Reducing meat consumption is critical for planetary health and could benefit public health. Restructuring food environments – such as increasing vegetarian meal availability – has been shown to influence dietary choices in online and university canteen settings. However, evidence from more diverse populations is limited. We conducted a stepped-wedge cluster-randomised controlled trial in six English worksite cafeterias over seven weeks. The intervention involved replacing one meat-based meal on the lunch menu with a vegetarian option. Likelihood of selecting a vegetarian meal (primary outcome) was analysed in a logistic regression with random intercept for cafeteria and fixed effect for time. Secondary outcomes included nutritional and sustainability markers (e.g., kilocalories, greenhouse gas (GHG) emissions), revenue, and food waste. Interviews with cafeteria staff and customers explored intervention acceptability and barriers. The intervention encompassed 26,170 meal sales over 42 site-weeks. The intervention increased the likelihood of selecting a vegetarian meal by 41% 95% CI: 28 to 55. Per-meal kilocalories were reduced by 26.1 95% CI -34.4 to -17.7, and per-meal GHGs by 0.16 kg CO2-eq -0.22 to -0.11. No significant negative impacts on business outcomes such as cafeteria revenue (-98.43 GBP -436.03 to 239.17) or food waste (-8.25 kg -48.38 to 31.87; -61.64 GBP -306.80 to 183.53) were found. Customers and staff reported that the intervention was acceptable, non-intrusive, and easy to implement, with concerns about customer dissatisfaction and food waste not confirmed by our data. Suggestions for improvement included greater attention to pricing and taste to further encourage uptake. Increasing the availability of vegetarian meals in cafeterias can significantly shift food choices and reduce environmental impact and calories consumed, without compromising business outcomes, and should be considered an effective strategy for sustainability and public health in diverse food service settings. This study is registered with the ISRCTN registry, ISRCTN36918695.
Becker et al. (Wed,) studied this question.