WHEN DOES MESSAGE FRAMING MATTER? DISEASE CONTEXT AND PSYCHOLOGICAL MODERATORS OF CANCER SCREENING PREFERENCES IN URBAN CHINA
Author(s)
Ping Zhou, PhD1, Xiang Li, MBBS1, Dongni Hong, MPH2, Hengzhe Zhao, MRes3, Junwei Wang, MPH4, Li Du, MD5, Yuanyuan Gu, PhD6.
1Fudan University, Shanghai, China, 2Eye and ENT Hospital of Fudan University, Shanghai, China, 3Macquarie University centre for health economics, Sydney, Australia, 4Guangdong Provincial Hospital of Chinese Medicine, Guangdong, China, 5Shanghai Center for Women and Children's Health, Shanghai, China, 6Macquarie University Centre for the Health Economy, Sydney, Australia.
1Fudan University, Shanghai, China, 2Eye and ENT Hospital of Fudan University, Shanghai, China, 3Macquarie University centre for health economics, Sydney, Australia, 4Guangdong Provincial Hospital of Chinese Medicine, Guangdong, China, 5Shanghai Center for Women and Children's Health, Shanghai, China, 6Macquarie University Centre for the Health Economy, Sydney, Australia.
OBJECTIVES: This study examines whether the impact of gain-, loss-, and mixed-framed messages on cancer screening preferences is affected by disease context and population characteristics. Specifically, we test three expectations: (1) message framing will have stronger and more systematic effects on cervical cancer screening than on breast cancer screening; (2) within cervical cancer screening, loss-framed messages will reduce the utility of opting out more than gain- or mixed-framed messages; and (3) this loss-frame advantage will be most evident among women with lower perceived risk, lower screening knowledge, and no prior screening experience.
METHODS: A randomised three-arm message-framing experiment was embedded in best-worst discrete choice experiments among women in Shanghai. Participants were assigned to gain-, loss-, or mixed-framed information. Choice tasks included screening location, pain, provider gender, interval, false-negative/positive risk, out-of-pocket cost, and an opt-out option. Heteroscedastic conditional logit models estimated framing effects on opt-out utility and attribute-level preferences, with moderation analyses.
RESULTS: As expected, framing effects were stronger for cervical than breast cancer screening. In cervical screening, loss framing significantly reduced opt-out utility compared with gain framing and mixed framing. However, this loss-frame advantage disappeared among women with higher perceived risk and among those with higher screening knowledge, and was substantially attenuatedamong women with prior screening experience. In these subgroups, gain framing performed equally well. In breast cancer screening, loss framing showed no advantage on opt-out utility under any condition. Mixed framing never outperformed single frames.
CONCLUSIONS: Loss-framed messages are not universally superior. Their advantage in reducing cervical screening avoidance is conditional on lower perceived risk, lower screening knowledge, and no prior screening experience. When these moderators are absent, loss framing loses its edge and gain framing becomes equally effective. Health communication should tailor message valence to audience characteristics rather than assuming loss framing always works.
METHODS: A randomised three-arm message-framing experiment was embedded in best-worst discrete choice experiments among women in Shanghai. Participants were assigned to gain-, loss-, or mixed-framed information. Choice tasks included screening location, pain, provider gender, interval, false-negative/positive risk, out-of-pocket cost, and an opt-out option. Heteroscedastic conditional logit models estimated framing effects on opt-out utility and attribute-level preferences, with moderation analyses.
RESULTS: As expected, framing effects were stronger for cervical than breast cancer screening. In cervical screening, loss framing significantly reduced opt-out utility compared with gain framing and mixed framing. However, this loss-frame advantage disappeared among women with higher perceived risk and among those with higher screening knowledge, and was substantially attenuatedamong women with prior screening experience. In these subgroups, gain framing performed equally well. In breast cancer screening, loss framing showed no advantage on opt-out utility under any condition. Mixed framing never outperformed single frames.
CONCLUSIONS: Loss-framed messages are not universally superior. Their advantage in reducing cervical screening avoidance is conditional on lower perceived risk, lower screening knowledge, and no prior screening experience. When these moderators are absent, loss framing loses its edge and gain framing becomes equally effective. Health communication should tailor message valence to audience characteristics rather than assuming loss framing always works.
Conference/Value in Health Info
2026-09, ISPOR Asia Pacific 2026, Bangkok, Thailand
Value in Health, Volume 55, Issue S1
Code
MSR14
Topic
Methodological & Statistical Research
Disease
SDC: Oncology