When clinical systems stop: continuity in healthcare IT

Researchers at UC San Diego measured what the July 2024 outage did to hospital networks, and published the result in JAMA Network Open. Of 2,232 hospitals with available data, 759 — 34.0% — showed service disruptions detectable by their scanning methods. Among 1,098 distinct disrupted services, 239 were directly patient-facing, including imaging platforms, staff portals for viewing patient data, and patient appointment portals.

Downtime procedures are clinical, not just technical

Paper forms that nobody has seen for two years are not a downtime procedure. A usable one specifies where the forms are kept, who holds the key, which identifiers are recorded so the records can be reconciled afterwards, how results are communicated without the usual system, and how the backlog is entered once service returns. The reconciliation step is the one most often missing, and it is where errors are introduced after the incident is declared over.

What to establish before it is needed

• —Which systems are life-critical, which are important, and which can wait a day — agreed with clinicians, not assumed by IT.
• —A read-only copy of essential patient information that survives the primary system being unavailable.
• —Printed, current downtime forms in each department, with a named owner who checks they are still current.
• —A communication path to staff that does not depend on the systems that may be down.
• —A rehearsed reconciliation process for entering downtime records, with a check for duplicates and omissions.

Health data raises the stakes on both sides

Health information is among the most sensitive personal data an organisation can hold, and Uganda’s Data Protection and Privacy Act applies to it. That creates a genuine tension during an incident: staff need access to continue treating patients, and the organisation still has to control and record who saw what. Emergency access procedures — often called break-glass — resolve it by allowing the access and logging it for review afterwards, rather than forcing a choice between care and compliance.

The measure of a health service’s technology resilience is not uptime. It is whether the ward can keep working on the morning the screens are blank.