Chen Chen and colleagues recently published Assessing the association between e-cigarette use and cardiovascular disease: A meta-analysis of exclusive and dual use with combustible cigarettes that found significantly elevated cardiovascular disease risk in dual users (people who use both e-cigarettes and cigarettes) and former smokers who had switched to e-cigarettes compared to people who had never used e-cigarettes or cigarettes.
Their comparison of sole e-cigarette use compared to people who had never used e-cigarettes or cigarettes showed an elevated risk, but it did not reach statistical significance.
Despite differences in methodology, their results are broadly consistent with our meta-analysis, which found elevated cardiovascular disease risks for current e-cigarette and dual use compared to people who were not currently using either product.
They make two important points in their Discussion, both of which contradict the FDA’s assumption that dual use is no worse (and perhaps less risky) than just smoking and that switching completely lowers CVD risk:
Our [Chen et al] results show a positive association between individuals who engaged in both e-cigarette and combustible cigarette use, which presented as 2–3 times increased odds of CVD compared to never use of either product. This finding underscores an additive nature when both forms of tobacco product use are concurrently practiced. Moreover, there was 2 times the odds of CVD observed among those who currently used e-cigarettes while having previously used combustible cigarettes compared to never use of either product, which shows consistency in the association between tobacco product use and cardiovascular health.
The FDA needs to update its risk assumptions about e-cigarettes and cardiovascular disease.
WARNING: The following discussion is for statistical nerds.
Risk of e-cigarettes alone
Chen et al’s failure to reach statistical significance for ecigs vs never users of ecigs or cigs was probably due to the fact that they used crude (unadjusted) odds ratios in their analysis, which means that the 95% confidence intervals for the individual studies were wider than in the estimates adjusted for potential confounders. (Age is probably the most important one, a point they touch on in their Discussion [see below].) Also, the fact that their reference group excluded former e-cigarette users and former smokers lowered the sample size, which increases uncertainty (i.e., broader confidence interval) and increases the chances of failing to detect a real effect (lower statistical power).
Age is a major risk factor of CVD and CVD is most common in people over 50 years. On average, e-cigarette users are much younger than smokers, so the unadjusted risks for (younger) e-cigarette users are expected to be lower. Chen et al recognized this fact when they noted that the failure to reach statistical significance for ecig only risks may have been due to not enough time passing for the relatively young ecig users: “The absence of statistical significance in this subgroup might stem from the potential risks associated with e-cigarette use not fully emerging or being detectable within the timeframe covered by our analyzed studies.”
In contrast, we used adjusted ORs, which provide more precise estimates reflected in narrower confidence intervals for the pooled estimate of risk and all the studies included in our analysis accounted for age (among other potential confounders). In addition, the fact that our reference group was people who were not currently using e-cigarettes or cigarettes meant we had a larger sample size. As a result, the confidence interval for the pooled estimate will be narrower when basing the analysis on adjusted ORs and smaller sample sizes.
This is exactly what the data show. In particular for e-cig only Chen and colleagues found OR = 1.24 (95 % CI: 0.93, 1.67) — which does not reach conventional statistical significance because the lower bound of the 95% confidence interval is below 1.00) — and we found 1.24 (1.05–1.46) — which does reach conventional statistical significance because the lower bound of the 95% confidence interval is above 1.00. The narrower CI is why we reached statistical significance based on the same point estimate of risk. Note that we treated stroke as a separate outcome. Including stroke (which they did) with the other outcomes, will also increase variability and widen the CI.
Another difference between their and our analyses is that Chen and colleagues used people who never used e-cigarettes or smoked cigarettes as the reference group and we used people who were not currently using e-cigarettes or smoking cigarettes as the reference group (which includes never and former users). In addition, Chen et al did a sensitivity analysis using the same reference group (non-current e-cig users and smokers) we did and found that the lower bound of the 95% confidence interval increased to 1.00 (OR 1.29; 95% CI 1.00-1.68), which is statistically indistinguishable from our result.
Risk of dual use
Chen et al’s estimate of the risks of dual use vs never use of ecigs or cigs was also similar to what we found (Chen: OR 2.56; 95% CI 2.11-3.11. Ours: OR 2.23 95% CI 1.59–3.14). Chen again did a sensitivity analysis using the same reference group we did (non-current users) yielded similar results: OR 2.35 (95% CI 1.86-2.97).
Risk of e-cigs among former smokers
Importantly, they found that current ecig use among former smokers vs never ecigs or cigs was increased: OR 2.02 (95% CI 1.58-2.58). (We did not do this comparison.)
Methodological differences between Chen et al and our paper
Here are the methodological differences between their meta-analysis and ours:
- The reference group in their main analysis was people who never used e-cigarettes and never smoked; our reference group was people who were not currently using e-cigarettes or cigarettes.
- They considered stroke part of cardiovascular disease; we treated it as a separate outcome.
- The literature searches were different and had different end dates. They searched through April 22, 2024; we searched through October 1, 2023).
- The inclusion and exclusion criteria for studies were slightly different.
- They treated different results from the same study for different outcomes (coronary heart disease, myocardial infarction and stroke) as if they were independent studies. Doing so violates the assumption that each study is independent of the others. We only selected one outcome per study to avoid possible double counting. (As noted above, we treated stroke as a separate outcome.)
- Our primary comparisons were of e-cigarette use to cigarette use and dual use to cigarette use (although we reported e-cigarettes and dual use vs. non-use); they did not do this comparison.
These differences are well within variations in approaches people take to meta-analysis. As noted above, the fact that the results were so similar in the two papers and that there are simple explanations for the difference in the e-cigarette only risk despite these differences is the important point.
Here is the Chen et al abstract:
Background: Growing evidence highlights the impact of e-cigarette use on cardiovascular health, prompting a crucial examination of its association with cardiovascular disease (CVD) in both exclusive e-cigarette and dual use scenarios with combustible cigarettes. This meta-analysis assessed the association between e-cigarette use and CVD by synthesizing the existing literature.
Methods: Pertinent observational studies were identified using multiple electronic databases, from August 22nd, 2006, to April 10th, 2024. A meta-analysis was conducted using random-effect models. Risk of bias was assessed using the National Institutes of Health (NIH) Study Quality Assessment Tools.
Findings: A total of 20 observational studies involving 8,499,444 participants were included in the meta-analysis. Dual use (e-cigarettes and combustible cigarette) increased the odds of CVD by 2.56 times (95 % CI: 2.11, 3.11) compared to never use of both. Current e-cigarette use combined with former combustible cigarette increased the odds of CVD by 2.02 times (95 % CI: 1.58, 2.58) compared to never use of either. Exclusive current e-cigarette use did not show a statistically significant association with CVD odds compared to never use of either (OR = 1.24, 95 % CI: 0.93, 1.67).
Conclusions: Dual use of e-cigarettes and combustible cigarettes was significantly associated with CVD, but results failed to show a significant association between exclusive e-cigarette use and CVD. Robust and longitudinal studies are needed to understand the long-term implications of e-cigarette use and CVD. Public health efforts should focus on awareness, smoking cessation, and regulating both e-cigarettes and combustible cigarettes.
The full citation is: Chen C, Huo C, Mattey-Mora PP, Bidulescu A, Parker MA. Assessing the association between e-cigarette use and cardiovascular disease: A meta-analysis of exclusive and dual use with combustible cigarettes. Addict Behav. 2024 Jun 8;157:108086. doi: 10.1016/j.addbeh.2024.108086. Epub ahead of print. PMID: 38917766. It is available here.
Stanton Glantz blog 