Alina Zalounina Falborg1, Peter Vedsted2, Usha Menon3, David Weller4, Richard D Neal5, Irene Reguilon6, Samantha Harrison6, Henry Jensen2. 1. Research Centre for Cancer Diagnosis in Primary Care, Research Unit for General Practice, Bartholins Allé 2, DK-8000 Aarhus C, Denmark. Electronic address: az@ph.au.dk. 2. Research Centre for Cancer Diagnosis in Primary Care, Research Unit for General Practice, Bartholins Allé 2, DK-8000 Aarhus C, Denmark. 3. Institute of Clinical Trials and Methodology, University College London, Maple House, 149 Tottenham Court Road, London W1T 7DN, United Kingdom. 4. Centre for Population Health Sciences, University of Edinburgh, Doorway 1, Medical Quad Teviot Place, Edinburgh EH8 9DX, United Kingdom. 5. Academic Unit of Primary Care, Leeds Institute of Health Sciences, University of Leeds, Leeds LS2 9NL, United Kingdom. 6. International Cancer Benchmarking Partnership, Cancer Research UK, Angel Building, 407 St John Street, Islington, London EC1V 4AD, United Kingdom.
Abstract
BACKGROUND: The routes to diagnosis and the time intervals along the diagnostic pathway affect cancer outcomes. Some data on routes to diagnosis and milestone dates can be extracted from registries or databases. When this data is incomplete, inaccurate or non-existing, other data sources are needed. This study investigates the agreement between multiple data sources on routes to diagnosis and milestone dates of cancer pathway. METHODS: Information on routes to diagnosis and milestone dates were compared across four data sources (cancer patients, general practitioners, cancer specialists and registries) for breast, colorectal, lung and ovarian cancers across the UK, Scandinavia, Canada and Australia. Agreement on routes to diagnosis and milestone dates was assessed by Kappa and AC1 coefficients and Lin's concordance correlation coefficient (CCC). RESULTS: 4502 patients were included in the analysis of routes to diagnosis. The agreement was almost perfect (kappa = 0.15-0.88, AC1 = 0.86-0.91) for breast cancer, substantial to almost perfect (kappa = 0.07-0.86, AC1 = 0.74-0.93) for colorectal and ovarian cancers, and substantial (kappa = 0.09-0.11, AC1 = 0.65-0.74) for lung cancer. 2287 patients were included in the analysis of milestone dates. The agreement was adequate for all cancer types (CCC = 0.88-0.99); highest agreement was seen for date of diagnosis (CCC = 0.94-0.99). CONCLUSION: We found a reasonable agreement between patient/physician questionnaires and registry data for routes to diagnosis and milestone dates. The agreement on routes to diagnosis was generally higher for breast cancer than for colorectal, ovarian and lung cancers. Lower agreement was seen on date of first presentation to primary care and date of treatment initiation compared to date of diagnosis.
BACKGROUND: The routes to diagnosis and the time intervals along the diagnostic pathway affect cancer outcomes. Some data on routes to diagnosis and milestone dates can be extracted from registries or databases. When this data is incomplete, inaccurate or non-existing, other data sources are needed. This study investigates the agreement between multiple data sources on routes to diagnosis and milestone dates of cancer pathway. METHODS: Information on routes to diagnosis and milestone dates were compared across four data sources (cancerpatients, general practitioners, cancer specialists and registries) for breast, colorectal, lung and ovarian cancers across the UK, Scandinavia, Canada and Australia. Agreement on routes to diagnosis and milestone dates was assessed by Kappa and AC1 coefficients and Lin's concordance correlation coefficient (CCC). RESULTS: 4502 patients were included in the analysis of routes to diagnosis. The agreement was almost perfect (kappa = 0.15-0.88, AC1 = 0.86-0.91) for breast cancer, substantial to almost perfect (kappa = 0.07-0.86, AC1 = 0.74-0.93) for colorectal and ovarian cancers, and substantial (kappa = 0.09-0.11, AC1 = 0.65-0.74) for lung cancer. 2287 patients were included in the analysis of milestone dates. The agreement was adequate for all cancer types (CCC = 0.88-0.99); highest agreement was seen for date of diagnosis (CCC = 0.94-0.99). CONCLUSION: We found a reasonable agreement between patient/physician questionnaires and registry data for routes to diagnosis and milestone dates. The agreement on routes to diagnosis was generally higher for breast cancer than for colorectal, ovarian and lung cancers. Lower agreement was seen on date of first presentation to primary care and date of treatment initiation compared to date of diagnosis.
Keywords:
Breast neoplasms; Colorectal neoplasms; Data accuracy; Early detection of cancer; Lung neoplasms; Ovarian neoplasms; Registries; Surveys and questionnaires
Authors: Lina Sulieman; Robert M Cronin; Robert J Carroll; Karthik Natarajan; Kayla Marginean; Brandy Mapes; Dan Roden; Paul Harris; Andrea Ramirez Journal: J Am Med Inform Assoc Date: 2022-06-14 Impact factor: 7.942