Just in time for Halloween:
I came across this 2-minute video by Brian Zikmund-Fisher, a professor of Health Behavior and Health Education at the University of Michigan, and I took a look because I was curious what he had to say. The video is called “Why aren’t we more scared of measles?” and has the following description:
Measles is one of the leading causes of death amongst children worldwide. In 2012, an estimated 122,000 people died of the disease according to the World Health Organization – equivalent to 14 deaths every hour. Yet talk to parents about this highly infectious disease, and the response is often a resounding “meh”. Why is this?
My first thought was most of the estimated 122,000 deaths are occurring in other countries but he’s looking at attitudes in the U.S., so it’s not clear how relevant this global figure is. So what are the numbers here? I did a quick google and found that there has been a sharp increase in measles in the U.S. this year (“A total of 288 confirmed measles cases” from Jan-May 2014, compared to 220 cases reported in all of 2011), of which half were in children, and zero deaths.
Or, as it might be put in the above blurb, this is equivalent to 0 deaths every hour!
So people’s attitudes on measles don’t seem so mysterious to me. There’s this disease that’s killing people in other countries but not in the U.S., and Americans aren’t worried about it. There have been some scattered outbreaks recently, and public health officials are rightly concerned, but it hasn’t hit the radar of most people in this country.
At this point I was wondering what the video was going to say. The online description continues:
It’s a fantastic introduction to why seemingly rational people sometimes behave the way they do toward vaccines.
According to Zikmund-Fisher, how we think about infectious diseases and risk is governed in part by the way our memories and feelings inform our perceptions – this is referred to by psychologists as the “availability heuristic”. It turns out that when we try and figure out how rare or common a disease is, we try to think of people we have heard of who have had it. If we know of people, we’re pre-programmed to feel more at risk than if we don’t. And surprisingly, the statistics – the actual numbers of people who get sick – don’t seem to matter.
Then I watched the video, and indeed the above paragraph pretty much describes what he said there. And it makes no sense to me.
We aren’t so scared of measles here because it’s not killing Americans and it hasn’t done so for decades. Now Zikmund-Fisher is saying we should be scared. Fair enough: there’s a risk if you don’t get the vaccine because then you can catch it some other non-vaccinated person, maybe from some other country where vaccination rates are low. All this makes sense but I don’t see the availability heuristic being the problem. People are actually making reasonable statistical inferences—no Americans are dying of measles and very few people are getting sick of it, so indeed it is appropriate for them to require additional information (in this case, CDC alerts spread by the news media) before changing their attitudes.
Here’s the point. Based on the information available to me, I agree with Zikmund-Fisher that everyone should get the measles vaccine. This decision makes sense based on the costs, risks, and benefits of vaccination. But I don’t see that availability bias comes in here. Part of the assessment of benefits is the existing rate of measles—and people are doing a pretty good job if they estimate this as near-zero.
Again, look at this from Zikmund-Fisher:
When we try and figure out how rare or common a disease is, we try to think of people we have heard of who have had it. If we know of people, we’re pre-programmed to feel more at risk than if we don’t. And surprisingly, the statistics – the actual numbers of people who get sick – don’t seem to matter.
But people are correctly estimating how rare or common the disease is here: almost nobody in the United States gets measles and nobody dies from it. Those risks might change, but to hold this view requires additional information, not just “the statistics – the actual numbers of people who get sick.”
What the health experts are saying—and it is completely reasonable—is that, in assessing the risks, people should not simply make a decision based on the actual numbers of people who get sick, but instead to look at look at new risks that have barely appeared here. That’s fine but it doesn’t have much to do with availability bias.
P.S. What’s really weird to me is that an expert in risk communication would summarize the risk as “14 deaths every hour”! Of all ways of summarizing, this just seems like one of the more confusing. I mean, why didn’t he go whole hog and give it a description such as, “If each measles death were a 10-pound bag of gold, it would be costing us 20 billion dollars a year.” Or a classic such as, “If you were to lay the measles victims end to end, they’d stretch all the way from Pittsburgh to Youngstown, and almost all the way back!”
It’s also true that heuristic =/= bias, necessarily. The heuristic inference, “If lots of American kids were dying from measles, I would have heard about it from friends, my kids’ school, or the news, and I don’t recall hearing it” is quite a reasonable use of availability.
What if we applied the same sort of heuristic to, say, Global warming? The “existing rate” of immediate impact from AGW seems pretty low.
We aren’t so scared of AGW here because it’s not killing Americans and it hasn’t done so for decades (and probably won’t for decades either, and I’m no AGW denier!).
In the measles case the cohort impacted is non-Americans, specifically perhaps certain impoverished nations. For AGW it might be future generations.
If it doesn’t make sense to feel a natural kinship towards one cohort, does it make sense to worry about the other?
I agree that it seems to have nothing to do with availability bias. We are in the good equilibrium: measles causes no deaths because (almost) everyone is vaccinated. If most people stopped getting vaccinated, the disease would start killing again. We’d be in the bad equilibrium.
Maybe this idea is too difficult to communicate to the public? That the reason why measles is not a problem is precisely because we are vaccinated.
Then again, the idea is no more difficult to grasp than: I am not hungry precisely because I have eaten. If I stop eating altogether, I’ll be famished.
“We are in the good equilibrium: measles causes no deaths because (almost) everyone is vaccinated”
“If most people stopped getting vaccinated, the disease would start killing again. We’d be in the bad equilibrium.”
“the reason why measles is not a problem is precisely because we are vaccinated.”
Be careful. My forays into the world of vaccine research left me very disappointed (the usual “significant p value” -> “my theory is true” business). I don’t feel like digging up references so I won’t say more here. However, I will ask that others supply references if they make claims. Do you have sources? I mean actual evidence that you are basing your beliefs on rather than “everyone knows”, “lmgtfy”.
I probably shouldn’t engage, but … what is your claim here? That the MMR vaccine doesn’t protect against measles? That measles would not spread (i.e. incidence would not increase dramatically) if we stopped vaccinating? That measles wouldn’t cause deaths in proportion to its incidence? Do you dispute the information on Wikipedia ( http://en.wikipedia.org/wiki/Measles#Prognosis ) ? (I’m not saying you should believe anything in particular, I’m just trying to figure out which part of the claim you dispute …)
How about this: http://www.cdc.gov/mmwr/preview/mmwrhtml/00053391.htm
Before measles vaccine was licensed in 1963, an average of 400,000 measles cases were reported each year in the United States (8). However, because virtually all children acquired measles, the number of cases probably approached 3.5 million per year (i.e., an entire birth cohort).
Since measles vaccine became available, professional and voluntary medical and public health organizations have collaborated in vaccination programs that have reduced the reported incidence of measles by greater than 99%.
…
There are a whole bunch of links and a lot of interesting discussion at the CDC site I link to, but of course, they gloss over the fact that vaccines cause autism (*).
* Furthermore, I am of the opinion that Andrew Wakefield should be in prison. http://en.wikipedia.org/wiki/Andrew_Wakefield
Ben and jrc,
Neither of you presented any specific evidence that has convinced you. The CDC link is essentially a review article (not even peer reviewed for those who think that matters), ditto the wikipedia article. I have never looked into the measles vaccine specifically. Like I said I don’t feel like digging up the references to support any claim of my own right now so I am abstaining from making one. My point here is that I feel claiming “measles vaccine works” without citing the specific convincing evidence is un-scholarly.
The main thing I would check in any paper is how they managed/attempted to disentangle changes in physician diagnosis behaviour from vaccine effectiveness.
1) Has the criteria for a measles diagnosis changed over time? Once lab tests become available it is plausible to think the criteria get more specific, obviously this could confound the treatment effect.
2) What effect does telling a doctor you were vaccinated and the “popularity” of the diagnosis have on how they interpret symptoms? If he/she knows you were vaccinated, are they more likely to diagnosis something else? What other illnesses share similar symptoms? Has the diagnosis rate of these risen?
So I point to a 99% reduction in measles cases in the immediate aftermath of vaccination programs and that is me responding “without citing the specific convincing evidence”? OK then: http://ije.oxfordjournals.org/content/39/suppl_1/i48.full
…
We identified three measles vaccine RCTs and two [quasi-experimental] QE studies with data on prevention of measles disease. A meta-analysis of these studies found that vaccination was 85% [95% confidence interval (CI) 83–87] effective in preventing measles disease…The literature also suggests that a conservative 95% effect estimate is reasonable to employ when vaccinating at 1 year or later and 98% for two doses of vaccine based on serology reviews.
…
As for your questions about diagnosis rates – most of the obvious theoretical unobserved effects point in the wrong direction (as in, identifying more cases than we would have before and thus under-estimating the effectiveness of the vaccine using only the time-series). The only two things that comes close to being a potential upward bias are the idea that doctors no longer diagnose measles in kids who have a measles vaccine and the idea that people wildly over-diagnosed in the past. But I think that between reliable diagnostic blood tests and the fact that measles causes a distinctive rash and associated symptoms, the current under-reporting argument is pretty unlikely (at least to outweigh the bias in increased monitoring/reporting that would have occurred in the last several decades). And I think that since everyone used to get measles and everyone knew what it was and looked like mean the latter argument is unconvincing.
Is that sufficient? Usually your question questions have some statistical intuition or content. Today you just seem to be saying “no, I don’t feel like believing that the measles vaccine works, and so I’ll write some pseudo-suggestive comments about diagnosis rates,” – that problem may be important in other areas (say, internal body cancers) but doesn’t make sense in this context.
And back to the main point – as soon as we start vaccinating people, measles rates plummet (like, 99% disappear). How is that not convincing? You want a t-test?
jrc,
Please don’t get upset with me for questioning (although here I did not even question, only asked for sources and speculated on possible confounds) holy cows. One of my motivations is that “if 9/10 people say something is correct, it is the duty of the tenth to attempt raising doubt”. I read that somewhere online once and liked it…
I mentioned above having never looked into measles vaccines. My skepticism is based on extrapolating from what I found regarding others and my own subfield of biomed (a lackadaisical attitude towards ruling out alternative explanations). I am going to investigate that paper, thank you. Your next paragraph includes many plausible assumptions but no data. It would be helpful to me if you found sources for what was said there. As to your final paragraph:
“And back to the main point – as soon as we start vaccinating people, measles rates plummet (like, 99% disappear). How is that not convincing? You want a t-test?”
This indicates you missed my entire point about alternative explanations. What else, if anything, happened concurrently? The diagnosis rates plummeting may be influenced by other factors.
Also:
http://www.ncbi.nlm.nih.gov/pubmed/22336803
We included five randomised controlled trials (RCTs), one controlled clinical trial (CCT), 27 cohort studies, 17 case-control studies, five time-series trials, one case cross-over trial, two ecological studies, six self controlled case series studies involving in all about 14,700,000 children and assessing effectiveness and safety of MMR vaccine. Based on the available evidence, one MMR vaccine dose is at least 95% effective in preventing clinical measles and 92% effective in preventing secondary cases among household contacts.
…
And measles is actually fairly easy to diagnose: “Koplik’s spots” http://en.wikipedia.org/wiki/Koplik%27s_spots are a relatively reliable diagnostic criterion (I am not a doctor …)
Ben,
In that review, MMR effectiveness for measles was only addressed in cohort studies. Here is what they say in the limitations section regarding those:
“A lack of clarity in reporting and systematic bias made comparability across studies and quantitative synthesis of data impossible.”
http://www.ncbi.nlm.nih.gov/pubmed/22336803
I have not bothered to investigate those papers further. Rather than cite review articles and meta-analyses, is there some specific dataset that people find convincing regarding the effectiveness of any measles vaccine?
As one whose sister gave him measles as a Christmas gift one year I can assure people that measles is ‘very’ easy to diagnose. And it is nasty even if you do not have permanent health imparement.
AFAIK I did not receive a vaccination, being just a little to old for the organized vaccination campaigns.
Replying to my own comment since nesting depth is limited …
The phrase “There was a lack of adequate description of exposure …” occurs in the section “Studies evaluating short-term responses”, so it isn’t particularly relevant.
However, there is some support for your point: “Out of the 12 cohorts and case-control studies assessing effectiveness of MMR vaccines in preventing measles or mumps, only three had a moderate bias risk. The remaining nine were characterised by poor methodological quality due to poor reporting or missing information about comparability between exposed or non-exposed groups.” — and none of the measles studies (looks like n=3 (papers) or 4 (separate studies), all cohort studies). (I don’t know which direction the bias would be expected to go.)
The authors comment on the bias risk problem … “As MMR vaccine is universally recommended, recent studies are constrained by the lack of a non-exposed control group. This is a methodologically [sic] difficulty which is likely to be encountered in all comparative studies of established childhood vaccines. We were unable to include a majority of the retrieved studies because a comparable, clearly-defined control group or risk period was not available. The exclusion may be a limitation of our review or may reflect a more fundamental methodological dilemma: how to carry out meaningful studies in the absence of a representative population not exposed to a vaccine that is universally used in public health programmes. Whichever view is chosen, we believe that meaningful inferences from individual studies lacking a non-exposed control group are difficult to make.”
I guess my question for you is as follows: given that (1) the epidemiological patterns are strongly consistent with the (admittedly patchy) evidence on the effectiveness of vaccine (in the past I have parameterized pre-vaccination models for measles in England and Wales, then used reported vaccine coverage rates (assumed 100% vaccine effectiveness) and matched the pattern of decline reasonably well); (2) there is still not a lot of (not any?) _watertight_ evidence for vaccine efficacy (I suspect that there are lot more studies that look at seroconversion, rather than measles infection, as an endpoint — it would be reasonable to take these into account as well, with some discounting for the unknown (but presumably high) protective effect of seroconversion); (3) I think you’d have to assume quite extreme changes in diagnosis and/or contact behaviour in order to get the same orders of magnitude of decline if you assumed the vaccine was not very effective; and (4) since the medical community is reasonably well convinced (for whatever reason) of the efficacy of vaccines, it would be hard both logistically and ethically to convince them to run a large trial with a placebo arm, what would you do next? Would you recommend that we stop vaccinating? Skepticism is fine, but I haven’t seen you present any data to support your alternative hypothesis (changes in diagnosis and contact patterns) …
Ben writes:
“…what would you do next? Would you recommend that we stop vaccinating? Skepticism is fine, but I haven’t seen you present any data to support your alternative hypothesis (changes in diagnosis and contact patterns)”
Ben I looked at two papers from the meta-analysis linked by jrc (Sudfeld 2010) and both included evidence for changes in diagnosis. See my post below. Also, in your link they reference Ong 2007 where it reads:
“A likely reason for this is that the case may have been misdiagnosed as a non-specific viral illness. Measles has become relatively uncommon in Singapore with two decades of widespread measles vaccination, and especially after the second dose policy was implemented in 1998. Many primary care doctors may not even see a single case of measles in a year. This makes diagnosis more difficult.”
http://www.ncbi.nlm.nih.gov/pubmed/17609829
So it looks like those problems are well known amongst people doing research on this, they just kind of “mention it” rather than attempt to account for it though.
As to what should be done? I think physicians should be encouraged to take another look at the literature with emphasis on the rarity/absence of blinding, etc. Some attempt should be made to determine what may be diagnosed instead of measles by doctors who have been biased by knowledge of vaccination state or rarity of the illness, and all future studies should also look at that.
Even in the 1960s when it would have been more ethical to perform blinded/placebo controlled studies they still failed to do so. I would also recommend some kind of report specifically on why past researchers seem to have had so much trouble performing methodologically sound research on this matter. If this is because it is unethical (or too difficult) to perform sound research, then this should be emphasized in the literature disseminated to the public.
Ruling out alternative explanations is not some after-thought for random internet people to be forced to speculate on, it is the duty of the people who promote the vaccines to address these issues.
jrc,
Did you look into the studies that meta-analysis (Sudfeld 2010) was based on? I looked at their results shown in figure 1. I checked the first paper and it says “not blinded” and they had difficulty in determining whether diagnosis was accurate.
Then I looked at the third reference (accounting for >90% of the “weight”) and found all sorts of oddities. They appear to not only be unblinded, but also have selectively excluded “likely to get sick” children from the vaccine-groups. Further, they report that doctors are much less likely to confirm a report of measles by parents if the child is vaccinated.
“This was not a blind study, since the investigators knew which children had received measles vaccine…
It seems probable that the occurrence of so much ‘measles-like’ illness in the vaccinated children was a reflexion of the difficulty in making a firm diagnosis of measles in the African child at one visit.”
Hartfield 1963
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2134550/
“Children were ineligible for participation in the trial if there was a history of fits, eczema, asthma, sensitivity to egg, treatment with chemotherapeutic agents or steroids, other current immunizations, or illness at the time they were called for vaccination…For the various reasons previously mentioned 14 % of the children allocated to receive killed and live vaccine and 11% of those allocated to receive live vaccine alone were considered ineligible. Fewer children in the control group were classified as ineligible, only 2 %, and this was because the decision on eligibility was usually taken when the child attended for vaccination…Of the total cases reported the doctor saw about 60%, and, of these, confirmed the parents’ diagnosis in 93 % in the control group, 64% in the killed/live-vaccine group, and 70% in the live vaccine group.”
Council by the Measles Vaccines Committee 1966
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1843609/
Also, vaccinated children were more likely to be ill:
“It is evident from Table IV that many children in all three groups were unwell and that the proportion was greatest in the live-vaccine group (61 %), less in the killed/live-vaccine group (54%), and least in the unvaccinated group (38%).”
Council by the Measles Vaccines Committee 1966
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1843609/
Two other things. First:
“Before measles vaccine was licensed in 1963, an average of 400,000 measles cases were reported each year in the United States (8). However, because virtually all children acquired measles, the number of cases probably approached 3.5 million per year (i.e., an entire birth cohort).
Since measles vaccine became available, professional and voluntary medical and public health organizations have collaborated in vaccination programs that have reduced the reported incidence of measles by greater than 99%.”
http://www.cdc.gov/mmwr/preview/mmwrhtml/00053391.htm
I thought I could perhaps follow up on the claim in the above quote (used by jrc), but the anonymous CDC author has failed to cite any dataset or paper. This is un-scholarly behaviour and reduces the credibility of that document in my mind.
Second, it is my understanding (“according to grandma”) that it used to be a somewhat common practice to purposefully spread measles amongst children, because it was thought the severity of symptoms often got worse with age. When I was young my parents did purposefully expose me to chicken pox so I imagine it was similar. If this is true, and the practice of purposefully spreading the virus is now less common, that would also confound any assessment of vaccine effectiveness.
Do you just want evidence for the decline in reported cases in the US? That’s not too hard to find, e.g.
* from Project Tycho: http://wiekvoet.blogspot.ca/2014/03/looking-at-measles-data-in-project-tycho.html (this appears to be global cases, but one could dig in and get the US totals)
* http://www.historyofvaccines.org/content/graph-us-measles-cases (refers to CDC summary, maybe you don’t like that?)
* http://www.iayork.com/MysteryRays/2009/09/02/measles-deaths-pre-vaccine/
etc etc etc
Ben,
As I see it there are two aspects to this research.
1) Can we reliably detect a phenomenon? Is it stable, ie similar effect size?
2) What is the most plausible explanation for this phenomenon?
From looking at these measles reports I do believe they are detecting a drastic decrease in rate of measles diagnosis amongst the vaccinated. The effect size also appears similar in each of these studies. So #1 seems to have been achieved.
However, I have not seen any evidence that this is due to the vaccine rather than some sort of behavioural change. Therefore #2 appears wide open. Looking further at that Ong 2007 paper (cited by your Cochrane review link), I found this claim:
“As only approximately 7% of the clinically-diagnosed cases of measles reported locally turned out to be measles by laboratory testing, there is a need for laboratory confirmation of measles to avoid misidentification of cases and improve disease surveillance.(2)”
http://www.ncbi.nlm.nih.gov/pubmed/17609829
I have no idea if this is unique to that area, but it certainly suggests that diagnoses can drop by 90% just by introducing lab tests. Obviously, if that is the case, comparing pre-lab test to post-lab test is not convincing evidence that the vaccine is effective. Therefore the info in those links should not be convincing to anyone.
I don’t know how easy it would be to study this since the lab tests were probably not deployed/utilized uniformly, rather it would be more gradual. Apparently they were around since at least 1966, but I have no idea how commonly used: http://www.ncbi.nlm.nih.gov/pubmed/?term=5331291
Anyway, it is quite amazing that my tiny sample of the primary literature (which was selected from reviews provided by others) has found support for my speculated confounds with plausible effect sizes as high as 90%. I really had never looked at the measles literature previously.
I see that sentence about only 7% of the cases clinically diagnosed locally were confirmed by the lab, but the context is really weird. I think what it refers to is this:
***
Active case detection was carried out throughout the school to identify recent absentees who had been clinically diagnosed for measles or were displaying symptoms and signs characterised by generalised maculopapular rash and fever, with or without cough, coryza or conjunctivitis. Cases were subsequently serologically tested to confirm that they were having an acute measles infection.
***
So I think they are saying that they grabbed anyone who had a rash and fever, and tested them, and only 7% of those kids with a rash actually had measles. I’d say that 100% of my rash/fever combinations were not measles, but, hey, maybe I’m wrong.
That said – the lab confirmation rate is a weird lesson to take from this paper. My takeaway would be this:
***
The attack rate was two of 171 (1.2%) students in the vaccinated group and 7 of 13 (53.8%) students in the unvaccinated group. The calculated VE for the MMR primary dose in the affected classes was 97.8%.
***
In the methods it explicitly differentiates between the two: “clinically diagnosed for measles or… [have rash/etc]”. I thought perhaps it was referring to data in their reference #2 but couldn’t find that one.
The problem I had with their VE estimate is they use the number of students in the classes as the denominator. It’s not clear this is a valid way to determine risk/exposure. They also say:
“The student from primary 6G and two of the students from primary 3D were siblings. The other two of their siblings who were studying in primary one and five in the same school were reported to be well. Another five of their siblings staying in the same residence were also well and did not show any symptoms of measles.”
They do not mention whether the siblings were vaccinated (I would guess not if it was already 0/3), but in that case we could add 7 to that denominator (why assume more exposure occurs at school rather than at home). It is possible to play endless games here, my point is just that the denominator is arbitrary so I wouldn’t put too much stock into those numbers.
Really, it sounds like they only checked absentees. In that case, perhaps the number of those should have been the denominator? If there is a high prevalence of vaccination in the general population you could keep raising the VE just by expanding your “population” even if the two are entirely unrelated.
This statement makes it sound like there were additional cases (not at the school) they did not include:
“The Ministry was first informed of four laboratory-confirmed cases of measles on May 4, 2004, from which two of the cases were primary six students studying in the same school. We informed the school administrator regarding the two confirmed cases…”
I would point out that Ebola has caused zero deaths of US citizens, but that hasn’t stopped a panic of sorts, although how much of that panic is news-cycle-driven and how much is “real” (extra credit: how do we tell the difference?).
The real problem with vaccination for almost anything is a combination of the availability heuristic and the free-rider problem: a substantial fraction vaccinated provides herd immunity to the rest. Individualized reluctance to vaccinate mostly needs to overcome not ignorance (as represented by the availability heuristic) but by tricking cunning rational calculators! The place the availability heuristic comes into play is not in disease prevalence, but in vaccine reactions, which are played up well out of proportion to their actual frequency.
“a substantial fraction vaccinated provides herd immunity to the rest”
Yes, but in the context of measles, that substantial fraction has to be extremely close to Measles is pretty much the most contagious disease of humans. It has a basic reproductive rate around 20. So if more than 5% of the population is unvaccinated, the effective reproductive rate will exceed 1 and epidemics will recur. Given the core of people who will refuse immunization for whatever reason and are not persuadable, maintaining herd immunity to measles require persuading nearly everybody who is on the fence to choose vaccination. A rational calculation accompanied by a sensitivity analysis will show that reliance on herd immunity to measles requires a very high tolerance for risk or a very sharp estimate of the rate of vaccine uptake.
The publicity that has attended adverse reactions to vaccines is certainly part of the problem. I think another issue is that for a long time, measles was just considered a garden variety childhood disease. The catastrophic neurologic complications and fatalities it can cause are relative rarities, and even back when measles epidemics recurred every few years, few people would have encountered them. I think many people perceive measles as a nuisance rather than a really serious illness.
Sorry. First sentence of second paragraph should read:
Yes, but in the context of measles, that substantial fraction has to be extremely close to 1.
“I think many people perceive measles as a nuisance rather than a really serious illness.”
I certainly did.
Yeah, I believe I had measles and mumps at the same time when I was four. I didn’t feel very sick and wouldn’t stay in bed. So I can see how it’s tough to convince people to vaccinate when a disease is often very mild, but in some cases can cause serious complications — either to the person with the disease, or to the people they come in contact with, if they are also not vaccinated.
Just be older and it is nasty. I had it when I was 18 and spent 4 totally miserable days in bed in a darkened room doing nothing but listening to the radio.
The lower incidence of vaccinations has cause small (N = 100’s) in Quebec and Wales in the last few years.
BTW I agree with the poster who suggest A. Wakefield should be in prison.
Jonathan:
> I would point out that Ebola has caused zero deaths of US citizens
http://en.wikipedia.org/wiki/Patrick_Sawyer
http://www.cnn.com/2014/07/29/health/ebola-outbreak-american-dies/
I didn’t know that, but my previous statement can be modified in two possible ways:
“Ebola has caused zero deaths of US natives.”
“US-contracted Ebola has caused zero deaths.”
Congrats! You have a theory of statistical inference named after you! (I prefered BDA3, but hey, I have a thing for acronyms).
i] This is true for likelihoodists as well. Since writing this 2 years ago, I’ve learned of Gelman Bayes which is not in terms of inductively obtaining posteriors and which uses a sampling distribution.
Re: P.S. this communication strategy is also known as fear mongering. The death of one person is tragedy, the death of a million is statistics. Which means in this context it is more scary if few people are dying in frequently rather than many rarely. It would be even scarier to say that somewhere in the world a person dies of measles every 4 seconds.
Sorry for numerous typos.
I have not watched the video but I suspect Brian is trying to make (poorly) is this: Global statistics show that measles are deadly. There are no ‘local’ statistics (or anecdotes) because measles is so rare in the US. The deadliness of measles is not available to parents locally as a result. The lack of availability makes it psychologically easier to adopt an anti-vaccination stance. From a psycho-logical perspective, that take home is pretty clear. From a hard Bayesian perspective, in which the proximal base rate is all that matter, it does not make sense.
You’re being quite subtle: he could have explicitly said “measles is potentially …”, rather than imply that, and the reference could have been then to show what the potential danger is, as demonstrated in the unvaccinated world. Your point is clearly correct and is a specific statement of the rule that we act rationally in one frame of reference when that is irrational in another frame.
I find that idea interesting in many applications. Consider the fear of clusters of leukemia: report a bunch of possible cases and everyone fears the worst form of contamination that they can’t control, that has already happened, that deprives them of choice and then put that in the frame that lack of clusters would be more suspicious. I like that one because it gets at the fears so clearly; as much as you understand clusters are normal (and that what looks to be a cluster isn’t on closer examination), you still worry “what if”.
I guess bringing up the availability bias would make more sense if people thought they were more likely to get ebola than measles in the US, because of the ebola reporting.
Some history. After the invention of the measles vaccine in the 1950’s, a conference was organized in 1961 to discuss mass vaccination against measles. Here are the published remarks of GW Wilson, a public health expert who participated:
>This conference is called to consider immunization against measles, and it is fitting therefore, that in introducing measles as a universal disease, I should refer to the subject of vaccination. There are certain general principles that should govern the policy of vaccination against any disease:
>(a) The vaccine should be harmless to the healthy child. In practice no vaccine has yet been devised that has not occasionally given rise to a severe and sometimes fatal reaction. The risk is much higher with some vaccines than with others. Unfortunately, for any given vaccine the risk can be assessed only by experience.
>(b) The disturbance caused by vaccination should not be greater than that caused by the disease itself. There is no doubt that in the prevention of smallpox the febrile eruption that follows primary vaccination is far less severe than that caused by the natural disease. In measles, however, this not so clear [sic]. Though at one time measles had a high case fatality rate resulting in a serious mortality, it has now in many parts of Europe and America become so mild that death is quite exceptional. In 1959, for example, an epidemic year in England and Wales, the total deaths from measles numbered only 98 in a total population of 45 million. Under these conditions, is the disease worth preventing, or should we concentrate on shielding infants and very young children from the risk of infection and protecting them with gamma-globulin when this is impossible? It is difficult to answer this question without knowing more exactly how much permanent damage measles does to the healthy child. In the tropics, of course, the position is different. There the case fatality rate for measles is high, and a much stronger case can be made out for vaccination. (Wilson 1962)
http://archpedi.jamanetwork.com/article.aspx?articleid=500099
As far as I know, it is still unclear why measles is a very mild disease in some populations, and quite deadly in others. When I dug into this over 10 years ago, there were a few theories, e.g., there has been some genetic selection in European populations to better tolerate the disease; or, in populations where measles is rare, adults do not gain immunity in childhood. If and when measles strikes, many adults become sick, and thus can’t care for the kids, or other adults, leading to high mortality rates. More reasons are summarized here (e.g., crowding, malnutrition, vitamin deficiencies, intensity of exposure):
http://jid.oxfordjournals.org/content/189/Supplement_1/S4.full
Here is someone claiming the other confound I mentioned earlier:
“Before the introduction of measles vaccines, measles virus infected 95%–98% of children by age 18 years [1–4], and measles was considered an inevitable rite of passage. Exposure was often actively sought for children in early school years because of the greater severity of measles in adults.”
Orenstein 2004
http://jid.oxfordjournals.org/content/189/Supplement_1/S4.full
Unfortunately no data or source. I’d like to know just how common that practice was and how this changed over time.
Pingback: A week of links | EVOLVING ECONOMICS