Invoking the precautionary principle is a favorite tactic of anti-vaccers, anti-GMO activists, and various other groups that are prone to opposing scientific advances, but there are numerous issues with this strategy. The exact definition of the precautionary principle is a bit amorphous and variable, but the general concept is that before taking an action that has potential risks (particularly with the application of something new), the burden of proof should be on the proponent to demonstrate the safety of that action before taking it. There is certainly an element of validity to this. I absolutely agree, for example, that new medications should be tested and shown to be safe before being made publicly available (indeed, that is part of why we have organizations like the FDA that regulate the approval of new drugs). Similarly, months ago when COVID vaccines were in early stages of testing, I and many other ardent pro-vaccers stated that while the vaccines were promising, we wanted to see the results of the tests before drawing any conclusions. That is a completely rational, evidence-based way to approach the situation. New drugs, vaccines, and technologies certainly have the potential to cause harm. So, it is not unreasonable to want safety testing prior to their use.

Where this becomes problematic, however, is in determining what constitutes reasonable evidence. Groups like anti-vaccers have a tendency to stretch the precautionary principle to unreasonable limits and use it as an excuse for science-denial, rather than a legitimate decision-making tool. Usually, what I see is an incorrect insistence that the precautionary principle demands absolute assurance of safety and that it should be used in a strictly binary decision tree where any possible risk means that the thing in question should not be used, regardless of the known benefits. This is very bad risk assessment and ignores important aspects of how science actually works. As I’ll explain, science is about probabilities, not absolutes, and you must consider both the risk and benefit of an action.

Note for clarity: Just to be sure that I have been clear, I am not attacking the general concept of the precautionary principle. Rather, I am taking issue with the way that many people apply it (and you could make an argument that they are not actually applying it at all and are instead using something else entirely that they are incorrectly calling the precautionary principle. Indeed, many definitions of the precautionary principle explicitly state that it only applies in the absence of scientific evidence).

Science is about probabilities, not proof

This is a really important concept that I have written about frequently. Science is an inherently probabilistic endeavor. It shows us what is most likely true given the current evidence, not what absolutely is true. It always leaves open the possibility that the current results are wrong or some other piece of evidence has been missed. Indeed, the statistics we use to determine things like drug safety and efficacy are inherently probabilities. When we say that a result is statistically significant, what we really mean is that if there is actually no effect (i.e., all results are from chance), there is a low probability that a result as great or greater than the one we observed could arise (i.e., if there is no effect of the thing being tested and we did the experiment again, we’d be unlikely to get such a strong result). Probabilities are inherent to modern science.

Ultimately, this is a good thing. It forces skepticism and avoids dogmatism. Unfortunately, opponents of science seize that inherent and entirely justified skepticism and erroneously conflate it with practical doubt. The fact that we can’t be 100% about a result does not inherently mean that we should have any practical doubt about it. Am I 100% certain that smoking causes cancer? No, but the topic has been so well-studied and the results are so consistent that I’m 99.999% sure, and for all practical purposes, there is nothing wrong with making a statement like, “smoking causes cancer.” Similarly, it is technically possible that scientists are wrong about gravity, but, to borrow from Tim Minchin, you would be pretty foolish to act as if there is any practical doubt about gravity when deciding whether to exit your apartment through your door or a second floor window.

Even so, for many vaccines, we have an extremely high level of certainty that they are safe and effective. Take the notion that vaccines cause autism, for example. As I’ve written about at length, this hypothesis has been extremely well-studied. Multiple studies with tens of thousands of participants have been conducted (e.g., (Hviid et al. 2019 [657,461 children]; Madsen et al. 2002 [440,654 children]; Anders et al. 2004 [109,863 children]; and Jain et al. 2015 [95,727 children]), including a meta-analysis with over 1.2 million children (Taylor et al. 2014), and no large, properly controlled study has found any evidence of vaccines causing autism. Nevertheless, I frequently encounter anti-vaccers who try to ignore that evidence by inappropriately invoking the precautionary principle and asserting that, “since we can’t be 100% sure that vaccines don’t cause autism, we should err on the side of caution and act as if they do cause autism.” That is extremely faulty logic and is nothing more than science denial dressed up as a cogent decision-making principle.

The other related issue is the never-ending string of possible mechanisms of harm. Anti-vaccers frequently concoct an ever-shifting litany of things that scientists need to test before they will accept that vaccines are safe, and they often do this under the guise of simply adhering to the precautionary principle. Sticking with autism for a minute, for a long time, mercury was the main anti-vax boogeyman (and it still is in some circles), and anti-vaccers insisted that the burden of proof was on pro-vaccers to show that the mercury wasn’t causing autism (fundamentally a precautionary principle argument; see note on the burden of proof). So, scientists did lots of tests and even removed thimerosal (the form of mercury in vaccines) from nearly all childhood vaccines. The results of those studies consistently showed that thimerosal doesn’t cause autism (Hviid et al. 2003; Verstraeten et al. 2003;  Taylor et al. 2014), and removing it did not reduce autism rates.

So the burden of proof has been met and the precautionary principle satisfied, right? Not according to anti-vaccers. According to them, maybe its actually the age at vaccination, and the precautionary principle says that we need to demonstrate that it isn’t age at vaccination before we consider vaccines to be safe. Then, when studies show that it isn’t the age at vaccination (Uno et al. 2015; Destefano et al. 2004; Smeeth et al. 2004; Madsen et al. 2002), they switch to number of antigens, and when studies discredit that (DeStefano et al. 2013), they switch to it being the number of doses, and when studies discredit that (Fombonne et al. 2006; Hviid et al. 2003), they switch to aluminum or countless other fantasies.

It’s like fighting the hydra. No matter how many arguments you test and defeat, more crop up to take their place. This is the problem with the practical application of the precautionary principle. There will always be other possibilities. No matter how many things we test, there will always be things that haven’t been tested. This is why the burden of proof is usually on the person making the claim, and by switching the burden of proof, the precautionary principle opens a can of unending worms (see note on burden of proof).

Again, to be clear, I agree with a reasonable level of testing before something like a vaccine goes to market, and even after it goes to market, if reasonable evidence arises that it is causing a problem, I agree that the evidence should be investigated and proper trials should be done. If there is a legitimate, science-based reason to suspect that a risk might exist, it should be investigated. That is a totally reasonable application of the precautionary principle. The problem is that many people try to use it unreasonably and insist that all possibilities must be tested, even if they can’t present any good evidence to show that a danger is likely. Actually testing all possibilities is, however, impossible. Thus, anti-vaccers can be immune to studies, because no matter how many we conduct, there will inevitably still be things we haven’t tested.

A similar issue arises with an insistence for “long-term” studies. As I’ve argued before, “long-term” is meaningless unless it’s carefully defined beforehand. We have studies that followed patients for multiple years (Idbal et al. 2013; Ferris et al. 2014; Vincenzo et al. 2014), which would fit the definition of “long-term” for most scientists, but this never seems to satisfy anti-vaccers. No matter how long the study is, they will always retort that the negative effects might come at some later age. This flawed reasoning is fundamentally just an argument from ignorance fallacy. It is saying that we don’t know for sure that there aren’t problems 35 years later, therefore we should act as if there are. That is bad logic unless we have some compelling reason to think that there would be negative effects 35 years later, which we don’t. We do, however, have plenty of evidence that there are enormous benefits to vaccines and the risk from not vaccinating is much higher than the risk from vaccinating. This brings me to the next topic: risk assessment.

Bad risk assessment

Good risk assessment needs to consider both the risks and the benefits within a probabilistic framework. All actions carry some level of risk, and people often ignore the fact that inaction can be riskier than action.

So, when it comes to topics like vaccines, we need to consider both the risks and the benefits. Vaccines do have side effects, but serious ones are very rare, and we know that the benefits far outweigh those risks because numerous studies have shown that vaccines are extremely beneficial at saving lives (Clemens et al. 1988; Adgebola et al. 2005; Richardson et al. 2010). Indeed, the WHO estimates that from 2000 to 2018, the measles vaccine alone prevented 23.2 million deaths! That benefit absolutely has to be considered when evaluating the risk of vaccines.

This is another place where anti-vaccers misuse of the precautionary principle becomes problematic. They focus on hypothetical potential risks that have somehow eluded all previous studies and argue that we should “err on the side of caution” before potentially injuring children with vaccines. That is horrible risk assessment because it totally ignores the massive, well-established risk from not vaccinating. In other words, the cautious approach is actually to use vaccines that have passed reasonable safety testing, because the known risk from not vaccinating is so high.

Is it possible that there is some unknown danger from vaccines that we have missed? Yes. As explained above, it is technically possible, but it is extremely unlikely. Meanwhile, we know that there are massive dangers from not vaccinating. Studies have repeatedly shown that vaccines save countless lives. Therefore, it is absurd to knowingly sacrifice those millions of lives out of a fear of some unknown danger which probably doesn’t even exist! That’s not erring on the side of caution, and it is not a proper application of the precautionary principle.

Similarly, when it comes to the issue of long-term studies, is it technically possible that there is some long-term effect that we haven’t found yet? Yes, but it is very unlikely. Meanwhile, we know that vaccines prevent diseases which are often fatal for children and can have long-term consequences for survivors. Here again, anti-vaccers are asking us to put more weight on an unknown and unlikely risk than on a very real and well-known risk. Further, even beyond the known risks, it is very possible that measles and other childhood diseases have additional long-term complications that we are unaware of. Indeed, based on our understanding of physiology and diseases, it is more likely that vaccine-preventable diseases have additional unknown long-term consequences than it is that vaccines do. Measles, for example, negatively affects children’s immune systems for years (Petrva et al. 2019; Mina et al. 2019; more details and sources here and here), thus opening the door for all manner of secondary infections and long-term complications. So if we are going to play this game of fearing the unknown, why should vaccines be the unknown that we fear rather than the dieses?

The answer given to that question often involves some variant of the appeal to nature fallacy, and asserts that we shouldn’t play God, or can’t improve on nature, or humanity survived for millennia with these disease, etc. These responses are all obviously flawed for numerous reasons that I have elaborated on elsewhere, so I won’t waste any more time here (see posts here, here, here, and elsewhere).

What about COVID vaccines?

Finally, let’s apply all of this to the new COVID vaccines, because I have recently seen countless people using some variant of the precautionary principle to argue against receiving the COVID vaccine.

First, it is true that, by the very nature of being new vaccines, the COVID vaccines have not yet received as much testing as most vaccines currently on the market*, but they have received just as much or more testing as those vaccines had when they first entered the marked, and the amount of testing they have received is very good. The Pfizer trials used over 43,000 participants (Polack et al. 2020), Moderna used over 30,000 (Mahase 2020), etc. Those are very large trials with good power to detect adverse effects, and they found that the vaccines were very effective (generally with effectiveness in the high 80s or 90s), with few serious complications (comparable to existing vaccines). This is very good evidence that the vaccines are safe and effective, and it meets any reasonable application of the precautionary principle. The problem is precisely that anti-vaccers’ application is unreasonable, so instead of accepting the results, they are spreading baseless fear about possible unknown dangers and future long-term consequences.

*Update 1-Sept-2021: This is no longer correct. The COVID vaccines are now extremely well-studied. See this post for details.

Again, I agree that those trials were necessary before the vaccines went to market. I don’t blindly support vaccines. Rather, I base my views on the evidence, and the evidence shows that these vaccines are safe and will save countless lives. Is it possible that scientists are wrong about these vaccines? Yes, but based on all the available evidence, it’s not likely. Nevertheless, many insist that because it is possible that we have missed something, we shouldn’t vaccinate and should instead apply “caution.” As I’ve been trying to explain throughout this post, that sets up a false dichotomy, because not vaccinating has substantially more risks than vaccinating.

Let’s just compare the two for a minute. The vaccines have passed large, well-conducted trials. Further, they are based on technologies that have been being researched for many years, and many of their components are the same as other vaccines that have passed numerous, repeated tests. All of this gives us very good reason to think that they are safe, and the probability that they are actually dangerous and we just haven’t found out yet is extremely low. In contrast, we know that COVID is highly contagious and highly deadly. Total global deaths are over 2.5 million, and in the USA, COVID is currently one of the leading causes of death (it spiked all the way to the number 1 spot during the large outbreak at the beginning of 2021). So, the known risk from not taking the vaccine is extremely high (see notes at the end before responding with the inane “it *only* kills 1 in 100” or “it only kills the elderly” arguments).

Further, if we want to go down this road of fearing unknown long-term complications, unlike the vaccine, there is very good reason to think that COVID will cause long-term problems. There is growing evidence that many patients have complications long after being infected (though the disease is recent enough that the data are still being collected and a clear picture hasn’t emerged yet), and given the damage that COVID is known to cause to the heart, lungs, and other organs, and our knowledge of other diseases, it is very reasonable to think that there might be long-term problems (Mitrani et al. 2020; Fraser 2020).

So, on the one hand, we have few known risks from the vaccine and no good reason to suspect unknown long-term complications, and on the other, we have an extremely high known risk from COVID, as well as good reason to suspect that there might be long-term damage.

Thus, proper application of risk assessment absolutely does not support avoiding the vaccine. Reasonable concerns have already been tested (thus appropriate precautions were taken), the known risks of COVID are substantially higher than the known risks of the vaccine, and there is far more reason to suspect unknown future complications from COVID than from the vaccine.

If your concern is really that there might be currently unknown long-term damage, then really think about which of the following is more likely to cause such damage: a vaccine that has been well-tested and simply stimulates your immune system and prepares it to fight a single pathogen, or a deadly virus that sets off a cytokine storm and is known to cause serious damage to your heart, lungs, and other organs. Which one actually seems riskier to you?

Part of the problem here is that we often perceive a decision not to take action as the safe option or “erring on the side of caution,” but that’s not always true. Not taking action still has risks, and taking action is not automatically “erring on the side of caution.” The precautionary principle no longer applies to approved vaccines because they have already passed testing. At this point, it is a simple matter of risk assessment, and the risks from not vaccinating are far higher than the risks from vaccinating.

See this post for details on why vaccines are highly unlikely to cause long-term side effects.

Note on the burden of proof: The burden of proof always lies with the person making the claim. In other words, the person saying something exist has to provide evidence that it exists, and the other person does not have to discredit its existence (e.g., the burden of proof is on someone claiming that bigfoot is real, not someone claiming it isn’t). The precautionary principle inherently flips this by saying that we need to provide evidence that a danger doesn’t exist, rather than the burden being on the person claiming that it does exist. As explained earlier, given the very real possibility of injury form something like a new medicine, some basic safety testing is rational prior to approving the drug. However, once reasonable testing has been done, the burden of proof then falls to anyone who claims that there is a risk that those tests missed. In other words, if you want to say that those tests were wrong, then the burden of proof is on you. This is fundamentally why anti-vaccine arguments fail. They try to shift the burden of proof rather than presenting actual evidence.

Note on the “it only kills 1 in 100” argument: I frequently hear people make the argument that we don’t need the vaccine because most people survive COVID. This is a very bad argument for numerous reasons, which I will only briefly outline (see detailed explanation here). First, 1 out of every 100 infected individuals is actually a very high death rate. Second, you also have to consider how infectious the disease is. Even if a disease only kills a minority of infected individuals, that can still result in millions of deaths if lots of people become infected. This absolutely is the case with COVID. Again, it is currently the leading cause of death in the USA. Third, this totally ignores all the non-lethal effects. Fourth, this totally ignores the massive economic harm being caused by the virus. Fifth, this totally ignores the fact that the odds of a currently unknown future side effect from COVID are still much higher than the odds of a currently unknown future side effect from the vaccine.

Note on the “it only kills the elderly” argument: First, no it doesn’t. Yes, mortality rates are much higher for the elderly, but they still exist for all other age groups. In contrast, the mortality rate from the vaccine is 0 for all age groups. So the vaccine reduces risk for all age groups. Additionally, if we want to protect the elderly, the rest of us need to vaccinate to reduce the spread of the disease (vaccines are often less effective for the elderly, not to mention that many vulnerable people simply cannot receive the vaccine). Finally, this argument suffers most of the same flaws as the “it *only kills 1 in 100” argument, so see that note as well. 

Related posts

• Future (“long-term”) side effects from COVID vaccines are extremely unlikely
• The “99% survive COVID” argument is deceptive and completely misses the point
• Understanding the reported risks of medicines, foods, toxic chemicals, etc.
• 10 hypocrisies/double standards of the anti-vaccine movement
• 15 Common Anti-Vaccine Arguments and Why They are a Load of Crap
• Measles infections weaken your immune system and increase your risk of other diseases. Vaccines prevent this
• The Rules of Logic Part 5: Occam’s Razor and the Burden of Proof
• When is it reasonable to demand more studies?
• Why are there so many reports of autism following vaccination? A mathematical assessment

Literature cited

(note: if these hyperlinks break and/or you have trouble accessing articles for free, see this post for suggestions about how to access them)

• Adegbola et al. 2005. Elimination of Haemophilus influenzae type b (Hib) disease from The Gambia after the introduction of routine immunisation with a Hib conjugate vaccine: a prospective study. The Lancet 366:144–150
• Anders et al. 2004. Thimerosal exposure in infants and developmental disorders: a retrospective cohort study in the United Kingdom does not support a causal association. Pediatrics 114:584–591
• Clemens et al. 1988. Mesles vaccination and childhood mortality in rural Bangladesh. American Journal of Epidemiology 128:1330–1339
• DeStefano et al. 2013. Increasing exposure to antibody-stimulating proteins and polysaccharides in vaccines is not associated with risk of autism. J Ped 163:561–567
• Ferris et al. 2014. Long-term study of quadrivalent human papillomavirus vaccine. Pediatrics 134: e657-665.
• Fombonne et al. 2006. Pervasive Developmental Disorders in Montreal, Quebec, Canada: Prevalence and Links With Immunizations. Pediatrics 118
• Fraser 2020. Long term respiratory complications of covid-19. BMJ 370
• Hviid et al. 2003. Association between thimerosal-containing vaccine and autism. JAMA 290:1763–1766.
• Hviid et al. 2019. Measles, mumps, rubella vaccination and autism: A nationwide cohort study. Annals of Internal Medicine.
• Idbal et al. 2013. Number of antigens in early childhood vaccines and neurophsychological outcomes at age 7–10 years. Pharmacoepidemiology and Drug Safety 22:1263–1270.
• Jain et al. 2015. Autism occurrence by MMR vaccine status among US children with older siblings with and without autism. JAMA 313:1534–1540
• Madsen et al. 2002. A population-based study of measles, mumps, and rubella vaccination and autism. New England Journal of Medicine 347:1477–1482
• Mahase 2020. Covid-19: Moderna vaccine is nearly 95% effective, trial involving high risk and elderly people shows. BMJ 371
• Mina et al. 2019. Measles virus infection diminishes preexisting antibodies that offer protection from other pathogens. Science 366:599–606
• Mitrani et al. 2020. COVID-19 cardiac injury: Implications for long-term surveillance and outcomes in survivors. Heart Rhythm 17:1984–1990
• Petrva et al. 2019. Incomplete genetic reconstitution of B cell pools contributes to prolonged immunosuppression after measles. Science Immunology 4: eaay6125
• Polack et al. 2020. Safety and efficacy of the BNT162b2 mRNA Covid-19 Vaccine. New England Journal of Medicine 383:2603–2615.
• Richardson et al. 2010. Effect of rotovirus vaccination on death from childhood diarrhea in Mexico. New England Journal of Medicine 362:299–305
• Smeeth et al. 2004. MMR vaccination and pervasive developmental disorders: a case-control study. Lancet 364:963–969
• Taylor et al. 2014. Vaccines are not associated with autism: and evidence-based meta-analysis of case-control and cohort studies. Elsevier 32:3623-3629
• Uno et al. 2015. Early exposure to the combined measles-mumps-rubella vaccine and thimerosal-containing vaccines and risk of autism spectrum disorder. Vaccine 33:2511–2516
• Verstraeten et al. 2003. Safety of Thimerosal-Containing Vaccines: A two-phased study of computerized health maintenance organization databases. Pediatrics 112:1039–1048
• Vincenzo et al. 2014. Long-term efficacy and safety of human papillomavirus vaccination. International Journal of Women’s Health 6:999–1010.