Identifying sepsis cases in administrative data: a validation study
UKHSA is working with the NHS to improve identification of sepsis in England. This study will help identify how many adults get sepsis in England and who is most affected.
Documents
Details
Overview
Sepsis is a life-threatening condition which occurs when the body’s immune response to an infection damages its organs.
There is currently no reliable number of how many people get sepsis in England each year.
UKHSA want to do better at estimating how many adults get sepsis and who is most affected.
What we aim to do with this study
In this study, UKHSA will use historical data and computer models to predict if someone had sepsis or not in order to calculate the number of people who had sepsis at specific time points. The prediction will be checked against a “true” diagnosis of sepsis and is considered accurate if the prediction matches the diagnosis most of the time. If accurate, UKHSA will be able to use these computer models to estimate how many adults get sepsis and who is most affected. This study aims to identify people who have had sepsis and not people who might develop sepsis.
How we will do this
Developing the gold-standard sepsis case criteria
A group of clinical experts, including infectious disease doctors and doctors from the Intensive Care and Emergency Department settings, from 7 different NHS Trusts, will together decide on a set of rules for how to identify sepsis from data included in the hospital electronic patient notes. This set of rules together is called the gold-standard sepsis case criteria.
Patients to be included
UKHSA analysts will select around 4,000 adult (16 years of age or older) patients at random using NHS hospital episode statistics (HES) data (obtaining summary admissions data for healthcare analysis, find out more: Hospital Episode Statistics (HES)) who stayed in hospital from April 2023 to March 2025 in 7 participating NHS Trusts:
- Royal Devon University Healthcare NHS Foundation Trust
- Hampshire Hospitals NHS Foundation Trust
- Royal Free London NHS Foundation Trust
- Guy’s and St Thomas’ NHS Foundation Trust
- University College London Hospitals NHS Foundation Trust
- Oxford University Hospitals NHS Foundation Trust
- Manchester University NHS Foundation Trust
UKHSA will securely send identifiers (NHS numbers, date of birth, hospital number and sex) of these patients along with a study-specific patient ID to NHS doctors at each participating NHS Trust. The doctors will use the identifiers to check these patients’ medical records. They will use the gold-standard sepsis case criteria to confirm if the person did or did not have sepsis – this is the “true” diagnosis of sepsis or not sepsis. The patient ID and confirmation of sepsis (or not) will be returned to UKHSA to develop the computer models. UKHSA will not see patient medical records.
Development of computer models
UKHSA will develop computer models using information contained in healthcare datasets that are used for administration and payment, such as NHS Hospital Episode Statistics data (HES). These datasets are often used to monitor outcomes in the wider patient population as the information is regularly updated, albeit collected for another purpose (such as managing patient care and hospital reimbursement) and require no additional NHS staff time to input data. The computer models will include some patient demographics (such as age, sex, ethnicity), clinical history (for instance, hospital and intensive care admission, prescribing or confirmed infections) and diagnosis codes. Diagnosis codes are used to translate clinical diagnoses recorded in patient records upon discharge from the hospital into the data used for administration and payment. The models will make a prediction of whether an adult patient staying overnight in hospital between 1 April 2023 and 31 March 2025 did or did not have sepsis. They will be compared with the “true” diagnosis sent by the participating NHS Trusts.
Accuracy of the computer model predictions
The models work well if the predictions match the” true” diagnosis most of the time. If they do, we can find out how many people in England get sepsis and who is most affected.
This study will not affect your care, but its results may help patient care and management in the future. By knowing and better understanding the total burden of sepsis and whether this differs by population groups, it will drive innovations in developing treatment and diagnostics to reduce unfavorable outcomes.
Patients and patient representatives have informed the development of the study.
Opt-out
If you don’t want to be part of this research, contact the study team on +44 20 8327 6214 or fill in this form. You will be asked to give your NHS number and date of birth so we can remove you from the data (UKHSA won’t have access to your name). Your details need to be kept until identifiers, like NHS number, will be replaced by a unique patient ID made specially for this research, at this point the patient identifiers in this research dataset will be destroyed. National Data Opt-Out already applied.
Contact details
Email sepsis@ukhsa.gov.uk;
Phone +44 20 8327 6214
Approvals
This study has received section 251 approval from the Health Research Authority Confidentiality Advisory Group (ref 25/CAG/0166) to use in-patient notes at the seven participating NHS Trusts to create a reference dataset for validation purposes
Links
You can read more about the study methods in the above-attached non-technical summary and how we will use and protect your data in our privacy notice. You can see a copy of the posters we have displayed in participating NHS sites below.
- Non-technical summary (includes the technical study protocol)
- UKHSA privacy notice
- Study poster