Case study

After the software is installed

Epic's software was installed and still failing at the accounts I inherited. Owning what happened after go-live meant root-cause analysis, re-implementation, and writing Caché plug-ins to close the gaps the base product left — then building the escalation path so regulatory changes and critical bugs stopped arriving as surprises.

Public MyChart app screenshot
MyChart — Epic's patient portal, and one of the two products I deployed for the 514-bed research hospital. EpicCare Link, its portal for referring providers, was the other.

A cancellation risk turned into a reference account, a 134-year-old health system taken online, and millions in federal incentives protected.

Installation is where most vendors declare victory. It is where the actual problems start: workflows that don't match how the department works, quality measures that won't report, and federal incentive money riding on both.

One 477-bed regional health organization had accumulated an outpatient issue list long enough that the account was a cancellation risk. Meanwhile PQRS and the other programs those incentives depended on changed on Washington's schedule, not the deployment's. There was no defined path for getting a regulatory change or a critical bug in front of the right person. 2

Treat post-install as the real job rather than the warranty period. Work the issue list to root cause instead of triaging symptoms, re-implement what was configured wrong, and write custom code where the base product genuinely didn't reach. Then make the escalation path explicit — from issue discovery and triage through customer notification — so the next regulatory change had a route instead of a scramble.

The accounts, and what came of each
Org profileWhat I ownedResult
477-bed regional health organization, an outpatient account at cancellation risk The inherited outpatient issue list: root-cause analysis, re-implementation, and Caché plug-ins where the base product could not reach 95% of the issue list resolved; the account became a reference
514-bed research hospital and health system, 134 years old The MyChart web and mobile portal and EpicCare Link interoperability deployment The health system went online
Eight client organizations — two outpatient software, six patient web-portal Post-install success, and co-lead for CMS PQRS reporting across them Kept on federal incentive programs worth millions in payments
100+ hospitals and organizations The hospital-facing curriculum and webinar library, and a monthly policy-and-IT call Quality-measurement requirements met without a support case

Deployed the MyChart web and mobile portal and EpicCare Link interoperability tools for a 514-bed research hospital and health system, and delivered plug-ins in Caché for the hospital, department, and individual physician needs the base configuration couldn't meet.

Two of the plug-ins I can still name: a checklist built into the visit workflow that gated Meaningful Use completion before an encounter could close, and automatic immunization scheduling driven by CDC recommendations. The first came out of reading Gawande's Checklist Manifesto. 12

1 Software installed and signed off 2 Issue list accumulates against real department workflows 3 Root-cause analysis instead of symptom triage 4 Re-implementation, or a Caché plug-in where the base product can't reach 5 Escalation path routes regulatory changes and critical bugs to a named owner 6 Quality measures report; incentive payments land

The PQRS escalation path became the model for handling regulatory updates and critical bugs, and I was recruited into internal area leadership four months in on the strength of the customer work.

Regulatory change Critical bug × a defined route Circulates no defined route Whoever happened to notice if anyone did
Before the escalation path, at Epic in 2013. A regulatory change and a critical bug entered the same way — by circulating until somebody happened to own them. Nothing here is broken in a way a status report would show: every step has someone touching it and no step has anyone accountable for it. The diagram below is the same two triggers after the route existed.
CMS changes the program a critical bug surfaces Discovery and triage Scope who is affected Named owner not a queue Customer notification before the deadline, not after
The PQRS escalation path, drawn from the route I built at Epic in 2013. Its whole purpose was to remove the step where a regulatory change or a critical bug circulated until somebody happened to own it — every trigger enters one route and terminates in a named person with a notification obligation. Every escalation design on this site descends from it.

The software was installed, signed off, and live — and still not doing the job, because nobody owned the period after go-live. Systems I have built since treat deployment as the beginning of the obligation, and the escalation path I wrote for PQRS is the direct ancestor of the escalation matrix in the AI governance work: define the trigger, the owner, and the notification before the situation arrives.

1 MyChart and EpicCare Link

MyChart is Epic's patient portal; EpicCare Link is its portal for referring and affiliated providers.

Product context for two of the implementations described here.

Epic Systems

2 What PQRS was

PQRS was a CMS quality-reporting program tied to federal incentive payments for eligible professionals.

The CMS fact sheet, dated inside the period described here, says PQRS used incentive payments to encourage eligible professionals to report specific quality measures. The program landing page this used to cite has since 404'd — PQRS was folded into MIPS after 2016.

CMS, “Physician Quality Reporting System (PQRS) Overview” (Aug 2013)

The organizations remain unnamed, so the size, age, 95% issue-list, and cancellation-risk claims come from my records. Contemporaneous documents support the PQRS role, escalation path, 100+ organization call, and move into leadership.