What this page opens (provisional cell-opener). The e-cigarette is a distinct exposure from combustible tobacco -> Smoking and Mortality, and the wiki held nothing on it. This page opens the e-cig -> cardiovascular cell on a single moderate SR+MA (Skotsimara et al., 2019), and its finding is mostly about which evidence state the cell is in, not a magnitude to act on. Confidence low; the value is stating the state honestly rather than letting the e-cig question sort into either «safe» or «established harm». The decision-change is a two-stratum split (below), not a scalar verdict.
The evidence is in three different states at once
| Endpoint | What the evidence shows | Evidence state |
|---|---|---|
| Acute HR / SBP / DBP (surrogates) | all three rise minutes after use, pooled + significant | harm signal (surrogate) |
| BP when a smoker switches (surrogate) | SBP/DBP fall vs continued smoking; HR unchanged | benefit vs the smoking comparator (surrogate) |
| MI / stroke / HF / mortality (hard) | one biased observational MI association; nothing else | INSUFFICIENT evidence |
The single scalar question «are e-cigs bad for the heart?» has no answer because these three states do not collapse — the direction depends on the endpoint and the comparator. (inferred from Skotsimara et al., 2019)
Acute: a real surrogate harm signal
Minutes after use, e-cigarettes raise all three haemodynamic surrogates (Skotsimara et al., 2019):
- HR +2.27 bpm (95% CI 1.64 to 2.89) across 11 studies (N=273), heterogeneous (I2=70%) but robust — dropping the high-weight outlier strengthens it to +4.48 (3.30 to 5.67) and removes the heterogeneity.
- SBP +2.02 mmHg (0.07 to 3.97), DBP +2.01 mmHg (0.62 to 3.39), both homogeneous (I2 ~0-16%).
These are surrogates, not outcomes -> Surrogate Outcomes: a moved marker is a signal of harm, not harm demonstrated. Their transmission to any patient-important CV endpoint is itself an unevidenced claim here — the acute rise is small in absolute terms and its clinical meaning over years is exactly what the hard-outcome row cannot supply.
Switching: BP falls — but against the smoking comparator
In the 3 studies (N=173) of tobacco smokers switching to chronic e-cig use, BP fell: SBP -7.00 mmHg (-9.63 to -4.37), DBP -3.65 mmHg (-5.71 to -1.59), HR unchanged (-0.03, -2.57 to 2.52) (Skotsimara et al., 2019). The authors frame it as the composite finding: «electronic cigarette smoking leads to increases in HR and BP acutely post-exposure but switching from tobacco smoking to chronic electronic cigarette use may beneficially affect BP» (Skotsimara et al., 2019). The comparator here is continued combustible smoking, so this is a harm-reduction signal relative to smoking — not evidence that e-cig lowers BP against clean air.
The decision-change: substitution valence flips by stratum — do NOT collapse
The same exposure carries opposite valence depending on what it replaces (the substitution rule — judge against the realistic alternative). (inferred from Skotsimara et al., 2019)
- Smoker switching (comparator = combustible cigarettes). The realistic alternative is continued smoking, whose CV hazard is enormous -> Smoking and Mortality (all-cause HR ~3). If e-cig is even modestly less harmful, switching is a harm-reduction move — the switch-arm BP fall points that way. But «less bad than smoking» is not «safe», the hard-outcome comparison is unproven, and full cessation dominates both.
- Never-smoker starting (comparator = nothing / clean air). The realistic alternative is no exposure. Here the acute surrogate harm has no offsetting benefit — starting adds a new CV signal for zero gain. This is the stratum the marketing-as-safe framing most endangers.
A single «e-cigs are safe / unsafe» verdict is wrong because it averages these two decisions. Skotsimara holds both apart — the switching benefit «does not suggest that the electronic cigarette should be marketed as a cardiovascular safe product» (Skotsimara et al., 2019).
Mechanism (directional, discounted)
Acute effects plausibly run via nicotine -> catecholamine release. But the mechanism does not close: in large trials nicotine-replacement gum «did not adversely affect blood haemodynamics or the risk for major adverse cardiovascular events», so the adverse e-cig effects «could be related to other substances … released with heating, such as heavy metals or other unknown ones» (Skotsimara et al., 2019). Consequence (): if the harmful moiety is a non-nicotine aerosol constituent, then nicotine-strength comparisons across brands do not bound the CV risk — a directional inference, not an outcome finding.
The gap that keeps this at insufficient-evidence
The cell cannot graduate from insufficient-hard-outcome on what is held. The only hard endpoint is a single observational MI association — OR 1.79 (1.20 to 2.66) vs 2.72 (2.29 to 3.24) for combustible — which the authors flag as «sensitive to non-random misclassification bias» (post-MI switching inflating it, i.e. reverse causation), and «there is no epidemiological data on the risk for stroke or heart failure incidence in electronic cigarette users» (Skotsimara et al., 2019).
G-gap (named, un-held): no meta-analysis or large cohort of e-cig -> hard CV events (MI, stroke, HF, CV mortality) is held or, as of this source’s 2017 search, existed. Until such evidence lands, the cell stays at insufficient-evidence -> The Insufficient-Evidence Statement — not «no effect» (the acute surrogate signal forbids the null) and not established harm (the hard endpoint is unproven). This is the «not yet» bucket under symmetric standards: e-cig gets neither a fashionable-safety pass nor an inflated hard-harm verdict.
Why confidence is low
- One moderate, surrogate-only source. The evidence «is only of moderate quality, derived mainly
from non-randomized observational studies» (Skotsimara et al., 2019); N=441 total, acute windows 5-30 min, only 3 switch studies. Tier
moderate(registry override at ingest). - Surrogate-to-outcome gap unproven — every meta-analytic endpoint is a marker.
- Fixed-effects pooling on I2=70% (acute HR) understates uncertainty; the outlier-drop sensitivity analysis partly mitigates.
- Layer-1 placement: for a smoker, this is a small refinement behind the big rock of quitting entirely -> Smoking and Mortality; for a never-smoker, a reason not to start. Either way the marginal decision is modest and the certainty low.
Self-critique [run 2026-08-30, before commit]
- No hard-outcome harm overclaim. The MI OR is presented only with its reverse-causation caveat and never as established harm; the page’s harm claims are explicitly scoped to acute surrogates, and the hard-endpoint row is labelled INSUFFICIENT throughout. The authors’ own not-safe conclusion (quoted above) is reported as their bottom line, not restated as a wiki finding of harm.
- No scalar safe/unsafe verdict. The page’s spine is the refusal to collapse — the three-state table and the two-stratum split both exist to block a single verdict; the switching benefit and the not-safe conclusion are held side by side, as the source holds them.
- Provisional-C, not banked. Single-source cell-opener; the finding (e-cig CV cell = insufficient hard outcome + acute surrogate signal + stratum-split valence) is RAG-reachable from this one source, so it is scaffolding-grade, not an emergent cross-source synthesis. Confidence low, flagged.
- Substitution valence is the source’s own structure, not an imposed frame — Skotsimara reports the acute and switch arms separately and states both the benefit and the not-safe conclusion; the wiki adds only the explicit never-smoker vs smoker comparator naming (tagged INFERRED).