Are alarm symptoms predictive of cancer survival?: population-based cohort study.

Abstract

BACKGROUND: Alarm symptom presentations are predictive of cancer diagnosis but may also be associated with cancer survival. AIM: To evaluate diagnostic time intervals, and consultation patterns after presentation with alarm symptoms, and their association with cancer diagnosis and survival. DESIGN AND SETTING: Cohort study using the Clinical Practice Research Database, with linked Cancer Registry data, in 158 general practices. METHOD: Participants included those with haematuria, haemoptysis, dysphagia, and rectal bleeding or urinary tract cancer, lung cancer, gastro-oesophageal cancer, and colorectal cancer. RESULTS: The median (interquartile range) interval in days from first symptom presentation to the corresponding cancer diagnosis was: haematuria and urinary tract cancer, 59 (28-109); haemoptysis and lung cancer, 35 (18-89); dysphagia and gastro-oesophageal cancer, 25 (12-48); rectal bleeding and colorectal cancer, 49 (20-157). Three or more alarm symptom consultations were associated with increased odds of diagnosis of urinary tract cancer (odds ratio [OR] 1.84, 95% CI = 1.50 to 2.27), lung cancer (OR = 1.76, 95% CI = 1.07 to 2.90) and gastro-oesophageal cancer (OR = 2.17, 95% CI = 1.48 to 3.19). Longer diagnostic intervals were associated with increased mortality only for urinary tract cancer (hazard ratio 2.23, 95% CI = 1.35 to 3.69). Patients with no preceding alarm symptom had shorter survival from diagnosis of urinary tract, lung or colorectal cancer than those presenting with a relevant alarm symptom. CONCLUSION: After alarm symptom presentation, repeat consultations are associated with cancer diagnoses. Longer diagnostic intervals appeared to be associated with a worse prognosis for urinary tract cancer only. Mortality is higher when cancer is diagnosed in the absence of alarm symptoms.