Gluten-free menu items often not so gluten-free

There’s a moment in the new Netflix documentary series Rotten^[1] when celebrity chef and restaurateur Ming Tsai says directly to the camera “If you don’t know what’s in your food, get out of the business—today.” He is addressing restaurant owners and managers who don’t know what allergens are in the food they serve or otherwise fail to implement proper controls to prevent cross-contamination. The entire episode of “The Peanut Problem”^[2] revolves around food allergies and their sometimes fatal consequences with a particular focus on food service establishments, and Tsai’s admonishment feels like the climax of the show. Indeed, surveys^[3] show that there is a lack of specific food allergen training in restaurants. This lack of training can be due to a variety of reasons, including high turnover rates among food handlers, a language barrier, or a perception that food allergies simply do not have a substantial effect on business^[4].

This lack of training also poses a problem in the context of celiac disease. Celiac disease is an intestinal disorder characterized by an intolerance to gluten^[5], which is a protein found most notably in wheat, but is also present in other cereal grains in small amounts. Select foods that contain gluten are shown in Figure 1.

Celiac disease affects approximately 1%^[6] of the population worldwide), and the only known effective treatment is strict avoidance of any dietary gluten^[7]. There is also a related disorder known as non-celiac gluten sensitivity^[8], which triggers an immune response when gluten-containing foods are consumed. This type of sensitivity may affect as much as 6% of the U.S. population.

In the U.S., the Food and Drug Administration (FDA) mandates that all consumer packaged food be labeled with their allergen content, but this does not extend to food served at restaurants^[9]. When eating out, individuals with celiac disease or other forms of sensitivity to allergens must communicate their dietary restrictions and rely on food handlers’ knowledge and expertise in controlling food allergens such as gluten to avoid exposure.

Gluten (cross-)contamination among food products that do not typically contain gluten or that are advertised as gluten free remains relatively common^[10]. However, most of the data on this topic to date comes from packaged retail goods, not restaurants. The data available for these types of establishments are usually limited to a single city such as Brasilia^[11] or Melbourne^[12], because the restaurants have to be close enough to the investigators or the university to ensure samples can be obtained and brought back to the lab for analysis. These types of studies also show some level of gluten contamination in meals advertised as “gluten-free.” The study under review is the first study to examine gluten contamination of supposed gluten-free foods in restaurants throughout the U.S. using crowdsourced data.

Dining out can pose serious risks and challenges to people with celiac disease. The study under review utilizes crowdsourced data to estimate the prevalence of gluten contamination of supposed gluten-free foods in restaurants throughout the U.S.

The researchers examined shared test data from U.S. users of a commercially available gluten detection device, called Nima. Users insert a pea-sized piece of food into the device, which has a high sensitivity for determining whether gluten is present above 20 ppm^[13], the current limit set for packaged food to be considered gluten-free by the FDA^[14]. Each test included a date, time, restaurant name, address, food item tested, the presence or absence of a gluten-free label, and test result. The researchers excluded tests with more than one food item in the test, and restricted analyses to only food types with greater than 100 tests. For example, pizza would only be included if there were more than 100 readings from pizza samples, but if there were only 20 samples of asparagus, that data would have been excluded.

Data on the gluten presence or absence in foods was collected by the means of a portable gluten detection device called Nima. For each gluten measurement, the volunteers also included information on the date, time, restaurant name, address, food item tested, and the presence or absence of a gluten-free label.

The analysis included 5,624 tests performed by 804 users, of which about 84% of the tests involved foods with a gluten-free label. The results indicated that 32% of all foods specifically labeled as gluten-free tested positive for gluten.

As you can see in Figure 3, out of all gluten free-labeled food, gluten-free labeled pizza and pasta had the highest rates of gluten detection at 53% and 51%, respectively. Vegetables had the lowest adulteration rate of 23%. The authors also categorized the restaurants into three categories: quick service (typical fast-food places like Wendy’s, Chick-fil-A, and Arby’s), fast casual (Panera, Chipotle), and casual dining (Chili’s, Outback, P.F. Chang’s). The data analyses revealed that, among restaurant types, fast casual restaurants had the lowest rates of gluten detection (22.3%), and that the rates of gluten detection differed significantly by meal time: during breakfast meals, 27% of gluten-free labeled foods tested positive, 30% at lunch, and 34% at dinner.

In a multivariate analysis of factors associated with the presence of gluten in gluten-free foods, “The West” (defined as Alaska, Arizona, California, Colorado, Idaho, Hawaii, Montana, Nevada, New Mexico, Oregon, Utah, and Washington) was less likely to have foods contaminated with gluten compared to other regions of the U.S. Additionally, in a multivariate analysis, dinner time still exhibited the highest rates of gluten detection, and both casual and fast casual dining establishments had reduced rates of gluten detection. Gluten-free pasta and pizza remained significantly higher in gluten than other foods.

Household income among the participants did not appear to be linked to gluten levels.

About 30% of foods that were labeled as gluten-free tested positive for gluten. Among these foods, pizza and pasta were found to be significantly higher in gluten than other tested foods. Dining establishments labeled as fast casual had significantly less gluten contamination than quick service and casual restaurants.

According to this data, it seems that, of the restaurant meals indicating they are gluten free, about a third nationwide test positive for some amount of gluten. Moreover, pizza and pasta dishes were the biggest offenders. A strength of this study is that the authors were able to include more data points than perhaps any other study involving gluten contamination in restaurant foods.

However, this study also has many limitations, one of which is the potential for bias. The data collected is not exactly randomized, as it is a self-selected group of Nima owners that have chosen to share their data. The authors note that Nima users may be more inclined to share gluten detected results rather than gluten not detected. This makes it hard to say that the numbers hold true for all foods in the U.S.

Moreover, a validation study on the Nima device^[15] published last year yielded mixed results. In that study, the Nima performed poorly in its ability to reliably detect gluten residue at 20 ppm in foods such as bread, pasta, and corn puffs, with a 47% detection rate. However, it was able to reliably detect gluten at the 20 ppm concentration in other foods, and also improved to 88% at 30 ppm in these foods, and 97.5% at 40 ppm. While sensitivity did improve above 20 ppm in these foods, 20 ppm is an important number, since it is the established threshold in the U.S. below which a food can legally be labeled as a gluten-free product. The product also demonstrated a low rate of false positives (3.6%), with the only failures occurring in corn puffs and oatmeal. It is also impossible for the authors to know if the device was used properly and in accordance with the manufacturer’s recommended procedure.

One other issue is that the findings in this study may be vulnerable to the multiple comparisons problem: the more comparisons that are made, the more likely it is that one or more results would appear significantly different, even if they weren’t truly different.

It may also be worth mentioning that many of the authors of this study were associated with Nima. In fact, one of the authors is the CEO of Nima. This is particularly problematic in view of the fact that the previously referenced paper^[15] that supposedly confirms the accuracy of the device was likewise sponsored and conducted by Nima.

The amount of data collected in this study makes it perhaps the largest study to date on gluten contamination of restaurant foods. However, drawbacks to this study include the high potential for bias as well as the questionable accuracy of the Nima device at lower gluten concentrations equal to what the FDA considers to be the limit for foods to qualify as gluten free.

Most individuals with celiac disease report that their gluten sensitivity impacts their quality of life^[16], particularly when it comes to dining out and traveling. Considering the levels of gluten contamination among gluten-free foods detailed in this study, it’s easy to see how eating out could be difficult with a condition like celiac disease. The few other studies that are available on the topic show similar results. For example, a study examining the gluten levels in beans^[11] sold at self-serve restaurants demonstrated that 45% of the businesses selected had at least one sample testing positive for gluten beyond 20 ppm. In another study^[17], 130 food samples labeled as gluten-free were collected from 25 bakeries in Brasilia, Brazil. Of those samples, 21.5% had gluten levels above 20 pmm, while two-thirds of the bakeries sold contaminated bakery products. A study conducted in Ireland^[18] found that 10% of gluten-free meals ordered contained greater than 20 ppm of gluten, most of which were greater than 100 ppm, with several samples exhibiting levels greater than 6000 ppm.

The Irish study mentions that even though only a minority of the waitstaff received formal training on the issue of gluten contamination, the majority of respondents were confident that the meal they provided was indeed gluten-free. Food safety training for restaurant staff does appear to be effective at reducing cross-contact. At least, that’s what observational data from Melbourne, Australia^[12] suggests. In that study, scientists found that most burger chains in the city were serving allegedly gluten-free meals that actually contained gluten. Only one year later, however, all of the burger chains were in compliance. The authors noted that the odds of compliance by any food business were reduced by 75% if the food business did not provide staff training. In one case, a pilot study^[19] was able to raise awareness of restaurant staff regarding common food allergies from 9% to 64%. In Italy, a regional celiac advocacy association has developed good manufacturing practices and conducts training courses for food businesses, and has developed a gluten-free certification program for those businesses that offer gluten-free options on their menu. One analysis of several gluten-free pizzas from certified establishments detected no cross-contamination, indicating that training and implementation of good practices can be an effective way to reduce cross-contact and create a safe place to eat for those that have celiac disease or other manifestation of gluten sensitivity.

While training and good practices are easy and effective means of reducing the risk of allergen cross-contact when preparing and cooking food, contaminated ingredients from food manufacturers are an issue that has to be addressed earlier in the food production chain. In fact, the most common hazards in manufactured and processed foods that are regulated by the U.S. FDA are undeclared allergens, according the fifth annual FDA report of the Reportable Food Registry^[20]. This could be caused by any number of factors and at any stage of the processing of the ingredients, ranging from the use of naturally gluten-containing ingredients over inadequate sanitation between changeovers, and changes in product formulations that were not updated on the labels, to insufficient separation of processes to control cross-contact from other production lines, or incomplete staff training on how to handle food allergens.

In any case, there appears to be a trend of decreasing gluten contamination^[21] over the past couple decades. Nevertheless, food establishments are still serving high numbers of contaminated, allegedly “gluten-free” meals. Based on the available evidence, it appears that the development and implementation of good manufacturing practices and the creation of food allergen training programs can significantly reduce instances of cross-contact.

Studies on restaurants and the general marketplace indicate that the presence of gluten in foods labeled as “gluten-free” is a common occurrence. Training programs on food allergens, when they engage cooks and servers, appear to reduce cross-contamination in restaurants.

Q. How does the Nima work?

The Nima is a wireless, handheld, battery-powered device designed to test food products. A fresh gluten test capsule is required to run a test. The test capsule is about the size of a Bic lighter and cannot be reused after testing. A pea-sized sample of food is placed into the capsule, and the act of screwing on the lid to the capsule grinds the food sample, if it is solid food like a bit of pizza crust. The capsule is then placed into the device. The test takes about three minutes to provide results.

There are some recommended guidelines^[22] to follow when testing certain foods. For instance, powdery foods like spices or seasonings should be diluted in water before testing. Gummy foods should be cut, chopped, or mashed before testing, and puffed foods should be crushed to remove the air before testing. Some foods cannot be tested at all, such as alcoholic drinks and fermented foods^[23].

New capsules are relatively expensive. They currently retail for $72 for a pack of 12 capsules^[24], or $59 for a subscription.

Q. Why does celiac disease develop?

Celiac disease is generally considered to be a genetic disorder, but some reports indicate that the incidence rates have increased^[25] over the years. This could be due to a number of factors^[7], including increased awareness of the disease, increased exposure to dietary gluten, better diagnostic methods, or other environmental risk factors.

Q. What is the difference between celiac disease and the ever-more prevalent non-celiac gluten sensitivity?

While celiac disease is a well-known genetic disorder characterized by an autoimmune response to gluten ingestion, non-celiac gluten sensitivity (NCGS) is a more nebulous condition. There are no known biomarkers for NCGS like there are for a wheat allergy or celiac disease. Therefore, diagnosis can be a challenge^[26]. In fact, since NCGS is a relatively new phenomenon, there are not even standard diagnostic methods for the disorder.

Some researchers have proposed a diagnostic method^[27] that consists of a double-blind dietary crossover trial that consists of a one-week diet with either eight grams of gluten per day or a gluten-free placebo added to the diet with a gluten-free washout period in between. This type of crossover trial has been utilized to demonstrate that only a minority of people who have self-diagnosed^[28] NCGS actually experience pain or discomfort when challenged with gluten.

Q. What happens when people with celiac disease eat gluten?

After eating gluten, people with celiac disease will typically experience abdominal pain, diarrhea, and nausea.

The gluten triggers an immune reaction in the small intestine that reduces absorption of vitamins and minerals, potentially leading to iron deficiency, folate deficiency, and reduced bone density^[29]. Over time, the reaction can cause villous atrophy, damage to the lining of the small intestine.

Other negative outcomes can occur due to the disease. Individuals with celiac disease are at an increased risk of developing non-Hodgkin’s lymphoma and small intestine adenocarcinoma^[30]. The risk of lymphoma is related to mucosal healing^[31].

Celiac disease is also associated with a number of negative pregnancy outcomes^[32] including stillbirth, preterm birth, intrauterine growth restriction, and low birthweight.

Dining out can be risky for individuals with celiac disease or another form of gluten sensitivity due to the gluten contamination among foods advertised as gluten-free. Several studies have been published on the gluten concentrations of consumer packaged retail goods, but little data is available on restaurant foods. The study under review is deemed to be the largest cross-sectional study published thus far on the prevalence of gluten contamination in restaurants.

Data on the presence or absence of gluten was collected via crowdsourcing the test results of a portable gluten detection device called Nima. The results of the study indicated that gluten-free pizza and pasta had the highest rates of gluten contamination. Multivariate analyses add that western U.S. and fast casual dining establishments had the lowest rates of gluten contamination and that the incidence of gluten contamination is higher for dinner than for breakfast meals. However, some limitations worth mentioning are the risk of bias at different levels of the data acquisition and processing in the study along with the inconsistent reliability of the device used in measuring gluten across all food types.

Implementing good food handling practices and establishing a food allergy training program in which restaurant staff are trained to understand common allergens are the biggest opportunities for reducing the high prevalence of cross-contamination that makes dining out one of the most quoted problems of people with celiac disease worldwide.