If you look at the actual paper, what they did was interview several hundred hospitalized people in France, and ask them whether they smoked. Six percent of them said yes. The authors say that this is much lower than the average smoking rate in all of France (25%).
Interviews are not an ideal source of data here. For one thing, some patients might deny being smokers because they want to make a good impression and perhaps avoid any triage situations.
Even worse, the researchers collected smoking data only from patients who were NOT in the ICU, because ICU patients are too hard to interview. I would also add that dead people are even harder to interview than ICU patients. Thus, the study tells us nothing about what makes people susceptible to serious illness or death. For all we know, almost all of the French smokers were in the ICU or had already died. It was irresponsible for the Economist to present this study as if it proves that smoking is beneficial.