[Openid-specs-heart] Purpose of Use

Tue May 16 20:00:38 UTC 2017

Adrian

Your points are certainly well taken too!  Especially regarding C and E below.  

Perhaps the point is that if we were to freeze the permitted value for Purpose of Use to patient access we’d actually limit the size of the problem we are trying to solve to a manageable size?

Nancy listed out several examples of POU where she writes  

The custodian organization uses the claimed purpose of use to interpret policy.  For instance, if the pou is ‘Treatment’ a complete record might be provided, but if the pou is ‘Coverage’ the policy may limit what is sent.  If the pou is ‘Research’ then the custodian organization might need to de-identify the data on the way out.

Where what is shared is defined by the POU.  Does it help us make progress to say – yes POU can have different values but for the purpsoes of this work group we are holding our work to the case where it is Individual Right of Access?

As you say – an UMA Auth server is not healthcare specific but I thought HEART was.  

I recognize the arguments that POU would be valuable to include but given what the original charter was does it make sense to consider working on a subset of possible POUs?

Aaron

Aaron Seib, CEO

@CaptBlueButton 

 (o) 301-540-2311

(m) 301-326-6843

From: agropper at gmail.com [mailto:agropper at gmail.com] On Behalf Of Adrian Gropper
Sent: Tuesday, May 16, 2017 2:29 PM
To: John Moehrke
Cc: Aaron Seib; HEART List
Subject: Re: [Openid-specs-heart] Purpose of Use

Aaron has stated his opinion on scalability quite clearly and I can give a privacy engineering perspective that attempts to explain the reason I come to the opposite conclusion from Aaron. HEART needs to be about 3rd party access.

>From a privacy engineering at scale perspective:

A - We need to minimize the number of copies of PII that are required in the normal course of business. This enhances both security (fewer places to secure and audit) and privacy (more opportunity to control and audit use).  

B - We can separate PII creators (lab, imaging facility, genome) from PII aggregators because the aggregators (institutional health records, relationship registries, outcomes registries, biobanks, hidden data brokers) combine PII over time and across various institutions. 

C - Some kinds of "lab" data are not conveniently copied to a patient-controlled location even if provenance is solved. For example, an MRI, CT, or genome can be 10's og GB and is more usefully streamed (like Netflix) or processed in-place (ask about a specific SNP in a genome). 3rd party access control allows for streaming direct from the source for scalability.

D - Even when a patient chooses to own a copy of her data as an aggregated health record, she still needs to provide for discovery of this record by appropriate third-parties. Typically, this means registering some metadata in a directory (e.g.: an HIE relationship locator service) and controlling access to that directory somehow. The AS can control 3rd party access to a directory.

E - Some healthcare registries are, by definition, coercive. For example, a Prescription Drug Monitoring Program. This kind of aggregated PII cannot be replaced by a copy under patient control and needs to be handled as 3rd party access.

F - A patient-owned aggregated health record, like the ones promoted by Aaron's organization, can be designed with compatibility for a HEART AS and will gain utility and marketability if they are. The reason is that an UMA Authorization Server is not healthcare-specific. Patients may want to provide access to location, environmental, social or banking relationships compatible with UMA that have nothing to do with FHIR and may not be compatible with software designed for healthcare. The assumption that a HEART AS is specific to healthcare reduces both patient empowerment and scalability.

Adrian

On Tue, May 16, 2017 at 1:42 PM, John Moehrke <johnmoehrke at gmail.com> wrote:

well... if not third party access, then normal OAuth will work just fine...  

See the fine work done leveraging this patient right of access, Sync for Science

https://www.healthit.gov/buzz-blog/health-innovation/nih-and-onc-launch-the-sync-for-science-pilot/ 

And if you are finding that normal OAuth or even lesser technology (username/password) is not working, then adding another layer of OAuth/UMA is not likely to fix the problem...

again, I am not arguing that Patient Right of Access is universally accepted and implemented... Just that adding more technology will not fix a fundamental misunderstand. A misunderstanding that HHS has tried over and over to fix.

Of course, I live in Wisconsin were there is no such data blocking...   :-)

John Moehrke
Principal Engineering Architect: Standards - Interoperability, Privacy, and Security
CyberPrivacy – Enabling authorized communications while respecting Privacy
M +1 920-564-2067 <tel:(920)%20564-2067> 
JohnMoehrke at gmail.com
https://www.linkedin.com/in/johnmoehrke
https://healthcaresecprivacy.blogspot.com
"Quis custodiet ipsos custodes?" ("Who watches the watchers?")

On Tue, May 16, 2017 at 12:19 PM, Aaron Seib <aaron.seib at nate-trust.org> wrote:

John

Your point is certainly well taken.

UMA is more powerful when it enables the user to manage third party access rights.  I am just not sure I subscribe to the notion that we should be building toward a world where there are still 3rd party access rights.  

I may be alone in seeing it this way but I had thought that UMA was a candidate for the consumer to convey to its app what the individual’s sharing preferences would be.  In the future the holder of a 3rd Party Auth would be going to their Consumer Controlled app to collect data for a given purpose of use.  I see how it could work both ways.

I may be speaking heretically here but do we really believe that a Provider Organization should be accountable for handling 3rd party auths?

I am speaking really practically here.  Do Provider Organizations have the bandwidth to handle 3rd party auths at scale?  I am not talking about the volume of these kinds of Auths today.  I am envisioning in the future when more consumers are authorizing sharing with researchers or for other purposes of use.

It seems to me that the Consumer Apps would be the place that would have the value add from building the computing capacity to manage those transactions.  Not the Data Source.  The benefit of the Data Source doing it is that you don’t have to solve the Provenance Problem but if I am following along the FHIR Resources are being defined to account for that.

At the end of the day it may be a policy question but I would argue from a policy perspective that in the best of all possible worlds the data should flow from the consumer to the researcher.  

There is probably more to it that others can point out that I am missing but from an SDO perspective what you say makes sense.

Aaron Seib, CEO

@CaptBlueButton 

 (o) 301-540-2311 <tel:(301)%20540-2311> 

(m) 301-326-6843 <tel:(301)%20326-6843> 

 <http://nate-trust.org> 

From: John Moehrke [mailto:johnmoehrke at gmail.com] 
Sent: Tuesday, May 16, 2017 12:24 PM
To: Aaron Seib
Cc: Justin Richer; HEART List

Subject: Re: [Openid-specs-heart] Purpose of Use

Aaron,

Thanks for illuminating. I have not been able to keep up with all of the current intent of HEART. It seems that you are saying that the current goal of HEART is always patient access. As you point out PurposeOfUse = “Patient Access”... If this is indeed the case, then it would be best to state that as a fixed PurposeOfUse. When stated that way, it is more clear. 

This means however that other PurposeOfUse are not accepted. I seem to recall use-cases that were being discussed where the HEART AS was intermediating Treatment use-cases. Where the patient would be authorizing specific Providers to access their data through HEART (UMA). This would be a PurposeOfUse of TPO. It would be important to be specific that "Dr Bob" has TPO access, but not research or other access.

Or uses where the Patient is authorizing access by a Research project... or where they are authorizing access for other uses.

I am not trying to say that PurposeOfUse is essential, but rather to point out that it enables an 'intent' vector.

I think limiting HEART to only the patient accessing their own data is too limiting. This is indeed an important use-case, isn't that a much more simple two party use-case? UMA is more powerful when it enables the "User" (aka Patient) to manage third parties access rights. Thus it seems way too limiting to just focus on patient access.

John

John Moehrke
Principal Engineering Architect: Standards - Interoperability, Privacy, and Security
CyberPrivacy – Enabling authorized communications while respecting Privacy
M +1 920-564-2067 <tel:(920)%20564-2067> 
JohnMoehrke at gmail.com
https://www.linkedin.com/in/johnmoehrke
https://healthcaresecprivacy.blogspot.com
"Quis custodiet ipsos custodes?" ("Who watches the watchers?")

On Fri, May 12, 2017 at 8:19 AM, Aaron Seib <aaron.seib at nate-trust.org> wrote:

Perhaps a very important qualification to add to this discussion is that the PurposeOfUse = “Patient Access” is the essential trump card in the entire paradigm as far as I am concerned.

This is what I mean.  If consumer wants their data the data holders (specific the Covered Entities of HIPAA) have absolutely no right asking what the consumers intended “pou” is and they are obligated by law to send their data to the designated endpoint indicated by the consumer.  

I am sure no one on this thread needs to hear that reminder but I am not exaggerating when I tell you every single time I am trying to help someone get - their data out in the real world - that I have to “educate” the data holders about this fact.  Every time I make a call the first question out of the orgs mouth is “What do you want it for.”

PurposeOfUse comes into play when the Institution Data Holder is pushing the data to a 3rd party independent of the consumer.  Perhaps it also has a role to play when the consumer directs the data holder to share the PHI with another third party on their behalf, such as when they direct a Provider to share their data with an a research repository – a workflow that I have seen proposed by a couple of programs in the country.

In a perfect world I argue that this not the best practice – the consumer should request the data to a consumer app that they control and then subsequently share that date with the destination(s) of their choice.  The programs that are building “middle-ware” that takes the consumer’s approval to share with their research end-point and presents that to the Institution where upon the data is transferred to the Program’s destination are inadvertently leaving a lot of value on the table as I believe the consumer should have access to this data in an app that they can use for other purposes.  Including making donations of their data to other research activities.

Of course the catch 22 for the research community is that not enough consumers have their own apps that can support that today but they will.  If in were in fact a perfect world the Program’s would be able to provide the consumer’s with a way to pick a consumer app of their choice to populate.  The consumer apps would have the enabling infrastructure to make the donation of their data to one or more research destinations of their choice and everyone would benefit significantly more.

But that is only a dream today and probably not relevant to this work group, right?

Aaron Seib, CEO

@CaptBlueButton 

 (o) 301-540-2311 <tel:(301)%20540-2311> 

(m) 301-326-6843 <tel:(301)%20326-6843> 

 <http://nate-trust.org> 

From: Openid-specs-heart [mailto:openid-specs-heart-bounces at lists.openid.net] On Behalf Of John Moehrke
Sent: Friday, May 12, 2017 4:20 AM
To: Justin Richer
Cc: duane decouteau; openid-specs-heart at lists.openid.net
Subject: Re: [Openid-specs-heart] Purpose of Use

PurposeOfUse is indeed a critical aspect in healthcare. It is the highest differentiation, higher than user-role. It indicates the broader context that the data is to be used within. For example a request for data in healthcare often is onbehalf of a broader use: Treatment, Coverage, Research, etc. It is not an attribute of the user, it is an attribute of the request for information. It is not uncommon for identity and context attributes to be conflated or simply communicated in one token; however that does not mean they really are the same, it just means that the environment has made a simplifying assumption to combine for ease of technology. It is most closely aligned with the broadest part of a OAuth scope. So it should be included in the request for authorization decision, and authorization token.

John Moehrke
Principal Engineering Architect: Standards - Interoperability, Privacy, and Security
CyberPrivacy – Enabling authorized communications while respecting Privacy
M +1 920-564-2067 <tel:(920)%20564-2067> 
JohnMoehrke at gmail.com
https://www.linkedin.com/in/johnmoehrke
https://healthcaresecprivacy.blogspot.com
"Quis custodiet ipsos custodes?" ("Who watches the watchers?")

On Thu, May 11, 2017 at 3:29 PM, Justin Richer <jricher at mit.edu> wrote:

The “pou” claim as it was specified in HEART does not fit this use case, then, and it’s appropriate that we removed it. This was a claim presented by the requesting party’s identity provider, and had nothing to do with the request being made by the client itself. That’s why I argued it wasn’t a good fit where it was. If we were to add it back in, it should go elsewhere in the protocol.

 — Justin

On May 11, 2017, at 2:01 PM, Nancy Lush <nlush at lgisoftware.com> wrote:

Hello all, 

Per our last meeting, I agreed to provide more information on the need for the pou claim.

The claim pou was recently removed from the HEART specs and needs to be restored.

I spoke with Duane Decouteau from the VA team and provide the following details:

Purpose of use drives policy in many electronic exchanges today.  The custodian organization uses the claimed purpose of use to interpret policy.  For instance, if the pou is ‘Treatment’ a complete record might be provided, but if the pou is ‘Coverage’ the policy may limit what is sent.  If the pou is ‘Research’ then the custodian organization might need to de-identify the data on the way out.

The pou is passed as a claim within the request. It is a determining factor in evaluating which policies apply to a request.  Pou is implemented in ehealth exchange as an underlying principal.  Duane feels that pou should be a cornerstone for patient consent.  It is fully implemented now in ehealth exchange at the VA, Kaiser and others.

The list of pou values can be found at this link:    <https://www.hl7.org/fhir/v3/PurposeOfUse/vs.html> https://www.hl7.org/fhir/v3/PurposeOfUse/vs.html

Respectively,

Nancy

Nancy Lush          

 <mailto:nancy.lush at lgisoftware.com> nancy.lush at lgisoftware.com

Lush Group, Inc

Office: (401) 423-9111 <tel:(401)%20423-9111> 

28 Narragansett Ave

PO Box 651

 <http://www.lgisoftware.com/> www.lgisoftware.com

Cell:(401) 965-9347 <tel:(401)%20965-9347> 

Jamestown, RI 02835

<image001.gif>

_______________________________________________
Openid-specs-heart mailing list
 <mailto:Openid-specs-heart at lists.openid.net> Openid-specs-heart at lists.openid.net
 <http://lists.openid.net/mailman/listinfo/openid-specs-heart> http://lists.openid.net/mailman/listinfo/openid-specs-heart

_______________________________________________
Openid-specs-heart mailing list
Openid-specs-heart at lists.openid.net
http://lists.openid.net/mailman/listinfo/openid-specs-heart

_______________________________________________
Openid-specs-heart mailing list
Openid-specs-heart at lists.openid.net
http://lists.openid.net/mailman/listinfo/openid-specs-heart

-- 

Adrian Gropper MD

PROTECT YOUR FUTURE - RESTORE Health Privacy!
HELP us fight for the right to control personal health data.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openid.net/pipermail/openid-specs-heart/attachments/20170516/745c7831/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.jpg
Type: image/jpeg
Size: 2369 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-specs-heart/attachments/20170516/745c7831/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image002.jpg
Type: image/jpeg
Size: 2369 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-specs-heart/attachments/20170516/745c7831/attachment-0001.jpg>