In the NewScientist of 6/30/12, in a special report on lung cancer by Linda Geddes, on p. 8:
… discovered that people who have never smoked survived for around twice as long with lung cancer as current or former smokers.
The never-smokers were more than twice as likely as current or former smokers to possess mutations in the EGFR gene and six times as likely to carry a mutation in ALK.
The unusual compound never-smokers compacts the phrase people who never smoked, to provide a counterpart to smokers ‘people who smoke (now)’ and nonsmokers ‘people who don’t smoke (now)’. But Adv + N compounds (incorporating the semantics of an Adv + V phrase) are uncommon.
There are many hits, including a piece on an NIH site, from the journal Addictive Behaviors in 2004, about the definition of never-smoker:
Defining a never-smoker: results from the nonsmokers survey.
The U.S. Centers for Disease Control defines a never-smoker as someone who has smoked < 100 cigarettes per lifetime.
… Our data indicate a graded effect but also suggest that 19 cigarettes per lifetime is a more conservative cutpoint than 99 for defining the never-smoker phenotype.
What never? Well, hardly ever.