Consumers often use cosmetics long after their expiry date, despite the health risks. This paper aims to understand why this is the case and to suggest policy changes that can promote safer practices in the use of cosmetics. This is the first study to examine risk perceptions in relation to expired cosmetics. Thirty-three semi-structured interviews were conducted with cosmetics users and employees of cosmetics companies in the UK and China. Perceived risk theory was found to be a useful analytical perspective. Eight risk factors emerged from the data, including two that had not been previously identified. It was also found that risk combinations were valuable in explaining consumer attitudes towards cosmetics expiry dates, suggesting that perceived risk factors interact with each other to create an emergent understanding of risk that requires an integrated understanding. While physical, functional, and personal brand risk can increase adherence to expiration dates, other risk factors such as financial and social risk can override such concerns and lead to expiration date neglect. Implications for supplier and regulatory policies and risk communication strategies have been identified that can help reduce risks to cosmetic users.
Using cosmetics past their expiration date can pose a significant health risk, as the quality and safety of many of these products decline over time, especially after opening (FDA, 2022). There are two types of cosmetic expiration dates: unopened shelf life and period after opening, or PAO, which is the shelf life after opening. This article focuses on PAO because, while unopened cosmetics can spoil, chemical changes and/or contamination of cosmetics usually occur after opening (Skowron et al., 2017). There is no global standard for expiration date labels on cosmetics, but similar approaches exist around the world. In Europe, cosmetics with a shelf life of more than 30 months must have a PAO date (Cosmetics Europe, 2023).
One of the reasons for using age as a criterion for selecting consumer respondents was our expectation of the role of age in decisions about disposal. In Section 3, we present several quotes where consumer age appears relevant (FR3, PHR1, PHR4, PSR1, SR2 and SR3). In some, they show that older women are more likely to adhere to PAO (PHR1, PHR4 and PSR1) and in others the opposite (FR3, SR2 and SR3). This suggests that although the influence of individual risk factors on specific disposal decisions may differ by age, this may not be the case systematically.
We conducted interviews in two countries. Overall, there was little difference in the content of the two sets of interviews, except that the ENR was obtained from interviews conducted in the UK rather than China, and we draw on previous work to support the idea that the importance of this dimension may be somewhat country-specific. We used individual and group interviews, a mixed-methods approach that may seem controversial, although it is generally accepted in qualitative research (Morse, 2009; Patton, 2002). Here, this method was useful when respondents knew each other well and insights were able to emerge that might not have otherwise emerged (e.g., SBCR4 and SBCR5 in Section 3.7). Researchers should be able to use multiple methods if their circumstances warrant (Morse, 2009).
While we focus on providing quotes from the second case, the salespeople were helpful in providing general comments that improved our understanding of the industry and consumer behavior and in providing a different perspective, as illustrated in quotes SR2 and SR3.
Similarly, interviews with two senior managers provided higher-level insights and helped validate our findings. As was common for much research at the time due to COVID, we used both face-to-face and online interviews, but found no significant differences in the insights gained from the two, although we agree with Wicklin et al. (2024) that analysis would be easier with video recording.
Our sample only included 33 respondents in two countries, but we feel we can make policy recommendations as our work identifies why consumers may ignore PAO expiration dates. Further research into the extent of infection from the use of expired cosmetics would still be useful. Our sample was exclusively female and further work could usefully include male respondents as well, as their attitudes and motivations for using cosmetics differ (Sweden & Dian, 2009). This research was conducted only in the UK and China, and research in other cultures, including those where religious beliefs and cultural traditions can influence cosmetic use, would be useful.
Our primary goal was to understand why cosmetic users might ignore expiration dates and not justify any particular risk perception framework, avoiding the challenge of confirmation bias. We add to this theory by noting that risk factors appear to interact, and here are examples of cases where an issue can be fully addressed by simply addressing more than one risk factor. Future quantitative work should explore