Saturday, March 3, 2012

Imaging and Meaningful Use 2 - First Impressions

Summary: Imaging and results now included; DICOM payload required for transfer to 3rd party, not necessarily for viewing. Transport mechanisms uncertain. Division of responsibilities between EHR and viewer uncertain in linked scenario.

Long Version.

In the second round, it seems that the Meaningful Use effort in the United States will address the matter of access to imaging as well as results from imaging studies, including a patient-initiated ability to download and transfer to a 3rd party. Excellent news, after having been excluded in the first round, assuming that the important stuff survives the public comment phase.

A summary and links to documents can be found at the the site, the relevant documents being:
The first of these contains the technical details of what standards are proposed to be required for certification, whilst the second proposes the objectives and measures for incentive payments. Both are scheduled to be published in the Federal Register on March 7th.

An initial review of these shows that imaging is addressed in three objectives:
  • "Use computerized provider order entry (CPOE) for medication, laboratory and radiology orders directly entered by any licensed healthcare professional ..."
  • "Imaging results and information are accessible through Certified EHR Technology"
  • "Provide patients the ability to view online, download, and transmit their health information within 4 business days of the information being available to the EP (Eligible Professional)"

I won't discuss ordering here further, except to note that the use of ICD-10-PCS for procedure codes is addressed under the discussion of "view online, download, and transmit", but not under the "use CPOE" certification criteria. It is certainly high time that there was more public debate about whether the ICD-10-PCS are sufficient for ordering or not (as opposed to billing), and the RadLex Playbook community will no doubt have views on this subject, but that is a debate for another day.

Imaging Accessibility (Viewing).

I quote the text of the "Imaging results and information are accessible through Certified EHR Technology" objective in full:

"Making the image that results from diagnostic scans and accompanying information accessible through Certified EHR Technology increases the utility and efficiency of both the imaging technology and the CEHRT. The ability to share the results of imaging scans will likewise improve the efficiency of all health care providers and increase their ability to share information with their patients. This will reduce the cost and radiation exposure from tests that are repeated solely because a prior test is not available to the provider.

Most of the enabling steps to incorporating imaging relate to the certification of EHR technologies. As with the objective for incorporating lab results, we encourage the use of electronic exchange to incorporate imaging results into the Certified EHR Technology, but in absence of such exchange it is acceptable to manually add the image and accompanying information to Certified EHR Technology."

Hard to argue with much of that, though it is a bit of a cop out to not include a requirement, as opposed to "encouragement", to use electronic means to get images into the EHR, particularly since it is so easy from the provider's side given the ubiquitous use of DICOM for payload and transport (any issues with which are not addressed by any "manual" alternative); I think the public comments should suggest a reversal of this stance, and require that electronic means be used.

For the Proposed Measure for this objective, the following text is included:

"For Stage 2, we do not propose the image or accompanying information (for example, radiation dose) be required to be structured data. Images and imaging results that are scanned into the Certified EHR Technology may be counted in the numerator of this measure."

Hmmm. For results, one can see the logic in this, since convincing radiologists to author structured content has been an uphill battle (esp., in the absence of tools and incentives). Efforts like the RSNA Radiology Reporting initiative are pragmatically focused more on structure narrative rather truly structured content (with codes for findings) in the sense that I think is meant here, and the return on investment in the MU2 timeline for demanding truly structured (coded) radiology reports is perhaps not clear enough yet.

But for the images themselves, I am not so sure. Personally, I think the MU2 objective should require "structured data" for images (as the input to the EHR at least; vide infra wrt. the viewing mechanism, as distinct from interchange), and the form of that structured data should be DICOM, and be specified in the certification criteria for this proposed objective (in the sense that a DICOM image with a "header" contains "structured data" about the patient, study, series, other demographics and acquisition technique, etc., whereas a JPG or GIF image does not).

I also think the objective should require that radiation dose information be structured, and that the structured form of that should be DICOM RDSR (in the certification criteria), since this is regarded as such a "hot" topic (if you will excuse the pun) (with the FDA and the state of California, at least), and it is high time that we more seriously engaged the EHR vendor community in helping with this (esp. wrt. a longitudinal patient record of structured radiation exposure and/or dose information).

The proposed measures further state:

"We define accessible as either incorporation of the image and accompanying information into Certified EHR Technology or an indication in Certified EHR Technology that the image and accompanying information are available for a given patient in another technology and a link to that image and accompanying information. Incorporation of the image means that the image and accompanying information is stored by the Certified EHR Technology. ... A link to the image and accompanying information means that a link to where the image and accompanying information is stored is available in Certified EHR Technology. This link must conform to the certification requirements associated with this objective in the ONC rule."

That sounds reasonable on the face of it too; a "link" rather than embedding "by value".

However, if one turns to the certification requirements in the other document, one finds:

"We clarify that the phrase “immediate electronic access” is intended to mean that a user should be able to electronically access images and their narrative interpretations directly and without, for example, having to login to a separate electronic system or repository. This access could be provided by multiple means, including, but not limited to, “single sign-on” and “secure identity parameter passing.”"

Way cool ! Nobody wants to have their workflow interrupted by having to sign on to multiple systems.

But ...

"We also note that there are data format standards for the transmission of imaging data (Digital Imaging and Communications in Medicine (DICOM)) that we reviewed for this certification criterion, but do not believe that the adoption of these standards is necessary to enable users to electronically access images and their narrative interpretations, as required by this certification criterion."

Hmmm. At first site this may seem strange, at least for the "images" as opposed to "narrative interpretation" part, but I think what they are trying to say here is that for the purpose of just viewing images in the context of an EHR, it is not necessary to use DICOM as an "image format" per se. And that is probably true, in that one may well have an interactive "application" that is part of the access mechanism (e.g., a proprietary thin or thick web browser client or plugin or JPEG or HTML5 or whatever type of viewer), and as long as one doesn't bother the user with having to sign in to a separate system (i.e., makes the experience transparent), then they are good to go.

It does beg the question though, of how interoperable this will be, given different browsers and platforms and local security settings and all that. Should there be elaboration on this in the certification criteria with respect to the need for a "zero footprint" viewer or similar, or the use of standard web-based access mechanisms like DICOM WADO ? Also, how does one go about certifying the mix-and-match combination of an EHR with a link to a different viewing system (like the web front end of a PACS or P/VNA univiewer) ... will every combination of EHR and PACS and viewer need to be certified, or can these be separately certified as components (and how can one do the latter without interoperability standards)?

Further, one needs to consider the question of the input format versus the viewing mechanism, as hinted at earlier wrt. "structured data", and this again relates to the matter of decoupling the viewer from the EHR (linking rather than embedding).

Another question that arises is the quality of the images viewed; in my opinion these should be held to the "diagnostic quality" criterion that the AMA imaging safety and standards panel made it clear was required, since without doubt, clinicians will be using images viewed in the EHR to make diagnostic decisions that affect patient management.

I think there is significant room for comment and improvement in the certification criteria for the image viewing requirement, and perhaps even a need to have separate requirements for the "embedded" approach as opposed to the "linked" approach.

Download and Transmit (Sharing, Interchange).

The suspicion that the viewing objective is distinct is confirmed by looking at the details of the other objective that is related to sharing and interchange, and which replaces the former generic objective for "timely access" and "electronic copy" for patients, though obviously there is overlap given the presence of the word "view".

Specifically, though the new "Provide patients the ability to view online, download, and transmit their health information" objective itself states nothing with regard to imaging per se, the certification criteria for this objective do specifically address imaging:

"We propose to require EHR technology to be capable of enabling images formatted according to the Digital Imaging and Communications in Medicine (DICOM) standard[11] to be downloaded and transmitted to a third party. We believe this specific capability has the potential to empower patients to play a greater role in their own care coordination and could help assist in reducing the amount of redundant and duplicative imaging-oriented tests performed. In fact, the National Institutes of Health has recently funded activities focused on personally controlled sharing of medical images[12] and published a solicitation notice on the same topic.[13]"

Good stuff, though perhaps there is a need for a few more specifics about the means of transmission of the DICOM images (as opposed to the payload being DICOM images), and this is a subject that has been debated recently in response to one of John Halamka's comments on his blog. During the public comment phase, it would probably be a good idea to clarify the role of IHE XDS-I and XDR-I in this respect, since those are certainly the mechanisms that most folks think are appropriate for this "CD replacement with the network transport" for imaging. In particular, the DIRECT mechanism that uses email to transport documents is not likely to scale well for the purpose of image transport, and though the proposed MU2 certification criteria talk about DIRECT, they do so in the context of the summary of care document. More information about DIRECT is available at their home page and their wiki.

I am not sufficiently familiar with the addressing and security-related aspects of DIRECT, which use DNS and LDAP to obtain X.509 certificates for encryption for intended recipients, for example, to know how well this will interact with using XDR-I as a "push" mechanism, nor whether the DIRECT idea of using a "push", as opposed to a "pull" using XDS-I (which has been widely promoted for this use-case up until now), or something in between like the RSNA's Image Share project (which is the one that the NHLBI solicitation referred to in the certification criteria is about), is more appropriate.

One other aspect of this objective (stated in the objective, rather than the certification criteria) is that:

"Transmission can be any means of electronic transmission according to any transport standard(s) (SMTP, FTP, REST, SOAP, etc.). However, the relocation of physical electronic media (for example, USB, CD) does not qualify as transmission although the movement of the information from online to the physical electronic media would be a download."

which I find interesting
  • because it to some extent overlaps with and/or perhaps contradicts the certification criteria which promote specific transport mechanisms, and
  • because it clearly defines that physical media does not satisfy the objective (which of course is the whole point of the exercise, to substitute the network for "sneakernet").
Anyway, arguably the "sharing" objective is much more interesting than the "viewing" objective, seems to expose some inconsistency between the objectives and the certification criteria (if not a division of opinion between the respective drafters), and without question in my opinion will require quite a lot more work on the certification criteria side at least to define how best to select a suitable set of standards to cover the imaging use cases.

In addition to the transport technology, it might also be appropriate to comment on how some of this is actually required to be implemented in terms of the actors involved and the sequence of operations in specific scenarios. For example, if the patient initiates a "transfer" of a DICOM study to a recipient, need it be required that those images then be importable into whatever system that recipient is using, automatically and electronically? Or to put this another way, since the certification requirements apply to the "EHR", what does this mean in terms of a required capability for the recipient's workstation or PACS? Anything ? Or nothing, since they may be out of scope of certification. This does open up a can of worms with respect to standards for "import" workflow (and the sort of issues that IHE addresses in the IRWF profile and the extensions to IRWF that we are working on this cycle).

Again, we should not forget to comment on the fact that "diagnostic quality" is required for the interchange objective, in addition to the viewing objective, since merely specifying DICOM as the payload is not sufficient in this respect, as opposed to, for instance "the original DICOM images as they were interpreted for patient care". Now there's a can of worms!

Retention (Not).

A further element of the proposed measures is:

"Meaningful use does not impose any additional retention requirements on the image."

OK, that seems fine to me too; who the entity of record is responsible for retention, and for how long, are things the ONC considers out of scope. We know that we are not getting a federally funded national archive of patient images any time soon, nor a mandate that there be similar archives at the state level or in the private sector.

Provider-to-Provider Image Transport Missing.

What I cannot see, at least with respect to imaging, is a mechanism for providers to exchange images, without the need to engage the patient to initiate and manage it.

E.g., where is the imaging correlate to the objective for "The EP who transitions their patient to another setting of care or provider of care or refers their patient to another provider of care should provide summary care record for each transition of care or referral."

Should there not be a similar objective:

"The EP who transitions their patient to another setting of care or provider of care or refers their patient to another provider of care should provide images for each transition of care or referral."

I.e., the current proposed objectives talk a lot about access with an EHR, and access by patients, but not so much about the scenarios in which the patient is transferred to other providers who do not have access via the EHR (or any other system) of the previous provider.

This omission seems to disregard the current trend towards using "cloud" based or simple pre-negotiated point-to-point push mechanisms as CD replacement solutions for provider-to-provider communication, without having to bother the patient with the details.

If two PACS talked to each other to satisfy such an objective, if it existed, then presumably they could be "modular certified" for the purpose of MU.


PS. Keith Boone has a nice mapping table of objectives to certification criteria on his blog at "MeaningfulUse Objectives Mapped to the ONC Certification Criteria" and also a "MeaningfulUse Certification Criteria Crosswalk to Standards". One strange thing that his table highlights is that "eligible providers" (EPs) are required to implement the "view online, download and transmit" objective, but "eligible hospitals and CAHs (critical access hospitals)" are not. It is not obvious to me what the reason for this distinction is for this objective, and what the specific implications are for imaging.


Keith Boone said...

David, CMS put View, Download and transmit in twice, once for EPs, and once for hospitals (see the column with §495.6 (l)(8)). Each criterion has slightly different wording and measurement, but is essentially the same.

Adrian Gropper said...

David, A superb review but squeezing the square hole of XDS into the round hole of MU is unwarranted. (The tortured architecture of RSNA Image Share can't be where this is going).

The combination of Direct, OAuth and DICOM provides a much simpler RESTful solution.

Our comment could suggest treating DICOM files as if they were BlueButton files and securing them with OAuth to facilitate single-sign-on and enable delegated access for physician-to-physician transfers. EHRs and Provider Portals will have streaming viewers supplemented by download and OAuth options for both the streaming and DICOM versions of a study. Secure Direct emails can be integrated with OAuth to enable a physician or patient to click a link in the Direct message and launch a streaming viewer or DICOM download.