by Corinne Keet
In this week’s New Yorker, Elizabeth Holmes, the CEO of Theranos, a $9 billion company trying to disrupt the world of laboratory testing, was quoted arguing against physicians as gatekeepers of lab tests: “The fact that in some states it’s illegal for someone to be able to get basic data about their body-for example, you’re pregnant or you’re not, you have an allergy or you don’t. Not a lot of sophistication has to go into the interpretation of that test.” [Emphasis added]. As an epidemiologist who studies allergies and an allergist, I have to wholeheartedly disagree with this statement, and the problems with this statement have profound implications for the overall business model of Theranos.
Allergy testing, and particularly food allergy testing done by blood (specific IgE), is actually very difficult to interpret. There are two key measures of how good a diagnostic test is: sensitivity and specificity. Sensitivity is a measure of how well a diagnostic test identifies those who have a disease. Specificity is a measure of how well it identifies those who do not have a disease. No diagnostic test has perfect sensitivity or specificity, but food allergy testing has especially poor specificity (meaning that a fair number of people without food allergy test positive). Making things worse, food allergy is relatively uncommon, making it more likely that a positive test labels someone as food allergic who doesn’t really have food allergy.
For example, if we took all children in the US, approximately 10% would test positive to peanut, while only about 1% have peanut allergy. That means that if you tested all kids in the U.S. for peanut allergy, 90% of positive tests would be false positives. Even if you used the higher cut-offs recommended in guidelines for interpretation of these tests, most positive tests would still be false positives, and with the higher cut-off you would miss a lot of real positives. False positives are not harmless, because they can cause fear, have social consequences and because unnecessary elimination diets can put people at nutritional risk. Of course, this doesn’t mean that food specific IgE is a useless test – just that we need to be judicious about who we test, and sophisticated about how we interpret the results in the context of other clinical data.
Admittedly, allergy testing is worse than most diagnostic tests, making it a particularly unfortunate example of the health benefits of direct-to-consumer testing*, but no laboratory test has perfect specificity. Even much better diagnostic tests have some false positives. Because the rate of these false positives depends on how likely someone is to have the disease before you test them, there are always going to be a lot more falsely positive tests if you expand diagnostic testing to low risk groups.
If the goal is to “redefine the paradigm of diagnosis away from one in which people have to present with a symptom in order to get access to information about their bodies to one in which every person…has access to actionable health information at the time it matters”, as Holmes puts it, then we’re going to have to accept a lot more false positives than the current system. As a society, we’ll need to think about how to deal with the consequences.
*Interestingly, it doesn’t look like allergy tests are actually on Theranos’ menu of tests.