Clinical Safety Officer and safety case
Estimate: 15%
The hazard log is drafted and its rules are checked by tests. Nobody has been named as Clinical Safety Officer, the hazards are not scored, and there is no safety case report.
This system must not be used with real patients at present. Each item below has an estimate. An item at 100% is done, with evidence. Every other item is still open.
Production means one thing here: a health organization runs the system for a live service, holding real patients' details, where a mistake can reach a person.
Being built, tested and published is not production. Neither is a demonstration on made-up data.
Getting there takes more than code. It needs a named clinician accountable for safety, a legal basis for the data, outside security checks, messages that really arrive, and an organization that runs the system and answers for it. Most of the remaining work is of that kind.
Each figure is a judgment, not a measurement: how much of the work that item needs is done, from 0% (not started) to 100% (done, with evidence). An item that needs a person or an outside party stays low however much code surrounds it.
The main repository gives the reason for each figure in the path to production, and the task list holds the detail.
Estimate: 15%
The hazard log is drafted and its rules are checked by tests. Nobody has been named as Clinical Safety Officer, the hazards are not scored, and there is no safety case report.
Estimate: 40%
A draft assessment, data flow and retention schedule exist. No Data Protection Officer has reviewed them.
Estimate: 0%
The project tests its own security headers and access rules. No outside tester has looked at it.
Estimate: 0%
Neither has been sought.
Estimate: 0%
This belongs to the organization that runs the system, and there is none yet.
Estimate: 85%
The supplied sign-in setup asks staff for a one-time code. An organization's own sign-in service must do the same.
Estimate: 70%
Built and tested, and the audit log records the verified user. It is not yet switched on by default, and the signing key has no secure store.
Estimate: 0%
The operator dashboard has no sign-in yet. It needs a decision first.
Estimate: 60%
A patient's account works only after a named staff member records proof of identity. There is no staff screen for this yet, and no digital identity service, such as a national login, is connected.
Estimate: 70%
A backup and restore drill runs on every change. No live service has backups yet.
Estimate: 40%
The system reports the figures to alert on, and the runbook names them. No alert has been set up.
Estimate: 0%
No target has been set.
Estimate: 90%
Built and tested at the one point every message passes through. The wording has not been reviewed with patients.
Estimate: 80%
The system's side is built and tested. It has not been tried with a real carrier.
Estimate: 10%
Every carrier would plug into one interface. There is no carrier contract, and no message has been sent through a real one.
Estimate: 10%
Every interface language has a draft. No first-language speaker has reviewed one, and Welsh must be reviewed before patients see it.
Estimate: 30%
Automated checks run in the browser tests, and a statement is drafted. Nobody has tested it by hand, and the statement is not published.
Estimate: 0%
Nobody has tested the system with patients or staff.
Estimate: 100%
Done: every dependency's license and known advisories are checked on each change.
Estimate: 60%
The full checks run on every change, but nothing yet stops a change that fails them from reaching the main branch.
Estimate: 100%
Done: each release comes with a list of everything it contains.