It feels as though serious NHS patient-safety scandals come along more often than they used to. Maternity services in particular have produced one after another, and the trigger for this project was another round of news coverage about a hospital service in 2026. I wanted to know whether that impression was actually right, and if it was, why.
Why it’s harder than it looks
“Are scandals becoming more frequent” isn’t one thing you can count. It splits into at least four questions that can move independently of each other:
- Incidence. Is serious harm actually happening more often?
- Detection. Are more existing failings being found?
- Visibility. Are failings that are found getting more press and political attention than they used to?
- Recurrence. Are the same kinds of failure happening again because the lessons aren’t sticking?
A simple count of scandals per year mixes all four together. So the project keeps two measures apart: a year-by-year count of named failings that got sustained national coverage, starting from the Bristol Royal Infirmary inquiry, and the underlying safety data (regulator ratings, negligence claims, incidents graded by harm). Whether those two move together is one of the questions, not an assumption.
How it’s being done
The research is split into nineteen separate documents, each testing one possible explanation: reporting culture and PSIRF, the growing size of the NHS, workforce and staffing, the media environment, the claims industry, public inquiries becoming the default response, and others. Two of them are deliberate controls, comparing England with other countries and with Scotland, Wales and Northern Ireland, to see whether anything England-specific is going on. A separate record logs every place an earlier conclusion was revised.
I set one boundary early on. Claims that harm was manufactured, or its disclosure timed for political ends, are out of scope, because nothing could ever count as evidence against them.
Some of the framing comes from what I saw from the inside, in an ICB quality team: the same kinds of incident recurring locally, with the same learning identified and not embedded.
Status
The research is done and the findings are drafted, but they aren’t published yet. Before they are, I need to decide how to handle naming trusts and individuals, and how firmly to state conclusions about policy. When it’s ready it’ll potentially go here, with every research prompt included as an appendix.