USDS Discovery Sprint Report

Mission Act:
Community Care

Chris Eldredge Lauryn Fantano Natalie Katesl Rick Lee I Sheri Trivedi Aaron Wieczorek

MAR 2019

TABLE OF CONTENTS

Executive Summary 

Establish a Veteran - Centric Vision l5

Cease Developing the Decision Support Tool 

Establish a Veteran Facing Tool for Eligibility  1

Support the Continuity of Scheduling Veterans? Ongoing Care I 3
Structure Contracts for?Continuity of Veteran Care I 

Partner with Another Federal Network LL

Use FHIR for the Exchange of Electronic Health Records (EHRs) 

Appendix A: In-Depth DST Assessment 

?When Veterans come them to get us
to say yes. Itget Veterans to
yes. That?s customer service.?

Secretary Robert Wilkie, Department of Veterans Affairs State of VA Employee Address

EXECUTIVE SUMMARY

Under the Veterans Access, Choice, and AccountabilityAct of 2014 (Choice Act), Veterans
became eligible to receive care from third?party healthcare providers under the Veterans
Choice Program, which created eligibility criteria centered around the timing and availability
of Department of Veterans Affairs (VA) medical care. Section 101 of the John 5. McCain 
Daniel K. Akaka, and Samuel R. Johnson VA Maintaining Internal Systems and Strengthening
Integrated Outside Networks Act Act) of2018 establishes the new, permanent
Veterans Community Care Program. The Veterans Community Care Program evolves the
legacy Veterans Choice Program substantially by expanding Choice eligibility criteria and
setting forth additional requirements around sharing medical records, establishing regional
medical networks, and maintaining continuity of medical care for Veterans receiving care in
the community. Per the MISSION Act, the VA must implement new legislative and regulatory
requirements by June 6, 2019, one year after the Act was signed into law.

At the request ofAssistant Secretary for Enterprise Integration, Dr. Melissa the United
States Digital Service (USDS) conducted a two?week independent discovery sprint on the

preparations for compliance with Section 101 ofthe MISSION Act, ?The Establishment

of Community Care.? Over the course of the sprint, the USDS team spoke with leadership

from the Office of Community Care, program stakeholders, congressional aides, Medicare
experts, leaders and project managers from the VA Office of Information and Technology (OIT),
practicing physicians at both the Martinsburg and Baltimore VA Medical Centers, and Veterans
who were eligible to receive community care under the Choice Act. Additionally, the team
spent several days embedded onsite with AbleVets, the contractor delivering the Decision
Support Tool (DST), the MISSION Act?s hallmark software for rapidly determining Veteran
eligibility. The team also shadowed eligibility and scheduling workflows with Community Care
clinical staff at VA Medical Centers.

While the USDS team expects the VA to fulfill the legal requirements for Section 101 at a
high level, doing so comes without alignment to a single, Veteran-centric vision for how to
fundamentally transform the VA health system to ensure Veterans have ?easy and reliable
access to care that the Veterans need when they need it.?1 To help guide MISSION Act

1] 

MAR 2019 Draft, Pre?Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 3

efforts ahead of the June 6, 2019 roll-out, the USDS team identified the most critical concerns
observed throughout the sprint. Each concern is outlined below, with corresponding
supporting observations and recommendations outlined in more detail throughout the
report.

- Redirect static eligibility determination away from the physician?s workflow. For
eligibility determination, the intended workflow and audience for the Decision Support
Tool (DST) is flawed and could significantly disrupt VA physicians? ability to see patients.

. Create a Veteran-accessible tool for static eligibility determination. There are clear
regulations outlining eligibility for the Veterans Community Care Program, but the
methods for determining eligibility based on the regulations are ambiguous and will create
confusion around whether Veterans are legally covered to receive care in the community.

. Support the continuity of Veterans? ongoing care when creating?scheduling processes
and guidelines. Frequent changes in administrative scheduling contracts created delays
and confusion in providing Veterans with access to community care.

. Structure new provider contracts to support the continuity of Veterans? ongoing
care. The statutory requirement that VA create networks of health care providers is
technically fulfilled with the use of TriWest to service regions while the VA engages
to contract with other providers, however, Veterans will experience continuity issues
while the VA transitions care between TriWest and new contracted networks. If at all
possible, modify the TriWest contract to allow any active authorizations created prior to
the termination of their contract to remain active. This will allow Veterans to continue
receiving community care from their provider after the new networks are established.

. Utilize national networks already operated by federal services that currently provide
community based care, such as the Department of Defense (D00) or Medicare. This
will minimize disruption to Veteran care while maximizing their network of available
providers.

. Prioritize the exchange of electronic health records (EHRs) using Fast Healthcare
Interoperability Resource (FHIR) APIs, a standard already in use by other parts of
the VA, Medicare, and the health industry as a whole. This will move the VA toward
the meaningful exchange of medical records for Veteran care while also simplifying the
proposed technologies and formats for community care and Department physicians.

MAR 2019 Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 4

MAR 2019 Draft, Pre?Decisional, FOUO

ESTABLISH A VETERAN - CENTRIC VISION

Within the Office of Community Care, stakeholders have roughly
described the MISSION Act as ?a natural extension of current
processes.? To that end, the VA has largely employed a divide?
and?conquer approach within the Veterans Community

Care Program and across OIT, leading to piecemeal updates

and evolutions to existing policies, contracts, and software
efforts. This approach has created missed opportunities to
evolve existing business processes and embrace the spirit
ofthe legislation, which is to fundamentally transform the

VA health system to ensure Veterans can reliably access care
when necessary. Little research has been done in the field to
understand how Veterans, physicians, and clinical staff are
currently providing and receiving care in the community through
the VA before new processes are established. Several individuals
in the Office of Community Care have great intention, however,
USDS hasn?t observed a shared vision for how the office can
holistically optimize the Veteran experience while also fulfilling
the statutory requirements ofthe MISSION Act.

Early research with Veterans highlighted both the need to
streamline the Veteran-facing points of contact while scheduling
appointments and receiving community care. Veterans want

to have more options to select the specific provider they are
being referred to whenever possible. Additionally, the existing
Patient Aligned Care Team (PACT) initiative is well understood
by physicians and clinical staff at VA medical centers, and USDS
heard from multiple people in those roles that they would prefer
that community care be an extension of the PACT initiative.

Instead of the VA saying ?60 to this
doctor?, I?d love to choose my own.
You may want to make your own
decisions. Oh yeah, I would much
rather do my own scheduling. When
I do research, I?m looking for the
reviews from people, then how long
they?ve been in practice, and how
long they?ve been certified.

Air Force Veteran from Hickory, Nil."

There is a misconception that the
primary care provider will co-
manage community care. I will
instruct every one of my PCPs not to
do thisr?fzmc'i Cu! 7 Dept}! .- (1.7, (.1 104

Med/1:11 is:

. Define what the Veterans Community Care Program should be from the Veteran?s
perspective. Senior leadership should fully engage on the ground, following user-
centered design principles to understand the unique experiences and challenges of
Veterans, VA physicians, clinical staff, and community providers, all of whom are crucial to
a successful evolution of community care. Specific members of the Office of Community
Care, as well as the AbleVets contractor, have expressed their strong desire to conduct
user research in the field, and leadership should encourage and support them in these

efforts.

. Minimize new community care efforts pending a holistic Veteran-centric vision. The
Community Care team has identified several low?effort contingency plans for MISSION Act
preparation. USDS recommends only implementing the minimal legally required efforts
for legislative compliance until the VA can articulate a singular community care vision.
Each current should be evaluated against that vision before efforts resume.

MISSION ACT: USDS Community Care Discovery Sprint Report 5

CEASE DEVELOPING THE CURRENT DECISION SUPPORT TOOL

The MISSION Act of 2018 expanded the complexity and number These people are out oftheir

of eligibility criteria for Veterans to receive community care minds. We aren?t housekeepers,
to include phySICIan recommendation, drive times, and
permutations of the existing timing and availability requirements,
among other smaller criteria. Actuarial data furnished by the

doorkeepers, garbage men. The idea
that I?m going to sit and have this

2018 VA Enrollee Health Care Projection Model (EHCPM) indicates conversation [about static eligibility]
that based on the proposed drive time regulations of 30 minutes is a silly concept.

for primary care and 60 minutes for specialty care alone, the

number ofVeterans eligible to receive care in the community is VA Primary Care Physic/an

expected to increase from 685,000 Veterans under the Veterans
Choice Program to 3.7 million Veterans under the MISSION Act.
Anticipating an influx of new eligible Veterans for the Community
Care Program, it?s in the best interests daily clinical
operations forthe department to develop a technological
solution capable of automating the eligibility approval process as
much as is technically and dependably possible.

Much of the data necessary to determine eligibility is currently housed across several legacy
VA systems that don?t interoperate, creating an inefficient and highly manual determination
process. Decision Support Tool (DST), the principal technology in progress forthe MISSION
Act rollout effort, could streamline the eligibility determination by connecting to these legacy
VA systems to gather data on the Veteran and produce a determination. However, following
extensive research around the technology and processes surrounding the development of
DST, and with active development only starting roughly 6 weeks ago, USDS has identified
significant risks surrounding software development timing, integration dependencies, and
usability.

An in-depth assessment of DST may be found in Appendix A.

. Stop development on DST as it is currently implemented. OIT requirements, User
Acceptance Tests (UAT) and Initial Operating Capacity (IOC) processes have allowed
only twelve weeks of actual development time for the software. Code lock for the June
6th release is scheduled to occur April 3rd, three weeks before initial ATO meetings with
OIT leadership. Should any issues be uncovered during UAT or IOC, such as the usability
concerns highlighted below, there will be no time for development modifications. From
an engineering perspective, the current implementation is brittle with the AutoHotkey
feature connecting the Computerized Patient Record System (CPRS) to DST. Minor user
preferences, like having multiple patient tabs open in CPRS or running Chrome in the
background, might completely stall the planned MVP implementation of DST. DST also
requires integration with six legacy VA systems, each with dependencies on other systems
to prioritize DST over their own backlogs. This creates a tremendous impediment to
completing development on time. Given the high level of risk surrounding the timeline,

MAR 2019 Draft, Pre?Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 6

planned engineering implementation and usability concerns, USDS anticipates that DST, if
released into production, will negatively disrupt the physicians workflow and reduce daily
appointment capacity within VHA.

. Redirect development efforts towards a reimagined DST that serves both VA
Clinical Staff and Veterans. The most complex eligibility criteria for clinical staff to
determine using manual workarounds are VA wait times and Veteran driving times. USDS
recommends reducing the DST MVP to only determine these criteria. Using only a user?
provided address and selection, the tool can determine static eligibility according
to wait time, drive time, and whether a VA medical center is located in the Veteran?s
state, covering the majority ofthe expected Veteran eligibility while also simplifying the
technical approach, eliminating all but one internal integration in favor of accessing
existing APIs (h . . and Bing Maps. USDS recommends
building this tool by creating a DST API and working with a bifurcated front end that
serves as a standalone entry point for VA staff to determining Veteran eligibility. Removing
the connections to CPRS provides the flexibility to have anyone within the patient?s
PACT perform this determination ratherthan the referring physician. Additionally, USDS
recommends integrating with the VA.gov ?My dashboard to show a Veteran whether
they?re eligible and allow a Veteran to proceed directly to community care. All of this work
should be achievable by June 6th, and AbleVets can add capabilities in future iterations
after the team conducts user research.

ESTABLISH A VETERAN FACING TOOL FOR ELIGIBILITY

The MISSION Act Veterans Community Care Program eligibility I?m going to do some research
criteria employs data?driven metrics, such as drive time, VA online, maybe canfi'nd a
provider wait time, and availability of VA services. However,
neither the law northe draft regulations specify to an adequate
level of detail how the VA should calculate drive times and wait

community primary care provider]
for longer than three or four months.

times to make these determinations. For example, calculating I haven?t had a steady PCPfor
drive time depends on an information system that uses mapping ayear and a haIf,just the Nurse
data of roadways, speed limits and other metadata'to compute Practitioner ofthe Month.

an optimal route and estimate the time reqUired. Differences

in specific technical implementations for gathering this type A,

of data will likely result in different estimates. Additionally, the
law and regulations do not outline how traffic congestion or
other considerations may impact an eligibility decision. These
nuanced details can affect whether or not a Veteran is eligible for
community care, and may lead to the public perception that the
criteria are arbitrary, inconsistent, or unfair.

This variability becomes particularly acute when a Veteran uses the publicly available
eligibility criteria specified in the legislation to estimate their own eligibility for services. Of the
primary care providers USDS interviewed, several noted that community care is not currently

MAR 2019 Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 7

offered to eligible Veterans unless they ask specifically for it during their appointment.
Therefore, it is critical for the Veteran to be aware ofthe Veterans Community Care Program,
and for the VA to mitigate public misconception by exposing a tool capability of determining
eligibility to Veterans.

. Establish clear, deterministic access standards to ensure that two parties don?t
arrive at different conclusions when determining a Veteran?s eligibility for
community care. In the upcoming regulations, the standards are based on drive time and
wait time, but there is no specific guideline for formulating drive or wait times, and no
recommendation on what data source to use.

. Provide static eligibility on the VA.gov dashboard. USDS recommends
exposing Veterans? potential static eligibility according to their current address and a
selection of specialty on their dashboard at VA.gov, creating a natural extension of
the actions currently available on this singular Veteran portal.

. Develop Veteran-centered guidelines to manage discrepancies in eligibility due to
data variation. Inevitably, there will be cases where a Veteran?s eligibility for Community
Care is variable due to uncontrollable fluctuations around the output for wait time and
drive time. For example, wait time for a cardiology appointment may teeter between
58 and 61 days within a given month, or drive time may vary between 25 to 35 minutes
depending on road closures. In these cases, USDS recommends the VA should adopt a
people-centered approach in the field, providing the most liberal interpretation possible
for eligibility criteria.

SUPPORT CONTINUITY FOR SCHEDULING ONGOING CARE

In 2013, the VA awarded contracts to two third-party 1 think it would be more effectivefor
administrators (TPAs), Health Net Federal Services (Eastern me to call the clinic myself, have my
region) and TriWest (Western region), to administer the Veterans
Choice Program (VCP) across the VA. Due to issues with serving
Veterans, as well as late payments to providers with service
agreements, the VA declined to exercise the option to continue
the Health Net Federal Services contract as of September

30, 2018 and expanded the TriWest contract to administer all
regions for both the Veterans Choice and Patient-Centered
Community Care (PC3) Programs. Veterans and VA community
care scheduling support staff alike highlighted that TriWest?s

schedule in front of me, rather than
go through a third partyYou should be assigned to one
person, and if you?re having

track record in scheduling appointments was mixed at best. problems in the non-VA care,
TriWest?s service level agreement (SLA) requires them to schedule you should be able to contact one
appointments for Veterans within 3-5 days. Clinical staff at person and that person should be
two VAMC locations reported that the TPA is often not meeting
that SLA, increasing the time the Veteran must wait to have an empowered to help you.

A m: 1/ ere/2:727 1?77)! 7: it? 1/21?) my"; 3'

MAR 2019 I Draft, Pre?Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 8

appointment scheduled and requiring VA staff to leverage homegrown workarounds for follow
ing up with Veterans and providers on their own.

Ultimately, the Office of Community Care plans to transfer appointment scheduling away from
TPAs and back into local medical centers, and based on observations in the field, USDS agrees
this is in the Veterans? best interest. However, the introduction of scheduling through TriWest
at additional VA locations earlier this year is contrary to the long term community care plan
and created shortsighted disruption for both Veterans and clinical staff as schedulers on the
ground were required to learn new software and processes.

. Create a thoughtful, seamless scheduling transition for Veterans moving from
Veterans Choice to Veterans Community Care. There have been, and will be, several
changes to the way?Veterans receive care in the community. Last September, the Health
Net contract ended. At the end of January, some facilities opted to offer scheduling
through TriWest. In June, the new Community Care Network established by the MISSION
Act will move scheduling back to the medical centers. These changes affect Veterans on
the ground.

. Complete a service design study to establish the best methods for managing and
scaling scheduling staff at VA medical centers while providing consistency for
Veterans. USDS learned in the field that scheduling teams had shifted their methods over
short periods oftime in attempts to manage their backlogs of consult requests awaiting
scheduling, including dividing an alphabetical list of Veteran last names, dividing consults
by medical specialty, and even ?dogpiling? several schedulers on one consult. Veterans
expressed that if their appointments would be scheduled for them, they would like to be
able to consistently work with the same scheduler.

. Allow options for Veterans to schedule their own appointments. Veterans reported
that regardless of being scheduled by the VA, TriWest, or Health Net, they were not really
offered a ?choice? of providers to see, but were rather told which provider they would
see. In some cases, the Veteran was given the name of a provider and told to make an
appointment on their own. They reported that they would prefer to research providers
and schedule appointments themselves, just as the general public does.

MAR 2019 I Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 9

STRUCTURE CONTRACTS FOR CONTINUITY OF VETERAN CARE

The MISSION Act requires that the VA establish networks of
healthcare providers to serve Veterans who are eligible for
community care. VA originally contracted with TriWest and Health

I was sent to a dentist down the
street from my house, who did the
preliminary X-rays and checkups.

Net Federal to create two massive provider networks spanning
the country. After the Health Net contract was allowed to lapse
due to poor service and nonpayment to providers, TriWest took
overthe region Health Net had previously covered, creating a
lapse in Veteran care due to unfinished authorizations left hanging
by the Health Net contract. The VA has since divided the country
into six regions for the new Community Care Network and is
awarding a different contract for each.

I got a call about a week later
saying, ?Due to non-payment by
the VA, we?re not accepting VA
patients anymore.? It took about

a month to straighten it The
[new] dentist requested approval
from the VA, but the VA kept
sending the authorization to the
original surgeon! I haven?t had the
appointment for the actual service

There are many areas of the MISSION Act statute and regulations
where Veterans could lose access to a needed doctor or service

during bureaucratic transitions, such as changes when:

- A medical service lines are deemed under quality and yet.

published in the Federal Register, potentially moving Veterans

. . Army Veteran from Neva-twrn, (I
from VA Medical Centers to Community Care, 

. VA Medical service lines become eligible for community care
based on how the VA implements quality standards, and

- When providers that are in network based on contractual transitions become out of
network when the contract for the region changes.

Poor implementation and administration ofthe contracts implementing the statute and
regulations pose real, life-threatening risks to Veterans. They may transition from one provider
to another as contracts are awarded, or lose access to their providers as they come in and out
of network during these regional transitions.

. Include language in all contracts to ensure providers remain in-network regardless of
contracting changes when a Veteran is in the middle oftreatment. USDS recommends
modifying the TriWest contract to allow any active authorizations created prior to the
termination of their contract to remain active, as well as including similar language in any
future provider network contract. Alternatively or additionally, make sure new contracts
that take over a Veterans care during an episode of care could cover the Veteran?s care
with their previous provider during that episode of care that spans contract periods. USDS
understands that in the past the Office ofthe General Counsel has blocked similar
efforts and recommends that senior leadership in VHA and OGC work together to figure
out the best path forward for Veterans.

. Ensure changes in eligible service lines are simple, communicated openly, and don?t
impact Veterans in the middle oftreatment. Minimize changes in eligible service lines

MAR 2019 Draft, Pre-Decisional, FOUO

MISSION ACT: USDS Community Care Discovery Sprint Report I 10

MAR 2019 Draft, Pre-Decisional, FOUO

to reduce the risk of disruption to Veteran care. During the Community Care team?s week-
long Face?to-Face Meeting in early February, the team recognized this fluctuation as a risk
but dismissed it as solvable by ?communicating to the Veteran that they have to switch

providers.? This is unacceptable and will harm the most vulnerable Veterans populations.

PARTNER WITH ANOTHER FEDERAL NETWORK

Numerous stakeholders, including those at the VA and on the Hill,
emphasized the importance of various provider network related
efficiency and consolidation efforts mandated by the MISSION
Act. The Act itself aligns VA community care programs with
Medicare payment rates and quality standards and also requires
that VA consult with Medicare and other entities on access
standards. It defines eligible providers by their participation in
Medicare or other federal programs such as those administered
by the Department of Defense.

Currently only one of the six new contracts needed to establish
the consolidated national network is operational. Two

are currently in protest and three others have yet to be
awarded. While the VA has the appropriate contingency plan

in place - to extend their current provider TriWest?s contract
through September 2020 - USDS believes that the additional
administrative burden ofcontinuing to pursue new contracts
arrangements outweigh their benefit. Based on past contracts,
outsourcing care coordination and provider networks introduced
new systems and players into the ecosystem, increasing the
potential for miscommunication and confusion from Veterans
determining appropriate point of contacts, without mitigating
any of the administrative burden from the VA.

. Cease efforts to develop robust payer infrastructure and
administer Community Care by exploring partnering with
another federal payer already serving Veterans. Aligning
with either Medicare or DoD?s TRICARE will prevent the VA
from creating new redundant and less efficient bureaucracy
that fails to capture the lessons learned over multiple
decades of operating existing government payer programs.
The especially has already served these Veterans,
and 48% of all Veterans are currently eligible for Medicare

according to data on 

. Move quickly to conduct market research to determine
the best federal program to partner with. USDS believes

I feel like I?m getting someone in
their field. You wonder whether VA
has the best doctors. When you go
outside the field you feel like you are
getting someone where that?s their
specialty.

Navy Veteran FL

It took forever to get an appointment
at the VA. I waited about 3 months
and dealt with the pain. Once I got
an appointment, it took another two
weeks for them to let me know I had
kidney stones and that they were
referring me to civilian health care to
handle it. The physician?s assistant
contacted me and said someone
would refer me out, and that a clinic
near my house would contact me.
After a few weeks, someone did
contact me, and they scheduled me
for sound blasting. I had contact
from the VA probably two weeks
after the procedure to schedule a
follow?up. They said it would be
three months for that.

MISSION ACT: USDS Community Care Discovery Sprint Report 

MAR 2019 Draft, Pre-Decisional, FOUO

that this approach is achievable under the letter of the MISSION Act (including the
requirement for a competitive bid process) via the Economy Act of 1932 and aligns with
the intent ofthe legislation to create an efficient, streamlined experience for Veterans
and the American taxpayer. VA should conduct market research on various public sector
providers and review the research across public and private sectors to determine the
best program to partner with. Then, the VA should focus and fast?track any current pilot
efforts in development with 000 TRICARE and Medicare to replace their current network

contracts.

Shift resources currently managing the VA network towards concierge services

that improve the Veteran community care experience. Based on conversations

with Veterans, providers, and stakeholders, there are clear gaps in communication and
continuity ofcare that must be addressed for the Community Care Program to successfully
serve Veterans, the most important ofwhich have been addressed elsewhere in this
report. A high-level of localization also means Veteran experiences can vary dramatically. 
For the Community Care program to succeed at scale, the VA must focus on improving
their services as a provider, not building another public sector national network.

USE FHIR FOR THE EXCHANGE OF ELECTRONIC HEALTH RECORDS (EHRS)

Section 101 ofthe MISSION Act requires the Secretary ensure the
?establishment of a mechanism to receive medical records from
non-Department providers.? The Community Care team currently
plans to meet this requirement through two mechanisms: the
use of Referral Manager (HSRM), a commercial
off?the-shelf software product, and through the manual, one-

off exchange of records in any format that a non?Department
provider prefers, such as fax, or secure mail.

USDS thinks based on its user research that it is unlikely that non?
Department providers will choose to adopt a VA?specific platform
for sharing medical data. As a result the majority of record
sharing will continue to require manual exchange.

. Prioritize the exchange of records using Fast Healthcare
Interoperability Resources (FHIR) APls, a standard
already in use in other parts of the VA, Medicare, and the
health industry as a whole. FHIR APIs would allow the VA to
integrate non-department records directly into a Veteran?s
existing file, instead ofas an attached image or note. FHIR
is already being used by the Project Special Forces team to
allow Veterans to connect their VA medical records to Apple
HealthKit. Some Office of Connected Care mobile apps are

I don?t know how they hand off
records for a consult. I?d like them

to have my history so they could
understand my I?m in the
middle of that right now getting my
Navy records.

a, .. 
XVIIUI V?jte.? 1137?. Hf,?

It seems like the VA doesn?t read their
own notes.

MISSION ACT: USDS Community Care Discovery Sprint Report 12

also using FHIR. The Centers for Medicare and Medicaid Services and the Office of the
National Coordinator have both recently proposed legislation backing the standard.
Cerner, as well as Amazon, Google, IBM, Microsoft, Oracle, and Salesforce among others
have also pledged to use FHIR for medical data exchangel We strongly recommend
investing in this infrastructure to develop a similarly streamlined process for medical
record exchange between the VA and community care providers instead of investing in
nascent technologies.

. While medical providers come up-to-date with the FHIR standard, continue using
current methods, including fax and secure email, to share medical records. Minimize
investment in new technologies that aim to exchange medical records in other ways.

 

MAR 2019 Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report I 13

APPENDIX A: DST ASSESSMENT

Under the Veterans Access, Choice, and Accountability Act of 2014, Veterans became eligible
for Community Care through a framework of eligibility criteria centered around the timing and
availability of medical care. The MISSION Act of 2018 expanded the complexity and number

of eligibility criteria to include physician recommendation and permutations ofthe existing
timing and availability requirements, among other smaller criteria. Because of this increase,
actuarial data furnished by a VA regulatory impact analysis indicates that the number

of Veterans eligible to receive care in the community based on the proposed drive time
regulations of 30 minutes for primary care and 60 minutes for specialty care alone is expected
to increase from 685,000 Veterans under the Veterans Choice Program to 3.7 million Veterans
under the MISSION Act. Anticipating an influx of new eligible Veterans forthe Community

Care Program, it?s in the best interests ofVA clinical staff for the department to develop a
technological solution capable of automating the eligibility approval process as much as is
technically and dependably possible.

Much of the data necessary to determine eligibility is currently housed across several legacy
VA systems that don?t interoperate, creating an inefficient and highly manual determination
process. Once clinical staff determines a Veteran is eligible, the staff must record that status
in the VA Computerized Patient Record System (CPRS). CPRS is a legacy system based on
VistA that presents its own update and modification challenges. Decision Support Tool (DST),
the principal technology in progress for the MISSION Act rollout effort, could streamline the
eligibility determination by connecting to these legacy VA systems to gather data on the
Veteran and produce a determination. However, following extensive research around the
technology and processes surrounding the development of DST, and with active development
only starting roughly 6 weeks ago, USDS has identified significant risks surrounding software
development timing, integration dependencies, and usability.

Software Development Timing

AbleVets, the software development contractor developing DST, is operating in a pseudo-
agile manner that shoehorns agile development sprints into an overarching VA waterfall.
OIT prescribed a four?month requirement generation process. AbleVets was only recently
able to begin active development of DST, and because of user acceptance testing, ATO, 508
approval, and other VA?driven procedures, they?ll be required to complete their production
release on April 3rd. The condensed timeline in which active development work is occurring
creates a significant risk that the engineering team won?t have access to all of the necessary
VA development and staging endpoints by that date. Because DST obtains information via API
calls into data services, it?s plausible that they will encounter integration problems that may
lead to DST system hangs, sluggish performance, or unrecoverable errors.

Key pre-release milestones are happening too late for AbleVets to be able to effectively
react to problems. Software development will have concluded roughly a month before the
first scheduled ATO meeting with Bill James on April 26th. VA DST project management has
casually discussed pursuing an IATT, but this has not been actively pursued. IOC testing, the
initial training and release to selected sites and the first testing done with actual users, is

MAR 2019 Draft, Pre?Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report I 14

scheduled soon thereafter between May 2nd and May 22nd. lfthis crucial first set of testers
find serious usability issues, or find Department designed the wrong product altogether, it will
be too late to pivot before the production release on May 23rd.

While AbleVets is under pressure to deliver DST on an expedited schedule, USDS finds that the
development team is motivated and capable of delivering a product that meets requirements
with a high standard of quality. USDS reviewed the source code, system architecture, agile
practices, engineering process and plans for production hosting and found no major issues.
Any delays in delivery are likely to result from delays in connecting and integrating to backend
APIs that provide information necessary for DST to function.

Integration Dependencies

There are two distinct parts to the planned DST rollout. First, there is a web application to

be deployed into the VA Enterprise Cloud. Second, there is a new AutoHotkey package to

be deployed on each workstation throughout all VA medical facilities that will provide an
integration between CPRS and the DST web application. The DST web application is not
intended to be used as a standalone application, and can only function as designed when
invoked from CPRS. This coupling binds the success of the DST web application to the success
ofthe AutoHotkey package and leaves no room for alternate user workflows.

From a technical perspective, DST is primarily a dashboard that aggregates information about
the Veteran from other systems so that clinical staff can make an eligibility determination.
While the business logic in DST is straightforward, the complexity and number of integrations
currently required to determine eligibility relies on complex alignment with other VA software
development teams in order to align priorities for DST release. Of the six integrations needed
for DST that require an interface agreement (MVI, PPMS, CDW, SEOC, and only
used for authenticating the physician, has been completed. Without all of the VA

service integrations functional, DST will be unable to provide any actual value in determining
eligibility. VA OIT has not yet filed most requests for access to these integrations. The lack

of completed interface agreements (IA) has already significantly hindered AbleVets from
completing sprint work. Despite a schedule indicating the requests are on track, the missing
lA agreements will realistically endanger the timeline going forward as government interface
agreements are the most often delayed. AbleVets tends to report Integration work as ?Green?
until it actively blocks development, although both USDS and AbleVets anticipate that the lack
of many of the necessary integrations will block planned development.

In addition, it is unlikely that the staging environments for the integrations are representative
oftheir production counterparts in their scale or complexity. In fact some data systems

such as the Corporate Data Warehouse (CDW) have no pre-production environment at all.
The VA requires an ATO for an application to serve production traffic, and the development
team will likely not have production access to the endpoints required to serve end users.

In order to populate the Veteran information and calculate an eligibility decision, DST calls
the various services in a cascading, serialized fashion. The continued inability to test these

MAR 2019 Draft, Pre?Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report I 15

service integrations with regards to response times and error rates at scale could lead to
severe latency, errors, or system crashes. AbleVets must be able to load test and stress test the
system in order to understand the bottlenecks.

Clinical staff requires the data they use to legally determine eligibility on behalf of the Veteran
to be trustworthy. Several ofthe planned integrations have known data quality issues,
including duplicative, missing, and out?of?date data, all of which could prevent the Veteran
from receiving an accurate eligibility determination. USDS has no recommendation to alleviate
these data quality issues, however, as it doesn?t appear that the VA has viable alternative data
sources.

Usability

The DST implementation depends on a screen scraping tool called AutoHotkey to determine
the appropriate time for DST to become active based on particular CPRS fields appearing

at the correct time and place. This implementation allows for no variability in the way users
may have configured their individual work systems, and the development team has no insight
into what will happen ifthe required conditions are not met. Furthermore, when AutoHotkey
monitors the user?s screen and creates a modal pop?up window used to calculate eligibility via
DST, the prompt actively blocks the user?s CPRS window, preventing them from continuing.
The modal initially presents a loading screen with no call to action while determining if further
action is necessary. It is unclear how long this initial step takes, but the AbleVets software
team estimated about 30 seconds. Due to the previously mentioned issues with testing viable
data, the development team is unable to test the possible behavior before production release.
More alarmingly, once the tool is rolled out to all locations on May 23rd, staff will begin to see
these pop-ups with no ability to use them until the MISSION Act goes into effect on June 6th.
USDS has no insight into how this workflow conflict will affect them during that period.

CPRS has usability issues that will only be exacerbated by the current implementation of DST.
Because AutoHotkey is looking for particular criteria, such as text in a dialog screen, or the
state of input controls, in order to activate DST, USDS suspects that clinical staff would only
be able to view one Veteran in CPRS at a time, which is wildly different from the way the team
observed staff using CPRS in the field. in addition, existing CPRS add?ons like Consult Toolbox
often block information that medical personnel need to view when completing toolbox forms.
As a workaround, they open multiple instances ofthe software on their machine and copy/
paste between them. DST will launch a new Google Chrome browser window each time

DST is triggered. It is unclear to USDS if all VA Medical personnel have the Chrome browser
installed. It?s also unclear what will happen in a perfectly plausible situation where the clinical
staff member has Chrome open in the background in order to check patient or provider
information. Various permutations such as these may lead to unpredictable behavior and
result in failure modes that result in confusion and frustration.

One of the most alarming usability issues for the DST MVP release for June 6 is the assumption
that eligibility would only be determined by the physician at the time the consult request

MAR 2019 I Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report I 16

is created in CPRS. Given the usability and responsiveness issues discussed above, adding
this eligibility work to the already time-constrained physician in a worst case could increase
each appointment by an estimated 5-10 minutes, forcing physicians to see approximately
three fewer Veterans each day, and ultimately decreasing the nationwide capacity by
approximately 75,000 appointments daily. This degradation goes against the spirit ofthe
MISSION Act to improve the Veterans experience and quality of care.

MAR 2019 I Draft, Pre-Decisional, FOUO MISSION ACT: USDS Community Care Discovery Sprint Report 17