We already know that the prospect of earning some cash helps people quit smoking.

But a new study from the University of Pennsylvania found that the way cash rewards are structured can make a big difference in how well such programs work.

People were far more likely to quit for six months if they had some skin in the game. The problem was getting them to invest some of their own money - they got it back if they stopped smoking - in a reward program.

The results were striking enough, though, that CVS Health, which helped fund the study, is launching a deposit-based smoking-cessation program for employees next month.

Called 700 Good Reasons, the program asks employees or family members who smoke to deposit $50 of their own. If they stop smoking, they will get the $50 back, plus $200 of the company's money at six months and $500 more at 12 months.

CVS stopped selling tobacco products in its stores last year.

Scott Halpern, deputy director of Penn's Center for Health Incentives and Behavioral Economics and lead author of the study, said smoking is a huge issue for employers, costing them an extra $200 billion a year. On average, smokers cost companies an extra $4,000 to $6,000 in health-care expense, absenteeism, and reduced efficiency. They effectively raise insurance premiums for everyone. The federal Medicare program faces a high burden as smoking-related health problems worsen in later life.

Halpern said 18 percent of Americans smoke.

Most companies already offer employees wellness programs to improve health behavior. Programs that offer financial incentives often assume that the amount is the key factor, Halpern said.

The new study, published online Wednesday in the New England Journal of Medicine, tested the premise that "how you distribute dollars may be as important . . . as how many dollars you distribute," Halpern said.

The study enrolled 2,538 CVS employees, family members, and friends from throughout the country. They were divided into five groups. One received "usual care," including information about quitting smoking and free smoking-cessation aids. Two other groups also received $800 rewards, based either on individual or group performance. The remaining two required participants to make a $150 deposit, with final rewards based either on individual or group performance. In those categories, successful quitters got their deposits back plus $650.

To their surprise, the researchers found that groups were no more successful than individuals. (In this case, the people in groups did not know each other.)

But the type of reward mattered. At six months, only 6 percent of the usual-care group were not smoking. Almost 16 percent of those in pure reward groups had stopped, compared with 10 percent of those in the deposit-based groups.

Yet those numbers mask an important fact, Halpern said. Ninety percent of people offered the $800 reward jumped at the chance, while only 14 percent of people assigned to the deposit programs actually paid the $150. Of those who paid, however, 52 percent were not smoking at six months - a remarkable number, because addictions are notoriously difficult to kick.

Obviously, people who paid the money were exceptionally motivated to quit. To control for that, the team did what Halpern called a "super-nerdy, highly technical analysis." It compared what happened to people who would have participated in any group. In that case, the quit rate at six months was 17 percent for the people who received only rewards and 29 percent for those who made deposits before they got the rewards.

That difference is what persuaded CVS to go with the deposit approach, but with a lower initial investment for the employee. "It's very clear that $150 was too much," Halpern said.

Asked why employers should not just make smokers pay more for their insurance until they quit, Halpern said some companies do that, but usually employees don't pay the price until the next calendar year and even then, the penalty is not explicit. To be effective, he said, either penalties or rewards need to be "visible and mindful."

Clearly, most smokers were still smoking even after the Penn programs. "Incentives are not a panacea," Halpern said. "We have yet to come up with a panacea for smoking cessation. It's a very difficult nut to crack."

In an editorial that accompanied the journal article, Cass Sunstein, a Harvard law and behavioral economics expert, said the study demonstrated the importance of "loss aversion."

People tend to dislike losing money more than they like getting it. The challenge, Sunstein wrote, is to get more smokers to conquer their fear of losing enough to enroll in the more effective program. If that's not possible, he said, "reward programs are much better bets."

215-854-4944 @StaceyABurling