Saturday, January 17, 2009

Cookbook Redux

I titled my last post a couple days too soon. Check out this link from the WSJ health blog. Dave Snow, CEO of the pharmacy-benefits manager Medco (I have no idea what in the hell that means), has an interesting health care reform proposal. Here, I'll let him describe his little stroke of genius:
Snow said the time has come for doctors to follow set protocols on how to treat patients, and to be paid based on whether they do it. Basically, ‘If X, then do Y,’ and ‘If Y, then do Z,’ sort of stuff.

Yeah, that sort of stuff. Sounds like a surefire solution. It's like algebra. If X (say, congestive heart failure) then one must implement Y (chest xray, diuresis, rule out MI, a million other things, etc). It's simple. And if X is accompanied by Y, Z, W and maybe a little lower case r, then one merely has to integrate the equation with respect to z squared and the solution will present itself. Here's more from Dave Snow (and check out that picture in the link, CEO Snow in all his big white toothed, used car saleman suavity):
“I have no patience for a doctor who says, ‘I’m above it all, I don’t want to practice cookbook medicine,’” Snow says.

Bravo. Because that's exactly what's wrong with American health care. Not enough doctors mindlessly managing their patients according to rigid, universal protocols without regard to individual needs or the subtle variances of the human species. All we needed to do all along was designate a bunch of "smart folks" (like a national Federal Reserve of Health Care!) to get together and hammer out a national health care cookbook for dumb, lowly physicians to follow. My God. It was right there in front of us all along!

Obviously this Chairman Snow character is too contemptible to waste much time on. His idea is vapid and nonsensical and completely detached from any sort of reality that I am familiar with. Let's just hope he doesn't gain any more traction than he already has with Tom Daschle...


Joseph Sucher, MD FACS said...

I am sitting here trying to figure out how to agree with you and yet educate your blog readers on why "cookbook medicine" is already here and YOU practice it. That's right we all practice by a book. Some of us do it significantly better than others. So let me try to get though this thorny terrain with the least amount of damage.

Dave Snow, the CEO of Medco, is clearly suffering from a major case of needing to feel better about himself by acting so smugly superior to physicians. Unfortunately the fact is Pharmacy Benefit Managers (PBMs) are what controls medicine prescriptions throughout the U.S. Buckeye, every little Rx you've ever scratched is captured, logged, tracked and stored by this guy. PBMs are the clearing house for all that is prescribed by you and me. So he has huge clout in DC (translation.. sleezy money). Its scary to think about it. I am disgusted and angry by his putrid comments. He completely lacks the understanding of the complexity of medicical practice.

HOWEVER. We the medical community continue to dig our heels into the ground as we grasp for the last bastions of days gone. You and I simply cannot keep up with the constantly changing, ever increasing complexities of medical knowledge. It is impossible. You are not a computer. Oops. Did I say computer? Here it comes.

I am convinced that we absolutely need the help of computers and computerized clinical decision support systems (CCDS) to help us practice safe evidence based medicine. The problem is idiots like Dave Snow AND so many doctors that are trying to defend their intelligence don't understand what the heck they're talking about when it comes to what this actually means. It is NOT "Cookbook Medicine". Look back at my wording. It is "Decision Support." Support is the operative word.

Physicians absolutely need CCDS. But it isn't ready for the masses, and it isn't necessary across the entire spectrum of medicine. How many times to you need to see cholecystitis before you know what to do? Not rocket science. But it is necessary for helping with the complexities of our practice (Advanced cancer, Shock, Unusual disease processes). Additionally it will help with advancing and testing evidence based medicine. I have published on this subject in the Journal of Trauma Feb 2008 and spent many years helping to develop methods for treating hemorrhagic and septic shock.

Let's be honest. Why are some physicians considered better than others? Because quite frankly, there are a bunch of idiots out there practicing medicine. Period. You know it! When I first started seeing septic shock in surgical ICUs over 15 years ago, I saw patient in bed 1 being treated radically different than the patient in bed 10 which was being treated completely different than the patient in bed 4. Not because they were so unique. But because every doctor had their own "style" of treating septic shock. What crap. That just meant that nobody knew what was right and what was wrong. So how do we figure out what we need to do if everyone is out there practicing their own 'style' of medicine.

Oh... and I forgot. I did say that you practice cookbook medicine. Of course you do. You see a patient, and take an H&P, review their medical history, check the vitals, do an exam, check the labs, order and review diagnostic tests, then make a diagnosis. Cookbook. You then take your pancreatic cancer patient, and by decision based logic you decide what the optimal treatment pathway is. The medical oncologist then delivers chemotherapy by "cookbook". I bet you even have the same surgical text that most surgeons have "Surgical Decision Making". COOKBOOK!. Holy Cow! Now let me calm down and get rid of that dirty term 'Cookbook'. Bury it, cut it up, smash it. It only serves to mislead what we do need. We need help. We need better decision support. And if you read my article, you will note that decision support never should replace EXPERT OPINION. Ultimately the physician must remain responsible (For Christ's sake, someone must be left on this earth who is willing to take on professional responsibility). So Dave Snow can go pound sand with his fanciful idea that tort reform comes from removing responsibility from physicians. IDIOT!

This has gone on too long. Maybe I'll add more if you allow me. But for right now I need to go operate on the 90 year old with an emphysematous gallbladder and 50 years of diabetes and peripheral vascular disease. I wonder what Dave Snow is doing right now? Hmmm, its only 10PM.

OHN said...

From a patients perspective, I would much rather my physician follow his own protocol, than the protocol of an insurance executive with a business degree.

As patients we hope our respective specialists have kept up their CEU's and are willing to try the G if X and Y aren't working.His thinking is the cookie cutter approach and let me tell you Mr. Snow(job) we are not all cut out of the same. I would never wish ill on anyone but I wonder how he would feel if someone that was dear to him was treated with the cookbook approach and slipped through that diagnostic cracks...he would probably be the first to make demands. (And, from a woman's perspective, that photo/pose of his is just plain creepy).

Joseph Sucher, MD FACS said...


I completely agree with you. Insurance companies should not dictate the 'protocols'. I absolutely agree that patients are unique. So let me try to make a big leap across to what I am an advocate for. Its not cookbook. Its not protocols. These words are mis-used, dis-used and mal-used. They incite fear from physicians and patients. They are terms that the Dave Snows cannot possibly comprehend how it will impacts healthcare.

I do believe in Decision Support. Specifically Computerized Clinical Decision Support (CCDS). CCDS is driven from evidence based medicine, is applied within the uniqueness of the local environment and is patient specific. What I mean by patient specific, is the algorithms allow for the unique characteristics of particular patient conditions. That being said. CCDS is resource intensive and currently not practical for application in most environments because it requires a team of experts that currently can only be supported in the academic environment. So its not a panacea solution.

For practical purposes, there remains the use of 'guidelines'. Guidelines are publications from expert working groups that synthesize the best known evidence and provide a basis of practice for given problems. Guidelines such as the one published by the SCCM for septic shock do not give specific binary branching protocols, but rather give somewhat generalized defintitions of problems that give the provider a treatment decision which is supported by evidence. This decision, again, is meant to be applied within the patient's particular needs in that particular local environment. This is what Dave Snow does not understand. Dave Snow has the impression that guidelines are concrete. They are not. Guidelines are vague and wishy-washy (technical medical term). So that's the big problem. You can't possibly hold anyone to a set of guidelines, because guidelines do not carry specific definitions with binary branching logic.

Bottom line is that I agree with you and Buckeye. I just want people to be a little enlightened about the possibility that much of medicine is cookbook that can EVENTUALLY be constructed into viable Decision Support algorithms that are necessary to help augment physicians ability to provide the most up to date, optimal care for their specific patient needs. Without CCDS we will continue to be mired in less than optimal care systems. Did you know that there are over 17,000 biomedical journals? Over 5,000 are indexed on PubMed. PubMed has over 16 million references with over 623,000 reference entered in 2006 alone. How is anyone supposed to keep up to date with this volume of information? We can't.

rlbates said...

Good post, Buckeye, and nice comments. I like and appreciate guidelines based on good research. Guidelines are just that -- guides, but not dictations of absolutes.

Frank Drackman said...

Oh Yeah?!?!?!?

So tell me how you put someone with documented history of Malignant hyperthermia, anaphylaxis to local anesthetics, egg allergy, to sleep for a CABG/AVR...and don't tell me its not your Job, you Surgeon, you. Oh yeah, theyve got a Malampati 6 airway....


Joseph Sucher, MD FACS said...

Great example! Thank you for asking.

First, think about this. You’re expert training has developed a logical sequence for how to solve this problem. That sequence is a binary algorithm of branching logic based on Bayesian analysis at each node. It happens in you’re brain so readily because of your experience and education. But yet…. You also have a “style”. Why? Because you know how to do it better than Dr. Gaspasser, right? Maybe. Maybe Dr. Gaspasser didn’t get you’re education. Or maybe his way is really better than yours. How the heck do you know? We will only know if we agree on definitions and thought processes and then implement them in a open loop formalized construct which allows for the expert to veer from the formalized pathway based on conditions that cannot be anticipated. As you so eloquently formulized your question, I believe that your point is that you cannot anticipate every possible variable. But in you’re example, you can:

Do to space and time considerations I will only give you a general idea of how this logic would flow. Sorry for the formatting issue.. but there is no way for it to look good in a blog. Please don’t ding me on my mistakes in choosing appropriate anesthetics in this very specialized example. I haven’t scrubbed into a CABG in over 10 years.

High risk assessment and management pathway for CABG/AVR
1. Operative risk assessment
a. Unacceptable -> end pathway
b. Acceptable ->
i. Go through assessment pathway of PA Cath
ii. Goto #2

2. Assessment of anesthetic risk
a. History of MHS
i. Yes
1. Do not use volatile anesthetics or succ… etc.
a. Branch to 2b
ii. No ---
b. Egg allergy
i. Yes
1. Induction with Etomidate, Ketamine
2. Ongoing case anesthetic with N20, Pentathol
ii. No ---

3. Airway Assesment
a. Branching logic based on Malampati score
i. This is great because it is a defined score and already has branching logic developed for the management of high risk airway.
1. Score 6 -> awake intubation with fiber optic
a. Can’t use local anesthetic
i. Use ketamine … blah blah blah.

This is a great example of how CCDS can be used. Thanks.

Frank Drackman said...


You Cancel the Case.

Frank, M.D.

Joseph Sucher, MD FACS said...


Wipe away the anger in your eyes. The very first part of the algorithm is..

"High risk assessment and management pathway for CABG/AVR
1. Operative risk assessment
a. Unacceptable -> end pathway"


I then proceeded to give you a very basic outline of dealing with the complicated problem of anesthetic choice algorithm. This is a type of scenario where CCDS is best. The more complicated, the better.

These are systems based on medical logic developed by expert physician practitioners. It does not come from insurance companies or the government. It comes from you.

If you do not become engaged in the shaping of medical information technology to augment your ability to care for patients, then you fail.

JF Sucher, MD

tyro said...


Algorithms are only as good as the people that make them. Same with decision support. They're made by computer engineers, keyed on the EMR, which is only as good as the data it has. Helpful, but limited, and no way ready to 'take over' for a doc's judgement.

We do practice 'cookbook' medicine in the sense that EGDT, for example, should be applied to the septic patient. But recognizing the septic patient is not the job of a computer, not yet. And there is no pathway for pelvic pain, respiratory distress, abscess, or AMS. We're trying. But it's not here yet. Until then we'll have to keep (gasp) thinking, and use algorithms when we can.

BTW, do you really think that way?

For follow up, read Atul Gawande's chapter in 'Complications' on CCDS and how it would miss the case of nec fasciitis his senior caught.

Alice said...

As far as I can see, what Dr. Sucher is describing is a written-down or computerized version of the decision-making that occurs in every doctor's head with every patient. The problem of too much knowledge is the reason for specialization; which is why I wouldn't try to answer Frank's question on induction, but I would hope by the time I finish residency not to need to look up a protocol on a patient with free air. The branching points should be in my head.

What Dr. Sucher specifically complains about, similar patients being managed differently by different doctors, may not be a knowledge deficit. Even in this era of evidence-based medicine, there are still many areas on which we do not have a definitive evidence-based answer. Having a computer tell us that would not improve the practice of medicine.

For the rest, the computers I've seen in action in healthcare are too dumb to do any good. I'm thinking of the kind of pharmacy program that objects whenever I try to give benadryl or phenergan to a patient with potassium running in their iv fluids, or that warns against ordering two different narcotics (long and short acting) for post-op pain control.

Joseph Sucher, MD FACS said...


This couldn't get any better. You are prompting me to continue to provide more evidence promoting CCDS.

You're are completely correct; 'algorithms are only as good as the people who make them.' This is paramount in my argument. We physicians need to take the lead in CCDS development. This was made clear in my article last year in the Journal of Trauma. I have created a blog entry here that contains the abstract and two figures (hope I don't get in trouble for any copyright infringement).

As far as sepsis.
1. We have created a CCDS algorithm that, in fact, manages the resuscitation of a patient with sepsis, severe sepsis and septic shock based on the SCCM EBM guidelines. This algorithm is based on a model first created by the East and Morris in Utah and transplanted to Houston prompting an incredibly sophisticated tool to manage the resuscitation of patients with hemorrhagic shock.
2. My partner has presented data at national surgical conferences detailing a nurse driven SIRS screening tool for a surgical population that is greater than 95% sensitive and specific.

You say that "recognizing the septic patient is not the job of a computer". You are correct. It is our job. But we can use tools that will help us do our job better. Again... the operative words are 'decision support'. It is not 'computerized medicine'

I also agree that CCDS can miss identifying a problem. Why not? People miss diagnoses all the time. There is a significant challenge with teaching people to simply identify sepsis. I have found that 3rd year surgical residents can have difficulty doing this. So if the brightest residents in the hospital have trouble learning this, how do we expect our counterparts to do a good job identifying this deadly problem early? More importantly how do we have our front line people (nurses, NPs, PAs) identify that something is wrong before it gets out of control? It boils down to building better systems. But again, CCDS is only as good as those that develop it.

I saw a very funny cartoon where a patient describes severe epigastric pain with associated back pain, high heart rate and sweating. His physician, sitting at the computer provides a diagnosis of cholecystitis. The patient has an arrow in his back. Classic. Its a very good illustration of many points. History and physical exam are not replaced by using a computer. Also, the computer can only give you useful decision support based on what it knows. If you don't enter the full dataset, it can't give you reliable help. Finally, CCDS is not best used for simplistic problems. It excels best with augmenting provider knowledge and decision support with complex problems.

You ask "do you really think that way?" I am not sure exactly what you are referring to, but in general I think that what I have written is exactly how I think. I honestly believe that there are two crucial things that need to change in medicine. First and foremost, our education system must change. Second, we absolutely need to invest in information technologies that will provide us with the ability to optimize patient care.

Finally, your point started with claiming that computer engineers are creating EMRs (which I interpret to mean that is not good). I have espoused that this is a problem. The only way to automate processes and to provide decision support is by building systems using the people that actually know what the job is.

platensimycin said...

The algorithm is meant to standardize diagnosis and treatment. On top of it there are case-to-case variations to consider, of course.

We all seem to start up on the same page but somehow end up getting on each other's throat. It's no wonder we "providers" are exploitable by regulators and insurance co alike. Very smart.

Anonymous said...

Here is an article that hopefully won't make us feel like insurers and cooperations' lab rats.

Anonymous said...

People, health, doctor, money and industry; NOT industry, money, doctor, illness and patients. Luck in US healthcare system?