Return to Website

Number Watch Web Forum

This forum is about wrong numbers in science, politics and the media. It respects good science and good English.

Number Watch Web Forum
Start a New Topic 
Question-- Studying the dead vs Studying the Living

To our bending author and others --

After reading all of the materials from Dr. Brignell, I want to check my results. My view of all epidemiology is that it is flawed from the start because it studies the dead instead of studying the living.

We gravitate to those who suffer / die and evaluate their characteristics and compile vast databases. We then create comparisons of mortality to evaluate the relative risks associated with those characteristics. We (the epidemiologists anyway) seem to ignore the survivability exhibited by the characteristics.

I guess this has been discussed before. Someone once asked "Why don't they record the mileage traveled successfully by motorcycles?"

The answer then was apparently root of the problem. Most motorcycles get to their destinations without any problems. It is only the last trip they shows the mistakes. Even the dead motorcyclist will have traveled many hundreds or thousands of miles successfully. It is always the last mile that kills them.

We focus on the last mile because we can. There is a clear indicator. 1 death. 1 motorcycle. In the case of Pfizer with its new drug, this completely backfired because someone forgot that Cholesterol is in our systems for a reason. People surviving high cholesterol well are way too numerous. People not surviving gives us data points... When you start tweaking a variable sometimes you don't see the other end because you missed the living focussed solely on the dying.

Re: Question-- Studying the dead vs Studying the Living

It's all about money.
To get the grant money, you must present the biggest,scariest numbers that you can find.

Data from the American CDC shows that average current smokers have a yearly lung cancer death rate of 7/10,000 and the never-smokers a rate of 2/10,000.

There is no money to be gotten from showing that current smokers have 99.95% of the never-smokers chances of not dying from lung cancer in any given year.
(9,993/9,998 = 99.95%)

So, it is best to find a subset of smokers that has a higher rate.

About 1/500 current smokers smoke 3-4 packs a day; but,their lung cancer death rate is about 25 times that of the never-smokers.

Sooo, omitting details,you claim that smokers have a lung cancer death rate that is 25 times higher than never-smokers.

Now that will get you grant money for more studies and programs to prevent the children from smoking!!!!

Re: Question-- Studying the dead vs Studying the Living

The approach isn't actually that daft in principle. Taking your motorcycle example, it is relevant because you can't predict in advance which mile will be your last. If you could, you wouldn't drive it.

Taking the smoking and lung cancer example, at the risk of coming out with a Rumsfeldism, cancer is a binary thing but it's more complicated than that. The way tumors develop, and the amount of genetic damage you accumulate on the way there means that you aren't running a discrete risk - as with riding your motorcycle of coming home alive (most times) or dead (maximum once per lifetime, probability greater than zero each time you go out on your bike). There's certainly no prospect of determining which cigarette tipped your growing neoplasm over the edge into a malignant tumor and avoiding that one cigarette (unless the previous one was your last) isn't likely to help.

Comparing the heaviest smokers with the never smokers gives you the biggest chance of finding a useful result, and you have to be wilfully blind not to understand that the evidence is such that a link between heavy cigarette smoking and lung cancer is a fact, for everyday values of fact (I sit here puffing on my pipe as I write this). As to what the risk distribution in between is, well, chances are it does not at any point on the curve reduce with greater exposure, but whether it is linear, exponential, square-root type, even hormetic at low doses, or a combination of all three at different levels of exposure, is not known.

GaryK's point about how statistics are displayed is valid, and there are always ways of showing numbers that make a point in a particular direction. Plus the oft-forgotten maxim that n times a very small risk is still a very small risk, and in some cases trumpeted despite being beyond the power of the study concerned to detect with any reliability.

Re: Question-- Studying the dead vs Studying the Living

JamesV -- I don't disagree with anything you say. I don't smoke and avoid smoke as much as possible. My primary reasons are: A. because I have a history of asthma and have had difficulties with smoke and B. because smoke stinks and makes everything around it stink.

The linkage between smoking and lung cancer is serious enough that I am happy to avoid it. I can't get myself to chastise others for smoking though. I just ask them to take it somewhere that it doesn't force me smell it.

The biggest factor to surviving on a motorcycle is attention. It is not the motorcycle that kills, it is failure to avoid bad situations and this is most often caused by the rider failing to be attentive. Motorcycle riders who keep living assume that every car they see is going to do something stupid (perhaps I exaggerate a little, but it is mostly true).

In all the cases though, there is much more to be learned from the living than from the dead. The dead just provide glowing markers for us to find. The living, the people who don't die from their behaviors, have to be sought out and are likely to provide little in concrete results. The motorcycle riders that are still alive may just not have made their last ride. The smokers may not have smoked their 1 too many.

If we could isolate the smokers who will never have 1 too many and determine the difference between them and the ones who do, we have a datum that helps us much more than the fact that some smokers end up dying a little earlier than non-smokers.

Find out why the motorcyclist who is always safe, manages to always stay safe and you have the potential training path to helping others always be safe (1st step, Motorcycle Training School!!!!).

Re: Question-- Studying the dead vs Studying the Living

The risk of death on a motorcycle is obviously going to be influenced considerably by the skill and experience of the rider. Treating motorcycling deaths in purely statistical terms isn't going to tease out that fact. The motorcycling death statistics, when we compare them to other methods of transport are also likely to overlook the fact that a lot of motorycle use is recreational in nature, and pursued by people with a high appetite for risky thrills. If you did an experiment and put average middle-aged Ford drivers on motorcycles, and put average motorcyclists in underpowered Fords, you would probably find the gap in the rate of death per distance travelled narrowing significantly.

So there is a certain amount of motorcycling death that would not be mitigated merely by banning motorcylces, for the motorcyclists would take up base jumping or something of the sort.

I think where your argument falls down is that everyone dies of something, and nearly everyone dies of something that could have been avoided under different circumstances - even if determining what those circumstances were is often complicated. It's also very easy to study the dead because they aren't going anywhere and can't de-volunteer themselves from your study. Even a well-designed and properly-managed interventional clinical trial will never recruit a representative cross-section of the population you intend to treat, for the highly risk-averse will never volunteer for your trial. For the same reason, your average high-street poll will never give a true picture of how many people would vote tory, labour, or liberal, not only are you selecting for people with either time or money (or more likely both) to be on the high street, but a lot of people simply won't answer your questions. Any systematic bias in that with respect to whatever you are polling will create a rift between the poll results and reality.

Two further things come to mind. Firstly the dead are, logically speaking, NOT living, so we can make inferences about the living from studying the dead - and take that further into conclusions about what causes people to die if, for example, we know that the proportion of motorcyclists and smokers among group DEAD=1 is higher than among DEAD=0. Secondly, the advantage of the dead is that they are no longer doing stuff that will influence their status of death. People start and stop smoking, buy and sell motorcycles (or gain and lose driving licenses). While these things will change in an individually unpredictable way among a sample of the living, they cannot change among a sample of the dead.

Re: Question-- Studying the dead vs Studying the Living

Seems to be that studying the living is only good for advertising policy.

To determine the effectiveness of prevention, you have to study the dead.

Preferably the deaths or not deaths should be limited to the behavior of one group of people.

Except for the odd pedestrian, motorcycle deaths only happen to motorcycle riders.

Never-riders and ex-riders will not die from cycle accidents.

This is unlike smoking; where, 70% of the deaths from diseases 'caused' by smoking are not 'caused' by smoking.

That sounds stupid; but,that is what passes for science in the anti-smoker world.

The American CDC says there are 1.3 million deaths from the 19 diseases that are 'caused' by smoking.

The CDC also says that there are 393,000 deaths from those 19 diseases that are 'caused' by smoking and that is only 30%.

Cycle mortality is very straight foreward.

Re: Question-- Studying the dead vs Studying the Living

I agree with everything you say here.

One clarification though. The dead include the people who are just ill. The people who are sick see doctors. People who aren't don't.

Re: Question-- Studying the dead vs Studying the Living

There indeed is the rub. I have smoked a pack per day of non-filtered cigarettes since I was 14 years old. X-rays and my doctor's stethoscope say that my lungs are exceptionally clear (I do a deep lung cleansing exercise several times a day) I am 76 years of age.

In addition, I also drank a lot of alcohol for a long time, and still drink a "wee drop" almost every evening.

I have found that by drinking a mix of one 11 1/2 oz can of V8 juice mixed with a 12 oz bottle of Negro Modelo beer every day, that this is a veritable potion of youth. I am 76 years old with as of now, a 55 year old body. I can even do deep knee bend exercises. Everyone who has tried it says the same thing. It makes their bodies physically younnger, and heals existing physical painful problems (and I don't get a single penny out of this).

Individuals are all a bit different from a phyysical standpoint. And who could know what other bad things those who contract cancer of the lung.

Another sigificant point: during the time the studies of cigarette smoking were done, the Kent cigarette company was making a filter out of asbestos fibers . . . for at least 12 years. Moreover, smokers like myself smoked one pack per day, while many were smoking 3 or 4 packs per day.

Re: Question-- Studying the dead vs Studying the Living

The plural of "anecdote" is not "data".

Re: Question-- Studying the dead vs Studying the Living

@Gary K, ex-cyclists will not die in motorcyling accidents because they are no longer exposed to the things that cause the death (like hitting the pavement at high speed). Ex-smokers however can still die of having smoked as the damage lingers on. The exposure and the effect can be separated by quite some time, which is not the case with motorcycling.

Re: Question-- Studying the dead vs Studying the Living

"The exposure and the effect can be separated by quite some time"

No more than 10-15 years at the most.

Health Care professionals tell us that 10-15 years after quitting a smoker's risk of cancer and heart disease is about the same as a never-smoker's.

Re: Question-- Studying the dead vs Studying the Living

One thing I don't buy on the smoking issue is that smoking causes cancer.

Don't get me wrong, I'm not going to pretend that smoking isn't harmful, I just don't buy the causes bit.

I suspect that there is something in each of us that makes some more susceptible to cancer than others. From there you get the increase in cancers which I would say are more correctly described as aggravated by smoking.

Don't all pounce at once..


Re: Question-- Studying the dead vs Studying the Living

But you are just paraphrasing our host...

You can only rarely get in trouble for paraphrasing. the host of a party. As long as the paraphrase is used positively...

Re: Question-- Studying the dead vs Studying the Living

While the plural of anecdotes is most definitely NOT data. Ignoring the implications of the anecdotes is also not science. The outliers do matter. The Japanese not dying from heart issues is worth researching (which IIRC turns out to have something to do with a bias against reporting deaths as associated with hearts). The people who thwart the "rules of healthy living" and thrive should make you ask why...

I believe our host has pointed to the most probable source of why -- stress --. Most specifically how each of us deals with it. Personally, I look at the trees and realize that they don't give a flying leap about my worries and realize that if my worries aren't important enough for the trees to worry about, they are probably aren't going to kill me.

A long time ago in an off beat classroom far far away, a teacher asked "How are a pineapple and a electrician the same?" A flippant student (not me, I am not that creative) responded, "They are both affected by thermonuclear detonations!" The truth of the answer was that he was trying to "satisfy" the requirements of participation. It has stayed with me all this time though. How do you measure the stuff that doesn't kill you other than reiterating pithy sayings like "If it doesn't kill you it just makes you stronger!"

Linear No Threshhold was beaten into me in Nuclear Power School. Combined with Time, Distance, Shielding, it was a very useful tool for planning. Always treat radiation carefully and it won't kill you. The application of this to operations got wacko. If there was some way that a person could get irradiated during an evolution, we prevented that path from happening. The problem with this was that we would spend an entire shift locking out everything to make an evolution possible because there was a chance that "a person might climb the ladder on the other side of the high bay, possibly in line with the portal that will be open for the removal of the remote arm. Any device that might put radioactive material in the position to stream through that portal must be secured. There is a chance that an earthquake might hit of such magnitude that the radioactive material will bounce from the transfer train up 6 feet to hit the transmission hole, therefore tag it out." We operated on the precautionary principle.. I can't say it was completely wrong for us to think that way. Tactically, it is actually sort of sensible. When moving radioactive material around, keep people away from it and you limit the problems that can happen. The cost of not thinking that way could be someone's life. When you are moving 40 ton casks around filled with highly radioactive material, move slowly and carefully but with efficiency.

All this is a careful means of avoiding the work I should be doing. I swear it is related to the discussion. Happily the people in this group tend not to jump on me as a denier of science, which happens on so many other sites.