Potentially avoidable delays in diagnosis among patients with colorectal cancer: A population-based study.

Abstract

122 Background: Colorectal cancers (CRCs) have the second highest cancer mortality for both sexes combined because approximately 60% of incident cases are not diagnosed until progression to later stages. Many patients with CRC experience cancer-related signs and symptoms up to a year prior to diagnosis, but they are often non-specific in nature, leading to diagnostic difficulty and delays in care. Few studies have assessed the timeliness of CRC diagnosis using Medicare claims data. This study examines pre-diagnosis healthcare utilization of patients with CRC to understand potentially avoidable delays in diagnosis and estimates the association between delay and short-term mortality. Methods: We used SEER-Medicare data from 2008-2017 to identify Medicare beneficiaries aged 66+ at the time of invasive CRC diagnosis. ICD-9 and ICD-10 codes on claims were used to identify alarm or “red flag” signs/symptoms (e.g., rectal bleeding) and general signs/symptoms (e.g., fatigue) based on a literature review. The diagnostic interval (DI) was defined as the count of days from the first claim for a sign/symptom of CRC in the one-year pre-diagnosis period until the diagnosis date. Avoidable diagnostic delay was defined as having a DI longer than the 80th percentile (320 days) or 3+ visits for the same sign/symptom at least 14 days apart in the year prior to the diagnosis. Logistic regression was used to estimate 1-year mortality odds ratios among symptomatic patients who survived at least one month as a function of the avoidable delay adjusted for age, sex, year of diagnosis, comorbidity score, cancer stage at diagnosis, and tumor site. Results: Among111,283 Medicare beneficiaries diagnosed with invasive CRC, 92,450 (83%) patients experienced at least one sign/symptom in the year prior to diagnosis. The most common sign/symptom experienced was anemia (42%), followed by abdominal pain (37%) while blood in stool was observed in 20% of patients. The median DI was 176 days (IQR: 41-302) and those with only alarm symptoms recorded had substantially shorter DI compared to those who experienced general symptoms (median 29 vs 207 days). 29% of patients were classified as having a potentially avoidable diagnostic delay: 12% had 3+ visits for the same sign/symptom, 6.5% had a DI >80th percentile, and 9.9% had both. The adjusted odds of 1–year mortality was 14% higher among those with potentially avoidable diagnostic delay compared to those that did not experience diagnostic delay (AOR: 1.14, 95%CI: 1.09-1.21). Conclusions: Many patients with CRC experience CRC related signs/symptoms up to one year prior to their diagnosis. Despite screening availability and use, a substantial number of patients with CRC have a diagnostic delay, and diagnostic delays negatively impact mortality. Therefore, efforts to identify patients early, before progression to later stages, and decreasing diagnostic delays is vital to improve CRC outcomes.