CCHIT Feature bucket

A central problem with CCHIT is the feature bucket.

CCHIT certification represents compliance with a list of hundreds of functional requirements. This would be great if that list of features were 100% a good idea, but the reality is far from the truth. From the FOSS perspective we feel that there is a considerable dumbing-down effect that the certification brings. It prevents us from maintaining meritocracy.

I want to focus on one CCHIT issue that serves to illustrate this issue: Passwords.

Here is item SC 03.10

When passwords are used, the system shall support case-sensitive passwords that contain typeable alpha-numeric characters in support of ISO-646/ECMA-6 (aka US ASCII).

The problem, VistA supports three user ids, one that is equivalent to a username, and two that are similar to passwords. Without getting over my head on the details, there are two possible password types so that you can have one that your admin user can know and reset for you, and one that no one knows but you. There are all kind of administrator abuse scenarios that this addresses, but the VistA username/password/password system is not certifiable out of the box because it does not support case sensitivity. Which, as you can see, is a requirement. Most people are only aware of the CPRS client for VA VistA but in reality there are several clients, all of which support the username/password/password mechanism.

So when any VistA-based EHR goes and gets CCHIT certified it has to make the password system -act- dumber (in compliance with SC 03.09), and add case sensitivity.

Then lets look at the ClearHealth Inc. projects opinion on the value of hashing passwords. They believe, essentially that it give a false sense of security and an admin overhead that should be avoided. I disagree with them, but I can see where they are coming from. This however is in contradiction with the following rule: SC 03.11

When passwords are used, the system shall use either  standards-based encryption, e.g., 3DES, AES, or standards-based hashing, e.g., SHA1 to store or transport passwords.

So one simple issue, we have considerable debate in FOSS systems about whether this is actually the right design at all. But CCHIT takes the position that their way is the ‘right’ way and will not certify a system designed in a different way.

I hope that this is helpful in understanding why the ‘feature bucket’ is a problem for FOSS. It is directly contradictory to the notion of meritocracy that rules our culture. The -best- ideas win, not the ideas that come from a vote of the committee.

What we need from CCHIT is to identify the boundaries of an EHR system, not the contents. This is an idea that I have heard so many times and from so many different people, the only thing I can be sure of is that it is not mine:

There are three obvious edges to an EHR system:

• The ability to report on quality metrics
• The ability to interoperate with other HIT systems
• The ability to monitor and track access (security)

There are published standards available for all of these that can be tested in an automated way. That defines what an EHR needs to be able to show the world, but not define -how- it needs to provide those services. Frankly, if a system is capable of improving the quality of the delivery of healthcare, sharing its data, and can limit access to private data, the implementation details are not as important.

While those details are still important and should still be subject to scrutiny and respect the freedom of users, they can become the subject of debate of people like me who obsess about these kinds of issues.

What are the advantages of this model?

• We do not need expensive juries, instead we can fully automate testing and make the certification cheaper for everyone.
• We allow for freedom to implement ideas differently as long as the results are the same.
• It is not biases against FOSS, proprietary or (for that matter) paper or other low-tech systems.

Just some thoughts.