Objectives: Cancer patients have a wide range of comorbidities that are important confounders for biomarker and clinical studies of prognosis and outcome. Comorbidities can be captured using the Charlson Comorbidity Index (CCI) through abstraction of medical records, but patient-reported outcome (PRO) questionnaires have also been used. The objective was to validate the PRO-CCI in a head and neck cancer (HNC) population, and to assess its level of agreement with the standard (std-CCI) method of chart review.
Methods: A one-page PRO-CCI was compared with the std-CCI obtained through independent abstraction in 882 HNC patients (2007-2010). Kappa statistics and associated measures (p(pos) and p(neg)) were used to assess agreement. Discrepancy for each comorbid illness was evaluated. Proportional hazard models compared the association of std-CCI and PRO-CCI with overall survival (OS). Adjustments were made and a modified PRO-CCI was re-evaluated in a new cohort of upper aerodigestive tract cancers patient.
Results: PRO-CCI was higher than the std-CCI (p < 0.0001). After adjustment, having at least two comorbidities according to either the std-CCI [HR 1.97 (1.38-2.80)] or the PRO-CCI [HR 1.62 (1.18-2.24)] was prognostic. Of the most prevalent comorbidities, agreement was high for most of the CCI elements (kappa 0.76-0.93), but poorest agreement for connective tissue disease (kappa = 0.29, p(pos) = 43%, p(neg) = 84%) and COPD (kappa = 0.48, p(pos) = 53%, p(neg) = 95%). When the connective tissue disease question was modified, agreement of this item improved (kappa = 0.47, p(pos) = 50%).
Conclusion: PRO-CCI can be an easy and effective tool in prognostic and outcomes research in HNC patients.
Copyright © 2012 Elsevier Ltd. All rights reserved.