Since our post on Monday, where we highlighted the potential impact to PHR adoption of the Boston Globe story on one consumer’s less than ideal experience with Google Health, there has been a number of other conversations worthy of note:
From John Halamka, CIO of Beth Israel Deaconess Medical Center, the source of the “bad data” we have his explanation of what went wrong and why. In addition to his summary on the issue, the comments left by readers are worth the time to read.
Then we have the doctor media darling and founder of Health 2.0 company, Hello Health, Jay Parkinson giving his own spin to the controversy where he completely, with some small justification, slams Google Health.
Google Health, nor any other PHR or other personal health data repository is responsible for the “bad data” sent to it from any source that a consumer may choose. Unfortunately. there are lots of “bad data” the question is, how do we manage it in such a way as to minimize harm? Google Health does hold some responsibility here for not having a process in place that acts as a weigh station of sorts, allowing the consumer to assess data integrity prior to dumping into their health account/PHR. This is but one of the many issues that Google has failed to address head-on as they continue to struggle with resource constraints. If Google does not start raising its stake in this sector and committing more resources to its Google Health initiatives, addressing such known issues as handling unstructured data, there is the very real possibility that Google Health will simply fade and become irrelevant.
Last, but certainly not least is a post by HealthVault’s chief architect, Sean Nolan where he goes into some detail as to just how challenging it is to create a personal health repository, dealing with questionable data and describing the process they have put in place at HealthVault to try and mitigate the problems and issues highlighted in the Globe article.
At the end of the day, virtually any PHR or health cloud such as Google Health, HealthVault and Dossia, will struggle with the challenges of dealing with messy, often conflicting and at times simply bad data. It will be rare that any data a consumer imports into their account is bad due to some malicious intent. No, the data is bad due to its incompleteness (claims codes are essentially useless for most clinical encounters), lack of notes providing the Gestalt of the clinician’s thinking during an encounter, or simply expediency (using ICD-9 codes that approximate).
The challenge the industry faces is how to effectively manage that “bad data”? There are many bright minds in this industry and more entering every day – we are confident that a solution will be forthcoming. But that solution, no matter how elegant and automated it may be, will still require human intervention, and who best to do such intervention than the Number One stakeholder, the account owner, the consumer.
Added to EHRLinks.com
Added to EHRLinks.com
Very good to see you again today, John. And this is a very good post.
I’ve learned a lot this week. One lesson is the amazing first-hand experience of seeing a story spin quickly out of control as second-tier reporters/bloggers misconstrued what was ACCURATELY stated in the Globe. In that context, your post is a balm because of its accuracy. 🙂
Another lesson is that indeed a lot of medical records data (not just billing data) is just plain bad, as you say. I got my hands on more details of my records, and I’ll probably write about it this weekend.
Methinks a lesson the healthcare world will need to learn, which was learned long ago in industry, is that it’s no use just installing a system. There must be enforced processes about how data gets into the system, how it gets edited, and ability to verify that it was all done correctly.
Industry applications like credit cards and phone bills worked that out decades ago. My impression is that it may take a generation to accomplish the same in healthcare. I hope it’ll be less, of course!