Beta This is a new service – email us your feedback to help us to improve it.


Learn from patient safety events service data principles

July 2021

Background and purpose

The purpose of the Learn from patient safety events (LFPSE) service (previously known during development as the Patient Safety Incident Management System - PSIMS) is to enable learning from patient safety events – incidents, risks, outcomes of concern and also things that went well. Our ability to protect future patients from harm depends on promoting a culture that welcomes and encourages the recording of events. It is essential to abide by these principles to ensure that we continue to successfully learn from patient safety events and reduce harm.

This document sets out the circumstances in which LFPSE data are the appropriate data source to be used and describes their appropriate use.

These principles emphasise the purpose and characteristics of LFPSE data, and promote consistency across data users. It is essential that users of LFPSE data understand and represent it appropriately, as inappropriate presentations of LFPSE data could discourage recording.

They apply to all stakeholders, including NHS England and Improvement staff using LFPSE data.

The principles

What do LFPSE data tell us?

  • They reflect the safety recording culture in an organisation.
  • They show the most frequent types of recorded patient safety events at national, regional and local levels.
  • They show patient safety events broken down as;
    • Patient safety incidents
    • Outcomes of concern where a patient safety incident may have contributed to the outcome but that is not yet clear.
    • Risks to patient safety (where no incident has occurred yet)
    • Examples of good care which users felt could be learnt from to amplify excellence
  • They show which types of patient safety incidents are recorded as most harmful, at national, regional and local levels.
  • They help us to understand how patient safety incidents happen.
  • They show how incident recording patterns change over time by degree of harm, by topic and by care setting or organisation.
  • They help us to segregate the data for different kinds of review, response and learning activities by different users.
  • They help us understand who is recording different events, and if they choose to do so anonymously, why.

What LFPSE data don’t tell us?

  • They do not provide the actual number of patient safety incidents or events occurring in the NHS or in a particular organisation.
  • They do not provide information on the full history and characteristics of patients involved in patient safety incidents or other events.
  • They do not tell us if an organisation is safe or unsafe.
  • They do not show an organisation’s level of compliance with safety guidelines.
  • They do not measure the efficacy of specific safety initiatives.

Data notes

Further data notes to consider when using LFPSE data:

Total numbers

LFPSE is a largely voluntary scheme for recording patient safety incidents, risks or outcomes of concern, so it does not provide the actual number of patient safety incidents, risks or outcomes of concern occurring in the NHS. The maturity of an individual’s or organisation’s safety processes and culture will affect what is recognised and recorded. So too will new treatments and procedures; changing standards of accepted clinical practice; publications or campaigns raising awareness of under-recognised risks; and the effectiveness of local systems for seeking out harm or risks through case record review and audit that would otherwise have gone undetected.

Consequently, patient safety events recorded to LFPSE are just that – an accurate source of events recorded to LFPSE, rather than an imperfect source of actual incidents, risks or outcomes of concern. LFPSE data users can help to ensure that this is understood by always referring to ‘recorded’ patient safety incidents, risks and outcomes of concern in records, tables and figures.

Incidents not errors

‘Patient safety incident’ is not a synonym for error. LFPSE will include records of harm unrelated to errors (for example, adverse reactions to medication that could not have been anticipated or prevented) and also records of situations with potential for causing harm that staff recognise and record before they can result in error. LFPSE data users can help to ensure this is understood by referring to quantitative LFPSE data as ‘recorded patient safety incidents/risks/outcomes of concern’, or just ‘recorded incidents/risks/outcomes of concern’ in contexts where patient safety is implicit (eg recorded medication incidents).

Dynamic database

LFPSE is a dynamic recording system and the number of events recorded as occurring at any point in time may increase or decrease as records are updated. Event classification, and particularly recorded degree of harm, can also change as local investigations progress and records are updated. Therefore, the date of the data extraction should always be included on tables and chart references.

Trends over time

Another limitation to the data can be that changes to recording requirements often result in changes to data series over time. This is particularly true when considering LFPSE data and data from its legacy database, the National Reporting and Learning System (NRLS). When looking at the data, data users need to bear in mind the dates of switchover to LFPSE recording from NRLS, changes in national mandatory recording requirements, or national guidance on recording to the LFPSE, as these may have a ‘one-off’ impact. Organisational change should also be borne in mind, as newly created and newly merged organisations take time to mature and set up their systems and processes.

When comparing data across time periods, it is important to compare data with the same time period in the previous year(s) to take into account known ‘seasonality’ in the data. ‘Incident seasonality’ is one cause of this: Research suggests healthcare activity and the types of conditions requiring treatment relate to the time of year and these changes may affect incident recording.

Therefore, when reviewing changes over time, we recommend that:

  • proportions or percentages are used rather than actual numbers (to allow for the differences in the underlying numbers of incidents and the underlying activity of the NHS)
  • either the same time period in the previous year, or a full year’s worth of data are used (to take seasonality into account)
  • it is checked that any ‘change/difference’ is not caused by new or amended national mandatory recording requirements or organisational restructuring.

This will not be possible in the first 12-18 months of LFPSE operation, as there is no historic data to compare to, but the principle should be borne in mind.

LFPSE data is not directly comparable to NRLS data, as the concepts and definitions within the taxonomy have changed, and care should be taken to avoid continuous comparison over time between one data set and the other.

Timely recording

We encourage organisations to record patient safety events as soon as possible. LFPSE is designed to allow for updates to records over time, with basic information captured first, and further detail and context added as it becomes available. The Care Quality Commission (CQC) guidance for recording death and severe harm incidents ( recommends recording ‘without delay’. However, in practice there is usually a delay between an incident occurring and being recorded.

Interpreting LFPSE records with a degree of harm = fatal

LFPSE prompts users to select a “fatal” level of harm ‘if the patient has died and there is at least a slight possibility the incident that you are recording may have contributed to the death, including stillbirth or pregnancy loss’. Mortality research has identified that attribution of fatal outcomes to incidents rarely clear cut. LFPSE data users can help ensure this is understood by not summarising recorded degree of harm with terms such as ‘causing’ death except where this is justified by the text of incident records after local or independent investigation.

There is a further (optional) field which allows users to provide a subjective assessment of the extent to which an incident may have contributed to the outcome; this data can be used in conjunction with the level of harm to help understand the impacts of individual patient safety events, and prioritise local responses.