Thursday, April 16, 2009

See You Get Care

Over at info.rmatics, Jonathan Abbett discusses the NYT article (warning, have BugMeNot ready) on a study about how adding a picture of a patient to a file makes the radiologists interpreting the files far away from the patient feel more empathy and write more detailed reports.

Jonathan then discusses how, when he was part of a team designing a Personal Health Record system, he considered having a picture of the person included, making the record a bit more human-oriented along profiles on social networks like Orkut and Facebook. He wonders whether this should indeed be part of the record to create better health care.

My answer to that would be: before you do that you really need more study. It could actually make things worse.

It's actually not a no-brainer, from the above study. The study used pictures of the people taken while they were sick. As such, they didn't just create more empathy, but they also became part of the diagnostic dossier: if the radiologists saw symptoms in the pictures, they would describe those too. This is very different from including the facebook profile picture, taken at another, perhaps happier time.

Sickness is actually a great equalizer. When we are in pain, when we are being examined in radiology wards, we all look the same: like crap, and usually surrounded by clinical gowns and tables and walls and objects. Our profile pictures do the exact opposite: trying to show our individuality, our vigor, our 'best' side, the interesting place we were, the cool clothes we wear, or how we are counter-cultural that week and never got around to changing that, or felt moody, or were going through a heavy metal phase, etc. The problem for me there is that this then becomes a place for the diagnosing remote clinician to hang their personal or cultural biases and have them creep into the health care process. Oh yeah, they have them. Don't tell me they don't. And when you then send these files halfway across the world to be interpreted in places where radiologists are cheaper? Off-shoring is happening in that area, yes.

Examples. I have fat friends, who dread going to the doctor, because no matter what they show up with (hangnails, glaucoma, broken arm, being bitten by a zombie) they know that at some point during the consult their medical problem will end up being all about how they are fat. Being fat is not something that will be 'fixed' during the consult (oh god if only), and they know they are fat already, trust me, and they still would like some health care while they are there, so they often end up wondering, would they get taken more seriously if they weren't fat? 'You have ten minutes to see me and we are talking about how I am fat, again?' I think these people should switch doctors but good luck finding a clued-in one in your network. Some of my female friends tell me the same about how all their health care problems get reduced to being about their female bits. And my fat female friends? Don't get them started.

The other side of that are the validated sociological studies that attractive (tall, lean, symmetrical faces and bodies) people get better, well, basically, everything. Hired faster, faster promotions, higher bonuses, more friends, quicker service, more freebies. Same for people with fairer skins in comparison with people with darker skins. Appearance becomes a detractor from actual capabilities, distorting the whole picture.

Health records don't currently have this element in them. Sure, the examining clinician and staff, the attending, the specialist, they get to see the faces, the bodies, some of the trappings of life that the patient walks in with. And these frontliners are also closer, presumably, to the culture and assumptions of the patient, although this is less and less true in hospitals in large metropolitan cities. But still, there is a higher chance of a shared culture, shared understanding of what tokens like hairstyles or life decisions mean, when the clinician and the patient are in the same place. They see, but they also understand.

Now put glimpses of those trappings in the file that goes out, sometimes far away, for interpretation. Where the clinical words 'morbidly obese' becomes instantiated as a very full-figured face, where the facial tattoos and piercings may get interpreted as something else than having had a career at Starbucks for too long, where a popped collar on a polo-shirt... you get the idea. It is far easier for humans to project narratives on pictures than on clinical notes in black & white, and you know what, we simply do not know if that is going to show up in the quality of the diagnosis.

We know that it does from the study, in a specific case of pictures that depict sick people, and when they are new. Do the interpretations of radiological data stay as detailed and long after two years of including pictures of everyone? Unknown. Will the interpretations stay as detailed if they are pictures taken of healthy times? Unknown. Will now everyone have to think hard about what picture to include, promulgating dreaded Internet Disease Profile Pictures? Well, people will anyway. Will 'prettier' pics get better or worse interpretations? Even health care? Unknown.

And quite frankly, this is not an area to experiment. Costs of getting it wrong are too high, and the costs will show up too far into the future to easily fix, nebulously, even if they are actually there. The current photo-less health record evolved over the last 200 years, and while it is far from perfect, the fact that during this evolution including a Polaroid instant shot just didn't become standard when it certainly would have been cheap enough to do so, should tell us something. As I said, 'More study is definitely be required' would be my recommendation.