Housing evidence / case study · working draft
San Francisco: incentives and homelessness
Rewards were linked to verified quitting, rather than attendance alone.
01 / THE APPROACH
What was tried
This San Francisco pilot recruited 83 adults experiencing homelessness through three safety-net clinics. Participants were randomly allocated to six months of escalating gift cards for verified quitting or to small payments for attending visits.
In the incentive group, rewards began at $13 and increased by $0.50 for each negative carbon-monoxide test. Comparison participants received $5 per visit regardless of smoking status. Repeated testing made the distinction between rewarding attendance and rewarding abstinence practical and measurable.
The study examined whether this longer reward programme was feasible and showed promise before a larger trial.
02 / THE FINDINGS
What was learned
At six months, 31.0% in the cessation-incentive group and 12.2% in the attendance-payment group were verified as not smoking during the previous seven days.
This is an encouraging result from a small trial, with uncertainty around the size of the effect and challenges retaining participants. It is a seven-day snapshot at six months, not evidence that participants remained continuously smoke free for six months. Delivery also required continuing testing and payment administration.
03 / TRANSFER TO WALES
What Wales could learn
Rewards depended on repeated carbon-monoxide tests, with a clear payment schedule. That required clinic appointments, testing capacity and a way to issue and track gift cards.
The promising result is a seven-day smoking measure at six months. Partners can learn about verification and administration without assuming the same reward schedule will work elsewhere.
Sources & review
Gu et al., Contingency Management for Smoking Cessation Among Adults Experiencing Homelessness: A Pilot Randomized Clinical Trial. JAMA Network Open 9(8):e2627917 (2026), DOI 10.1001/jamanetworkopen.2026.27917. Open-access full study. Source checked 23 October 2026. Seven-day abstinence included CO verification; it is different from the study’s CO-only measure. Six-month follow-up was available for 42/83 participants.
