Reply to Campagna and Caci: Taking for Granted Conclusions from Studies that Cannot Prove Causality of Respiratory Symptoms and Vaping

From the Authors:

Campagna and Caci raised the concern that smoking history was insufficiently adjusted for in our study (1). Although we appreciate their interest in our work, their concerns were, however, unfounded. To evaluate whether e-cigarette use was associated with respiratory symptoms, we started the analyses with assessing an e-cigarette use status (never, former, current) adjusted for cigarette smoking, other tobacco product use, and secondhand smoke exposure, along with other sociodemographic and behavioral covariates. Rather than a binary measure of current smoking status (as the letter suggested), cigarette smoking was specified as never, former, current some days, or current every day. The letter failed to acknowledge that the analyses in our paper were repeated among respondents who had never smoked combustible cigarettes. All respondents in this subgroup analysis would have zero pack-years of cigarette smoking, virtually eliminating the influence of cigarette smoking history. The positive associations between e-cigarette use and respiratory symptoms remained in this subgroup analysis, and the magnitude of the associations was even stronger, presumably due to lower incidence rates in the reference group.

The letter compared our results with those from the Sargent and colleagues study (2) indicating that exclusive e-cigarette use was not significantly associated with respiratory symptoms after adjustment for pack-years of smoking. However, the null associations after adjustment for pack-years of smoking apply only to cross-sectional associations between e-cigarette use and respiratory symptoms, both measured at wave 2. When evaluating worsening of respiratory symptoms over time (i.e., asymptomatic at wave 2 to symptomatic at wave 3), a more similar approach to our analyses, Sargent and colleagues actually showed a significant positive association of exclusive e-cigarette use with respiratory symptoms (relative risk [RR], 1.63; 95% confidence interval [CI], 1.02–2.59) when a respiratory index cutoff ⩾2 was applied and a similar, albeit statistically nonsignificant, result for a cutoff ⩾3 (RR, 1.58; 95% CI, 0.84–2.96). Moreover, Sargent and colleagues reported that the RRs for exclusive cigarette smoking only and dual cigarette and e-cigarette use were 1.93 (95% CI, 1.50–2.50) and 2.20 (95% CI, 1.67–2.89), respectively. These estimations of longitudinal association were, in fact, very similar to what we found in our study examining joint cigarette and e-cigarette smoking patterns (exclusive e-cigarette use, RR, 1.62; 95% CI, 1.23–2.12; cigarette smoking only, RR, 2.07; 95% CI, 1.75–2.46; dual use, RR, 1.88; 95% CI, 1.41–2.51).