Assessing the Accuracy of Registry-based Tobacco Use Status and Utility for Patient Recruitment into Tobacco Trials

Authors:

Paul Krebs (Presenter)
New York University School of Medicine

Erin Rogers, New York University School of Medicine
Howard Wong, New York University School of Medicine
Jamie Ostroff, Memorial Sloan-Kettering Cancer Center
Behnoosh Momin, CDC
Jane Henley, CDC

Public Health Statement: Proactive tobacco treatment approaches have been shown to increase the use of tobacco treatment and long-term abstinence. Cancer registry data offers potential for identifing patients who smoked at time of cancer diagnosis to offer them connection to cessation services. Accurate tobacco use data, however, is essential to inform a successful proactive care approach.

Purpose: To compare cigarette smoking status from the cancer registry with the patient’s electronic health record (EHR).

Methods/Approach: Registry data for patients diagnosed from June 2014 to December 2016 were obtained from 2 cancer centers affiliated with New York University, and smoking status was ascertained. For patients identified as current cigarette smokers in the registry, study staff used data abstracted from EHRs to independently code smoking status.

Results: Of 10,316 cancer patients at one study site, 577 (5.6%) were documented as current cigarette smokers using cancer registry data. Among those categorized as smokers via the registry, data in the EHR identified 71% as current smokers, 25% as former smokers, 5% as never smokers, and 4 as deceased. The EHR listed cigarettes per day for 443 patients and years of smoking for 360 patients. Comparisons at the second study site could not be accurately made for 85 eligible patients as the EHR had neither consistent nor reliable record of tobacco use status.

Conclusions/Implications: Cigarette smoking status in the cancer registry and the EHR were consistent for most, but not all, cancer patients. For registries to be useful in identifying tobacco users, accuracy must improve. Standardized fields and validated tobacco use questions could aid providers in reliably documenting tobacco use in EHR and registry data, thus improving their utility in future studies.