Internal document reveals more frequent computer problems at Spokane VA than previously known

Jul. 20—WASHINGTON — On the morning of July 13, health care providers at Mann-Grandstaff VA Medical Center in Spokane were greeted by a familiar message in their inboxes.

Because of a "significant degradation" in the electronic health record system they rely on to do their jobs, the email said, clinicians who couldn't use the computer system should follow a "downtime contingency plan."

After more than 18 months serving as de facto beta testers for the troubled system, for which the Department of Veterans Affairs is paying Cerner Corp. at least $10 billion, VA employees in Spokane knew exactly what the email meant. Instead of using the computer system to track patient information and order prescriptions and follow-up care, they would need to document everything on paper.

"It basically just shuts everything down to a crawl," said Gary Bilendy, an urgent care nurse at Mann-Grandstaff who has experienced dozens of similar Cerner degradations. "You just can't function that way. Somebody's going to slip through the cracks."

In response to questions from The Spokesman-Review, VA Press Secretary Terrence Hayes said there had been a total of 24 outages and 48 "performance degradation events" in the Cerner system since it was launched at Mann-Grandstaff and its affiliated clinics across the Inland Northwest in October 2020.

But a document obtained by The Spokesman-Review suggests those numbers underestimate the true frequency of disruptions in the system. The document includes more than 180 incidents classified as degradations, "downtime" and full or partial outages that have affected the system's users just since September 2021.

Problems with the Cerner system have risked veterans' safety and left health care workers exhausted and demoralized. A report published Friday by the VA Office of Inspector General revealed a feature of the system caused long delays in care when referral orders were effectively lost, resulting in 149 cases of harm.

After The Spokesman-Review obtained a draft of that report, the VA announced June 18 it would delay the system's launch at facilities in Western Washington from August until March 2023. But despite the VA concluding the system does not have "adequate reliability" to be deployed around Puget Sound, it continues to be used at facilities serving tens of thousands of veterans in Washington, Idaho, Oregon and Ohio and is scheduled to launch in Boise on Saturday.

While poor training and flaws in the system have caused problems even when it is up and running, VA officials have admitted the system has been partly or completely unusable dozens of times since it was launched.

In May, the top VA official in charge of the system's rollout, Terry Adirim, told The Spokesman-Review there had been nine "unplanned outages" and 42 "unplanned degradations" as of April 20.

Adirim defined an outage as an "unscheduled event where a clinician is unable to use the electronic health record because the entire system is down." A degradation is "when all systems and applications are available, but all clinicians experience a similar issue, including the system running slower than normal."

Other incidents that don't meet those definitions — including "system errors, latency and application incomplete functionality" while "portions of it were still working" — were not included in those numbers, Adirim said in a statement at the time.

Hayes, the VA press secretary, did not give different definitions to explain why the numbers he provided as of July 6 — 24 outages and 48 degradations — were so much higher than the figures Adirim revealed in May. Most of the problems have occurred since September 2021, Hayes said in a statement: 20 outages and 29 degradations.

Of the 24 outages, Hayes said 22 were caused by a component or system belonging to Cerner and two were caused by the Department of Defense, which is implementing a similar Cerner system in its medical facilities that shares a database with the VA. Of the 48 degradations, he said, 29 were caused by Cerner, 15 by the Defense Department and four by the VA. Cerner was acquired by Oracle for $28.3 billion in a deal that closed in June, making it a division of the tech giant.

In response to The Spokesman-Review's reporting, top Democrats on the House and Senate VA committees sent questions to VA Secretary Denis McDonough on June 7 requesting detailed information about each of the incidents Adirim described by June 21. A spokesman for the House VA Committee's Democratic staff, Miguel Salazar, said the committee agreed to give VA officials until July 1 to provide answers. Despite that extension, Salazar confirmed, the VA had not responded as of June 19.

Bilendy, the urgent care nurse, said the system stops working so often that Mann-Grandstaff employees know exactly what they need to do.

"We panicked a little more the first two or three times," he said. "But now it's just, 'Cerner's down,' and we start filling out pieces of paper."

But he emphasized that not having access to the system limits what health care providers can do, leaving them without access to a patient's medical history and other important information. It also means they have to enter handwritten records into the system when they can access it again.

"We're completely dependent on this system that just, when it decides to fail, it fails," Bilendy said. "We have workarounds, but they're not efficient at all. You basically just have to give people the care they need and the documentation just falls to the wayside."

Five other clinicians at Mann-Grandstaff, who asked not to be named because they fear retaliation, corroborated Bilendy's description of the computer problems and their impact.

Cerner, now known as Oracle Cerner, is required to meet contractual obligations for "uptime," the percentage of time when the system is working. Hayes provided figures showing 100% uptime in October 2021, January 2022 and February 2022. That number fell to just over 97% in March, far below the industry standard, but still doesn't capture most of the problems employees face at Mann-Grandstaff and the other VA facilities using the system.

In October 2021, when uptime was officially at 100%, an incident occurred in which users who had logged out of the system couldn't log back in, according to the internal document. When asked about that incident and others the VA apparently does not count as downtime, Hayes declined to say how the department classified the incidents. He said VA officials could not verify the incidents listed in the internal document, because they "do not recognize the terminology used in the spreadsheet."

While all "major degradation events" are tracked by the VA, Hayes said, the department does not track every incident that causes health care workers to use "downtime procedures," a decision that is left to each medical center's director.

Deborah Hellinger, Oracle's senior vice president for corporate communications, said in a statement the company is "confident that the VA system will be the standard bearer for the industry."

"Nothing we have discovered since completing the Cerner acquisition has caused us to waver from that conviction," she said. "We are already fixing the issues that have caused past outages, and we expect to overcome any subsequent obstacles with equal urgency and expertise."

The Senate VA Committee will hold a hearing Wednesday about the Cerner system's rollout with Adirim and other VA officials set to testify, along with an Oracle executive and a representative from the VA Office of Inspector General.

Orion Donovan-Smith's reporting for The Spokesman-Review is funded in part by Report for America and by members of the Spokane community. This story can be republished by other organizations for free under a Creative Commons license. For more information on this, please contact our newspaper's managing editor.