2nd Edition of Public Health World Conference (PHWC) 2026

Abstract

Background: Sugar sweetened beverages are an important source of free sugar intake and contribute to the burden of noncommunicable diseases. In Hong Kong, nutrition information for prepackaged products is mainly provided through back of package labeling and voluntary initiatives, while evidence on front of package sugar labeling for beverages remains limited. As governments increasingly consider interpretive nutrition labels and pricing policies, local evidence is needed on how consumers respond to different label formats when choices involve competing products rather than a single beverage viewed in isolation. This study examined how adults in Hong Kong trade off price, sugar content, icon based sugar labels, and warning labels when choosing beverages.

Methods: A cross sectional, web based discrete choice experiment was conducted among 536 Hong Kong adults who had consumed sugar sweetened beverages in the past month. Quota sampling was used to approximate the age and sex distribution of the Hong Kong adult population. A D efficient fractional factorial design generated in Ngene produced 48 choice sets, divided into four blocks of 12 tasks. Participants were randomly assigned to one block. Each task presented five beverage alternatives, including soft drink or soda, flavored juice, sweetened tea, sweetened coffee, and energy drink, plus an opt out option. Product attributes included price, sugar content low, medium, or high, icon label traffic light or teaspoon icon, and high sugar warning label none, text warning, or graphic warning. The final dataset comprised 6432 choice sets. Choices were analyzed using conditional logit models with an opt out alternative. The main model jointly coded sugar content and warning label status because warning labels were applied only to high sugar beverages. Relative importance scores, predicted probabilities, and subgroup analyses were used to support interpretation.

Results: Price and the sugar warning profile were the strongest determinants of beverage choice. Compared with high sugar beverages without warnings, high sugar beverages with a text warning had lower odds of selection OR 0.875, p 0.047, while graphic warnings showed a larger reduction OR 0.775, p less than 0.001. Low sugar beverages were more likely to be selected than high sugar beverages without warning OR 1.187, p less than 0.001, whereas medium sugar beverages were less preferred OR 0.767, p less than 0.001. Higher price reduced the likelihood of choice OR 0.976, p less than 0.001. The teaspoon icon was also associated with lower choice odds than the traffic light icon OR 0.895, p less than 0.001, although icon format contributed less to overall attribute importance. Relative importance analysis indicated that the sugar warning profile accounted for 63.2 percent of the explained utility range, followed by price 20.8 percent and icon label 16.1 percent. Predicted probability analyses suggested that high sugar beverages became less competitive when warning labels and lower sugar alternatives coexisted in the same choice set, with the graphic warning producing the lowest predicted probability among high sugar profiles. Preference heterogeneity was more evident for baseline purchase propensity and price sensitivity than for warning label preferences. In post task feedback, 73.5 percent of participants agreed that high sugar warnings would influence their choice, and 75.7 percent supported government implementation of health warning label policies for beverages.

Conclusions: This study provides local evidence from Hong Kong on how consumers trade off sugar information, warning messages, and price when choosing beverages, addressing an evidence gap in Asian urban settings where mandatory front of package sugar warning policies remain limited. By modeling warning labels as threshold based signals applied to high sugar beverages, the findings offer policy relevant insight into how such labels may work in realistic choice environments rather than as isolated information tools. The results suggest that high sugar warning labels, particularly when combined with pricing strategies and the availability of lower sugar alternatives, could help shift consumer preferences away from high sugar beverages. These findings can inform the design of future beverage labeling policies and broader sugar reduction strategies in Hong Kong and comparable settings.