Comparison of Face-to-Face Interview Questionnaires and Medical Records Data for Smoking Habits in Lung Cancer Patients

Background: This study evaluated the accuracy of smoking habit from the data obtained from the medical records of lung cancer patients against the data obtained form face-to-face interview questionnaires Methods: The smoking habits of 225 lung cancer patients were categorized into never smoked, ex-smoker and current smoker in face-to-face interview questionnaire and medical record taken at the time of admission for a diagnosis. The overall agreement between two sources was evaluated. The factors affecting the disagreement between two sources and the level of data omission of the smoking habits in medical records were analyzed suing multiple logistic regression. Results: The smoking habit between two sources showed moderate overall agreement(Kappa (kappa)=0.60). The lowest agreement was observed in the ex-smokers(kappa=0.49). Multivariate analysis revealed an age of 65 or older to be a statistically significant factor associated with the increasing disagreement risk compared with those 64 or younger (OR 3.02; 95% CI 1.58-5.80). The omission rate of smoking habits in the medical records was 18.2%. Adenocarcinoma was shown to be a statistically significant factor of associated with an increasing omission rate compared with squamous cell carcinoma (OR 3.00; 95% CI 1.19-7.59). Conclusion: The smoking habits obtained from medical record moderately reflect their true behavior. However, the smoking habit data from medical record should be used with caution when being used in a clinical study or cohort study of lung cancer.