Our objective analyses using the POS system revealed that POP health information in supermarket stores could encourage customers to purchase vegetables. Purchased vegetables are expected to be consumed by the customers themselves and probably by their relatives. Thus, POP health information could help in healthy behavior modification in the field of diet and nutrition.
The intervention effect on
sales was most apparent for fruit and flower vegetables. In Japan, tomatoes, eggplant, cucumbers, and okra are very popular fruit and flower vegetables, also known as summer vegetables. The production and distribution as well as sales of these fruit and flower vegetables decrease during the fall and winter seasons. In this study, although sales of fruit and flower vegetables decreased over the intervention period in both stores, the rate of the decrease could have been diminished by the POP information about recipes that use these vegetables. Although we do not have sales data on individual vegetables, we speculate that our POP information might have affected sales of tomatoes. This is because tomatoes can be eaten raw and therefore can be served easily in daily meals.
A significant intervention effect was also found for leafy vegetables. Most leafy vegetables are low in calories, easy to prepare, very popular for use in salads, and available throughout the year. These characteristics might have encouraged customers to choose leafy vegetables when they bought vegetables. However, the intervention effects were positive for all varieties of vegetables, although the intervention effects for some varieties were not statistically significant. Therefore, we believe that our POP information positively affected the decision to purchase vegetables in general.
In previous studies, the effects of nutritional environment intervention through POP information have been inconsistent. In the review by Seymour et al. [9
], the success rate of increasing targeted items was higher in worksites and universities than in grocery stores. The authors suggested that the limited food choices available for workers or students in on-site eating establishments and vending machines might have contributed to the successful results in worksites and universities. Although the authors concluded that intervention in grocery stores appeared to be the least effective, five of eight studies that used information strategies in grocery stores reported increased sales of some targeted items. Most studies featured in that review chose multiple target items for POP information, whereas ours emphasized the health effects of vegetables. The fact that this study chose a clear target might have contributed to its finding that the intervention was successful.
Buscher et al. [6
] reported that 89% of cafeteria users in a university recognized POP information at the entrance, whereas only 7% of users recognized POP information at a good display space. We placed POP information not only on the vegetable display floor but also at the checkout counters. These display procedures that targeted all customers could also have affected even those customers who were not initially interested in consuming vegetables. The paper by Buscher et al. found that POP information failed to encourage customers to purchase vegetables, unlike in our study. This difference in results could be explained in part by differences in target population and methodology. The target population of POP information in that paper was university students, whereas ours was general customers living near supermarket stores. Our target population could be more sensitive to health-related POP information than young university students because they would have responsibility for preparing meals not only for themselves, but also for their family members. The duration of POP information displayed for vegetables also differed between the two studies. Buscher et al. changed the target foods of POP information every week, and POP information for vegetables was displayed during the first week of their 6-week intervention period. In our study, intervention effects of POP information were not observed during the first 4 weeks of the intervention period. In order to alter customers’ purchasing behavior for vegetables, continuous provision of POP information for at least 4 weeks would be necessary.
Hirayama et al. [10
] reported that, historically, consumption of green–yellow vegetables was associated with reduced risk of mortality from cancer of the stomach, colon, lung, cervix, and prostate. A recent Japanese large-scale cohort study also reported reduced risk of gastric cancer associated with vegetable intake [11
], although this study failed to demonstrate reduced risk of cancer of the lung, colorectum, and prostate. According to some Japanese cohort studies, vegetable intake is negatively associated with risk of mortality from total CVD [12
] and stroke [13
]. Therefore, the population strategy to promote vegetable intake through POP information at food stores could contribute to reduction of mortality from gastric cancer and CVD in the Japanese population.
While we mainly targeted vegetables for promotion, fruits also have the potential to contribute to prevention of cancer [1
] and CVD [3
]. However, promoting fruit consumption is not a priority measure in Japan’s public health strategy [4
]. According to observational studies, a fruit-rich diet is negatively associated with obesity [14
]. However, these studies found that subjects who ate more fruits got less energy from other food items. In the study by Vioque et al. [15
], total energy intake was adjusted in multivariable models. Consequently, their study suggests that a high level of fruit consumption could contribute to obesity prevention if total energy intake were to remain unchanged. There is no guarantee that adding energy intake from fruits without reducing energy intake from other food items would not lead to obesity. Fructose in fruits can also cause visceral adiposity, dyslipidemia, and insulin resistance [16
]. These are the reasons why we prepared fewer posters on fruits than on vegetables and placed them only at the fruit display space. Cautions against overeating were also provided. Consequently, as we expected in advance,
sales did not rise even at store I. How fruit consumption can be promoted without causing adverse effects owing to overeating will be a future concern.
Another aspect of this study is the use of the POS system as a source of research data. The system immediately records all sales data by reading barcodes on purchased goods at checkout counters. Using this database, one can analyze complete sales data at a store, as we did for vegetables and fruits. Furthermore, since the data include no personal information about customers, the ethical barrier in collecting such data for research is low. Thus, the POS system is expected to be highly useful in epidemiological study in the future.
Our study has several limitations. First, the number of stores in the study was too small: only one store was given intervention and one store was observed as a control. To confirm the universality of our results, larger-scale randomized studies are necessary.
Second, the long-term effects of the POP health information were not analyzed. This would have been difficult for us, because store C started to provide POP information after our study was completed. Therefore, the long-term effects could not be investigated owing to the loss of control data. Since it is ethically difficult to leave customers of control stores blind for useful information, analyzing the long-term effects remains a methodological challenge.
sales of fruits was positive at store C, whereas it was negative at store I throughout the study period. We cannot deny the possibility that some sales promotion had been carried out at store C before this study was started, although there remain no records or memory of such promotion. However, we asked the stuff members of both stores not to conduct different promotions between the stores during the study period other than our POP information display in store I. For the present study, the most important finding for fruit sales is that the changing patterns of
sales were not significantly different between the stores. This finding shows that the POP information for fruits in store I had no significant effect on the purchasing behavior of customers.
Fourth, we cannot conclude that our procedure would be applicable to a non-Japanese population, because types of diet, grocery shopping systems, and the educational backgrounds of customers vary between countries.
In conclusion, as we have shown in the case of purchasing behavior for vegetables, POP health information could help promote healthy dietary habits. Larger randomized studies, not only in Japan but also in other countries, would be necessary to confirm the universality of our results. The POS system can provide objective and accurate research data for such epidemiological studies.