Lack of Cancer Recurrence Data in Large Databases: A National Survey of Hospital Cancer Registries.

Background: Cancer recurrence information is not routinely collected by the US cancer registries. Prior research suggests hospital characteristics, staff qualifications, and chart access may be contributing factors but this has not been explored nationally. This study aimed to understand issues underlying poor collection of recurrence information and to identify targets for improvement.

Methods: A survey was sent to Commission on Cancer hospitals to investigate reasons for variations in recurrence data collection, examine resources allocated, and assess coding variability. Descriptive and multivariate analyses were performed.

Results: Eight hundred and forty-five of 1417 surveys to Commission on Cancer hospitals were analyzed. Sixty-nine percent reported annually examining charts for recurrence ("investigating" hospitals). They more likely had experienced registrars (91% versus 84%, P < 0.05), integrated electronic medical records (75% versus 68%, P < 0.05) and chart access to in-network hospitals (80% versus 73%, P < 0.05). Thirty-seven percent reported ability to follow-up using medical records on <50% of patients. Reasons for noncollection included inability to accurately collect (37%), limited resources (44%), and low priority (18%). Odds of being an investigating hospital increased with the percentage of patients who could be followed up with medical records (90%-100% OR = 6.72). There were minimal differences among hospitals in registry caseload and resources. 79.5% reported that without prior recurrence documentation, they would code the patient as not having a recurrence, 8.8% would change all recurrence variables to "unknown," and 7.2% would leave them blank.

Conclusions: Those tasked with collecting recurrence information report significant barriers concerning data access, data quality, adequate resources, and coding variability. A unified effort is needed to improve collection.