E403: Inconsistent Clinical Documentation is a Barrier to Capturing Recurrence in National Cancer Registries

Historically, cancer recurrence data has not been publicly reported by the National Cancer Database (NCDB) due to concerns of data missingness and underreporting. The expansion of electronic health records and changes in registry coding practices may have improved reporting. As a step towards sharing this data in the NCDB, our objective was to assess contemporary rates of missingness of recurrence data and identify remaining challenges to abstracting high-quality data.

Methods:

A mixed methods study was conducted among all Commission on Cancer (CoC) accredited sites. Stage I-III breast, colon, and lung cancer cases diagnosed in 2016, with 60-months of follow up, were retrospectively evaluated from the NCDB for missingness of variables associated with the date and type of recurrence. Multivariable analysis was used to examine patient, tumor, and program factors associated with missingness. Cancer registry staff at all CoC-accredited programs were surveyed in October 2023 to investigate data abstraction challenges and potential solutions.

Results:

Of 178,574 breast/colon/lung cancer cases identified, recurrence information was complete in 93.5%/90.5%/94.1% of cases, respectively. On multivariable analysis, patient demographics and tumor characteristics were not associated with data missingness. Greater distance traveled and hospital region were associated with missing data (Table).

The registrar survey was completed by 354 (24.5%) of 1,445 programs. Inadequate physician documentation of no evidence of disease (71.6%) and lack of resources (52.6%) were the most common reasons for missing data. Sixty percent of programs reported many patients receive surveillance care at other facilities and endorsed difficulty accessing this information. Many registrars (62.7%) also raised concerns regarding discrepancies between clinical documentation of recurrence and registry coding rules. A majority of respondents (75.1%) indicated that required synoptic physician documentation of disease status would greatly improve abstraction.

Conclusions:

Although contemporary missingness for recurrence rates is low, there remains high variance across programs with registrars identifying inconsistent clinical documentation as a significant problem, along with data access, and inadequate resources. Current efforts are underway to further examine rates of recurrence in the NCDB and to develop a synoptic reporting tool for physician documentation of cancer disease status which may promote abstraction of high-quality data along the continuum of cancer outcomes.