All 4 books by Edward Tufte now in
paperback editions, $100 for all 4
Visual Display of Quantitative Information
Envisioning Information
Visual Explanations
Beautiful Evidence
Paper/printing = original clothbound books.
Only available through ET's Graphics Press:
catalog + shopping cart
Edward Tufte e-books
Immediate download to any computer
connected to the internet:
La représentation de l'information
quantitative 200 pages $12
La Representación Visual de Información
Cuantitativa 200 páginas $12
Visual and Statistical Thinking $2
The Cognitive Style of Powerpoint $2
Seeing Around + Feynman Diagrams $2
Data Analysis for Politics and Policy $2
catalog + shopping cart
Edward Tufte one-day course,
Presenting Data and Information
Denver, July 31
Seattle, August 3, 4
Portland, August 6
Indianapolis, November 9
Columbus, November 10
Cleveland, November 12
San Francisco, December 14, 15, 16
Medical information exchange: The patient, doctor, computer triangle

The usual concern is that patients feel left out with the physician is "looking" at the computer. How is this different than the physician looking at the paper chart? While in medical school many years ago I was taught not to make notes while seeing the patient and to do the recording afterward. In today's environment there is far more data to contend with than there used to be. Achieving ever more stringent theraputic targets are demanded of the physician.

The suprise has been the level of interest that patients show in seeing what is on the computer screen. Patients are participating in the data entry and editing of the data. I have heard that physicians are not that interested in graphs. They want to see the data. I think that this is a mistaken response to not haveing seen good graphs. The even bigger suprise is how interested patients are in seeing graphs of their lab results. This is not reported yet but in conversations with physicians from various countries this observation seems to be consistent.

-- Ray Simkus (email)

Response to Patient - Physician-Computer Triangle

Depending on when you obtained your initial medical training, you'll find that the patient-physician and patient-bioinformatic dynamics have evolved.

First, the prevailing model is (or should be!) a collaborative relationship. Medicine is increasingly interactive, with patients demanding more feedback and decisions at each stage of diagnosis and care.

Next, bioinformatics has aggressively metastized into medicine in the past two decades. Blame this in part on the increased power and decreased size of computing devices, as well as a generalization of the scope of computing. Just as medicine was first infiltrated by the natural sciences in the 19th and 20th centuries, it is now being infiltrated by bioinformatics.

I think that the patient is no longer overly offended by a provider taking notes during a visit. Like the attentive waitron taking order-related notes, the patient will no doubt now interpret note-taking as a positive sign of the provider's attention to detail and care. I for one wouldn't want the provider keeping it all in his/her head for later compilation. That being said, it is important to interact with the patient, and to deal directly with note-taking anxiety. In short, the note-taking should be in harmony with a parallel verbal dialog with the patient.

Now: graphics versus data, and the patient versus data. Don't drown the patient in the data. A good graphic tells the data's story without distortion. More important in the patient's mind is the meaning of the data. If you can relate to the patient what the data means vis-vis their status, then you shouldn't have any problems. The key challenge in the patient-bioinformatic relationship is to ensure that the data presented adds to the patient's knowledge. The challenge is in bridging the gap between the isolated clinical findings and the larger medical picture.

-- CJ Alverson (email)

Response to Patient - Physician-Computer Triangle

What are the low-tech solutions?

Lets assume for a moment that most of your patients can comprehend any graphic you can. I'm trying to get into medical school so I've observed a number of clinical settings and I think the best conversations about data occur when people are shoulder to shoulder looking at the data. It's a matter of reading: sitting 90 degrees or 180 degrees around from text or graphics inhibits communication.

In an orthopedic clinic every resident went in to the exam room, took the history, went back to the office and briefed the attending while they both looked at the x-ray or MRI on the wall-mounted lightbox. The resident would take down the film, they'd both walk to the exam room, and the attending would put the film up on the lightbox so the patient, resident, and attending could all see and discuss it. As radiology has gone digital, the computer screens are now on desks but usually positioned to facilitate multiple viewers. Rarely is a desk now in the middle of the office, it's usually up against a wall so the screen faces into the entire office. One challenge with this is that people often have to reach across the computer screen to manipulate the image via the mouse. A good presentation mouse reduces this problem; I use this one from Fellowes. Despite plummetting cost I haven't seen doctors putting flat screen monitors up on the walls to replace light boxes, and I think this is will soon become a net loss in terms of time, money, and reputation.

When radiologic studies aren't involved, the exam room layout can present a challenge. Patients assume they're supposed to sit on the exam table and stay there after the exam is over. An alternative is to invite the patient to stand next to you (when possible) and use the exam table as a table to lay down the record so both of you can review the numbers. If the patient is bed-ridden, holding the chart up as though you're reading a child a bedtime story is a logical choice.

That's a good analogy. Do you feel closer to your children when they're sitting across from you at the dinner table or when you're reading them bedtime stories? If you don't have children, how did you relate to your parents in this situations when you were a child? Being shoulder-to-shoulder does reduce eye contact. That doesn't mean eye contact goes away. Glances are exchanges when appropriate as a natural part of the conversation. There's also a sense of collegiality and inclusion that comes from the shoulder-to-shoulder position.

-- Niels Olson (email)

Response to Patient - Physician-Computer Triangle

There are, I believe, some empirical studies of doctor-patient interactions. Supposedly, an early study reported that doctors on average made their first interruption after the patient had talked for 18 seconds. The story has it that this 18-second finding produced appropriate hand-wringing, and medical schools began to teach doctors to listen, listen, listen. Years later, the study was redone; the first interruption now came after 24 seconds. My knowledge is at the urban legend level; perhaps a Kindly Contributor knows the literature.

Here is a straightforward way to improve doctor-patient communication. In advance of the meeting, the patient prepares a typed-out list of all the issues to be covered at appointment with the doctor. This list should include causal speculations by the patient: "This pain on my right side might be a gall bladder issue. A grandparent and my father had gall bladder problems around my age." The patient should make several copies of this list-agenda paper and bring them to the appointment.

At the beginning of the meeting, the patient hands a doctor a copy of the list. The doctor, who did not get to be a doctor by being a slow reader, can read about 3 times faster than the patient can talk. After handing the agenda-list over, the patient should look down at her/his own copy, hinting that it is time to start reading. Or perhaps saying, "Here it all is, read this." This may finesse a little joke by the doctor, "So, what's with the paper dress?" (as the recent New Yorker cartoon had it).

The list gets everything the patient initially has to say out on the table, without interruption. As the appointment continues, the list helps set an agenda and a schedule for the allocation of time during the meeting. It also helps to make sure that the patient does not abandon lower-level issues that should be discussed--because there the issues are, already written out. Each item on the list is, in effect, checked off as the appointment moves along. Perhaps the patient should even ostentatiously check off the first point on the list after it is discussed to indicate that this list is what we're going to march through. The idea is that the doctor is not only going to be looking at the computer and at the patient, but also at the list. Because the patient keeps looking at the list.

The patient should bring several copies to the list to the appointment, since the patient may see several medical staff members during the appointment. Each medical person gets the agenda-list. For example, my doctor often has a medical student in training who handles the initial discussion and who then goes off to describe the situation to the doctor, who shows up later. Both the student and the doctor get copies of the agenda list. Multiple copies are particularly effective in the emergency room (if it is possible to prepare a list before going the ER!), since the patient will surely see a good many medical staff members at the emergency room.

The list-agenda enhances the efficiency, accuracy, and resolution of the information presentation made by the patient. It also helps reduce socially or situationally-determined answers to the doctor's questions; instead some of patient information has already been prepared in advance, free of social pressure.

This list goes into the patient file and also assists the doctor in preparing notes for the patient record. Maybe someday the patient provides the list via email in advance of the appointment; even so, the patient should still bring paper copies to the appointment itself.

Perhaps now and then such an agenda-list will even improve the effectiveness of medical care.

-- Edward Tufte

Response to Patient - Physician - Computer Triangle

A good bibliography

Searches for "patient doctor interruption 18 seconds" in Google and Google Scholar.

The Pubmed abstract for the original 18 second article: Beckman HB, Frankel RM. The Effect of Physician Behavior on the Collection of Data. Ann Intern Med. 1984 Nov;101(5):692-6.

The Pubmed abstract for the original 23 second article: Marvel MK, Epstein RM, Flowers K, Beckman HB. Soliciting the patient's agenda: have we improved? JAMA. 1999 Jan 20;281(3):283-7.

Recent work by the same authors: Schirmer JM, Mauksch L, Lang F, Marvel MK, Zoppi K, Epstein RM, Brock D, Pryzbylski M. Assessing communication competence: a review of current tools. Fam Med. 2005 Mar;37(3):184-92

Pubmed searches with intra-team combinations of these author names yield the most targetted additional results.

-- Niels Olson (email)


The paper agenda-list goes after the key problem mentioned in research on doctor-patient interactions:

". . .physicians frequently choose a patient problem to explore before determining the patient's full spectrum of concerns."

"Physicians often redirect patients' initial descriptions of their concerns. Once redirected, the descriptions are rarely completed. Consequences of incomplete initial descriptions include late-arising concerns and missed opportunities to gather potentially important patient data. Soliciting the patient's agenda takes little time and can improve interview efficiency and yield increased data."

Another virtue of the paper list-agenda is an improvement, a large improvement I believe, in interview efficiency.

-- Edward Tufte

The blogger of Over My Med Body! introduces an interesting idea: pocket-sized pictures to illustrate to patients the anatomy of their particular disease. In specialty clinics the drug and device sales reps solve this problem by providing models and wall charts, but on the wards where patients have multi-system problems, that approach doesn't work.

-- Niels Olson (email)

PING is an open-source project to create a patient-controlled medical information record which might be of interest to readers of this thread. Perhaps some of you Kindly Contributers will choose to help steer the development of this project even closer to its noble goal.

-- Matt Huyck (email)

The sample screens shown are extremely thin on content, with a relentless sequences of low-resolution screens. The metaphor seem to be computer programming, not medical information.

-- Edward Tufte

On radiology, which I'll assume is going under clinical notes (would that also include surgical reports?) You'll want to include a visually interactive java applet, like Google Maps, totally intuitive, that supports viewing of dicom images (standard format for MRI, CT, radiographs, PET, etc). The radiology reports should include a links to these dicom images.

Hospitals have doctor-oriented databases with visual displays (many of which are quite horrid). Presumably a lot of what you're doing is expanding the doctor-oriented displays full of acronyms into patient-oriented displays using full words. I would recommend looking long and hard at the more successful layouts of doctor-oriented material and seeing what design elements can or should be replicated in this patient version. I would recommend you have a link in the top to a medical dictionary, there are many available free online. Alternatively, use the rollover to embed definitions of all words on each page that are found in some adopted medical dictionary (maybe get a site license to Dark's Medical Dictionary, which is free). Perhaps next to the left of Help should be an Education button, with a link to a page of links: a couple dictionaries, PubMed, etc -- resources for the intelligent patient.

-- Niels Olson (email)

A "master problem list" is a recommended feature of medical records, whether paper or electronic. (See a low-tech example for people or for animals.) Sending this list to a patient prior to an appointment would jump start the formation of an agenda-list for that particular appointment. Rather than creating an agenda-list from scratch, the patient would be reminded of past issues and could reassess their status and importance, and would also know what compaints his/her physician is or isn't aware of.

-- Alison Anderson (email)

The master problem list exists for every patient going to a routine physical. Interestingly, the notions of five-year and pre-employment physicals were brought about by a campaign at the beginning of the 20th century, first by the public health movement then happily adopted by the American Medical Association. However, the young and healthy don't really need the physicals, so the master problem lists were developed to tease something billable out of this rather large population group. The military's master problem list is DD2807, Report of Medical History. Service members fill out a 2807 before every physical (All DD forms). I don't know if the two methods, form and blank page, have been compared head to head, but the 2807 starts with the list and leaves about two thirds of its second page blank for comments.

Paul Starr provides an excellent social history of the routine physical, along with every other aspect of American medicine, in his Pulitzer Prize-winning, very anti-AMA, The Social Transfomation of American Medicine, published by Basic Books in 1984.

-- Niels Olson (email)

I've also seen master problem lists in some ERs, where the physicians actually have bins of forms they use to guide their interview and diagnosis based on the complaint (car wreck, gun shot, psych, head & neck, thorax, upper limb, abdomen, pregnant, gastrointestinal, urinary, genital, etc. The bins are actually stacked in anatomical order head to toe, with the high priority ones, like car wreck, on top). Neither of these, routine physical or ER, are the same as a typical clinic appointment where the typical middle-aged adult goes to the typical internist with the typical complaint of persistent heartburn and the typically present but sytemically separate history of chest pain on exertion three times in the last six months.

-- Niels Olson (email)

An example of one of the great axioms of human behavior: "It's more complicated than that.": http://nytimes.com/2005/08/14/health/14patient.html

-- Edward Tufte

This thread has been quite interesting, and I can offer my own take. I practice in a primarily pediatric subspecialty practice, and as such, have the luxury of being able to impart information to the parents and other caregivers, in addition to the patient (if he or she is not an infant). One of the most useful things I have found if actually going over my notes with the parent.

First, I show them the growth chart, which shows in a very clear, graphical manner, how the child has been gaining weight and growing taller. This is especially useful in cases of patients I have been following serially.

If appropriate, I will go over the "family tree", as often the conditions I deal with have a familial basis. This also often helps clarify why I ask detailed questions about seemingly distant relatives. Seeing the tree graphically seems much easier than resolving a mass of text about "the mother's sister's second son...".

Additionally, our clinic visit charts are formatted to be a "checklist", such that normal findings can be checked off, and only abnormals need to be expanded upon. Parents, seeing the array of checks, instantly understand the depth of the evaluation, as well as the degree to which things are normal or not -- even without being able to clearly read or understand the details. They see that they are there, and that is often assuring.

Finally, our exam rooms have large posters with the body parts of interest (in our case, hearts) that we can point to. "Your child has a problem with this valve...", or "There is a hole here...". Seeing this up on the wall is often helpful for the understanding of teh parents, and of older children.

Of course, many of these strategies are specific, but I think may be applied in other clinical contexts.

-- Fraz Ahmed Ismat, MD (email)

Backward stammer softly the words, 'til reels the mind.

Robert E Hirschtick. Cut-n-Paste. JAMA. 2006;295:2327.

-- Niels Olson (email)

Stepping back from the serious world of medicine to the more trivial one of the sales counter; I'm hard of hearing, so a display of how much the sales clerk is asking for would be much better than a mumble. But the tendency of small shops to take the price displays of their tills and hide them behind piles of small goods or little advertising displays is extremely irritating. Then they look at me strangely when I crane my neck to see what the charge is!

Large stores are better, presumably because of some top-down diktat based on human factors researchers. "The customer would really like to see what the description of their goods is, and how much the price is tallying up to." Well yes, that's basic; you're a sales clerk, not a high priest conducting a special ceremony behind a screen.

Taking this rant back to medical displays, if you are going to have a face-to-face doctor-patient arrangement, for goodness sake provide a duplicate display device that is visible to the patient and shows everything that the doctor is seeing. Nothing more annoying and scary than watching your doctor stroke his beard as he scans your information (*your* information) on a display you aren't allowed to see.

-- Derek Cotter (email)

Assessment of relative risk in medical interventions is often difficult for patients and professionals. In this article in PloS Medicine, a new format is offered to make the decisions clearer.

The Roulette Wheel: An Aid to Informed Decision Making

I am not sure, however, that this makes things more clear.

-- Fraz Ahmed Ismat, MD (email)

My husband recently tried the "paper list of medical issues" idea at an appointment with his primary care physician (before I had read this thread!). He was rebuffed quite brusquely. Not only did the doctor -- whom he had not seen before -- show no interest in accepting or reading the list, she did not even give him time to summarize it for her. The appointment lasted only a few minutes. My husband has long suspected that at least some of his various troublesome symptoms are related by some underlying cause, but has been constantly frustrated in his attempts to find a doctor who will explore larger issues instead of just treating the most active symptoms.

Surely much of the blame for this episode lies with the particular doctor involved, the environment of the practice, and the procedures of our HMO. But it makes me think that Dr. Tufte's excellent suggestion will face many barriers in becoming widespread, even if patients embrace it.

-- Jennifer (email)

In my neck of the woods (which is to say, Canada) we are discouraged from bringing multiple complaints to the clinic. It's one complaint per visit, due to the billing structure and the need to keep appointments to a manageable duration. I understand some clinics are accommodating, allowing patients to book multiple concurrent appointments for multiple complaints. That said most visits to the clinic are for a runny nose or a sprained ankle, where problems are pretty independent.

A good doctor should explore all symptoms when investigating systemic complaints. Perhaps a good approach would be to address a "list of symptoms" for such cases. This differs from ET's suggestion only by its label: it doesn't imply a separate cause for each "problem".

-- Ashley Gadd (email)

Several have commented on taking a list of concerns to the doctor. I can offer this perspective.

I learned of lists as a child. When I needed something from Dad, any request agreed to would be recorded on a legal pad. This was called "making the list." Nearly all items on the list were eventually accomplished, but I learned coaxing was required to expedite my urgently needed items. My later questions of "where exactly" within the list was my item was met with an evasive "relax, you're lucky you made the list" or "it's on the first page". Dad controlled the list, its length, and its ordering.

Even though as an academic physician I spend considerably more time thinking than practicing, my patient care experience still yields me many lists from patients. From my pespective, some of these lists help move the visit satisfactorily along, and others do not.

Harsh reactions from physicians to patient generated lists may have to do with ownership of the visit. Within my own medical school class, it was telling that many of our promising talent show skits never eventually worked because classmates fought over control of the script. Acceptance of a patient list may seem physicians to represent loss of control of the scope and length of the visit. Even physicians who regard patient lists with disdain would welcome a chance to preview a certain complaint early in the visit as opposed to catch a "Did I mention that I'm having chest pain?" in the 14th minute of the visit.

My biggest problem with patient generated lists; these lists often seem like grocery store lists. This is not to say that there is no order, but rather that the order is not apparent or not medically useful. The grocery list example can be carried further: such lists are generated chronologically (out of cheese on Monday, out of soap and bread on Tuesday), categorically (pantry items, bathroom items, refrigerated foods), personally (items Susan needs, items Margaret needs, etc.), or by priority (Milk for the morning!). While each of these shopping lists helps, the best, most-efficient list for shopping is geographically organized (working the store from right to left). These lists are harder to generate, as it takes an imaginary trip up and down the aisles to translate the list into this format.

Generating a perfectly efficient list for a medical visit may be impossible for the patient. A patient generated list often requires translation into a medical hierarchy. What I often find myself missing is why is an item on a given list. Often I find myself spending more time on a given item than the patient desired ("Oh that doesn't bother me than much, I just wanted to know if it meant that I had cancer"). An explicit description as to why an item is on the list, though, would help me. What about an organization of: things I need today (refills, signatures on screening forms, or a disability parking permit); symptoms that are bothersome or impair me that I would like some help with (i.e., my knee gives out on me); symptoms I am worried about (I wake up three times at night to urinate, I found it difficult to speak for about 10 minutes yesterday and now I'm fine). That would allow me to delegate some tasks (other staff to assist with refills) and reorder prioritize others (temporary loss of speech rather than nocturia) and to save others for later, if needed.

-- Ted Johnson (email)

One aspect that I wonder about

Patient-doctor communication relies on adequate doctor-doctor communication, which is assumed but perhaps deserves another look, particularly at how medical students are taught to communicate with their colleagues. Increasingly, medical treatement requires many different physicians to contribute to the outcome, and the difficulty in communicating is one cause of increased medical "transaction costs" as well as having a direct impact on patient care. Does anyone have any data, references or ideas on doctor-doctor communication?

-- Anne von Bergen (email)

For more on lists, see the thread Lists: theory and practice

-- Niels Olson (email)

Response to Medical information exchange: Ownership of the visit?

Ted said:

Harsh reactions from physicians to patient-generated lists may have to do with ownership of the visit.

Kurtis replies:

Begging your pardon, but what do you mean by "ownership of the visit"? It seems to me that the patient—you know, the person who is coming to you for help—"owns" the visit. That doctors usually think otherwise is the primary reason I refuse to go to one except in true emergency situations … or "take this urine test to keep your job". I'm grateful that I can count such incidents on my ten little fingers.

-- Kurtis Kroon (email)

Today's NY Times is running a story titled, "The Doctor's Wold: Socratic Dialogue Gives Way to Powerpoint."

I'm feeling better already. Next slide, please.

http://www.nytimes.com/2006/12/12/health/12docs.html? em&ex=1166072400&en=00fcaec4afb78c5e&ei=5087%0A

-- Curt LaFond (email)

Permalink: The Doctor's Wold: Socratic Dialogue Gives Way to Powerpoint

-- Niels Olson (email)

Regarding the Socratic Dialogue article Curt LaFond so kindly provided: I have only attended one grand rounds where patients were present, in Dermatology at Bethesda Naval Hospital. It was after returning from sea duty, while working on my prerequisites for medical school. My sense then and now is that the real value of the format was the indelibility of the memories. They are seared into the mind, even the minds of the audience members. For the discussants, particularly the residents making the presentation. The room absolutely crackles with energy as the participants walk down the path of reason, toward truth, but with social maneuver ever at play. Special evolutions on Navy ships have a similar feel, where the participants are not only responsible for the safe execution of dangerous maneuvers with enormous ships propelled forward too quickly for certain mistakes to be undone, but it is also played out in front of a sizable fraction of the crew and wardroom. The experiences form the mind, usually for the better, though perhaps only in the long run.

PowerPoint may give the illusion of greater factual density, but I have never heard a PowerPoint slide whispering behind a resident's words as they teach me about heart murmers. The ghosts of patients and attending physicians, however, I can hear screaming like banshees.

-- Niels Olson (email)

Doctor, doctor, may I please have an aspirin? Sorry Mr. Tuffle, but it's too inexpensive.

"The More You Pay, the Better the Care? Think Twice"
By EDUARDO PORTER, The New York Times, December 16, 2006

-- Edward Tufte

I recently attended a talk with AMA trustees Ardis Hoven and Robert Wah. Wah, in particular, just came from serving as the head of information systems for Navy medicine. He projected EMRs are going to become web-based databases maintained by either the government or commercial custodians. I think, tacked onto that, there will be custodial and transmission costs involved and providers may have access fees for the data, which will be ported into the interface of their choice, much like we all currently have our choice of e-mail systems: Gmail, Outlook, etc. This raises at least three sorts of questions: how to pay for it, how should the interfaces be designed, what other interfaces, besides an EMR interface, like epidemiology or budgetary, could be designed to look at this database?

-- Niels Olson (email)

The Veterans' Health Administration has launched a web-based portal to a shared medical records system where the patient can update his records via the Personal Health Journal application and view his medical history and readings in graphical displays. It's called MyHealtheVet and it's at http://www.myhealth.va.gov/.

-- Anne Carroll (email)

I was interested to see an article in The Guardian on 4 January, reporting on a meeting the correspondent had with Steve Jobs in 2002, where the reporter joined a delegation from the National Institutes of Health. The NIH delegation were keen on pushing the idea of a tablet computer by Apple, one would assume for usage of designs such as you have proposed. Jobs said that they were not pursuing this for several reasons: one, the wireless bandwidth for huge images plus the security required was not anywhere on the horizon; second, the screen resolution would not be high enough for that required by NIH; third, any device designed to work in the medical field also had the risk of attracting significant liability. This explanation seems entirely reasonable but from almost 7 years ago. It would be interesting to hear an updated consideration... the Guardian article can be found at: http://media.guardian.co.uk/newmedia/story/0,,1982965,00.html. However, you might need to register (for free).

-- Will Oswald (email)

(from the perspective of an emergency department physician) 1. Graphs. Love 'em. Graphs are only occasionally useful for ED evaluations, but for inpatients they can vividly display trends. Unfortunately, our hospital uses three different software programs for the recall of lab values, and only one will graph data. This requires 3 or 4 successive clicks and only allows the display of one trendline at a time - so no direct visual comparisons are possible.

2. Lists. I think some of the physicians' negative reactions may be linked to two problems with lists. Ted Johnson accurately points out the organizational problems of patient-generated lists in his post above. In the emergency department, lists of problems appropriate for primary care physicians are too inclusive, and force one to wade through years (as they usually chronologic) of minor problems only to come across the recent heart attack as the last item. An ER physician wants the relevant information, not all the information. Reading through lists of data which are disorganized (for our purposes) increases the cognitive load I desperately seek to minimize at every opportunity. The second problem is that, while 25% of lists are helpful in summarizing a patient's complicated medical history, the other 75% belong to patients with a touch of hypochondria. By this, I mean a patient who has carefully catalogued years of vague and transient symptoms who expects an emergency department visit will yield an Answer. By "Answer," I mean a rapidly diagnosed, single, unifying, definitive diagnosis with single, cheap, safe treatment, preferably in the form of a once-a-day pill. A bit cynical, I know, but I have run into many lists and found them to be a marker for trouble more often then not.

3. Socratic dialogue gives way to PP. Almost all of our presentations are delivered in PP format. Almost all of our engaging and interesting talks are not in PP. As stated elsewhere in the forums and by ET, a good speaker can overcome PP but a poor speaker uses PP as a crutch. I've tried spreading the gospel (even ordered "The Cognitive Style of Powerpoint" for distribution) but have found that there's no cure for the ease with which PP produces "acceptable" presentations.

4. I love that medical liability helped scare Steve Jobs off. Scared, as well.

5. How would data from an individual hospital make it onto a third-party managed web-based EMR? Through the hospital, or the patient?

-- Dan Vining (email)

Dan Vining asks

How would data from an individual hospital make it onto a third-party managed web-based EMR? Through the hospital, or the patient?

There are technical and social issues here. First the technical, then I'll phase in social issues.

Like e-mail, the data would be sent as messages transmitted to a server. "Server" might mean server farm(s?) in this case, much as e-mail messages through swizzle.stick@gmail.com are sent to one address, but bits may be flipped on a hard disk in Menlo Park, Miami, Madrid, or Mumbai. Certain elements would be "prime keys". A prime key is a database field that uniquely identifies each of whatever you're interested in. Many e-commerce sites have switched to using e-mail addresses as their logins, their prime keys for their transaction database, because swizzle.stick@gmail.com is unique across the entire Internet. Social Security numbers are prime keys. Time stamps would also be required, and presumably the messages would be many and small: you would want to transmit every time the data entry person exits a field. This means AJAX or something like it. Based on what I've seen, this is in fact what existing the webbased instances of VistA. A message, which no human would probably ever see (like you rarely if ever see all the administrivia wrapped around e-mail messages), might be myhealth.va.gov 20070107 14:25:07.0147 [crypto]
987-65-4321 sez 123-45-6789 accucheck @ 20070107 11:45 = 499 [/crypto]

Which would be a message, which needs to be encypted, sent by Doctor DeBakey (987-65-4231) from whatever computer he is on ( to myhealth.va.gov saying Alfred Thayer Mahan's (123-45-6789) blood glucose, using the accucheck test, was 499 mg/dL (the system ought to know glucose is always be in mg/dL) when I measured it at 1145 this morning. When you hear computer nerds talking about AJAX, this is the kind of stuff they're doing. Sending lots of tiny messages, no more than absolutely necessary.

The system should also note that 499 is the upper limit of the accucheck test and that it is only reported by the machine if the actual level is even higher, so it should present the number in red. I can see a sparkline . . .

Who would put data in? Probably a lot of different people. Some of the 'who-can-see-and-do-what' issues may still need to be solved for this particular category of social networking, but the basic challenge of multi-user input to controlled-access web-based databases has been thought out before. Look at any social networking site. Do all doctors have access to all records all the time? Do all insurers have access to all doctor records and all patient records? Do any insurers have access to all anything? More likely, data entry would be controlled by permissions, which would be automatically set by role. Once Alfred Thayer Mahan designates Michael DeBakey as his doctor, then Dr DeBakey would have access to see Alfred's record. As an administrative condition of extending the offer of a particular insurance plan to Mr Mahan, Humana would require Mr Mahan designate Humana as his insurer. If Mr Mahan is a jigelo living off of Mrs Mahan and wants to enjoy spousal coverage under her plan, Mrs Mahan would have to specify A.T. as her husband. And so it goes. The really interesting thing is that ultimately the custodians, the sysadmins, the programmers, will write the rules, as Lawrence Lessig pointed out in Code and Other Laws of Cyberspace and Aaron Swartz summarized here using the Wikipedia database as his example.

There are several companies growing their way into solutions to these problems right now. When A.T. needs insulin, Dr DeBakey orders it in the room and A.T. can elect to pick it up at a pharmacy or have it mailed to him, either there in the office, if Dr DeBakey has time, or A.T. can make that decision via his own patient interface himself, later. The pharmacy will also have to have a sort of industrial strength instant messanger service that tells them when new orders come in to their pharmacy, be they the mail-out pharmacy or the Walgreens down the street. Flickr has solved this problem for photo prints, surely the largest private sector of the American economy can figure this out for drugs and medical devices.

In-patient? Correct me if I'm wrong but this is where most EMR systems were first fielded, despite the fact they are the most difficult, time-sensitive cases, from a purely medical standpoint. In-patient notes? Dictate them electronically. NPR recently reported that Dragon Naturally Speaking 9.0 is good-to-go out-of-the-box.

So, there are interfaces designed for the the doctor. I haven't seen, though they may exist, web-based interfaces for the patient, the pharmacy, the hospital, the call-center insurance representatives on the phone trying to explain to the family that the deductible really is $10,000 for emergency surgeries, based on the plan they subscribed to, but those are the big players in the game: patients, doctors, hospitals, insurers, and pharmacies. Pharmacies are really a 10%, borderline, barely-big player in the two trillion dollar American healthcare industry. There may be interfaces for epidemiology and public health, but my sense is these people will, like they are now, pulling data sets provided by the sysadmins into hardcore analysis programs.

Presumably, one can easily extend the doctor interface to other healthcare providers—out-patient physical therapists, occupational therapists, etc, etc. All payors can have the same interface, though the big payors will certainly want additional features. On the other side, hospital billing would need an interface, which could be extended to all billing entities. Thus for the rural practitioner, the doctor might go from room to room logging on each time as Michael DeBakey, while his office manager would be sitting at the front desk and log on once in the morning and see a 'billing' interface. Presumably it would be good for patients to be able to access their records in the waiting room, from their cell phones or consoles in the office.

It's all very not intimate, very not doctor-patient-relationship. Maybe moving the administrative angst and recording accountability to a computer and out of the exam room would allow a real focus on doctor-patient relations while they're in the room.

-- Niels Olson (email)

From the Boston Globe:

First, do no assuming -- A doctor urges his peers to think differently
By Sam Allis, Globe Columnist | January 28, 2007

"Fact: A doctor in this country interrupts a patient, on average, in the first 18 seconds of a visit..."

The article features observations by Dr. Jerome Groopman of Beth Israel Deaconess Medical Center, including:

"While the patient safety movement has led to major improvements in protocols to avoid systems errors, he points out, nothing has been done to address a more profound issue: how doctors think. And bad thinking is what causes countless mistakes. "No one talks about this stuff," says Groopman."

-- Jim Linnehan (email)

Here in British Columbia the government-run Medical Services Plan has recently issued a request for proposals through which 6 (why 6 I don't know) Electronic Medical Record vendors will be chosen to have the purchase of their EMR software, by doctors like me, subsidized by the MSP.

Sadly, I have attended EMR demos faithfully over the years (and the salesmen have brought their laptops to my office just as frequently - nothing kills my enthusiasm for a new EMR like the demo just not running at all....)but I have never seen an EMR which incorporates sparklines, or the graphical patient info format proposed in Tufte's Lancet article, or the general principles of design I find in the four of Tufte's books which I own.

Is anyone aware of an EMR in actual use which uses Tufte's principles?

Will Johnston

-- Will Johnston (email)

I was sufficiently intrigued by the use of sparklines in summarizing medical data as to attempt to implement them in my EMR. See http://synapsedirect.com/forums/permalink/3020/3016/ShowThread.aspx#3016 for a graphic.

I am not sure how much it adds though. In the graphic above, the patient had an episode of vasculitis hence the high crp/esr then. But that high result masks the smaller variation that occurs thereafter.

-- Graham Chiu (email)

Patient privacy in the Internet age has become a hot topic. Every healthcare provider has a HIPAA postscript in their e-mail. Millions of these postscripts are transmitted along with additional millions of patient data points every day. It is often easy enough to simply send "That patient with the regurg? His O2 sat is stable". However, as formal means of electronic collaboration develop and the affordances of these electronic collaborations become more expected, this problem of privacy over open networks has the potential to become a major seam in the security of patient privacy. I have drafted some thoughts on these postscripts for the upcoming weekly medical blog carnival Grand Rounds. I would greatly appreciate the review of anyone reading Ask E.T.. I think the long-term solution is a community-owned standard, so I invite everyone to improve on Wikipedia's entry HIPAA compliant email postscript.

Plot-spoiler: the experts seem to agree that PGP is the answer.

-- Niels Olson (email)

Regarding patient privacy over open networks for transmitting EHR, as raised by some contributors.

The buzzwords of the day are RHIOs ("Ree-Ohs"), or regional health information offices. They might be managed by 3rd parties - several states such as Michigan have offered funding for proofs-of-concept. The Nationwide Health Information Network (NHIN) is also a player, (http://www.hhs.gov/healthit/healthnetwork/) as is the Office of National Coordinator for Health Information Technology, run out of the Dept. of Health and Human Services.

The issues of federated identity management are hugely complex, and debates over whether a RHIO would store the actual patient data or simply (?) manage pointers to data stored in provider and payer systems are not likely to be resolved soon.

Unfortunately, until the government comes up with a standard there isn't any near-term resolution to EHR transmission... nobody will take the risk of not meeting the (eventual) Medicare market.

-- Gordon Fuller (email)

Chatty doctors chat about themselves

Gina Kolata in The New York Times here

-- Edward Tufte

The study in The Archives of Internal Medicine (summarized in the Times article above) relied on bugged actors as surrogate patients. It's worth taking a moment to ask if actors are reliable and valid surrogates in such research. For example, the 18-second study cited above relies on real patients, not actors.

Clinical skills researchers generally agree that actors are valid and reliable. It is convenient for them to agree on this, but is their assertion correct? Some of the foundational work they cite seems relatively weak compared to the standards of clinical trials. For example, this study is the basis for validity for use in second-year medical students; it has 10 subjects in each arm of the study, equivalent to Phase I clinical trials for safety. Phase III pre-marketing trials last for a year or more and typically have hundreds or thousands of patients in each arm of the study. The researchers also seem to feel that while real patients are valid and reliable, it is much more difficult, perhaps insurmountably more difficult, to compare the interactions of one doctor and his panel of real patients, to another doctor and her panel of real patients.

This review article has a good list of the historically significant studies:

Wallace, J., Rao, R. & Haslam, R. (2002) Simulated patients and objective structured clinical examinations: a review of their use. Advances in Psychiatric Treatment, 8, 342-348.

-- Niels Olson (email)

Google and Microsoft Look to Change Healthcare

I suspect this will go much like I described in my January 12th and 16th posts above.

-- Niels Olson (email)

Also: Google Health Prototype, from the Google Operating System blog.

-- Niels Olson (email)

Have a look at the little favicon for the Google links in the post above. It is interesting to see that it is an image. I wonder if the Max Diving picture could be made into a similar size image and placed as a favicon for the ET website.

-- Tchad (email)


With a little looking around I found an even better favicon reference at W3C. Purportedly, a 24-bit 32??32 PNG should work, though it should be checked in a variety of browsers to see how good a job each does at displaying and resizing it.

-- nathan vander wilt (email)

Image as logo

The W3C Reference is helpful - the 16 August post was specific to Google's use of an image that is not just a bmp outline of a logo. Google uses a big "G" on the normal google site but the page that Niels linked has a small image.

GOOGLE HEALTH PROTOTYPE LINK - For ease of reference

The detail of the image is too complex to be made out and it is in the complexity that it is unique - much as it was on the iPhone picture of Max Diving. This discussion should probably migrate to the thread on "Image as Logo" if it is to continue.

IMAGES AS LOGOS - Another Ask ET Discussion Thread

See also


-- Tchad (email)

One young physician hangs his shingle on the internet:


The big idea seems to be to grab a patient panel of up-and-comers: he's only taking patients between 18 and 40 in the New York area who are web-savvy.

-- Niels Olson (email)

Monitoring what they do to you and yours


-- Edward Tufte

The selection of treatment for breast cancer depends on many variables, including whether or not the cancer is invasive, the stage of disease (0, I, IIA, IIB, IIIA, IIIB, IV), the number of axillary nodes involved (0,1-3,>3), whether or not the tumor cells express hormone receptors and on specific gene expression (e.g. HER2). These factors are determined through clinical observation, imaging, biopsies, surgical exploration, and diagnostic tests.

The multiple combinations of these factors create myriad treatment paths. Treatment elements include surgery (from lumpectomy through to complete mastectomy & reconstruction), chemotherapy, radiotherapy, hormone therapy, and targeted drugs (e.g. Herceptin).

I would appreciate any ideas about how to clearly communicate (on paper) these pathways, helping the reader to understand the relationships between diagnostic tests and therapeutic decisions that create the forks in each of these many paths.

-- Rob Wishnowsky

Interestingly enough, AHRQ (US Health and Human Services quality research agency) has a funding opp right now for what they call "dense information design" for the display of medical information, over time, graphically, to providers and patients. They note that "not having all pertinent information on the same screen" can increase errors (although they do not talk about poor and poorly integrated visual displays, which are rampant), and seek experts in "dense display of data and other innovative information design principles" (I would link to the grant app but it's actually open only to those with a specific contract vehicle; it's called "Use of Dense Display of Data and Information Design Principles in Primary Care Health IT Systems" and you can ping me if you're interested....I'm just a bystander in the process as well as a lurker here...)

-- Stanley Chin (email)

Evidence Based Medicine

Dear ET,

I have been digging around looking at evidence based medicine. One of the very interesting websites is the James Lind Library that has a collection of materials on the evolution of fair tests of treatments from 2000 BCE to the present. It contains key passages and images from manuscripts, books and journal articles along with commentaries, biographies, portraits etc. As an example here is a page from a Chinese medical text from the Song Dynasty (960-1279 = http://tinyurl.com/6pb5oy.

Best wishes


-- Matt R (email)

I previously referred to the Veterans Administration's electronic medical record system, VistA. Having now used several medical record systems I stand by my statement that we owe those folks a debt of gratitude. Here is a VistA developer's opinion of the current controversy around Google Health and other online records custodians (which is the model I reported the AMA trustees envisioned would develop): Fred Trotter, In all Fairness.

In a related note, VistA code, being developed by government programmers, is available under the Freedom of Information Act and as such is continuously ported to the open source project WorldVistA. WorldVista developers have started working with the OLPC project to develop a Linux client. This has the potential to make widely accessible patient databases easily deployable all over the world, including in the immediate aftermath of natural disasters.

-- Niels Olson (email)

Patient records in healthcare


Mainly anecdotal evidence here; hope this is not a cherry-picked bogus-trend story.

-- Edward Tufte

The article this press release is referring to is from a recent edition of Cancer. It speaks to many issues discussed across Dr. Tufte's forum topics. Present the appropriate information thoughtfully and patients will make better decisions.


Ben-Zion Wasserman MD

-- B Wasserman (email)

I recently went to a new dentist, so was obliged to get new x-rays.

It was a standard office, nothing fancy, but came on good recommendation.

Without leaving the chair, the technician took about six shots. As each processed digitally, they appeared horizontally on a large monitor that was tilted at a premium viewing angle above me. I didn't even have to sit up to look at them!

She briefly explained what portion of teeth I was looking at by referencing an image and then lightly tapping my teeth with the cleaning tool.

She was then able to show me a few spots that had some tartar and even zoom in on them.

As she was cleaning them, I had a good visual of what she was working on and a general sense of how long it might take -- which wasn't long thankfully. I'm also happy to report I had no cavities.

I was impressed not only with how much faster X-rays are now, but the added efficiency the tools seemed to give in providing the treatment.

-- Rob F (email)

Threads relevant to interface design:

Threads relevant to medicine, public health, and biology: