Why no one will ever “prove that GMOs are safe”

Water has never been proven safe. Image by Konstantin Stepanov via Flickr.
Water has never been proven safe. Image by Konstantin Stepanov via Flickr.

When discussing transgenic crops, I regularly get asked to provide a paper that “proves” that GMOs are safe. Whether you want proof that biotech crops, organic bananas, or conventional peaches are safe, I cannot provide you with such a paper. Safety is a relative term and is generally defined as the absence of risk or harm. As such, asking for proof of safety is, in essence, asking someone for proof of the absence of risk. The risk of what ever is being evaluated is measured in relation to other options, not against a theoretic idea of “perfectly safe”. Relative risk is scientifically determined by examining the evidence at hand: experiments are performed to determine the impact of a substance on health, environment, etc and the data from these experiments are assessed to determine if the substance causes harm.
Scientifically, nothing is truly 100% safe. To explain why, we’re going to do an exercise and try to prove that water is safe. The first thing to keep in mind is that there are many aspects to water safety: impact on health, water transportation, water treatment, proper water storage, etc. For our example, we’re going to select “impact on health”.
Then, we have to come up with a null hypothesis. It may seem counter-intuitive and the double-negatives in the explanation below suck, however the baseline for much of research is that there’s no impact or no difference. It’s the researcher’s responsibility to disprove that hypothesis, ie. to show that there is a difference or that there is an impact. For example, if you want to find out if kids who watch TV are more prone to hitting, your null hypothesis could be “watching TV does not increase aggression”. So for our exercise, our hypothesis will be “Drinking water does not cause cancer”.
Next, we narrow down the hypothesis to a question that we can actually test. For example, “children aged 2-4 who watch 1-2 hours of TV a day do not hit their parents more frequently than children who do not watch any TV.” For our study, we’ll consider our question to be “Individuals who have lived in the San Francisco Bay Area for 10-20 years and drink 2-4 cups of tap water daily do not have a greater incidence of breast cancer than the national average”. We conduct our study and gather data which will probably take a few years. Then we apply the proper statistics. If our study finds a difference, then we’ve disproven our null hypothesis, much hoopla will be made, and our findings will be published in the “Journal of Awesome Research”. If there’s no difference, then our null hypothesis still stands and our study will likely be published in the “Journal of Flibbity-Flooba”.
If we find no difference in the incidence of breast cancer in our study, have we “proven” that water is “safe”? No. All we’ve done is add data to the body of evidence that suggests that drinking water does not cause cancer and that it’s safe to drink it. As previously mentioned, nothing can be truly proven safe, even water; too much or too little of it can kill us, let alone if it is improperly purified. Yes, water can be considered dangerous: water-borne illnesses are one of the leading causes of death worldwide, water causes floods, and it can damage homes. But at the same time, we need water to survive and its intake in the proper amounts promotes health. Science has helped us define the harm that water can cause and the proper steps that we can take to ensure that harm is minimized.
Let’s go beyond our water analogy: here are a few other hypotheses along with a more narrow question of what will be tested:

  • Broad: The MMR vaccine does not cause autism. Narrow: There is no significant difference in the incidence of autism in caucasian children who have received Merck’s MMR vaccine in the San Jose Bay Area and children who are not vaccinated.
  • Broad: Eating transgenic crops does not harm the gut. Narrow: There is no significant difference in the relative abundance of bacterial species in the intestinal flora of pigs fed a diet consisting of 30% genetically modified Bt-corn for 30 days compared to a control diet.

Again, let’s say that you are unable to disprove your null hypothesis. Does that mean that you’ve proven that the MMR vaccine doesn’t cause autism? No. Have you proven that GMOs do not impact the bacteria in the gut? No. What you’ve done is add data to a body of evidence that suggests that the MMR vaccine doesn’t cause autism and that crops developed through biotechnology don’t cause harm. You’ve been able to demonstrate a relative lack of risk between the substance (the genetically engineered crop) and a control (the conventional crop) in the specific area examined.
Until someone comes up with a study showing that A causes B, then the null hypothesis is what we turn to: A does not cause B. Otherwise you could come up with any crazy hypothesis and people would have to “prove you wrong”. You could state that that earthquakes are caused by invisible dragons jumping at the same time and demand evidence showing that it’s not true. Or that wearing a watch causes carpal tunnel syndrome due to the radiation emitted by the watch. Or that computers leach dangerous toxins that cause brain tumors. That’s not the way it works. Dragons don’t cause earthquakes and computers don’t cause brain tumors, until you can prove otherwise. The onus is on the person making a claim to provide evidence supporting its existence. Therefore, if you claim that invisible dragons cause earthquakes, it is not my responsibility to “prove you wrong”. Rather, it is your responsibility to provide evidence demonstrating that these beings caused the earthquake. A similar analogy, known as Russell’s Teapot, was coined by the philosopher Bertrand Russell. To illustrate where the burden of proof rightfully belongs when someone makes an unfalsifiable claim, Russell laid out this example: “If I were to suggest that there is a teapot, too small to be detected by a telescope, orbiting around the Sun somewhere between the Earth and Mars, nobody would be able to disprove my assertion. But if I were to go on to say that, since my assertion cannot be disproven it should not be doubted either, then I should rightly be thought to be talking nonsense.” This example highlights how impossible it would be to “prove him wrong”; therefore, the burden of proof would lie with him, as he is the one making the claim.
Whenever someone makes an unfalsifiable claim, the burden of proof lies with the person making the claim. Without a credible hypothesis for harm, and after decades of study, the burden of proof for the people claiming that biotech crops could be causing autism or cancer, lies with those making these claims.  As Carl Sagan said: “Extraordinary claims, require extraordinary evidence.” Claiming that “we just don’t know enough” is a not hypothesis that can be tested and is not in itself evidence of harm: it’s just idle speculation.
The absence of single papers demonstrating safety is often used to invoke fear and doubt, and impossibility of proving a negative is often capitalized in anti-GMO rhetoric (this recent article by a medical doctor in the New York Daily Mail is a perfect example of such arguments): “Do we know that GMOs don’t cause cancer? Do we know that they don’t cause male infertility? etc.” Well, no… We don’t… But in the many feeding studies that have been conducted, there’s absolutely no evidence to suggest that it DOES cause cancer, there’s no logical mechanism proposed by which this might occur, and the null hypothesis still stands. You could virtually make the same argument about anything. “Do we know that eating pomegranates doesn’t case male pattern baldness? Do we know that typing on a keyboard doesn’t cause STDs?” No… We don’t… I don’t think anyone has ever done those studies. But strangely enough, no one has proposed a ban the use of keyboards until someone proves that typing on one doesn’t cause herpes, nor has anyone suggested that I should uproot the pomegranate tree in my backyard. Remembering that safety can never be proven and that we can only demonstrate a lack of relative risk can allow us to view such claims with skepticism. The null hypothesis also allows us to weed out superstitious nonsense from cohesive scientific arguments. Some of the conspiracy theories circulating right now include the idea that the ebola virus is man-made and that AIDS is not caused by HIV. We can stand against such harmful ideas by stating “show me the evidence suggesting that ebola IS man-made” or “here’s a plethora of data indicating that AIDS IS caused by HIV”.
When you ask for “proof that GMOs are safe” or ask for a paper that has this evidence, that is absolutely the wrong request. In fact, it’s a loaded question, whether the person asking realizes or not, the “proof” is impossible to provide, no matter the subject. Ask a specific question and then try to find the data showing that it DOES cause harm. Unfortunately, I can’t provide you with such data because I haven’t read a well-designed, well-executed, reproducible study demonstrating that GMOs have a negative health impact.
THIS is why scientists stress the number of studies that have examined genetically engineered crops. THIS is why scientists stress the statements made by reputable scientific institutions about GMOs. THIS is why meta-analyses and literature reviews are important. Because no single study proves safety: its the sum of the studies, the body of data, the totality of research that’s been done which suggests that the current GMOs on the market are safe relative to their conventionally bred counter-parts. By examining the body of data, scientists develop a consensus: the overwhelming majority of mainstream working scientists in the relevant field stop debating an issue because the direction that the evidence points has become clear. Although there are many matters on which a clear consensus has not yet emerged, on the topic of genetically engineered crops the consensus is that the current crops on the market place pose no greater risk to health than their non-GMO counterparts.
My last point is this: receiving funding for a study is very difficult, and the institutions that fund these grants want to see proposals that

  1. have a logical mechanism: in the context of this article, if a researcher wants to determine if a substance can cause harm, then the granting agency will want to know how that harm could potentially occur. In our exercise regarding the safety of water, we’d never get the money for our study unless we could outline a biological mechanism by which water could cause breast cancer.
  2. have a high likelihood of generating positive results or disproving the null hypothesis. Negative data or being unable to disprove your hypothesis doesn’t really build a career for a research scientist in the current academic system, nor does it validate the work of the granting agency. As a consequence, many researchers will not pursue a path where they don’t see fruitful results nor will granting agencies fund such research. Imagine working on a project where you don’t expect to make a difference, waste tax-payers’ money, and burn away your time. Why on earth would you start such a project? While it is possible that this approach will miss some harms that we did not understand, in a resource constrained world, it’s counter-productive to pursue every speculation whether it makes “sense” or not. It’s far more productive to pursue the ideas that make sense before worrying about things that don’t make sense.

Whether or not you agree with the system, this is what scientists in the public arena have to work with. Scientists do not want to waste their time and valuable resources on a study where they don’t expect to demonstrate anything new. Instead of recognizing this simple explanation, many chose to believe that research is being silenced or that scientists are being paid off. Very nearly none of the scientists working in the relevant fields think that there is any reason that long term feeding trials of transgenic crops will produce a demonstration of harm, yet many anti-GMO activists would choose to believe that such research is not being conducted because “Big-Ag” drives funding at academic institutions. Rather than acknowledging that the absence of studies examining a link between Round-Up Ready corn and Alzheimer’s is a result of the fact that experts in the field cannot envision a credible mechanism for such harm based on the evidence at hand, many would prefer to believe that it’s because “Big-Ag” is suppressing the data or that Monsanto will break scientists’ kneecaps if they publish results suggesting that GMOs can cause harm.
But if your definition of “safety” is a 5 year longitudinal study on 1000 rats examining a link between Bt-corn and gluten sensitivity, by all means: spend 10 years of your life in school earning less than minimum wage, and then try to find a granting agency that will fund your study based on whatever evidence and reasoning you have. Best of luck in your future career path!