Smoking and healthspan: what quitting actually recovers

Smoking occupies an unusual position in population health research. Almost every other risk factor is contested at the margins — how much alcohol is too much, whether saturated fat matters, what the optimal BMI is. Smoking isn't. The effect size is large, the dose-response relationship is clear, the mechanism is understood, and the finding has replicated in essentially every population studied for seventy years.

The scale of the effect

The landmark evidence here is the British Doctors Study, which followed a cohort of male physicians for fifty years. Its headline finding, published by Doll and colleagues in 2004, was that persistent cigarette smoking from early adult life shortened life expectancy by around ten years compared with never-smoking. Later analyses in larger and more diverse cohorts, including work published in the New England Journal of Medicine in 2013, have produced broadly similar estimates — roughly a decade of life expectancy for lifelong smokers.

For healthspan specifically the picture is, if anything, worse. Smoking is causally implicated in conditions that are strongly disabling rather than immediately fatal: chronic obstructive pulmonary disease, peripheral vascular disease, and the long functional decline that follows a non-fatal stroke. These conditions do not simply end life; they consume years of it.

Why smoking costs healthspan more than lifespan

Consider what smoking does mechanistically. Combustion products damage vascular endothelium throughout the body, drive systemic inflammation, and destroy lung tissue in a way that does not regenerate. Each of those pathways produces conditions with long symptomatic courses.

COPD is the clearest example. It is progressive, largely irreversible, and its defining feature is breathlessness that constrains everyday activity for years or decades before it contributes to death. A person with moderate COPD may live a long time — but a meaningful portion of that time involves a materially reduced capacity to walk uphill, carry shopping, or sleep flat.

This is why smoking carries the largest negative weight in the Helixspan model even though it's not the factor with the largest effect on raw mortality in every cohort. The question the model asks is about healthy years, and smoking is unusually efficient at destroying those specifically.

What quitting recovers, and when

The recovery evidence is the genuinely encouraging part of this literature, and it's underappreciated. Risk does not simply stop accumulating at the point of quitting — a substantial portion of accumulated excess risk reverses over time.

An important asymmetry: vascular risk reverses substantially; structural lung damage largely doesn't. Emphysematous destruction of alveolar tissue is permanent. Quitting halts the progression rather than repairing what's gone. This is a real argument for quitting sooner rather than a reason not to bother — the amount of irreversible damage is a function of how long you continue.

The age-40 finding

One result gets cited constantly and deserves precise statement. Analyses of large cohorts — including work by Jha and colleagues published in 2013 — found that people who stopped smoking before roughly age 40 avoided the large majority of the excess mortality risk associated with continued smoking, with some estimates putting the figure around 90%.

Two clarifications, because this finding is frequently overstated in both directions. It does not mean quitting before 40 makes you equivalent to a never-smoker — residual excess risk remains, particularly for lung cancer. And it does not mean quitting after 40 is futile: the same body of work shows meaningful gains from quitting at 50, at 60, and later. The gradient is continuous. Earlier is better; later is still substantially better than not at all.

"I only smoke a little"

The dose-response curve for smoking is real but importantly non-linear at the low end. Light and intermittent smoking carries substantially less risk than heavy smoking, but considerably more than the linear intuition suggests — the harm per cigarette is highest for the first few. Several large analyses have found that smoking even one to five cigarettes a day is associated with meaningfully elevated cardiovascular risk, at a level well above one-twentieth of a pack-a-day smoker's excess risk.

The plausible explanation is that cardiovascular harm involves mechanisms — endothelial dysfunction, platelet activation — that saturate at low exposure. There is no established threshold below which cigarette smoking has been shown to be free of cardiovascular risk.

Where the evidence is genuinely unsettled

We'll be direct about the limits of what's known. E-cigarettes avoid combustion, which removes a large fraction of the toxicants responsible for smoking's harms, and short-to-medium-term biomarker studies generally show substantial improvement in smokers who switch completely. What does not exist is long-horizon cohort data — the kind of forty-year follow-up that established the harms of cigarettes in the first place. Anyone quoting a confident figure for the lifetime healthspan cost of vaping is extrapolating.

The Helixspan model does not attempt to score vaping, because we don't think the evidence supports a defensible weight. Users who have switched from cigarettes to vaping are in a genuinely ambiguous position relative to the model's categories, and we'd rather say so than invent a number.

How the model scores it

The calculator applies four smoking categories:

• Never — no modifier. This is the reference state.
• Former — a −2 year modifier, representing average residual risk across a population of ex-smokers with varying quit durations and prior intensities.
• Light — a −5 year modifier, reflecting the non-linear low-end curve described above.
• Heavy — a −8 year modifier, the largest single negative weight in the model.

The "Former" category is the crudest of the four, and worth flagging. Somebody who smoked five a day for two years and quit at 25 and somebody who smoked a pack a day for thirty years and quit last month both land in it, despite being in very different positions. A more sophisticated model would take pack-years and years-since-quitting as separate inputs. We've kept the simpler version because self-reported pack-year estimates are unreliable, but the imprecision is real and you should read a "Former" result with that in mind.

If you're currently in the light or heavy category, this factor is almost certainly the highest-leverage change available to you — the calculator's recommendation engine will surface it first for exactly that reason.