for all cancers, or for lung and breast cancer, however was associated with reduced hazard of death for colorectal (6%) and prostate cancer (10%). The patterns were consistent with adjustment for potential confounders. 8.6% of the cohort had missing practice referral data, of which the majority were not registered practice (63%). Those with missing GP data/not registered had substantially increased hazard of death (HR 1.51; 95% CI 1.50-1.53).

Consequences

Primary care registration and activity demonstrates reduced cancer patient mortality. Specifically higher practice referral ratios and detection rates were significantly associated with reduced five year mortality. The previously reported association was sustained over more recent and longer time periods, for all cancers combined and the main subtypes. The absolute differences in reduced cumulative mortality approaches the magnitude of important differences in survival between England and comparable countries. Urgent referral pathways demonstrate different effectiveness based on cancer site. Further work to understand the association between urgent referral and stage at diagnosis and the effect of the 2015 updated NICE referral guidelines is warranted.