I wish I could find it now. Stupidly, I didn't print it out when I first ran across it back in 2003 or 2004, [but it's since been deleted — note: see my post #22 below]. I know I read it. And Charles Krauthammer mentioned it several times in passing, both in print and on the air. And my internist mentioned the study to me. So I'm sure it wasn't a figment of my imagination.
The article appeared in the New England Journal of Medicine, which is peer-reviewed and generally regarded as authoritative. The authors, whose names escape me, were all well-regarded epidemiologists.
The principal point was that, if everyone quit smoking, or tobacco had never existed, and all other epidemiological factors were held even, the national health care bill would be about 18% higher than it was at that point, with 20% of adults smoking.
The majority of our health care expenses are concentrated in end-care, the last year(s) of life. And smokers tend to die cheaply: lung cancer (few operations, and the chemotherapy regimes are relatively inexpensive), critical cardiovascular or cerebrovascular events (heart attacks and strokes, which require intensive but extremely short-term care), etc. Smokers tend not to die of Alzheimer's disease, Parkinson's disease, and other long-term degenerative conditions, which require long-term, 24/7, extremely expensive care.