Sunday, November 18, 2007

Things I Learned in Yale

Okay, in Yale Alumni Magazine, that is. Glenn is an alumni of the law school and yesterday, while rather bored, I picked up the magazine and thumbed through it. Here is what I learned. First, most doctors can't do stats:

Almost every medical school student takes a course or two in biostatistics to learn how to understand research data. But Donna Windish, an assistant professor at the School of Medicine, has shown that the information often doesn't stick. "A significant percentage of physicians-in-training do not understand the statistics they encounter in the medical literature," she says.

In her own teaching, Windish had seen that trainees often read only the abstracts, or "ignored the statistics and skipped right to the results." This practice turns out to be common throughout the medical profession -- and potentially troubling. "An abstract usually says little about methods of design, conduct, and analysis," says Windish, citing an earlier study that showed frequent data mismatches between the abstract and the paper.

"Doctors don't necessarily need to know how to do the mathematical calculations," Windish says. "They need to understand the concepts and how to use them."

Funny, talk to most doctors and they will tell you that only MDs can prescribe because they "know all that calculus, stats and stuff." Really? I've never seen a doctor do any calculations to write a prescription. Now, I've learned that many of them them don't know how to interpret a piece of research thoroughly. That really breeds confidence.

Next, I found out in the magazine that:

Morning people are more likely to be emotionally stable than their "night owl" counterparts. Yale psychology postdoctoral researcher Colin DeYoung and his colleagues studied 279 students in an introductory psychology class at the University of Toronto and found a moderately strong correlation between "morningness" and character traits associated with stability.

Uhmm, okay, but I am not sure I buy this theory for the population at large. Undergrads are notorious for partying at night and sleeping during the day. When they have kids or get older and have to be at a job, I wonder if this still holds true? Can't researchers ever get away from studying undergraduates, who are such a peculiar type of cohort that findings may not carry over to other people at different points of life or in different environments?

Finally, I learned that working at Starbucks can save your life. There was an interesting book reviewed in the magazine with the intriguing title, How Starbucks Saved My Life: A Son of Privilege Learns to Live Like Everyone Else. Apparently, the author, who is a member of the Yale class of 63, got a plum job with J. Walter Thompson ad agency only to lose it at the age of 53. Then he goes on to have a number of misfortunes including impregnating a mistress, getting kicked out by his wife, a brain tumor etc. He loses his job and ends up working at a Starbucks in Manhattan and loves it. Naturally, the book review makes fun of this fact, describing the author as a "Starbucks sycophant" but whatever.

The book sounds fun and interesting enough to consider for my collection of reading material that I can never get to since I am too busy reading magazines such as the one described for no other reason than it was sitting beside me on the coffee table.


Blogger Parker Smith said...

Speaking of calculations and doctors - I've never gotten the impression that medications, either prescribed or over the counter, take things like body mass, age, and gender into account in any significant way.

Or does a given dose of ibuprofen really affect a 95 pound teenage girl the same way it does a 215 pound middle age male?

The directions on the bottle would make it seem so...

9:47 AM, November 18, 2007  
Blogger TJIC said...

From the few things I've read on the topic, expert systems do a much better job at diagnosing people and prescribing medicine than do doctors...but doctors are quite jealous of their professional privilege and the "unquantifiable" art to it, and so resist these tools.

10:31 AM, November 18, 2007  
Blogger Helen said...

Parker Smith,

Yes, I found that after I had my heart attack, docs gave me the same meds they were giving 200 pound men. Needless to say, I was sick as a dog from them and they had to be cut out or back. I do think size, gender and general body chemistry have a lot to do with how we process medication.

10:39 AM, November 18, 2007  
Blogger DADvocate said...

I once had a psychiatrist lecture me on how doctors understood statistics that laymen couldn't. I took a little more than the required statistics in college. Plus, working in marketing research, I've become more familar and have co-workers that can make your statistical head spin. None are MD's.

The statistic that the psychiatrist was lecturing about? That the FDA had not approved Zoloft for use with kids. He was asserting that he knew better than the FDA.

10:50 AM, November 18, 2007  
Blogger Cham said...

Yes, I have seen a doctor do calculations, once, only once. And that was at my last doctor visit at Target. The good doc got out her calculator and several reference books to make sure the prescription was correct for my body type. I was confused as to what she was doing so she showed me.

7:36 PM, November 18, 2007  
Blogger David Foster said...

The inability to do simple statistical reasoning is pretty common. One of the networks (CBS, I think) recently ran a series on the "high suicide rate" among Iraq veterans. Problem is, if you use comparable age and gender groups, the rate among these veterans isn't really atypical.

8:21 PM, November 18, 2007  
Blogger ck said...

Parker Smith,

Depends on the medication. Barring any kidney or liver problems, most OTC medications have a very large Therapeutic Window. This is the range where you are taking enough medication to get an effect, but not enough to cause toxicity (95% CI I believe).

Now with prescription drugs, few doctors actually look through all the data as Dr. Helen was pointing out. Your best bet is to be nice to the pharmacist, maybe even ask for a consult, to see if you have any reason to be concerned.

When it comes to drugs, the pharmacists know the ins and outs.

9:42 PM, November 18, 2007  
Blogger Willys said...

"Or does a given dose of ibuprofen really affect a 95 pound teenage girl the same way it does a 215 pound middle age male?"


On describing to my Doc a pain I no longer recall and that I take 'an' Ibuprofen for relief he asked 'Just one?' and his face screwed up the likes of which I wouldn't suspect a Doctor to show and said... 'that's enough for... a twelve year old.'

So now, when necessary, I just pop 'em by the handful.

'Life is good'

10:18 PM, November 18, 2007  
Blogger JorgXMcKie said...

After quite a few years of teaching statistics to polisci grads and undergrads, I have come to the conclusion that if I can get them to accept the statement that, "descriptive statistics should *NEVER* be used to drive policy" I have been moderately successful, since fewer than 10% will ever remember even the baby stat of crosstabs.

A few learn multiple regression and a bit more. They may well run things someday.

10:18 PM, November 18, 2007  
Blogger Chad said...

You know that's funny because in Hospital Corps school (10 Weeks long) in every Hospital Med orientation program, in Independent Duty Corpsman School and in various follow on training we had to learn how to calculate med dosages by weight and in some cases age. If Drs. cant do it we have a lot of problems.

10:40 PM, November 18, 2007  
Blogger Jim C. said...

Some abstracts are bad? Who writes them? Maybe better qualified people should.

11:30 PM, November 18, 2007  
Blogger Chrisnicel said...

Helen -
I am a biostatistician at a Medical School and I can personally attest to MDs being undereducated in the areas of study design and analysis of clinical research. In general, they are also unwilling to admit it.

The med students that I teach generally treat stats as a nuisance and roll their eyes at the nuances of study design. I haven't found a way to convince them that it is equal on par with the anatomy lessons or biochemistry lessons.

Terri in Texas

12:53 AM, November 19, 2007  
Blogger a psychiatrist who learned from veterans said...

"A significant percentage of physicians-in-training do not understand the statistics they encounter in the medical literature," she says.

As Bill Clinton said, it depends on what the meaning of 'understand' is. If it mens understanding that 'p < 0.05' means the result has a less than 1 in 20 chance of occuring by chance alone then they understand. If you mean understanding why the Archives of General Psychiatry went from accepting Analyis of Varaince to whatever it is they want now and how to perform the statistical calculation well no. It is more important to understand that when the FDA says a drug is indicated for a disorder that means it is statistically effective and the side effects are generally reasonable relative to the burden of the illness. If you understood that then you woudn't prescribe Wellbutrin, Buspar and Cymbalta successively for panic disorder, something I encountered in recent weeks. We should be so fortunate however that if they were going to prescibe something other than that having an indication, they 'would encounter the medical literature' through Pubmed to see if thre are any published articles on the subject. Happy everybody is into knowledge though. It's going to make my morning drive easier everybody learning how automatic transmission works before they hit the road tomorrow. I'll be driving 'standard' which is intellectually easier.

2:21 AM, November 19, 2007  
Blogger Unknown said...

Glenn may be an alumnus (male singular) of Yale Law, but he is not an alumni (plural).

2:34 AM, November 19, 2007  
Blogger Unknown said...

Dosage of drugs has NEVER made sense to me. How can my 4'8" 90 lbs wife and I (6'1" 205 lbs) get the same dosage? It is weird!

I'd say that very few people know enough statistics to use statistics. This appears to be worst of all in the media where it appears that they do not know anything about statistics.

4:13 AM, November 19, 2007  
Blogger David Foster said...

An example of poor statistical reasoning is the common argument that "your chances of being killed by a terrorist are much lower than your chances of being killed in a car wreck; therefore, we should spend less $ on counterterrorism and more $ on highway safety." One example of such an argument (by someone who should know better), and my response, here.

10:46 AM, November 19, 2007  
Blogger Earnest Iconoclast said...

I've always thought it amusing that pediatricians dose by the kg of weight for their patients (kids), but other doctors dose the same for all adults. I'm sure that some drugs are more weight dependent than others, but I've never had a doctor dose me by weight (and I'm quite a bit above average).

The biggest problem I see is the resistance to collecting data on performance by doctors for fear that it might be used against them in a lawsuit... while I understand their fear, it seems like a bit like covering your eyes to avoid seeing the scary monster that's about to eat you...


1:07 PM, November 19, 2007  
Blogger Serket said...

I agree that doctors should know if a study is legitimate before endorsing it. Somebody needs to be checking on these studies. I definitely prefer to sleep in, but I'm not sure how this translates to my emotional state. I do seem to feel physically better when I wake up earlier.

2:10 PM, November 19, 2007  
Blogger Carthoris said...

Well, Dr. H, you sort of have a point, in the same way that the hooligans over at /. have a point when they laugh at some ruling by a judge that was profoundly ignorant of the nature of Internet mail protocol. We could all do better jobs if we were better informed about each and every branch of human knowledge. Perhaps law schools should teach more about how computers and the Internet works, these days, and medical schools should teach more about epidemiology, or least statistics.

But...medical training is already brutally intense and enormously long and expensive. Just what would you have them drop, so they can spend much more time on the subtlety of statistics? Less time dissecting? Less time reading up on infectious diseases and how to control them? Less time in the clinic figuring out how to cope with real patients in the real world (with all its distractions and routine screw-ups)?

One of the problems with folks from an academic background is that they are a little fuzzy on real-world economics. There's a tendency to just prescribe the best conceivable solution to any problem -- regardless of whether it's at all practical or economical. Medical education is not designed to give students all possible useful knowledge: it's designed to select the maximally useful set given the constraints of time and human stamina. If you don't like the priorities, that's fine, but realistically you can't propose adding something without proposing subtracting something, because there's just no wasted time left to exploit.

Medical education has historically been very heavy on the investment in clinical training, because, let's face it, no one wants to be a new doctor's very first case of [insert your medical condition here]. Hence the emphasis on getting the protodoctor the maximum amount of clinical experience under supervision as fast as possible.

Recently we've added more and more 'theoretical' background knowledge, on the grounds that pure clinical experience and knowing the standard care in your medical textbooks isn't enough: with more rapidly changing medical tech, and more options in how you treat complex diseases, you need to know enough about the 'concepts' behind the tech to be able to make independent judgments. What you're saying is to add on to this burden, make doctors more independent judges of the results of clinical drug trials, etc., instead of relying on standards of care.

Well, fair enough, but realize something has to give, and it will probably be that clinical experience. You'll be having doctors who know more statistics concepts, but who have less 'real-world' experience with your disease. You sure that's a trade-off you like? Think long and carefully. I realize your personal situation is kind of a statistical outlier, but for most people the major determiner of outcome is the clinical experience of the initial diagnoser, the ability to match real-world symptoms with a relatively small range of typical medical conditions. That is, from a purely public-health perspective, it matters more that doctors have the clinical experience to recognize COPD in all its wide variety of clinical presentations than that they are first-rate critics of a Phase III trial reported in JAMA.

3:27 AM, November 20, 2007  
Blogger Serket said...

Yesterday I went to a clinic and was given a presciption of a protopic ointment for eczema. I picked it up from the pharmacy and in the instructions it also included an a statistical analysis of the study. Perhaps the pharmacists know this stuff better. I know a little about statistics since I took a class in highschool and college.

1:07 PM, November 20, 2007  
Anonymous Anonymous said...

Birth. Education enough to contribute to the tax base after school - paid for by others' taxes. A life of paying taxes. Death. More taxes after death.

Who needs any more stats than that?

7:18 PM, November 20, 2007  
Anonymous Anonymous said...


3:32 AM, June 08, 2009  

Post a Comment

<< Home