Fake News (Rarity, Part Two)

I was diligently trying to understand the prevalence of ataxia compared to other diseases, and I didn’t expect to find numbers that were inflated by an order of magnitude. After I posted my article, I started noticing references to 150,000 (people in the U.S. with various kinds of ataxia), when in fact the high estimate is around 16,000 (people in the U.S. with spinocerebellar ataxia).

This bogus number of 150,000 has been around, unchanged, for over 20 years, while the population of the U.S. has gone up about 24% (about 60 million). I see this bogus number as the great ataxia urban legend.

The world population in the early 1990s was around 5 billion. At that time, the worldwide prevalence of SCA would have been reasonably estimated to be about 150,000 (3 per 100,000). I think it’s likely that someone got their wires crossed back then, and that misuse has stuck now for decades.

In 2017, I see knowingly hanging on to these massively inflated numbers as having a Trumpian disdain for the truth; it’s shameful.

Tiny numbers

The number of people in the U.S. with SCA is tiny. The maximum estimates are around 16,000 cases, with the lower estimates being around 3,000. I took these particular estimates—in terms of 1-5 per 100,000—from here, on page 7:

https://ataxia.org/pdf/2009_Managing_Ataxia_and_Managing_Your_Neurologist.pdf (removed from website)

Note: I later noticed neither the word prevalence nor incidence is used. More on this later.

I have never seen any contradictory information on ataxia in terms of per 100,000 numbers. I’ve certainly never seen any claims that it is 50 per 100,000 instead of 5 per 100,000, which is essentially what people are claiming when they refer to the mythologically high number of 150,000 in the U.S.

Everyone in the U.S. with a health care education (researchers, neurologists) should know instantly what 1 per 100,000 amounts to in the U.S., so clearly they know that 150,000 is wildly bogus. Every tiny startup company who publishes these bogus numbers must know what the actual numbers are; they must know the size of their potential customer base is 1/10 of what they say publicly. One wonders if they are honest at least with their investors.

In other words, many people out there must realize when the numbers are completely bogus, yet this myth has persisted for over 20 years.

It boggles the mind that seemingly no one ever attempts to correct this error. There seems to be an unwritten rule that using a highly inflated, bogus number is synonymous with hope, and that correcting someone is off limits rather than in everyone’s best interest.

Other sources

Rather than belabor the point, here are a few more web pages that corroborate low prevalence numbers for SCA and Friedreich’s ataxia (most are even lower than the numbers I’m using):

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3105349/#S6
https://www.ncbi.nlm.nih.gov/books/NBK1138/
http://www.dizziness-and-balance.com/disorders/central/cerebellar/sca.htm
http://www.prnewswire.com/news-releases/repligen-receives-orphan-drug-designation-from-the-fda-for-rg2833-for-friedreichs-ataxia-94737209.html
https://www.news-medical.net/health/Friedreichs-Ataxia-Prevalence.aspx

As in my original article, I’m using 317 million as the population of the U.S., though for rough estimates sometimes I use a rounded 300 million.

Wikipedia

This page says, “An estimated 150,000 people in the United States are diagnosed with [s]pinocerebellar [a]taxia.”

https://en.wikipedia.org/wiki/Spinocerebellar_ataxia

This is blatantly incorrect. A number of 16,000 may be reasonable for an upper estimate. I traced the Wikipedia entry to this sloppy and error-filled edit by a rogue user on 2012-09-05:

https://en.wikipedia.org/w/index.php?title=Spinocerebellar_ataxia&diff=510914186&oldid=504136874

National Ataxia Foundation

This page says, “An estimated 150,000 Americans are affected by heredity [sic] or sporadic ataxia.”

https://ataxia.org/fact-sheets/

The NAF’s wording is a bit nebulous and contains a grammatical error. My guess for their justification of using such a high number is that for every person with ataxia, there are many more affected by that one instance. That’s perhaps a bit misleading, yes, but arguably not blatantly incorrect.

In my view, propagating misinformation like this poisons the well of meaningful information out there and serves mainly to misinform both those looking for answers and those casually absorbing data.

Their justification may involve them having the correct constituent numbers elsewhere on their website.

Can we get the numbers up a little?

There’s already a big difference between the prevalence of SCA vs. the combined prevalence of hereditary (dominant and recessive) and sporadic ataxias. Let me try to be up front and show a table of prevalences based on using 1-5 per 100,000 for autosomal dominant ataxia, 2-3 per 100,000 for Friedreich’s ataxia (I’m generalizing this to represent autosomal recessive ataxia), plus the ratio of sporadic to hereditary being 2:1 (which I’ve come to think is a bit on the high side; it may be closer to 3:2).

Low est. Avg est. High est.
Dominant 3,170 (1) 9,510 (3) 15,850 (5)
Recessive 6,340 (2) 7,925 (2.5) 9,510 (3)
Sporadic 19,020 (6) 34,870 (11) 50,720 (16)
Total 28,530 (9) 52,305 (16.5) 76,080 (24)

I created an interactive user interface to allow easy computation of numbers like those in the table above. See here.

The international view

I dearly hope that the NAF’s justification for publishing the 150,000 number doesn’t include treating Americans as any inhabitant of the Americas, referring to the countries comprising the entire Western Hemisphere. Currently, the population of that region is about one billion, and 150,000 is a reasonable ataxia prevalence for that region.

United Kingdom

https://www.ataxia.org.uk/

This site says that there are over 10,000 with ataxia in the UK. The population of the UK was about 64.1 million in 2013. Thus, their number puts ataxia prevalence at about 16 per 100,000. If they used the same prevalence that the NAF erroneously uses, then they would claim 30,000 cases (instead of 10,000). If the NAF used the same prevalence that the UK site used, then they would claim about 50,000 (instead of 150,000).

Note that the prevalence used on the UK site is quite reasonable (about mid-range), considering that—like the NAF—it’s trying to represent hereditary and sporadic cases combined.

Australia

http://www.scars.org.au/ataxia.html

This site addresses only SCA and says there are 6 per 100,000 cases. The population of Australia was about 23.1 million in 2013. The given prevalence means that there are about 1,386 people there with SCA.

The Netherlands

http://www.ataxie.nl/

Specifically, see http://www.ataxie.nl/wp-content/uploads/2013/09/Huisartsenbrochure_ADCA.pdf.

The brochure addresses autosomal dominant ataxia only. It gives a prevalence of 2-4 per 100,000, which is also given in terms of the actual number of about 700. The population of the Netherlands was about 16.8 million in 2013. They used the upper estimate (4 per 100,000) for their calculation, and they did the calculation correctly.

Sweden

http://www.sca-network.com/fakta-sca3/

This page gives a fairly standard prevalence of 1-3 per 100,000, uses a (correct) population of about 10 million for Sweden, and correctly computes an upper number of 300.

Does it matter?

Yes, I think it matters.

  • International embarrassment. Is the ataxia problem in the U.S. more prevalent and more important than everywhere else? No, our head is in the sand.
  • Ataxia awareness, my foot. The people with the disease are systematically misinformed and are being used to spread misinformation.
  • It warps perception of other rare diseases. ALS. Huntington’s disease. Ataxia. Though different, they are similar in a lot of ways, including prevalence. But ataxians are taught that their disease is special and ten times more prevalent than the others, when it’s not. This breeds contempt. People on the NAF Facebook page think it’s unfair that the other diseases get so much attention.
  • Confusion. Ataxians wonder where all the fellow ataxians are hiding, rather than understanding the rarity of the disease. The reason that ataxians don’t come out in droves is that the droves don’t exist.
  • More confusion. A prominent misconception is that authorities like the NAF know the true numbers and that therefore 90% of the people with ataxia (and dying from it) don’t know they have it.

Prevalence vs. incidence

The terminology here should be entirely unambiguous and is instead used in a whirlwind of inconsistency.

Prevalence is how many people are known to have something right now. Incidence is the rate of new cases known to be occurring, and so incidences must include a time frame, such as being per live births (because that implies annual).

If you say the prevalence of something is 5 per 100,000, that is unambiguous for a particular population. If you say the incidence of something is 5 per 100,000, even for the same population, that is totally ambiguous; you could make it unambiguous by saying 5 per 100,000 got something last year (in which case it applies to an entire population) or by saying that the rate of new cases is 5 per 100,000 per live births.

What is the incidence of SCA? In 2010, the birthrate in the U.S. was about 1,350 per 100,000. Factoring in SCA at 1-5 per 100,000 live births, that’s about 13.5-67.5 per 100,000,000, or about 41-203 new SCA cases that year.

https://en.wikipedia.org/wiki/Demographics_of_the_United_States#Birth.2C_growth.2C_and_death_rates

Note the vastness in difference in meaning between saying 5 per 100,000 have something now versus 5 per 100,000 new people got something last year. It’s the difference between saying 16,000 have a faulty gene (correct for SCA) vs. saying 16,000 got the faulty gene last year (incorrect for SCA; it’s closer to 200).

If the prevalence of ataxia were 150,000 in the U.S. today (as many have stated erroneously), then that would be about 47 per 100,000, and I’d expect a similar ratio in live births. If instead the incidence per live births were 5 per 100,000 (which probably is correct), then that would be a phenomenal decline! Instead, ataxia has a low incidence and a low prevalence.

It’s perhaps almost fine if people refer to incidence and prevalence of ataxia as being the same (i.e., 1-5 per 100,000), as long as they mean the incidence per live births and prevalence per population.

Examples of confusion

The most disappointing word misuse is here (references to incidence when prevalence is meant), because the document itself is truly a treasure trove of valuable information:

https://ataxia.org/wp-content/uploads/2017/07/Evaluation_and_Management_of_Ataxic_Disorders-An_overview_for_Physicians.pdf

Here’s a page that says, “Ataxia-telangiecstasia occurs in one in 40,000 to 100,000 live births in the U.S.” This sounds about right.

http://www.chop.edu/conditions-diseases/ataxia-telangiectasia

Wikipedia contradicts this with, “The incidence world-wide is estimated to be between 1 in 40,000 and 1 in 100,000 people.” They mean prevalence.

https://en.wikipedia.org/wiki/Ataxia_telangiectasia#Epidemiology

Wikipedia gets it wrong again. And wrong again, here:

https://en.wikipedia.org/wiki/Friedreich%27s_ataxia

The first paragraph says, “Its incidence in the general population is roughly 1 in 50,000.” Whoever wrote that meant prevalence.

Crosschecking the numbers

Let’s think about what it would mean if 150,000 people in the U.S. had ataxia. Ataxia would be nearing the limits of even being considered a rare disease, as the threshold is 200,000 in U.S. law. It would put the prevalence of ataxia an order of magnitude higher than the prevalence of ALS, Huntington’s disease, and muscular dystrophy, when in fact its prevalence is about the same as or less than ALS.

Facebook

The National Ataxia Foundation‘s Facebook page has 3,073 likes as of today. My feeling is that most people with ataxia in the U.S. have heard of the NAF, and so many in the U.S. are on Facebook. If 150,000 people had the disease, clearly more people would “like” the page.

The Huntington’s Disease Society of America‘s Facebook page has 15,446 likes as of today. This is more in line with the number of people with the disease, which is perhaps only twice as many.

The ALS Association‘s Facebook page has 340,791 likes as of today. This number is unusually high (outside the realms of a rare disease) and has probably been skewed a bit by the popularity of the ALS Ice Bucket Challenge of 2014. Considering there are only upwards of 20,000 with the disease, that’s about 17 likes per instance of the disease.

The weirdest math ever

The problem isn’t just with the math. It’s with assumptions made and an apparent lack of context for intuitively knowing that one’s initial conclusions are massively wrong. Here’s the most egregious example I’ve ever seen:

https://www.facebook.com/permalink.php?story_fbid=1002439456443754&id=970145149673185

The proposed justification for the high numbers is that most with ataxia don’t even know it. People resort to mythology to make the numbers work. It’s mystifying.

The post above makes the wildly bizarre conclusion that 150,000 people die yearly from ataxia and that 150,000 people each year are born with ataxia, in the U.S. That is a staggeringly strange and twisted take on the already invalid 150,000 number.

(150,000 is divided by 365 to get about 400. This computation is totally nonsensical.)

Mathematically, the actual ataxia birthrate is 41-203 per year in the U.S., as explained here. Assuming the mortality rate is similar to the birthrate is probably fine. In other words, upwards of 200 die per year (with SCA) rather than 400 per day—three orders of magnitude difference.

  • In the U.S., about 1,350 per 100,000 are born in one year. That was about 4.347 million in 2010.
  • In the U.S., about 800 per 100,000 die in one year. That was about 2.576 million in 2015.

The Facebook post above would like us to believe that 6% of all yearly deaths in the U.S. are due to ataxia. Again, that number is off by about three orders of magnitude, being closer to 0.008%.

What do I mean by not having a context for understanding why the conclusions are massively wrong? Let’s look at deaths in the U.S. caused by the leading causes of death:

https://www.cdc.gov/nchs/fastats/leading-causes-of-death.htm

At 150,000 deaths per year, that would put ataxia below cancer (#1) but above heart disease (#2). The absurdity of this is so extremely high, but it hints at why people wonder why so many other diseases get more attention than ataxia.

Online support groups

One active online support group is this one, with 1,784 English-using members worldwide (update: number of users is no longer exposed):

https://www.livingwithataxia.org/

There are perhaps 476 members with SCA, and maybe 30 or so with SCA3. This is hard to nail down because of inconsistent free-form terminology, including many misspellings.

This site seems to confirm an interesting situation in the world of ataxia, which is that most cases of ataxia are sporadic. At least, the people with sporadic ataxia are questing for answers, and support groups can help them on their quest. Contrastingly, if you have hereditary ataxia, then you’ve likely seen a parent go through it already and know what to expect in yourself without input from a support group.

The general population

To help estimate the quantity of people with SCA in a given U.S. state without looking up specific population numbers, let’s use the same numbers and ratios that the Electoral College uses. In other words:

If there were 150,000 with ataxia in the U.S. (and there aren’t), then California might have 55/538 * 150,000 = 15,335 cases. If there were that many, there would probably be more than just three support meeting locations in the state mentioned on the NAF site. There could be a group in every major county across the state (there are 58 counties) to accommodate the supposedly many thousands of people with the disease.

https://ataxia.org/support-groups/

California (the most populous state, by the way, at over 38 million people) has an estimated maximum of 55/538 * 15,850 = 1,620 people with SCA (low estimates being around 324). That’s 162-810 people in each of northern and southern California. That’s maybe a few dozen or a few hundred people with one of the more common kinds of ataxia. In other states, the numbers go down to as low as 3/538 * 15,850 = 88 (with low estimates being around 27).

In other words, there may be in reality a few dozen to a few hundred people in a U.S. state with SCA, not the thousands per state that some would like us to believe.

I have some guesses

My first guess is that the 150,000 number simply was taken from the valid SCA prevalence in the early 1990s for the entire world, population 5 billion. My second guess is as follows:

Perhaps the 150,000 number, which is blatantly incorrect, was erroneously “borrowed” in the early 1990s from the prevalence of spinal cord injuries. I’m imagining a journalist desperate for a statistic to bolster his article who ends up using a number that is entirely wrong yet ends up persisting as part of the legend for 20+ years.

Here’s a PDF from 1992 that says there were 150,000 with spinal cord injuries living in the U.S.:

http://www.cs.cmu.edu/~khaigh/ILSAEXTERNALWEBSITE/content/publications/1992-HumanFactors.pdf

Note that this was around the time the SCA genes were being discovered and articles were being written. For example, here’s an article from 1996 about SCA2, which gives the erroneous 150,000 number:

https://www.ninds.nih.gov/news_and_events/news_articles/press_release_spinocerebellar_ataxia_2_103196.htm (removed from website)

But this accidental prevalence-borrowing is just a guess for a possible explanation. The 150,000 number is propagated in every human-interest story I’ve seen, sometimes giving it as the number of people with hereditary ataxia, and sometimes giving it as the number of people with any form of ataxia. It’s just an arbitrarily large number that for whatever reason people have been comfortable using for decades now.

Update, 2015-08-19

Someone finally decided it was time to be bold and increase the number from 150,000 to 250,000. It continues to disappoint me that using inflated prevalence numbers is synonymous with raising awareness.

Widow raising ataxia awareness

Update, 2015-09-18

Here’s an article with a rare, accurate reference to per-100,000 prevalence:

http://abc7.com/health/research-offers-hope-for-ataxia-patients/991451/

Update, 2015-12-11

The international view is often more correct than in the U.S., but not here:

https://www.vernonmorningstar.com/life/living-with-ataxia-and-sharing-awareness/

The article claims that 25,000 have ataxia in Canada. The population of Canada is about 36 million.

  • 9–24 per 100,000 is 3,410–9,240, not even close to 25,000.
  • 25,000 in Canada is about 69 per 100,000: even more wacko than the numbers often used in the U.S.

Update, 2016-12-22

This is the first time I have ever seen accurate numbers given “in print.”

http://www.prnewswire.com/news-releases/biohaven-enrolls-first-patient-in-pivotal-trial-of-bhv-4157-in-patients-with-hereditary-spinocerebellar-ataxia-300382831.html

SCA is estimated to affect approximately 10,000 to 20,000 people in the United States.

On the same day, this super-disappointing usage came out, boldly raising the SCA prevalence to 250,000 (78 per 100,000), which puts it beyond the realms of rare diseases entirely:

https://www.facebook.com/jason.wolfer.7/posts/1186884801377887

Perhaps a pattern is emerging. The worldwide prevalence of SCA in 1992 could have been reasonably estimated to be about 150,000 (2.7 per 100,000). The worldwide prevalence of SCA in 2017 could be reasonably estimated to be about 250,000 (3.3 per 100,000). It might surprise some U.S. Americans that the U.S. makes up only 4.4% of the world’s population, not as close to 100% as some might imagine.

Whoops

I didn’t mean to encounter so much inconsistency. In fact, I may have fallen victim to some of the misinformation myself in my original article. I’m a big fan and user of Wikipedia, and this one bad experience won’t change that. But I won’t know for sure if I can trust the numbers I used (for all diseases) until I review them all in detail.

Another point well taken by me is the overwhelming presence of sporadic ataxia. The fact is, the majority of people with ataxia have sporadic ataxia.

Leave a Reply

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