Validation and quality assessment of the Kilimanjaro Cancer Registry.
Zullig, LL; Muiruri, C; Nyindo, P; Namwai, T; Karia, F; Oresto, M; Silayo, E; Msomba, A; Bartlett, J; Zafar, Y
Published in: Journal of Clinical Oncology
266 Background: The global cancer burden has increasingly shifted to low and middle-income countries and is particularly pronounced in Africa. However, there remains a lack of comprehensive information about cancer incidence due to limited development of cancer registries. In Moshi, Tanzania, a regional cancer registry exists at Kilimanjaro Christian Medical Center (KCR). The quality of the registry data is unknown. Our objective was to evaluate the completeness and quality of the KCR. Methods: In October 2015 we conducted a retrospective review, validating internal consistency of registry records compared with paper-based medical and pathology records. We randomly sampled approximately 3% of over 3000 registry cases (n = 100), focused on cases abstracted from 2014-15. Four independent reviewers not associated with the KCR manually collected and compared key data elements (e.g., cancer site, morphology, diagnosis date). Data were collected using Microsoft Excel. Data management and analysis were conducted in Stata. Results: The KCR contains information on over 3000 patients from 32 tribal groups and 8 geographic areas in the Kilimanjaro region. All of the 100 reviewed cases had complete cancer site and morphology in the registry. Six had a recorded stage. For the majority of cases (n = 92) the basis of diagnosis was pathology. Pathology records were found in the medical record for 41% of cases; for the remainder, pathology records were stored separately in the pathology department. Of sampled cases, the KCR and medical records were 98% and 94% concordant for primary cancer site and morphology, respectively. For 28% of cases, recorded diagnosis dates were within 14 days of what was found in the medical record, and for 32% of cases they were within 30 days. Conclusions: Although training is needed to standardize diagnosis date definitions, this study suggests that KCR data is high quality. Research is needed to determine effectiveness at ascertaining all patients in its catchment area. The registry may have utility for resource planning and research purposes.
