How Accurate Are NCDB Comorbidity Assessments in Lung Cancer and Other Resected Tumors?
Posted: Monday, January 14, 2019
A study by Melisa L. Wong, MD, MAS, of the University of California, San Francisco, and colleagues found that the National Cancer Database (NCDB) systematically underestimated comorbidities in patients with surgically resected lung, breast, and colorectal cancers. Although their analysis suggested that this underestimation did not drastically affect prognostic models for post-operative 5-year overall survival, the authors concluded that care should be taken when using NCDB comorbidity data in comparative effectiveness studies. Their findings were published in the Journal of Oncology Practice.
“The NCDB does not differentiate between no documented comorbidities and missing or unavailable comorbidity information. The lack of a missing value category…limits the ability of researchers to identify and analyze these patients differently from patients who truly have no comorbidities,” wrote the authors.
In this NCDB Special Study, de-identified medical records of 9,640 patients with stage I to III lung cancer who were treated with definitive surgical resection in 2006 and 2007 were reabstracted by cancer registrars. A total of 10 eligible patients (with no postoperative disease and adequate follow-up) were included from each American College of Surgeons Commission on Cancer–accredited facility. Data were collated from 1,214 facilities and were compared with the Charlson-Deyo Comorbidity Index (CCI) scores reported in the NCDB.
In total, comorbidities were underestimated in 36.2% of patients with lung cancer and overestimated in 11.9%. Relative to breast and colorectal cancers, comorbidity was highest (13.2% NCDB CCI 2+) and postoperative 5-year survival was lowest in patients with lung cancer (62.9% for stage I, 44.4% for stage II, and 35.3% for stage III).
Disclosure: The study authors’ disclosure information may be found at ascopubs.org.
