Abstract
BackgroundData from centralised, population-based statutory cancer registries are generally considered the 'gold standard' for confirming incident cases of cancer. When these are not available, or more current information is needed, hospital or other routinely collected population-level data may be feasible alternative sources. We aimed to determine the validity of various methods using routinely collected administrative health data for ascertaining incident cases of colorectal or lung cancer in participants from the 45 and Up Study in New South Wales (NSW), Australia.MethodsFor 266,844 participants in the 45 and Up Study (recruited 2006–2009) ascertainment of incident colorectal or lung cancers was assessed using diagnosis and treatment records in linked administrative health datasets (hospital, emergency department, Medicare and pharmaceutical claims, death records). This was compared with ascertainment via the NSW Cancer Registry (NSWCR, the 'gold standard') for a period for which both data sources were available for participants.ResultsA total of 2253 colorectal and 1019 lung cancers were recorded for study participants in the NSWCR over the period 2006–2010. A diagnosis of primary cancer recorded in the statewide Admitted Patient Data Collection identified the majority of NSWCR colorectal and lung cancers, with sensitivities and positive predictive values (PPV) of 95% and 91% for colorectal cancer and 81% and 85% for lung cancer, respectively. Using additional information on lung cancer deaths from death records increased sensitivity to 84% (PPV 83%) for lung cancer, but did not improve ascertainment of colorectal cancers. Hospital procedure codes for colorectal cancer surgery identified cases with sensitivity 81% and PPV 54%. No other individual indicator had sensitivity >50% or PPV >65% for either cancer type and no combination of indicators increased both the sensitivity and PPV above that achieved using the hospital cancer diagnosis data. All specificities were close to 100%; 95% confidence intervals for sensitivity and PPV were generally +/−2%.ConclusionsIn NSW, identifying new cases of colorectal and lung cancer from administrative health datasets, such as hospital records, is a feasible alternative when cancer registry data are not available. However, the strengths and limitations of the different data sources should be borne in mind.
Highlights
Data from centralised, population-based statutory cancer registries are generally considered the 'gold standard' for confirming incident cases of cancer
Statistical methods Sensitivity, specificity and the positive predictive value (PPV) of ascertaining cases of cancer for February 2006 to December 2010 in non-New South Wales Cancer Registry (NSWCR) datasets compared to Colorectal cancer (No cases in NSWCR: 2253)
There were 2253 people with a diagnosis of colorectal cancer and 1019 with lung cancer recorded in the NSWCR over the period February 2006–December 2010
Summary
Data from centralised, population-based statutory cancer registries are generally considered the 'gold standard' for confirming incident cases of cancer. Prospective linkage of cohort study questionnaire data to administrative, routinely collected population health datasets provides an effective method for identification of health outcomes, such as incident colorectal and lung cancers. This ascertainment of cancer cases allows for powerful and highly efficient investigation of the factors influencing cancer incidence, mortality, treatment and survival, which are important for research, evaluation and planning. Such linkage provides almost complete participant follow-up for health outcomes, without the cost, time, burden, and inaccuracies of self-report
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.