Electronic medical records: Assessment of recorded cancer stage in support of quality of care.
129 Background: Electronic medical records (EMRs) are being increasingly adopted in part driven by reports of their positive impact on patient’s quality of care. An underlying assumption is that data recorded will be relatively complete. As a field of primary importance, this study assessed the frequency with which cancer stage was recorded within an EMR data field during a historical and recent 12-month period. A random sample of records with missing stage was assessed to identify at a qualitative level reasons that stage may be omitted. Methods: Two datasets were constructed. The first comprised of oncology EMRs from 77 practices covering 476 sites of care across 34 states from 1/1/2000-12/31/2010. The second dataset from 58 practices covering 391 sites of care across 37 states. Inclusion criteria required patients to have a valid visit (i.e., not simply ‘scheduled’) and ≥ 1 diagnosis of a primary, malignant, neoplasm (except brain or spine). All data fields utilized to record stage (stage I, II, etc.) or from which stage could be reliably derived (T, M, N fields) were defined as "recorded." Practices were not required to exist in each dataset. Recorded stage by age, gender, state, and payer type was also assessed. Results: Reasons reported for absent stage within the data field included: consult visit only, written in the progress notes, text present in a scanned report, stage X (insufficient information), continuing treatment initiated elsewhere, and missing entry error. Conclusions: A significant proportion of cancer patients may not have stage recorded in the designated, searchable, data field within an EMR. The frequency of recorded stage is increasing over time. Reasons for unpopulated stage field(s) include use of nonsearchable text entries, scanned reports, and short episodes of care. Further research is needed to validate the observations in this study, determine root causes, and employ appropriate solutions. [Table: see text]