Doctor Data made to order

Recently, Health and Human Services (HHS) Centers for Medicare and Medicaid Services (CMS) released a request for public comment on how they should handle the reversal of an injunction that prevented them from revealing specific information about how doctors perform.

The mere request by the federal government for feedback on how best to use their new capacity to leverage Big Data to improve the healthcare system is a breath of fresh air. It is a welcome change from Snowden/NSA news and is a welcome indication that somewhere in the Federal Government there exists someone who understands technology policy well enough to not have “Aim gun at own toes” on their todo list. So we should all enjoy that for a moment.

In more “reasonableness from where I was not expecting it” news, it turns out that the American Medical Association chose not to continue to fight for the doctor data gag order, from the RFC:

The parties each had until July 30, 2013 to appeal the court’s decision, and no appeal was filed.

The parties that could have filed an appeal, but chose not to, were the Florida Medical Association and the American Medical Association. I suspect that almost all of the state-level medical associations and the AMA will participate in this RFC. This is a much more refined manner to discuss how Doctor Data should be made available. That’s right, folks–the American Medical Association and the Federal Government are participating in an open discussion about how to move healthcare forward using doctor data. I am a little dizzy with joy and surprise.

In order to reward the Federal Government for its reasonableness, I would strongly encourage people who know something about doctor data and/or doctor privacy issues to respond. I have already been working with a subset of top-tier healthcare journalists (which means the ones who will take my emails) to submit a cogent FOIA request now that they injunction has been lifted.

That FOIA request is fairly complicated, but I intended it to be a template of what HHS should consider as it makes these new policies. The heart of my FOIA request could be summarized: Deidentification is dicy but Aggregation is your friend.

Many patients and journalists frequently ask me if I can use DocGraph data to track the footsteps of a single patient through the healthcare system. I am always delighted to explain to them that my data is not just deidentified, it is aggregated. That means that I see “counts” of patients in my data, but never individual patient data.

Patients and journalists frequently come to my doctor dataset seeking justice about a particular case or incident. I have to frequently explain that I am not interested in looking at a single “at bat” for doctors. I want to know their batting average. (I am now somewhat associated with the Moneyball in Healthcare concept, so I can now abuse sports analogies with impunity. )Everyone screws up occasionally. Everyone hits a home run occasionally. Right now we judge doctors based on their screwups and their victories, without considering their average performance. We need to stop lionizing the surgeons who take the hardest surgeries, and ostracizing perfectly good doctors who have the misfortune of being sued too many times.

We need to find ways to reward the doctors who consistently get the boring stuff right. We need to reward the doctors who get it right even on Monday mornings at 7:00 a.m. when reasonable human beings should be asleep. We need to reward the doctors who get it right at 8:00 p.m. on Friday evenings when reasonable human beings should be having the first beer of the weekend, or at 3:00 a.m. Sunday morning. To be a great doctor, it is more important to be consistent then it is to be brilliant. It is more important to be humble enough to honor data and science as the “boss”, than it is to be a “leader”. Our culture tends to reward “rock star” doctors who generate unpredictable but sometimes brilliant results. Thank God that real hospitals do not work like “House M.D.” or “Grey’s Anatomy”. (They actually do work like “Scrubs”, sadly, but as a patient I much prefer “careful tragic comedy” over “sloppy bad drama” any day)

The only cure is good data, and CMS is asking us, the public, to tell them exactly what they want. Please consider submitting something to this process. If you are unsure that your thoughts on how open doctor data should work, feel free to leave them here as comments on this article or to contact me online (I am easy to find). I would me more than happy to help you improve your good idea, or encourage to abandon your embarrasingly terrible one.

This offer goes double for those who believe that doing this will violate doctor privacy. I am all for any privacy rules that do not interfere with researchers and data geeks like me sorting out who the bad apples are. As far as I am concerned, if some researcher made a case somewhere that the color of a doctor’s car might be correlated or causally connected to even a trivial patient outcome, I would suggest that every doctor in the country register their car make, model, and color when they update their address in the NPI database. In some senses, we need a generation of doctors to step up and say “We really have no idea what lifestyle and personal choices are actually connected with patient care; its time to turn the microscope on ourselves.” A hundred years from now, we can start making conclusions about what “does and does not matter” for doctor performance, and we can allow doctors to regain some privacy. Until then, we can certainly protect their home addresses and phone numbers (please stop uploading them to NPPES) and other “really” private data, and please beg some latitude as data scientists figure out what makes the best doctors tick.

I am going to pester the Strata RX people to have some kind of session about this at the upcoming conference in Boston, but even if nothing comes of that, I can promise you that we will be discussing this issue carefully in the hallway track after Janos and I talk about using Gephi with DocGraph.