Please do not panic: The research abstract on the Nima Sensor and gluten-free restaurant meals presented at the ACG annual meeting is misleading

Please do not panic: The research abstract on the Nima Sensor and gluten-free restaurant meals presented at the ACG annual meeting is misleading

Yesterday evening I was just about to close up shop when three messages arrived almost simultaneously asking about a research abstract presented at the American College of Gastroenterology annual meeting entitled “Gluten Contamination of Restaurant Food: Analysis of Crowd-Sourced Data.” According to the abstract, study authors, including physicians from the Celiac Disease Center at Columbia University “analyzed data from a portable gluten detection device (Nima), collected across the USA during an 18-month period by users (n = 804) who opted to share results of their point-of-care tests.” One of the attention grabbing statements in the abstract reads, “One-third of restaurant foods labeled GF contained at least 20 ppm of gluten.” The entire abstract is available at

Okay, so what’s the problem with this abstract?

  1. A “gluten found” reading on a Nima Sensor does NOT mean that the pea-size sample placed in the gadget contains a level of gluten at or above 20 ppm. Nima Labs has previously stated that when a sample contains a level of gluten below 2 ppm, the Nima Sensor is reporting gluten found approximately 8% of the time. Based on third party testing data this gadget is reporting gluten found approximately 35% of the time when the level of gluten is 5 ppm and approximately 56% of the time when the level of gluten is 10 ppm (Note: from a practical standpoint this means that if a sample contains a gluten level of 10 ppm there is about a 50:50 chance of getting either a smiley face result OR a gluten found result).

Under the FDA’s gluten-free labeling rule for packaged foods, “gluten-free” means a level of gluten below 20 ppm. It does not mean zero gluten. Folks can disagree with FDA on what gluten-free should mean but detecting some level of gluten is very different than quantifying gluten at a level of gluten at or above 20 ppm. Details matter especially in research abstracts presented at medical conferences. For more information see

  1. This gadget has not been scientifically validated in the peer reviewed scientific literature by an independent third party (*). It is not a medical device and it does not require FDA approval. The Food Allergen Research and Resource Program (FARRP) has evaluated the device (Nima Labs funded the research but FARRP is credible). Among many concluding remarks, researchers stated:

“In our opinion, use of the Nima device will protect the health of gluten-sensitive consumers, if properly used on foods with reasonably uniform gluten distribution. The Nima device did perform poorly in detection of the critical 20 ppm on certain categories of foods, including bread, pasta, and corn puffs (47% detection).” The full evaluation is available at

  1. This gadget is not well suited to finding cross contact in restaurant meals. Why? Only a pea-size sample is tested. According to the FARRP report:

“The sample volume taken into the Nima device is a small pea-sized portion. Based upon the results, the small sample provides reliable results when the gluten is well distributed in the tested food. However, the presence of gluten-containing particulates could be missed with this sampling device. The sampling problem with particulates is a key issue with the Nima device. Consumers would need to take multiple samples to increase reliability when particulates are suspected.”

There are serious concerns about gluten cross contact in restaurant settings. Cross contact must be evaluated using scientifically validated methods. Using the unvalidated Nima Sensor in a “scientific” study as the basis for determining whether a restaurant is accurately stating a food to be gluten-free is irresponsible.

*Here is what the FDA has to say about scientifically valid methods for the purposes of evaluating a gluten-free claim, “a scientifically valid method for purposes of substantiating a ‘‘gluten-free” claim for foods matrices where formally validated methods (e.g., that underwent a multi-laboratory performance evaluation) do not exist is one that is accurate, precise, and specific for its intended purpose and where the results of the method evaluation are published in the peer reviewed scientific literature. In other words, a scientifically valid test is one that consistently and reliably does what it is intended to do.” Source: Gluten-Free Labeling of Foods: Final Rule


“In the enforcement of its regulations, FDA routinely uses scientifically valid methods that have undergone an independent multi-laboratory performance evaluation where the results have been published in the peer-reviewed scientific literature to ensure that the results obtained are accurate and reliable. Source: Questions and Answers: Gluten-Free Food Labeling Final Rule



Share this post

Comments (7)

  • Al Reply

    Excellent! I was waiting to hear your thoughts on this. Thanks!

    October 9, 2018 at 6:15 pm
    • Tricia Thompson Reply

      Thanks, Al. I’m sure you saw GIG’s response as well?

      October 10, 2018 at 3:02 pm
      • Al Reply

        Yep. 🙂

        October 11, 2018 at 1:23 am
  • Angelica Reply

    I appreciate your going over this again. But what’s the solution? Can you post about approved / better devices and why those are better? I’m getting lost in the minutiae. First, it’s too sensitive, then the sample size is too small (pea sized), so it might miss it. I don’t consider “too sensitive” to be a problem for me. If it detects above 5ppm, I think that’s a good feature. 20ppm was an agreement reached after debate, iirc, and GIG’s standard for certification is 10ppm, NCA’s cert is below 5 ppm…. etc I do consider it to be a problem if it doesn’t have a good way to tell if something has CC. Can you elaborate on the reason why too much sensitivity is bad?

    October 10, 2018 at 2:33 pm
    • Tricia Thompson Reply

      Hi Angelica, It isn’t necessarily that too much sensitivity is bad but rather that it is not consistent sensitivity. The device should have a consistent limit of detection (LOD). It should not be the case that depending upon the sample tested, a 10 ppm level of gluten will result in a gluten found result approximately 50% of the time and a smile result 50% of the time. It also should not be the case that a 20 ppm level of gluten will be detected less than 80% of the time. For more information, see the table at

      Pertaining to this particular research abstract, the authors (and many others) appear to be under the impression that a “gluten found” result using a Nima Sensor means that the sample contains a level of gluten at/above 20 ppm. This is simply not true. There is also the issue of false positives which GIG elaborates on in their statement.

      In terms of cross contact, think about a plate of food. Gluten present in a food item due to cross contact will not be evenly spread throughout the product. Cross contact is spotty by nature. Whether a restaurant patron is lucky enough to choose the pea size sample that just happens to contain gluten is slim too none.

      The reason you and many others are confused is that there are many issues with the Nima Sensor–it is too sensitive and not sensitive enough at the same time. Testing for gluten in food is not any easy process.

      October 10, 2018 at 3:01 pm
      • Angelica Reply

        Thanks, that’s very helpful. There’s also the issue of “hydrolyzed” gluten, but I’m unclear on whether all fermented foods have hydrolyzed proteins (including, if present, gluten), or if it’s a specific process that only brewers go through. In that case, I didn’t think anything could tell you accurately if there is any gluten. And since breads now are more often fermented, even gluten free ones, that makes any test a particular problem for bread. Do you know if someone has a list of the foods that are a particular problem for the Nima to test properly? And what’s the solution? Should I order food from a restaurant and then send it to a lab for testing? I feel like I don’t have any practical tools and that’s why Nima and similar devices seem so very helpful. Is there another device that’s better for CC detection, etc?

        October 12, 2018 at 4:38 pm

Leave a Reply

Your email address will not be published. Required fields are marked *