Comorbidity in Lung Cancer: A Prospective Cohort Study of Self-Reported versus Register-Based Comorbidity

    loading  Checking for direct PDF access through Ovid

Abstract

Objective

Comorbidity is a debated prognosticator in lung cancer. Results are conflicting, and the variety in study designs and inclusion criteria hampers direct comparison of available studies. So far, methods for generation of data on lung cancer comorbidity have attracted little attention. We evaluated whether self-reported comorbidity and register-based comorbidity are of comparable quality.

Methods

In a prospective study, we evaluated paired data sets on self-reported versus register-based comorbidity to detect whether any data asymmetry could be related to the data collection methods. We evaluated the Charlson comorbidity index and the simplified comorbidity score as predictors of overall survival (OS) in lung cancer to test whether our findings would affect outcome.

Results

In a cohort of 336 patients with lung cancer and 125 patients in whom cancer was suspected but who were determined to be cancer-free, we demonstrated a significant underreporting of self-reported comorbidities compared with register-based comorbidities. Furthermore, we demonstrated that an association between comorbidity and OS can be detected only by using register-based data and only by the simplified comorbidity score (hazard rate = 1.6, 95% confidence interval: 1.0–2.5, p = 0.03). The results remained significant after adjustment for the relevant clinicopathological characteristics (hazard rate = 1.8, 95% confidence interval: 1.1–3.3, p = 0.028). An association between comorbidity and OS could not be detected by using either the self-reported data or the Charlson comorbidity index.

Conclusions

In this cohort of patients in whom cancer was suspected, comorbidity estimations varied depending on the data collection method used; similarly, the data collection method could affect the association between comorbidity and OS.

Related Topics

    loading  Loading Related Articles