RESUMO
BACKGROUND: The Blood Cancer Network Ireland and National Cancer Registry Ireland worked to create an Enhanced Blood Cancer Outcomes Registry (EBCOR). Enhanced data in acute myeloid leukaemia (AML) included an extensive data dictionary, bespoke software and longitudinal follow-up. AIMS: To demonstrate the utility of the database, we applied the data to examine a clinically relevant question: Charlson comorbidity index (CCI) usefulness in predicting AML patients' survival. METHODS: A software designer and consultant haematologists in Cork University Hospital worked together to standardise a data dictionary, train registrars and populate a database. One hundred and forty-one AML patients underwent enhanced data registration. Comorbidities identified by chart review were used to examine the capability of the CCI and age at diagnosis to predict mortality using Kaplan-Meier curves, Cox regression and receiver operating characteristic curves. RESULTS: In regression analysis, a dose-response relationship was observed; patients in the highest CCI tertile displayed a greater risk (HR = 4.90; 95% CI 2.79-8.63) of mortality compared to subjects in tertile 2 (HR = 2.74; 95% CI 1.64-4.57) and tertile 1 (reference). This relationship was attenuated in an analysis which adjusted for age at diagnosis. The area under the curve (AUC) for the CCI was 0.76 (95% CI 0.68-0.84) while the AUC for age at diagnosis was 0.84 (95% CI 0.78-0.90). CONCLUSIONS: Results suggest that the CCI provides no additional prognostic information beyond that obtained from age alone at AML diagnosis and that an EBCOR can provide a rich database for cancer outcomes research, including predictive models and resource allocation.
