Presentation (CTI)

OBJECTIVES: Ascertaining nicotine/tobacco product use (e.g., cigarettes, e-cigarettes, smokeless tobacco) is critical for studying respiratory- and cancer-related outcomes but can be difficult to measure using secondary data sources. Standardized ontologies capturing nicotine/tobacco use are routinely used but have limited granularity. We evaluated concordance of nicotine/tobacco use from standardized ontologies recorded in electronic medical records (EMRs) with self-report survey data on nicotine/tobacco use using the All of Us (AoU) database.
METHODS: We conducted an observational retrospective analysis of adult (≥18 years) AoU participants with completed self-report questionnaires and linked EMR data. Nicotine/tobacco use was defined using ICD-9-CM, ICD-10-CM, and SNOMED CT concepts for nicotine/tobacco dependence. Concordance between self-reported nicotine/tobacco use and codes was evaluated, along with degree of missingness of nicotine/tobacco codes among those with self-reported use.
RESULTS: Among 571,624 patients with survey data between June 2016-September 2023, 33,665 (5.9%) had EMR data with nicotine/tobacco diagnosis codes within one year of surveys. Prevalent codes were “nicotine dependence, unspecified, uncomplicated” (52.9%), corresponding to self-report cigarette, e-cigarette, and smokeless tobacco use of 87.1%, 39.1%, and 17.7% respectively, and “nicotine dependence, cigarette, uncomplicated” (61.4%), corresponding to similar self-report product use. No codes explicitly captured e-cigarette use. Cigarette codes had 87.8-97.7% agreement with self-reported cigarette use. Poor agreement was observed for other products: 27.2-51.2% with non-specific nicotine/tobacco codes and 70.5-72.4% with chewing tobacco codes reported e-cigarette and smokeless tobacco use, respectively. Among patients with EMR data and no recorded nicotine/tobacco codes within one year of surveys (N=90,637), 30.8%, 15.5%, and 8.6% reported cigarette, e-cigarette, and smokeless tobacco use, respectively.
CONCLUSIONS: Standardized nicotine/tobacco dependence codes underestimated cigarette, e-cigarette, and smokeless tobacco use and rarely distinguished specific nicotine/tobacco products, limiting accurate product-level exposure ascertainment. Standardized ontologies alone should be used with caution to discern nicotine/tobacco exposure levels in real-world studies. Researchers should additionally consider triangulation with unstructured EMR or survey data.