Zero downtime PACS tech refresh during COVID-19 pandemic: experience at a major cluster of Singapore hospitals

Purpose: The goal at Singhealth was to achieve a PACS tech refresh and upgrade with zero downtime and minimal rollout delay. Singhealth is a very large health system, performing 1.6 million radiology exams a year and having 2360 PACS users (comprising 2000 clinical users and 360 radiologists). As PACS functions as our image distribution system for clinical users and also as our radiology reporting platform, uptime was critical.

Methods/Materials: Our PACS was designed for high availability and disaster recovery (DR). The setup at each of the four major sites comprises production PACS and backup (BK) PACS for high availability. The DR site concurrently serves as our cluster central archive and image distribution system from which clinical teams access images for patient care. Direct connections between each sites’ production PACS caters for DR downtime.
The new PACS servers were initially set up offsite due to heavy restrictions on access to our hospitals by non-medical personnel during the pandemic. These were only delivered onsite prior to go-live.
Decentralised user acceptance testing was performed at all sites concurrently. The DR site was upgraded first, followed by each site's BK PACS. Comprehensive end-user testing of the upgraded BK systems was followed by transitioning selected users onto the BK system to surface any remaining software bugs or configuration issues. Finally, the production PACS were upgraded sequentially.
Meetings and updates were by videoconference, WhatsApp chats or email. Project teams only met physically during the go-live transition.

Results: During the DR upgrade, clinical teams accessed cross-site images using federated search and radiology users by direct cross-site PACS access. During production upgrades, radiology users accessed BK PACS for a seamless transition with near zero effective downtime.

Conclusions: The COVID-19 pandemic disrupted many aspects of life and work; in some places, disrupting planned system upgrades. Nevertheless, during the pandemic, a major cluster PACS tech refresh was successfully executed by detailed preparation, extensive pre-upgrade testing and careful execution, leveraging on our high availability BK PACS and DR PACS to achieve near zero downtime.