Identifying bowel cancer stage from routinely collected electronic data sources in New Zealand

Abstract

Background: Cancer stage data is critical for bowel cancer quality indicator reporting and to understand and monitor bowel cancer inequities. Increasingly electronic data are available through routine health care delivery. While obtaining routinely collected electronic data may be more timely and cost-effective than paper- or registry-based data sources, it is unclear whether such data are also accurate, complete, consistent, and equitable. Aim: To determine the adequacy of bowel cancer staging data from routinely collected electronic data. Method: A systematic review of international literature was conducted to determine the feasibility of identifying bowel cancer stage from routinely collected electronic data sources. Routinely collected electronic data for patients living in the Auckland District Health Board (ADHB) catchment area with a first bowel cancer diagnosis in 2017 were audited. Patients were identified from the New Zealand Cancer Registry (NZCR). Data recorded on both the NZCR and ADHB multidisciplinary meeting (MDM) datasets were compared using a suite of bowel cancer quality indicators, developed to assess stage (tumour, T, node, N, and metastasis, M, separately or complete, TNM), data accuracy, completeness, and consistency. These indicators were reported in total and by prioritised ethnicity (Māori, Pacific, and non- Māori/non-Pacific (nMnP)). Results: Only two studies were identified (both from Australia) that explored the adequacy of bowel cancer data extracted from routine electronic data sources. These studies found that it was feasible to extract stage from electronic pathology reports, but data were incomplete. A total of 190 ADHB bowel cancer patients were identified for the audit. Their median age at diagnosis was 68 years (interquartile range = 22 years; quartile one, quartile three = 56, 78 years), and they comprised 9 Māori, 10 Pacific, and 171 nMnP patients. From the NZCR dataset alone, complete stage could be extracted (electronically and manually) for 96 patients (50.5%). Among the 142 patients (74.7%) with a recorded MDM, 44 patients (31%) had complete TNM stage in the MDM records using electronic and manual extraction. M stage data were available (either indicating presence or absence of metastasis) for 72.1% of patients on the NZCR and 33.8% of patients in the MDM records. Concordance of TNM stage among patients with complete TNM stage in both the NZCR and MDM datasets was 77.3% (17/22). Māori patients were less likely than nMnP to have complete, electronically extractable TNM stage on the NZCR (Māori 11% (1/9), nMnP 53% (90/171)) and concordant TNM stage (electronic and manual extraction) between the NZCR and MDM records (Māori 0% (0/1), nMnP 80% (16/20)). Conclusion: The extraction of complete bowel cancer TNM stage from the routinely collected electronic data sources was feasible, but health services will need to undertake further work to enhance the accuracy, completeness, concordance, and equity of such data before they can be used reliably for bowel cancer quality indicator reporting.