Follow-up of Non Critical Actionable Findings - Proposal

From IHE Wiki
Jump to: navigation, search


1. Proposed Workitem: Completion of FUNC (Follow-Up of Non-Critical Actionable Findings)

  • Proposal Editor: Teri Sippel Schmidt/Marquette, Steve Langer/ Mayo Clinic, Tessa Cook, MD PhD/ Univ of Pennsylvania, Kinson Ho/Change
  • Proposal Contributors:
  • Editor: Teri Sippel Schmidt/Marquette and Kinson Ho/Change
  • Contributors:
  • Domain: Radiology

2. The Problem

The FUNC profile was underestimated in 2016. As a result, Results Distribution (RD- HL7 v2.5.1 ORU message) was separated from the FUNC profile and was completed 2017. Despite a valiant effort, the FUNC profile itself remains incomplete, but the problem it addresses are no less relevant or important than when the workitem was originally selected.

Current state of FUNC

The FUNC supplement text is currently ~100 pages in length with RD removed.
FUNC Volume 1: The use cases and background material are fairly well developed and have been reviewed several times by the IHE Rad Technical Committee. These will need to be re-reviewed in light of RD having been separated, but should ultimately be decreased rather than increased.
FUNC Volume 2: The basic architecture has been determined; after intense and lengthy discussions; the "Alert Reporter" actor is effectively also a FHIR server. Prior assumptions about keeping all intra-radiology department transactions as HL7 v2.x messages should be reopened for discussion as FHIR has evolved significantly in the last two years.

FUNC clinical problem statement: (Taken from Volume 1)

In a University of Pennsylvania (HUP) two-year study presented at RSNA 2016 looking at recommendations for follow-up based on abdominal imaging, researchers found that 14% of such exams recommended follow-up imaging for non-critical, actionable findings (Cook2016). A substudy looking at six months of these recommendations noted that there was a 4:1 ratio of in-system to out-of-system physicians ordering abdominal imaging performed at this large academic medical center. Combining these results, it is estimated that, at least for abdominal imaging, approximately 3% of exams contain non-critical, actionable findings that need to be communicated outside a single health system. As such, these patients are automatically at a much higher risk of not having their findings communicated (because the communication may currently rely on faxing printed results or calling ordering physicians’ offices). Using the large, academic center in the study above as an example, this translates to approximately 30,000 at-risk patients every year at HUP.

The original FUNC 2016 Profile Proposal can be found here: Critical_Finding_Follow-up_and_Communication

The current FUNC supplement under development can be found here: Draft of FUNC Supplement for Public Comment

3. Key Use Case

This profile focuses on alert notifications between affiliated facilities, for example, from a hospital to a referring provider group, or within a Canadian provincial healthcare domain, and between known enterprises, for example, from the Medical College of Wisconsin and the University of Wisconsin, where there is significant patient population overlap and exchange of patient information. This profile is not intended to cover notifications sent between distinct unaffiliated facilities.

Specifically, FUNC addresses communication and feedback for ACR Category 3 ("non-critical actionable") findings. ACR Category 1 and 2 (urgent and emergent) findings require immediate human intervention (i.e., a phone call to the physician caring for the patient) and are not addressed by this profile.

The six use cases currently defined in Volume 1 are:

X.4.2 Use Cases: Follow-Up Alerts of Non-Critical Actionable Findings
X.4.2.1 Use Case 1: Simple case of follow-up within a single healthcare system
X.4.2.2 Use Case 2: Multiple providers to be notified
X.4.2.3 Use Case 3: Affiliated healthcare systems
X.4.2.4 Use Case 4: Multiple alerts within a plan
X.4.2.5 Use Case 5: Follow-up rejected by provider
X.4.2.6 Use Case 6: Expiration of follow-up alert plan

4. Standards and Systems

The transaction within the Radiology Department, to set up the alert notification, is currently an HL7 v2.5.1. order message (ORM). (RAD-Y2 and RAD-Y5) which needs to be migrated to FHIR v4.

The transactions to communicate this alert throughout the enterprise are FHIR CommunicationRequest and Communication Resources. (RAD-Y3 and RAD-Y4)

The Alert Report actor does not yet exist, at least en masse, using standards in the real world today. (note: the existing supplement has the Alert Reporter as all-FHIR based transactions already.)

The FUNC Actor Transaction diagram is as follows. The Report Manager and RAD-Y1 (RAD-128) transaction have been moved to the Results Distribution (RD) profile.

FUNC.AT diagram.png

5. Technical Approach

See current supplement at: FUNC Profile

Existing actors

  • See Volume 1: Report Manager

New actors

  • See Volume 1:
  • Follow up Source
  • Alert Reporter - new to Radiology, from mACM
  • Alert Aggregator - new to Radiology, from mACM

Existing transactions

  • See Volume 1:
  • RD RAD-128 (Send Imaging Results)

New transactions (standards used)

  • See Volume 1:
  • RAD-Y3 Send Followup Alert (FHIR)
  • RAD-Y4 Acknowledge Followup Alert (FHIR)
  • RAD-Y2 Initiate Followup Alert Plan (HL7v2 ORM or FHIR)
  • RAD-Y5 Close Followup Alert Plan (HL7v2 ORM or FHIR)

Impact on existing integration profiles

  • none

New integration profiles needed

<Indicate what new profile(s) might need to be created.>

Breakdown of tasks

Transactions

  • Trigger transaction: (MinUE)
  • RAD-128 Send Result - complete, may need a CP, but not aware of any
  • assume done
  • Radiology department transactions: (MinUE)
  • RAD-Y2 Initiate Followup Alert Plan (decision: FHIR v4)
  • SP effort 2
  • SP complexity 2
  • SP uncertainty 2
  • RAD-Y5 Close Followup Alert Plan (decision: FHIR v4)
  • SP effort 2
  • makes assumption that RAD-Y5 is very similar to RAD-Y2
  • makes assumption that semantics exist in HL7 v2 ORM message in current draft FUNC supplement
  • Cross enterprise transactions: (MaxUE)
  • RAD-Y3 Send Followup Alert (FHIR v4 w/ mACM)
  • SP effort 2
  • SP complexity 2 (cross enterprise messaging -authorization)
  • SP uncertainty 2 (specifically going to v4 and mACM yet to be re-published with v4 updates)
  • RAD-Y4 Acknowledge Followup Alert (FHIR v4)
  • SP effort 1
  • SP complexity 1


Profile

  • We estimate that FUNC Volume 1 development (writing) is 60% complete to get to Public Comment.
  • SP effort 3
  • SP complexity 1
  • SP uncertainty 0
  • One primary Use Case with variants
  • SP effort 2 (MinUE)
  • SP complexity 3 (MinUE -2; Max UE-3)
  • Options
  • none


Topics/Debates

  • how close to follow mACM? (MinUE)
  • SP uncertainty 1
  • define boundaries of robustness (ie., fallback mechanisms; set up conscious boundary) (MinUE)
  • SP uncertainty 1
  • defining scope of "cross-enterprise" (MaxUE)
  • SP uncertainty 1
  • potential new use cases (MaxUE)
  • SP uncertainty 2


Estimates for additional time and effort required to complete FUNC include:

  • editor: 8 h to clean up Volume 1 and re-align
  • editor: 20 h to clean up Volume 2 and re-align (assume moving v2 ORM to FHIR)
  • TC: 8 more TC sessions 2h each to re-review Vol 1 and complete Vol 2 transactions (1- ORM?, 3- FHIR); vote for PC (16 HOURS TC)
  • editor: 3 h to clean up Volume 1 and re-align
  • editor: 15 h to clean up Volume 2 and re-align
  • doc specialist: 5 h to publish for Public Comment
  • TC: 6 h each member to review independently and submit comments for PC
  • editors: 5 h to clean up and organize PC comments
  • TC: 12 h of PC comment review (qty 6 - 2h sessions); vote for approval to TI
  • editors: 20 h of PC comment clean up
  • TC or individually: re-review supplement to vote for TI: 4 hours
  • doc specialist: 5 h to to publish for Trial Implementation

Summarized TC committee efforts:

  • To get to publish for Public Comment: 16 hours
  • To review Public Comments: 12 hours
  • To get to vote to publish for Trial Implementation (final review): 4 hours

For MINIMUM USEFUL EFFORT we propose the following: aka: Scope Management

Complete RAD-Y2 and RAD-Y5 to get the communication of the Actionable Finding out of the Radiology Department to the EMR (i.e, the Alert Manager). The methods by which the Alert Manager completes the communication to the end user would be undefined (in other words: RAD-Y3 and RAD-Y4 would not be included). The analogy is the Radiology Department sending Certified Mail to a healthcare system and receiving a signature that the envelope was received. Hopefully, we could do one step better than a simple signature "ack" and have the "ack" defined to be different levels of acceptance other than just "received", but those could simply be an enumerated value set of response codes.

6. Support & Resources

  • Need to continue to have support from FHIR knowledgeable developers (e.g.., Elliot Silver).

7. Risks

Challenges during the prior work included underestimation of the level of complexity, scope, global differences, as well as technical challenges in adopting a new and evolving technology (FHIR).

  • Need to decide whether or not to introduce FHIR server into use cases/architecture; go contained resources everywhere; or a hybrid
  • Scope creep beyond current A/T diagram
  • FHIR STU-3 is currently frozen. FHIR DSTU-4 is ??. The re-write between DSTU-2 and STU-3 was nearly 100%.
  • FHIR Communication and Communication Request is a FHIR Maturity Level 2.
  • Current existing products may not use FHIR, but probably do not use HL7 v2 new message either.
  • Very specific technical issues such as differences of agreement on Contained Resources.

8. Open Issues

  • would need committee decision to move elusive HL7 v2 ORM-ish message to FHIR transaction

9. Tech Cmte Evaluation

Technical Evaluation: Sept 7, 2018

  • MaxUE = 30 SP 45%
  • MinUE = 18 SP 30%


Effort Evaluation (as a % of Tech Cmte Bandwidth):

  • xx% to complete through Trial Implementation publication for FUNC
  • MUE xx% - x hours at each of 3 TC meetings
  • Plan:

Candidate Editors:

Teri Sippel Schmidt and Kinson Ho