What CASCADE is.
CASCADE makes an organisation's own safety history usable at the moment a decision is being made, before and after harm. A named person makes every decision.
Watch the 1-minute overviewWhat does each stage ask?
- Connect
- What related concerns and learning should we consider together?
- Remember
- What have we tried before, and what happened?
- Decide
- What action is justified by the available evidence?
- Act
- Who will deliver the change, and what support is needed?
- Evaluate
- Was the change implemented, and is it making a difference?
The same harm keeps coming back
Your problem
About a minute, no sound, with captions. Esc closes it.
Read the captions as text
- Organisations look into harm, act and close the action. Then the same harm comes back.
- Describe a safety problem in plain words, or pick an example. Here: Maternity.
- CASCADE finds what was tried before, how strong it was, and whether it worked.
- Fetal monitoring taught twice: low effectiveness, put in place both times, did not work.
- It shows what was tried, how strong it was and whether it worked. A named person decides.
- Who to involve and how we'll know it worked go in the action log, to be checked later.
- Made-up data only. Try it yourself, or book a 30-minute walkthrough.
Questions people ask about CASCADE
What is CASCADE?
Soughtout SafeSystems builds CASCADE™, patient safety learning software that helps health and care organisations and systems use their own safety history, from incidents, complaints and reviews, at the moment a decision is being made. It works before harm too: near misses, staff concerns and rising trends go through the same steps. It shows what was tried before, how strong it was and whether it worked. A named person decides.
Who is it for?
People who review incidents, people who lead safety governance, and people who own safety actions, in health and care organisations and the systems around them.
Does it replace our incident system?
No. It works alongside the system you already use, such as Datix, Radar or Ulysses, reading exports from it. Staff keep reporting as they do now. How CASCADE fits
Does it use AI?
No AI model runs in this demonstration. One worked example shows AI suggestions generated in advance on made-up data, each labelled. In a pilot, AI reads each record and suggests its theme, contributing factors and the type of each action; a named person accepts, edits or rejects every suggestion. It never scores or decides. Safe and legal
How an AI would read a record: one worked example
What stage is it at?
A working prototype on made-up data. No real patient, staff member, incident or organisation is represented.
Who is behind it?
Soughtout SafeSystems Ltd (UK, 17119617), founded by Aderonke Opawande. aderonke@soughtoutsafesystems.co.uk
What does this edition not show?
This edition shows what CASCADE does: the problem, the previous learning it returns, the decision a named person makes and the resulting action log. It does not show how records are matched: the matching runs on the server and stays there. Every record is synthetic.
What we already knew
Look up previous learning first. Back to your problem
You asked: · Change
More: why some didn't hold, what was searched, the same theme across divisions
How these records connect
What was searched, and what was left out
The same theme across divisions
A learning-only view
Your decision
Look up previous learning first. Back to your problem
CASCADE shows what was tried, how strong it was and whether it worked. A named person decides. Earlier actions on this theme needed help from outside the team.
Show the reasoning
Next records your decision in the action log. You can come back and change it.
Putting it in place
Look up previous learning first. Back to your problem
How we'll know: a worked example
Made-up data, filled in from your decision. Read-only.
Your own: how we'll know
From the PSIRF safety action development guide. Every field is optional here.
A low-effectiveness action can be the right interim step. It needs closer checking and a plan to strengthen it.
Who needs to be involved?
Consider involving these people, so the action is put in place and lasts. Tick who you will involve; it goes in the action log.
Did it make a difference?
Look up previous learning first. Back to your problem
Three questions, asked of every action
More: who else should know, when to stop, your action log
Who else should know?
Action log
Built in your browser from what you did. We do not keep it.
Your judgement of each record shown
What you just saw
What you just saw: a problem; what was tried before, how strong it was and whether it worked; a named person's decision; and how to check it at 6 and 12 months. All on made-up data.
With your own records, CASCADE adds: your own history in front of the person deciding; actions checked until they hold; learning shared across services.
What's next (planned work): learning at the point of reporting; a team memory that stays when people move on.