Friday, December 9, 2016

Libertarian passionately defends special interests

Megan McArdle has a recent piece claiming that Americans should continue paying high prices for drugs to incentivize pharmaceutical innovation. To briefly summarize, her argument consists of hurdle rates 101. This is all well and good, but aren't libertarians supposed to care about all of society, not just pharmaceutical innovation? Let's apply reductio ad absurdum. There is no market mechanism setting prices for on-patent drugs, just drug companies saying "we did some market research so we think the price should be X" balanced with a vague threat that Senator Elizabeth Warren will cordially invite pharma CEOs to the Senate (and say mean things on Twitter) when they overreach on prices. Without a market mechanism, how do we know the "correct price?" Suppose a drug costs $1,000 per pill. Is this the correct price? What about $10,000 per pill? Or $100,000? When does it become too much? If we pay 1000% more, will we get 1000% more pharmaceutical R&D spending in return? When I say too much, I want card-carrying libertarians to get my meaning, so I mean "crowding out innovation and growth in the rest of the economy." But of course I also mean "too much" in the usual sense, i.e. "this pill is crowding out the growth of my bank account." Ha.

The NIH budget is $31.3 billion in FY2016, while American prescription drug spending stands at $425 billion. If we decreased the NIH budget by $15 billion, what do you think would happen to the long-term rate of new drug discovery? If we decreased pharma revenue by $15 billion, what would happen? On the flip side, what if we increased the NIH budget to $60 billion? What if we increased pharma revenue to $455 billion?

Here's one model of drug innovation: the big discoveries happen mainly within academia and small drug companies. At this point it is known to some degree that the drugs might work. Large pharmaceutical companies buy up the rights to these promising discoveries, and primarily provide the service of managing regulatory risk on the (hopeful) way to market. They are also quite good at phase III clinical trials, post market supply-chain management, and derivatizing small molecules within their existing intellectual property umbrellas. Following FDA approval, they provide the service of naming a price that they think the health system will bear. This step might carry the greatest regulatory risk of all: name a price too high for an essential drug, and the federal government might start stepping in to control prices industry-wide.

I don't think McArdle would like that.

Tuesday, December 6, 2016

What’s at stake on both sides of the vaccine-autism debate?

People on either side of this debate will agree: our children’s health is at stake. But my goal with this post, more or less, is to take a close look at the epistemological issues that can arise when worried parents confront the scientific community. I also want to briefly examine the medical-legal implications of this confrontation, or what Michel Foucault referred to as Biopower.

Part I: Ways of Not Knowing

Anchoring bias. Publicized in Thinking Fast and Slow by Daniel Kahneman, anchoring bias is the tendency to hang on to a piece of information simply because it is the first fact encountered. For example, the fraudulent 1998 Andrew Wakefield MMR-autism paper may have stimulated a degree of anchoring bias within the vaccine skeptic community. Even while the paper was made up, its media sensation tapped into real fears that continue until this day.

Framing. Also discussed in Kahneman’s recent book, framing is the tendency to interpret the same facts in different ways depending on how they are presented. The term “vaccine-autism debate” (the very title of this post) is a type of framing that ignores several important issues. For example, it overlooks the strong possibility that autism is caused by environmental factors other than vaccines as well as diagnostic substitution. For more information, see “known unknowns vs. unknown unknowns” and “diagnostic substitution,” below.

Part II: Ways of Knowing

Expert consensus. Many, if not most individuals look to experts before making a decision. In this case, the experts are the FDA, the CDC, medical associations and the broader scientific community who have formed a consensus opinion that currently recommended vaccines are safe and effective, i.e. have a favorable benefit-to-risk ratio, and do not cause neurodevelopmental disorders. Indeed, childhood vaccination rates are quite high in the United States, suggesting that many parents are comfortable with the expert consensus.

Risks vs. benefits. Parents who express concern about vaccines tend to emphasize the risks of vaccination, whereas healthcare providers place a stronger focus on the benefits (alternatively, you could say that healthcare providers are highly focused on the risk of infectious disease). Given this extraordinary divergence of emphasis between vaccine skeptics and mainstream medicine, it is safe to say that only one community can be right (with the essential caveat that there are many different vaccines, with potentially different benefit and risk profiles).

Known unknowns vs. unknown unknowns. This term was coined by Donald Rumsfeld, but he may have been borrowing from Mark Twain who supposedly said, “It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so.” In the vaccine-autism debate, we have many studies of vaccines and autism but many fewer studies of unknown environmental toxins and autism. Why? Those studies are very hard to do. Parents who are worried about autism would do well to support broad-based research into the causes of autism, rather than fixating on vaccines alone. (Note: I’m not implying that all or even most parents worried about autism are fixated on vaccines, just that some are.)

In data we trust. Having discussed a number of psychological biases that can affect our reasoning, including our reasoning when we interpret data, let’s have a look at the role of data in determining the risks vs. benefits of vaccination. Note: we should not always trust data!

Part III: Knowing from Data

Correlation vs. causation. The developmental regression of autism tends to occur, or first be noticed, around the time that children receive a lot of recommended childhood vaccines. This correlation does not imply causation (nor does it disprove causation), but it is understandable from this fact that some parents could attribute autism to vaccines. More broadly, several new childhood vaccines have been added to recommended vaccine schedules at the same time as an increase in the number of autism cases diagnosed. To vaccine skeptics, this correlation suggests causation, but from a logical standpoint it is not exactly a smoking gun. The correlation itself is also unclear— try looking at national and state-level autism data in conjunction with vaccination data and you’ll see what I mean. I may attempt to do this exact thing in a future post.

Causality in epidemiology. It can be very challenging to determine causation from epidemiological time-series data. It is true that autism rates have increased over the last 20-30 years, but the reasons are unclear. Anyone claiming that a specific vaccine was associated with an autism spike is selling you snake oil. For example, in his book Evidence of Harm and in the Huffington Post, the prominent vaccine skeptic David Kirby points out that the Haemophilous influenzae type B (Hib) vaccine was introduced in 1988, which coincided with an increase in autism diagnoses. However, he doesn’t tell you that a new version of the Diagnostic and Statistical Manual of Mental Disorders (DSM) featuring a radically expanded definition of autism was released the very year before, in 1987. Traditional epidemiological data cannot tell you which of these effects was more important.

Diagnostic substitution. As discussed immediately above, the recent (over decades) increase in autism cases is partly due to increasing awareness as well as changing and expanding diagnostic definitions. For an in-depth look at this issue, check out this excellent blog as well as this article.

Randomized controlled trials. Partly due to the above issues, the medical community favors randomized controlled trials to address important clinical questions that can be confounded by diagnostic definitions and other factors. However, because of the risks of creating preventable childhood illness, it’s generally regarded as unethical within the medical community to randomize children into a non-vaccine group.

Part IV: Knowing from Science

Constructive reasoning. Constructive reasoning is the idea that if we have a reasonably good mental model, we can extrapolate the model to situations that we haven’t previously studied (this is the Kuhnian approach). For example, we know that vaccination has been very successful at eliminating devastating viruses such as smallpox and polio, as well as viruses that occasionally have terrible effects such as the measles virus. This leads to the notion that we should continue developing and expanding the arsenal of vaccines.
          On the “vaccine skeptic” side, we know that organic mercury is highly neurotoxic and should be maintained at very low levels within the human body. Therefore, it seems very reasonable that thimerosal has now been removed from nearly all childhood vaccines in the United States. (The thimerosal preservative was only present at very low levels, but the added expense of removing it from vaccines seems worthwhile from a constructivist, precautionary standpoint.)
  It is not difficult to find flaws in the constructive reasoning of vaccine skeptics. For example, Lundy Bancroft has a list of vaccine ingredients on his blog, many of which are labeled neurotoxins. Yet from a constructivist standpoint (in my role as a medical student) it is obvious that most of these ingredients are safe and harmless. For example, he lists potassium chloride as a neurotoxin, but that’s only if you inject massive amounts – not the tiny quantities found in vaccines. (Potassium and chloride are abundant within the body. Most of your potassium is inside of your cells. Injecting too much potassium can throw off that balance. Injecting a vaccine does not come close to that level. If you eat one banana, you will receive roughly 70 times more potassium than is contained in a typical vaccine. I'm happy to provide a calculation if provoked.) Here are some discussions of Lundy’s other claims:
  • "Squalene is a neurotoxin”: squalene is naturally present in your body as a biosynthetic precursor to prostaglandins, which are chemicals that make you hurt (really). It’s included in some vaccines because it slightly stimulates the immune system, leading to improved efficacy. “Stimulate the immune system” might sound scary; however, if you walk into a coffee table, you will also naturally generate squalene and prostaglandins as a bruise forms. 
  • "Formaldehyde is a carcinogen”: yes, in larger amounts. However it’s also naturally present within your body and dietary fruits at levels higher than you receive in vaccines.
  • “beta-Propiolactone is a carcinogen”: this one might be true, so aligning with my constructivist viewpoint it might be good to remove it from vaccines. However it also eliminates bacterial spores which keeps vaccines safe. Also, note that grilled meat is a carcinogen. Lots of things are carcinogens! Hot tea is a carcinogen, and so is alcohol.
  • “Polymyxin is a neurotoxin”: polymyxin is an antibiotic that has some neurotoxic effects at higher doses. It’s present in some vaccines to keep bacteria from growing, but not at a level that is likely to cause human neurotoxicity (keep in mind it’s not mercury, just an FDA-approved antibiotic).
  • “Neomycin is an immunotoxin”: neomycin is also an FDA-approved antibiotic, which causes allergic reactions in some people.
  • “Gentamicin sulfate is a nephrotoxin”: this is also an approved antibiotic with some known nephrotoxicity at therapeutic doses. The doses in vaccines are fine. (We are talking about orders of magnitude of differences in dosage levels.)
  • “Monobasic potassium phosphate is an immunotoxin”: you have lots of potassium and phosphate in your body. The small amount in a vaccine changes nothing.
There are more ingredients on Lundy’s list, but I think you can see where this is going. The list is not well curated and has potential to generate a lot of unnecessary hysteria.

Scientific proof. Karl Popper famously wrote that science should be falsifiable. (His arguments were actually much more nuanced, but oh well.) For example, the proposition “vaccine X is safe,” generally favored by the scientific community for currently approved vaccines, is falsifiable – you only have to demonstrate that vaccine X isn’t safe. Yet, vaccine skeptics haven’t successfully falsified “vaccine X is safe” for any of the currently recommended childhood vaccines (historically there have been some vaccines pulled from the market). However, the related proposition “vaccine X is not safe,” favored by vaccine skeptics is much more difficult to falsify. Why? There are many ways that vaccine X can be unsafe, and it is laborious to “disprove” all of them. Instead, the FDA simply demands studies and deems the vaccines safe if the apparent rate of complications is very low. This is not adequate for vaccine skeptics, who tend to demand more and more studies of safety and/or abstain from vaccination altogether.
  In case the above paragraph contained too many double negatives, here’s another way to look at the problem. Vaccine skeptics have floated a number of falsifiable hypotheses regarding the safety of vaccines. For example, there’s the Wakefield-MMR-GI symptoms-autism hypothesis. This hypothesis was falsified, so vaccine skeptics migrated to the thimerosal-autism hypothesis. But then thimerosal was removed from vaccines (in the United States) and autism rates persisted and/or continued to increase (caveat: read my section on “diagnostic substitution”). Yet, now that the thimerosal hypothesis has been somewhat falsified (let’s say it’s on its last legs), vaccine skeptics have floated additional hypotheses about additional vaccine ingredients, such as the ingredients in Lundy’s list. It is very difficult and time-consuming for mainstream science to refute all of these hypotheses individually, especially when there is limited research funding and there is a pressing need to study other potential causes of autism.

Part V: Ways of (Not) Arguing

Weasel words. Weasel words are a technique to issue a poorly substantiated claim at a subconscious level. For example, Lundy Bancroft titles his blog “What the CDC and the drug companies don’t want you to know about vaccines.” Lundy goes on to make some fair points, but these points in no way substantiate his title, which implies a broad-based conspiracy within the medical and scientific community. It is also possible to find weasel words on the pro-vaccine side of the debate. For example, a 2009 article in PLOS Biology by Liza Gross uses the heading “Evidence-Resistant Theories” to imply that vaccine skeptics do not take evidence into account.

Straw men. A straw man is useful as a scarecrow for farmers and gardeners, and even more useful for vaccine debaters who wish to win an argument on dubious grounds. The fraudulent 1998 Andrew Wakefield paper that claimed a connection between the MMR vaccine and autism has now become something of a straw man within the pro-vaccine community. In other words, citing the Wakefield paper does not prove vaccine safety, it only proves that one particular study questioning vaccine safety turned out to be fabricated. I am personally guilty of using this straw man on multiple occasions, but I’ve now committed to change my ways.

Follow the money. Lundy Bancroft makes extensive claims of financial conflicts of interest at the CDC, which I responded to here. In my opinion, Lundy’s concerns about the CDC are somewhat warranted. However it’s important to notice that Lundy is committing the ad hominem fallacy. In other words, Lundy seeks to discredit all of the vaccine recommendations of the CDC by discrediting one or more members of its Advisory Committee on Immunization Practices, or ACIP. Logically speaking, this is a non sequitur. There is an entire scientific community backing the safety and efficacy of childhood vaccines; it would require a vast conspiracy if this consensus were created from conflicts of interest alone.

Part VI: Biopower

Opt-in vs. opt-out. We have something of an opt-out pediatric vaccination system in the United States. This is to say that vaccines are required of children who attend public schools (usually thanks to state laws), but it can be fairly easy to obtain personal or religious exemptions to vaccines in some cases. It is widely believed that opt-out policies lead to higher participation rates than opt-in policies. For example, this appears to be the case for organ donation.

Forced vaccination. Vaccine skeptics tend to be very critical of the opt-out vaccination system and accuse the medical-legal community of performing forced vaccinations. This charge has some merit, given that requiring children to vaccinate before they attend school is rather coercive. However, most states allow religious and/or philosophical exemptions to vaccination, which somewhat undermines the argument that forced vaccination is a common practice in the USA.
  Looking worldwide, the oral (attenuated) polio vaccine provides a major example of forced vaccination with known risks, not only because it has been coercively administered but because the vaccine is capable of propagating from certain individuals to others through a fecal-oral transmission route (just like the ordinary poliovirus). Yet, according to the World Health Organization, this somewhat contagious vaccine has prevented more than 13 million cases of polio. We are now getting to a point where it seems potentially advisable to globally switch to the inactivated polio vaccine (but I am not an expert on this subject).

Herd immunity. Herd immunity is the notion that it is difficult to incubate a significant level of a virus or bacterium within a population if most individuals are immune. This, in turn protects immunocompromised individuals and those who cannot receive vaccines for medical reasons. In the most extreme example, smallpox has been eradicated worldwide thanks to herd immunity combined with relentless vaccination efforts. Herd immunity is frequently cited as a justification for coercive vaccination policies.