Main Article Content
Dispensing claims are used commonly as proxy measures in pharmacoepidemiological studies, however, their validity is often untested.
Objectives and Approach
We quantified the level of ascertainment and potential biases arising when using dispensing claims to identify incident cancer cases in cohort studies. We used Department of Veterans’ Affairs client data linked with the New South Wales (NSW) Cancer Registry and Repatriation and Pharmaceutical Benefits Scheme data. We included clients aged ≥65 residing in NSW between July 2004 and December 2012. We matched clients with a cancer diagnosis to clients without a diagnosis based on demographic characteristics and available observation time. We used dispensing claims for anticancer medicines dispensed between July 2004 and December 2013 as a proxy for cancer diagnosis and calculated sensitivity, specificity, positive predictive values and negative predictive values compared with cancer registry data (gold standard), overall and by cancer site. We illustrated the potential for misclassification by the proxy in a cohort of people initiating opioid therapy.
We identified 15,679 incident cancer diagnoses in 14,112 clients from the cancer registry and 62,663 clients without a diagnosis. The proxy’s sensitivity was 30% for all cancers and ranged from 10-67% for specific cancers. Specificity was >90% for all cancers. The proxy correctly identified 26% of people with a cancer diagnosis who initiated opioid therapy, failed to identify 74% those with a cancer diagnosis, and was most robust for clients with breast cancer (61% were correctly identified).
Conclusion / Implications
Use of anticancer medicine dispensings for identifying people with incident cancer diagnosis is a poor proxy. Excluding people with evidence of anticancer medicine dispensing from cohort studies may remove a disproportionate number of women with breast cancer. Researchers excluding or otherwise using anticancer medicine dispensing to identify people with cancer in pharmacoepidemiological studies should acknowledge the potential biases introduced to their findings.
This work is licensed under a Creative Commons Attribution 4.0 International License.