Background Systematic reviews conclude personal financial incentives (PFI) are effective at changing health-related behaviours. However, PFI may be publically unacceptable, meaning they are unlikely to be widely implemented.

We answered two questions using a discrete choice experiment (DCE): what are the relative preferences of UK adults for attributes of PFI for four health-related behaviours? Do preferences vary with age, gender or educational attainment?

Methods DCEs describe interventions according to their ‘attributes’ (e.g. PFI type), and ‘levels’ of these (e.g. cash, shopping voucher). Participants are asked which of 2–3 ‘scenarios’, combining different levels of each attribute, they prefer. Relative preferences for attribute levels can then be determined.

Participants were adult members of a market research panel, living in the UK, selected using quota sampling (based on age, gender, educational attainment, current smoking and physical activity). Preferences were examined for PFI for: smoking cessation, regular physical activity, and attendance for vaccination and screening. Attributes (and levels) were: PFI type (none, cash, shopping vouchers, lottery tickets); value (continuous variable); schedule (same value each week, value increases as behaviour change is sustained); other information provided (none, written information, face-to-face discussion, both); and recipients (all eligible individuals, people living in low-income households, pregnant women).

Data were collected anonymously online (n = 356) and analysed in Stata v13.0 using conditional logistic regression. Results will be presented as marginal utility values compared to reference for each behaviour, with p-values (confidence intervals are difficult to calculate).

Results Cash or voucher-based PFI were as (ps > 0.05), or more (ps < 0.05), acceptable to participants than no PFI for all behaviours. PFI offering lottery-type rewards and those targeted at particular groups were less acceptable than no reward and untargeted rewards for all behaviours (ps < 0.05). Preferences were inversely related to PFI value. Preferences for additional information provided alongside PFI varied between behaviours. Preferences for no PFI, additional information alongside PFI, and PFI not targeted at particular groups increased with age (ps < 0.05). Men were more likely than women to prefer any PFI over none, and PFI targeted at those living in low-income households (ps < 0.05). There were no consistent differences in preferences by educational attainment.

Conclusion We used a convenience sample, but increased representativeness with quota sampling. PFI for four healthy behaviours were as or more acceptable than no incentives to UK adults. To maximise acceptability, PFI should be in the form of cash or shopping vouchers, and not lotteries; be of low value; and available to all.