Time from first presentation in primary care to treatment of symptomatic colorectal cancer: effect on disease stage and survival

Abstract

Background: British 5-year survival from colorectal cancer (CRC) is below the European average, but the reasons are unclear. This study explored if longer provider delays (time from presentation to treatment) were associated with more advanced stage disease at diagnosis and poorer survival.

Methods: Data on 958 people with CRC were linked with the Scottish Cancer Registry, the Scottish Death Registry and the acute hospital discharge (SMR01) dataset. Time from first presentation in primary care to first treatment, disease stage at diagnosis and survival time from date of first presentation in primary care were determined. Logistic regression and Cox survival analyses, both with a restricted cubic spline, were used to model stage and survival, respectively, following sequential adjustment of patient and tumour factors.

Results: On univariate analysis, those with <4 weeks from first presentation in primary care to treatment had more advanced disease at diagnosis and the poorest prognosis. Treatment delays between 4 and 34 weeks were associated with earlier stage (with the lowest odds ratio occurring at 20 weeks) and better survival (with the lowest hazard ratio occurring at 16 weeks). Provider delays beyond 34 weeks were associated with more advanced disease at diagnosis, but not increased mortality. Following adjustment for patient, tumour factors, emergency admissions and symptoms and signs, no significant relationship between provider delay and stage at diagnosis or survival from CRC was found.

Conclusions: Although allowing for a nonlinear relationship and important confounders, moderately long provider delays did not impact adversely on cancer outcomes. Delays are undesirable because they cause anxiety; this may be fuelled by government targets and health campaigns stressing the importance of very prompt cancer diagnosis. Our findings should reassure patients. They suggest that a health service's primary emphasis should be on quality and outcomes rather than on time to treatment.

Figure 1

Spline curve of logistic regression of provider delay (weeks) on odds of having later stage (C+D) disease at diagnosis. Model 1 (solid line): unadjusted model with four knots (provider delay of 4 weeks used as reference). Model 2 (dashed line): adjusted for place of presentation, patient age, patient gender, smoking status, Carstairs deprivation quintile and Charlson index. Model 3 (dotted line): further adjusted for tumour grade and type of admission. Model 4 (long dash and dotted line): further adjusted for symptoms and signs of constipation, rectal bleeding, abdominal pain, anorexia, weight loss and abdominal mass.

Figure 2

Spline curve of logistic regression of provider delay (weeks) on odds ratio (shaded area are 99% CIs) of having later stage (C+D) disease at diagnosis after multi-adjustment for potential confounders. Reference category is 4 weeks delay. Model adjusted for age, gender, smoking status, Carstairs deprivation quintile, Charlson index, place of presentation, tumour grade, type of admission, and for symptoms and signs of constipation, rectal bleeding, abdominal pain, anorexia, weight loss and abdominal mass.

Figure 3

Spline curve of Cox proportional hazards regression of provider delay (weeks) on hazard ratio for mortality. Model 1 (solid line): unadjusted model with four knots (provider delay of 4 weeks used as reference). Model 2 (dashed line): adjusted for place of presentation, patient age, patient gender, smoking status, Carstairs deprivation quintile and Charlson index. Model 3 (dotted line): further adjusted for tumour grade and type of admission. Model 4 (long dash and dotted line): further adjusted for symptoms and signs of constipation, rectal bleeding, abdominal pain, anorexia, weight loss and abdominal mass.

Figure 4

Spline curve of Cox proportional hazards model of provider delay (weeks) on hazard ratio (shaded area are 99% CI) for mortality after multi-adjustment for potential confounders. Reference category is 4 weeks delay. Model adjusted for age, gender, smoking status, Carstairs deprivation quintile, Charlson index, place of presentation, tumour grade, type of admission, and for symptoms and signs of constipation, rectal bleeding, abdominal pain, anorexia, weight loss and abdominal mass.