Testing the Effects of Sodium Warning Menu Labels on Sales in a Hospital Cafeteria

Background: Sodium warning labels are required on chain restaurant menus in New York City and Philadelphia. Other cities are considering this policy to reduce sodium consumption, but no real-world evaluations exist. This study evaluated the real-world effects of sodium warning menu labels on sodium content of purchased items, and whether effects differed by the labels' sodium threshold. Methods: We conducted a 6-month intervention in a hospital cafeteria (5-week menu cycles were repeated 5 times between March and August 2019). We collected daily sales data for all items. No warning labels were displayed in the cafeteria during the 1st, 3rd, and 5th menu cycles. During the 2nd menu cycle, we added sodium warnings to menu boards and placards for menu items with >1500 mg sodium (recommended daily limit for at-risk groups, including people with hypertension and adults >50 years old). During the 4th menu cycle, we added sodium warnings to menu items with >1150 mg sodium (half the daily recommended limit for the general population, 2300 mg). We used interrupted time series analyses to estimate level and trend changes in mean daily sodium purchased from specific menu categories after labeling compared with the counterfactual. Results: Our primary analyses examined grill items purchased (n = 83,514) during the study period, because all grill items were offered daily and represented a large proportion of entree sales. After implementation of >1500 mg sodium warnings, we noted a -45.89 mg or 6% level decrease in sodium per grill item purchased (95% CI: -70.64, -21.13), followed by an increasing trend of 1.41 mg sodium per grill item purchased per day (95% CI: 0.19, 2.63) over the 5 post-i mplementation weeks. By the end of the >1500 mg sodium intervention, sodium per grill item purchased was 3.6 mg (95% CI: -37.5, 44.6) higher than the counterfactual. Implementation of >1150 mg sodium warnings was associated with weaker changes in sodium purchased that were not statistically significant. Conclusions: After implementation of >1500 mg sodium warning labels on hospital cafeteria menus, mean sodium per grill item purchased decreased modestly at first but returned to preimplementation levels by the end of the 5-week intervention period. Later implementation of >1150 mg sodium warning labels did not significantly impact sodium purchased.