BestHeartSurgeryBook Appointment

IdentityOps Lessons From a Fictional Hospital Admin Team

Healthcare Technology

Picture a mid-sized hospital back office that never touches a patient chart. Registration clerks, billing staff, scheduling coordinators and a small systems group keep the building running behind the scenes. This fictional team is a composite we use to explain IdentityOps, the routine work of keeping an accurate roll of every sign-in identity and the evidence behind each one. Nothing here describes a real organisation, and nothing below is clinical or medical guidance.

The story starts on an ordinary Monday, when a coordinator named only “the Tuesday leaver” in the team’s notes left the hospital after a notice period. Her manager told HR, HR closed the record, and everyone assumed the rest would follow. It did not follow, and the reasons are a tidy tour of the discipline.

Where the first assumption broke

The systems group had a habit of treating one event as the whole job: the directory account was switched off, therefore she was gone. In the vocabulary of the Ops Atlas entry on IdentityOps, that is the classic gap between a disabled account and an offboarded person. The Atlas stresses that a leaver flow is a chain, running from the HR termination to the identity provider, then to session revocation, then to the provisioning push that tells each connected application, with a timestamp recorded at every hop.

The team had the first hop and a half. What they lacked was the rest of the chain, so for most of a working day she retained a browser session in the scheduling tool. Sessions can outlive accounts, and the Atlas reports that access tokens commonly live for an hour or more by default. Switching off the account stops new sign-ins; it does not always cancel the ones already open.

Measuring the miss the way the analysts would

Rather than argue about blame, the group decided to measure. They pulled the quarter’s leavers and counted two things. First, how many had their directory account disabled inside the agreed window. Second, how many had no live account left anywhere, including the small departmental tools nobody had connected to the directory.

The Atlas offers a worked example of exactly this comparison. With forty leavers, thirty-six switched off on time, and a trio among them keeping a stray login in some minor application, the directory-only figure reads 90.0 percent while the end-to-end figure falls to 82.5 percent. Our fictional team ran the arithmetic on its own numbers and found a similar spread. The lesson was less about the percentage and more about which number leadership had been shown: the flattering one.

  • Directory-only compliance answers “did the main switch flip?”
  • End-to-end compliance answers “can this person still do anything anywhere?”
  • Only the second figure belongs on a risk dashboard.

The apps nobody had registered

The audit turned up a surprise. Two small tools used by the billing staff handled sign-ins on their own, with local usernames. Because they sat outside the directory, no leaver flow could reach them. Someone had also created a shared login for a reporting job, used by several humans over the years. One login, many hands: the Atlas flags that habit as a recurring weakness, and the team saw their own reporting job in it.

The fix was unglamorous. They wrote down every application, who owned it, how it authenticated, and whether it could accept automated deprovisioning through the SCIM standard. The Atlas notes that some vendors reserve SCIM for higher subscription tiers, so a few applications would need a conversation about licensing before they could join the chain. For those, the owner agreed to a manual removal step with a deadline.

A recovery script instead of a favour

A second episode involved a forgotten sign-in method. A registration clerk lost access to her authenticator and phoned the help desk, which, as it had always done, restored access after checking her employee number. The systems group realised that an employee number is something many colleagues know. The Atlas describes the pattern in terms of social engineering through help desks and says the safer route is a documented recovery path in tiers: self-service with a second method first, then a one-time pass handed over only once the manager has vouched face to face or on a live call.

The team adopted a short script and a rule that every recovery generates a notification to the person concerned. It slowed down the first few cases and then stopped feeling slow.

How the story fits with neighbouring work

It is worth saying what this team did not take on. They did not redesign who may open which patient-facing module; that is permission design, a neighbouring concern. They did not rebuild HR’s offboarding form either. IdentityOps sat in the middle: it consumed the HR event as a trigger and made sure the technical consequences followed, completely and on a clock.

Our fictional group read the cited incidents too. In the Atlas telling, the breaches it reviews were won against routine, not against encryption: an old test environment missing a second factor, a secret parked in someone’s private cloud drive, a support desk too generous with resets. No clever mathematics was involved, only a gap in habit.

What the team did in ninety days

They followed a simple sequence rather than buying a platform. In the first month they named an owner for each application and wrote the register, set up hardware security keys for the few administrators who could change identity settings, and wrote the recovery script. In the second month they connected one population to the HR feed and extended automated deprovisioning to the ten applications that mattered most. In the third they tested session revocation, ran their first leaver drill with a volunteer from another department, and added alerts for password resets followed immediately by changes to federation settings.

Reading list for a hospital operations group

Teams in similar back offices may find these local posts a useful companion. A piece on how virtual follow-up visits work after cardiac procedures shows why more staff and more remote sign-ins make a clean roster matter. Another, about selecting a hospital programme for heart failure recovery, touches the coordination between departments that identity housekeeping quietly supports. Both are informational reading, not advice.

Takeaways from the fictional week

An analytical read of the story gives three practical points. Measure the end of the chain, because the beginning always looks good. Register everything that authenticates, including humble tools and shared logins. And replace the favour-based recovery call with a script that treats the caller as unverified until proven otherwise.

Money is rarely the obstacle here. What helps is a calendar, a few owners, and a willingness to look at the lower number on the dashboard. The Atlas suggests a cadence that smaller groups can borrow: reconcile daily, sample recoveries weekly, review machine-credential expiry monthly, and run a leaver drill each quarter. Our fictional team put those four dates in a shared diary and kept them, which is most of the work.