Abstract
Introduction
A Korean National Health Insurance Service (NHIS) cohort study reported higher lung cancer incidence among former smokers who used e-cigarettes after quitting than among those who quit without them. Interpreting that association requires knowing which product was used and how long ago participants quit; the cohort measured neither directly, but the contemporaneous Korea National Health and Nutrition Examination Survey (KNHANES) did.
Methods
KNHANES 2018, a cross-sectional national survey, used the NHIS umbrella e-cigarette term, a separate heated tobacco product (HTP) item, and months since cessation. Former smokers using either product (n = 43) were compared with those using neither (n = 1,305), with replication in 2019–2023.
Results
Of adults affirming current umbrella e-cigarette use, 64% also reported HTP use (former smokers, 50%); 79% of product-using former smokers reported HTP use. Median time since cessation was eight months among product users and twelve years among others (Kolmogorov–Smirnov D = 0.76, P < 0.001). The gap was replicated in every wave through 2023 (D = 0.63–0.76) and persisted in the long-term quitter stratum (pooled 2019–2023 medians, 6 vs 18 years), in Kim et al.’s high-risk subgroup (2 vs 12 years), after age matching (residual 9.0–11.0 years) and among umbrella-item respondents only (D = 0.73).
Conclusions
The reported association is consistent with residual confounding by cessation timing and with product misclassification. It also rests on few exposed events: 16 in the long-term stratum, and 31 of 71 retained when exposure was updated at the next screening. Because at most half of the exposure was exclusive liquid e-cigarette use, an e-cigarette-specific estimate would need about four times as many exposed events for the same precision. Repeated-screening designs that update cessation and product use at every screening, already used for cessation in the same database, would retain the full sample and allow more precise, product-specific dose-response estimates.
Full Text
The Full Text of this preprint is available as a PDF (492.7 KB). The Web version will be available soon.