1024 — Determinants of Time to Diagnosis of Colorectal Cancer in the VA System
Fisher DA (Durham VAMC, Duke University), Zullig L
(Durham VAMC), Grambow SC
(Durham VAMC, Duke University), Abbott D
(Durham VAMC), Provenzale D
(Durham VAMC, Duke University)
Objectives:
To determine predictors of 1) time to diagnosis of colorectal cancer (CRC), and 2) stage at diagnosis.
Methods:
An ancillary study of VA CanCORS: Cancer Cares Outcomes Research and Surveillance, a 14-site observational study. VA medical record data, up to 24 months prior to CRC diagnosis, were abstracted for the 487 CRC patients, and were used to determine patient demographics, comorbidity (ACE 27), and cancer stage (I /II vs III/IV). Patients were classified as 1) screen-detected, 2) symptom-detected, or 3) incidental –i.e. the result of evaluating an abnormal test from an unrelated medical concern. Time to diagnosis was defined as the time from the initial event (abnormal screening test result date or first medical consultation of a symptom or abnormal test result date) to date of tissue diagnosis. Patients who presented with obstruction or perforation were excluded. We used Cox proportional hazards models to determine the impact of diagnostic category and covariates on time to diagnosis and logistic regression models to determine the impact of time to diagnosis and other variables on stage at diagnosis.
Results:
We excluded 21 subjects who presented emergently and 12 with missing demographic data (10) or stage (2) leaving 454 subjects; 98% male, 66% Caucasian, 53% married, 53% with no or minor comorbidities. Stage was I or II for 43%. Diagnosis was by screening for 39%, symptoms for 27%, and incidentally for 34%. Median times to diagnosis (days) were: screen-detected 90, symptom-detected 69, and incidental 73. Older age was the only predictor of diagnostic delay (HR 0.89 (0.81-0.97)). Time to diagnosis was not associated with stage, but screen-detected patients were more likely to be diagnosed at stage I or II (OR 0.40 (0.24-0.65)) than symptom-detected.
Implications:
Median times to diagnosis were not prolonged and relative delay was not associated with more advanced disease at diagnosis. Consistent with other studies, screen-detected cancers were diagnosed at an earlier stage.
Impacts:
Increasing the proportion of screen-detected cancers may reduce the incidence of late stage CRC. More than half of the patients were not diagnosed as part of the screening process. This should be further investigated.
