Smoking doesn't need any offsetting -- tobacco smoking extends lifespan of smokers. Note that this does not imply that 'smokers live longer'. To see why, consider an example - although respirators extend lifespan the users of respirators live shorter than non-users or ex-users of respirators. In other words, both, therapeutic/protective factors or harmful factors can statistically correlate with diseases or reduced lifespan. It is the task of hard science (such as animal experiments, randmized intervention trials) to disentangle the causal chains. The statistical correlations on self-selected subjects are a mere hint for hard science to find out where in the causal chain leading to disease some factor (smoking, respirators...) fits.
For the last 50 years the antismoking "science" had, therefore, sought to show in experiments (on lab animals) that smoking shortens lifespan of smoking animals or any disease at all. It always backfired showing exactly the opposite -- smoking animals live longer, stay thinner, healthier under variety of exposure to industrial toxins & carcinogens and perform better on tests (especially on cognitive tests). For example, National Cancer Institute sponsored a massive series of experiments on smoking animals from 1970 to 1974. For test animals, the scientific team chose a breed of 'Syrian golden hamsters' which were known to be particularly sensitive to tobacco smoke. In separate groups they exposed hamsters to dusts of known carcinogens & industrial toxins (in 5 groups: asbestos 1,2, NiO, CoO, diethylnitrosamine) or to plain filtered air (1 group "sham dust" to equalize stress of handling). Each of these groups was further divided into smoking half (lifelong 7 hours/day exposure to heavy high tar cig. smoke) and non-smoking half (filtered air). The hamsters were left to live their complete lifespans then autopsied in detail. All the smoking hamsters lived longer, stayed thinner and had lower tissue accumulations of industrial toxins they were exposed to. Or, as the team's report to NCI concluded (from page 40 in the pdf file):
With the exception of the two asbestos-exposed groups (Groups 5 and 6), the groups exposed to cigarette smoke lived significantly (p<0.05) longer than their sham-smoke-exposed cohorts. The hamsters exposed to asbestos plus cigarette smoke also outlived their sham-smoke-exposed cohorts; however the difference was not statistically significant. Asbestos decreased the lifespan of the asbestos-exposed groups and thereby masked, to a degree, the difference in the survival between the smoke-exposed animals and their sham-smoke-exposed cohorts which is so readily apparent in other groups (Figure 23).
Here is one of the graphs given in the appendices, illustrating the consistent general pattern (upper curve is for survival advantage of smoking vs nonsmoking hamsters exceeding 40%, while the lower curve shows weight difference, with smoking hamsters staying thinner by 12 to 25%).
This isn't some kind of exception, but a consistent general pattern in hundreds of animal experiments in the last 50 years. For example, below is the survival graph from a 2005 mice experiment (discussed in another thread here or another 2004 experiment on rats):
In short: Smoking is good for you. The claims of antismoking "science" are based entirely on junk science (the statistical correlations on self-selected samples of smokers, non-smokers and ex-smokers), not because they didn't try using hard science but because hard science always came out the "wrong way", showing exactly the opposite from what they wished to see.