It’s true that no particular death can be called the effect of smoking. All that can be said is that a particular person who died from a given disease and who smoked probably died because of their smoking. Deaths from smoking are thus like not like deaths from automobile accidents, deaths from shooting, deaths from falls from a great height, etc.
Overall it is possible to estimate the number of deaths from smoking though. Many studies have been done showing that smokers die from many different diseases at a rate that’s considerably greater than nonsmokers from those diseases. Please note that this is many different diseases, not just lung cancer. Lung cancer isn’t even the majority of the extra deaths among smokers. Many kinds of cancer, many kinds of heart and lung problems, and many other diseases are indirectly caused by smoking.
What is done (or at least should be done) is the following: For each disease for which it has been shown that smokers are more likely to die than nonsmokers, the number of smokers who died in the given year from that disease is worked out. Then it’s calculated how many of them would have died if they were nonsmokers. The difference is the number of extra deaths caused by smoking. Then add up these numbers for all the diseases which cause extra deaths among smokers. This is the number of extra deaths among smokers that year.
If you’re saying that this isn’t satisfactory to you, well, tough. That’s the way it works for conditions like smoking that increase the number of deaths but don’t appear as the “cause of death” on death certificates. Being eaten by a shark may be a direct cause of death, but it’s very rare. Dying from the effects of smoking may be an indirect cause of death, but it’s very common.