Safety first: A story on restaurant food poisonings

Adil Yalcin
Jan 9, 2015 · 6 min read

Rich, Visual Exploration and Reporting of Food Poisoning Outbreaks in England & Wales, 1992–2009

Dining out is a common form of entertainment, but can it also be risky? 667 cases of food poisoning outbreaks were reported in England & Wales between 1992 and 2009. That is 1 outbreak every 9 days on average. Using this data, how can we make informed decisions about our dining experiences and reduce the risks of unpleasant next-days? Keep reading to find out more.

To explore the data interactively, click here.

Based on the reported data of food poisoning outbreaks in UK, our story explores the outbreak trends from multiple perspectives:

The results are generated using the interactive explorer at http://keshif.me/demo/food_poising_outbreaks.html. You can visit the explorer webpage to ask your own questions by simple selections, and find more insights that are customized to fit your taste.


1 — National cuisines summary shows that Chinese, Indian, British and Italian cuisines are the most commonly reported cases. But what if Chinese food is your favorite type of food? It is possible that there are more (Chinese) restaurants, thus more of the reported outbreaks. When making your cuisine decision, it’s still a good idea to consider restaurant ratings, setting, and your eyes to make sure that food is clean as well as delicious.


2 — Many people know that chicken is a food that needs to be handled very carefully. Bacteria loves surfaces in contact with raw chicken. Using this outbreak data, we can once again confirm that chicken (poultry meat) is indeed a risky food type, so maybe choose red meat or vegetables next time, which have fewer cases of reported outbreaks. The dataset also includes 61 unknown cases, which suggests that the surveying agency should consider improving the data collection.


3 — Behind the outbreaks, a bacteria called Salmonella is the most frequent cause. Viral infections are also very common, which may be harder to treat. While all we can do is to stay away from bad restaurant/food choices, hospitals may reallocate their resources to better fight common infection types. On the other hand, restaurants should also pay attention to the trends of contributing factors. This data suggests that restaurants should increase their efforts in proper food handling, while keeping their staff well educated about personal hygiene.


4 — Does your food type decisions affect poisoning characteristics? To reveal patterns, for example, we can select poultry meat (by mouse-over), and visualize percentages of poultry meat outbreaks per each category. Continental / European kitchens seem to be surprisingly worse at handling chicken (31%), followed by Indian (30%), Chinese (21%), and British (18%) cuisines. We can follow the same approach for other food types as well. For those in love with the Italian desserts, cakes and pastries, 33% of poisoning cases in Italian restaurants are reported for these delicious treats, possibly among the menu highlights. Maybe the next time you order some sweet treat in an Italian restaurant, take another look at how the food is handled.


5 — How does your cuisine selection affect your risks? Since Chinese is the most common, let’s look at its outbreak characteristics. First, the food type we order is likely to have an influence. Rice, chicken and mixed food are more associated with the outbreaks in Chinese restaurants, possibly following the popularity of these menu selections. Second, we notice that Salmonella bacteria accounted for most outbreaks by far (133/175), and the second most common (non-Salmonella) was Bacillus spp. (15/175). We also see the distribution of contributing factors, which seems to closely follow the common distributions. But, is there any effect to how Chinese restaurants differ than the rest? To answer this, we analyse relative frequencies, in percentage.

As you can see, Chinese restaurants, which contribute to 25% (175/677) of the outbreaks, seem to be doing worse with hygiene related factors (%34 - %39 among all restaurants). The best factor they keep up with is with infected food handlers, where their share for this factor (25%) follows their share in overall outbreaks.


You like what you see and you want to explore your own data? Discover how at www.keshif.me, an open-source toolkit for the web that makes your rich data easily explorable. Comments or feedback? You can reach me at @adilyalcin.


Note: This entry has been submitted to The Kantar Information is Beautiful Awards Mini Challenge — Food Poisoning, with the data separately published at http://bit.ly/1zNNCo4. The interactive approach does not focus on just one story or one-data. It enables reaching many stories and many insights intuitively. It shows that all the available data can be visualized in one display without an overwhelming design. Probably this exploratory interface will be able to regenerate all stories and findings in other submissions based on this shared data. I’m looking forward to checking out the rest of the creative submissions : )

The data was made available in aggregated summaries. Information specific to each specific outbreak case was not available. To enable analysis of trends, the outbreak table was generated using the reported statistical characteristics. The relations between food type, pathogens, contributing factors, and the setting are not available at the competition or published data summaries. Thus, any such relationship shown in using the interactive browser should not be considered to reflect the distribution of the collected data. Also, the time dimension is missing in the available data, as only high level overviews were reported (per year / season, and per specific cuisines only. I tried to contact the authors of the paper (Gormley et al.) to access the complete tabular outbreak data, but the authors have left their office, so emails bounced. I have not received a reply from the current maintainers of this database so far as well.

Thanks to Sally Flynt for editing the text, and Tak Yeon Lee for some design suggestions.

    Adil Yalcin

    Written by

    Enabling Data-Driven Insights for Everyone | Founder-CEO at @keshifme ~ Visual Data Analytics and Exploration, Design, Engineering, Entrepreneurship