Introduction

To estimate the cost of sepsis in acute and critical care in NHS hospitals, we compiled data on the number of hospital admissions with a diagnosis of sepsis, together with:

• costings for acute general wards from the Payment by Results (PbR) data from Secondary Uses Services (SUS) data
• Hospital Episodes Statistics (HES) data
• National Cost Collection (NCC) data
• literature costs for critical care

To estimate the wider cost of all infections in NHS hospitals, we compiled data on admissions due to conditions from a preliminary list of ‘all infections and associated long-term sequelae’ developed by the UK Health Security Agency (UKHSA) and costed using a similar approach.

This guidance explains how the data has been used to arrive at the estimated costs used in the Sepsis modern service framework.

Definitions

Sepsis and the wider, ‘all infections’ list are defined using a combination of International Classification of Diseases, 10th Revision (ICD-10) codes, which are used to translate medical diagnoses and causes of death into a standardised format.

Find the codes used to define sepsis and all other infections in the technical annex: hospital admitted patient care and critical care activity.

Methodology

We included spells in the analysis if the primary diagnosis for any of the finished consultant episodes (FCEs) comprising the spell was included in the sepsis ICD-10 codes.

We focused on the primary diagnosis, as it reflects “the main condition treated or investigated during the relevant episode of healthcare” (as described in the National Clinical Coding Standards; NHS England, 2023). We did not include diagnosis fields other than the primary diagnosis in the analysis.

We further consolidated primary diagnoses for each episode of care into a single primary diagnosis for the entire admission, based on the number of times it occurred in the admission. This consolidated primary diagnosis can be used to form disease groupings.

Due to methodological differences, the total number of spells used to estimate the cost of sepsis differs slightly from those used elsewhere in the Sepsis Modern Service Framework. Further details can be found in the accompanying technical annexes.

We costed activity in acute general wards by merging the Payment by Results (PbR) data from Secondary Uses Services (SUS) with Hospital Episodes Statistics (HES).

PbR describes the main payment mechanism for hospital services in England (by patient activity), which uses Healthcare Resource Groups (HRG) and other adjustments, such as Market Forces Factor. For activity not paid by the PbR (for example, low-volume or specialised services), local arrangements are agreed with commissioners and specific hospitals. Those spells are costed using the National Cost Collection (NCC), applied at the FCE level using their HRG code and mode of admission, and adjusted for day cases. Where a spell is not matched to PbR or the NCC, we estimated costs for the spell by multiplying the length of stay by the average cost per day for a general ward from Jalilian and others. (2024).

In addition, we supplemented the cost of spells in general acute wards with costs in critical care. We calculated the number of days in critical care from SUS and applied a flat rate per bed day sourced from the literature (Guest and others, 2020).

For all bed days, where a patient is admitted and discharged on the same day, we applied an assumption of 0.5 bed days. We adjusted costs to 2023/24 prices using the GDP deflator.

Results

A total of 128,208 spells with a primary diagnosis of sepsis were identified in the 2024/25 financial year. We produced the following estimates:

• total costs
• cost per bed day
• cost per admission
• minimum and maximum cost per admission for acute wards and for acute wards and critical care combined

We also produced the breakdown per ICD-10 codes.

For this sample:

• 82.2% of spells were costed using the PbR
• 15.1% using the NCC
• 2.7% using a bed day cost sourced from the literature

For acute wards, the estimated cost per admission was £6,689, and the total cost was £858 million. Overall, 9.1% of admissions contained a critical care component. The total number of bed days in an intensive care unit (ICU) for this sample was 71,030.

When critical care costs were added to acute ward costs, the cost per admission increased to £8090 and the total cost increased to £1,037 million.

All the counts (bed days and admissions) shown in the results are rounded to the nearest 5.

We drew this analysis from a broader analysis of the cost of infections on NHS hospitals (covering all infections and their long-term sequelae), using the list of codes for ‘all infections’, as detailed in the accompanying technical annex, which found that the burden of all infections on acute wards in the same financial year was just over 2 million admissions, with a total cost of £6.7 billion.

Limitations

We relied on ICD-10 codes to identify sepsis cases; however, sepsis is both under-recognised and inconsistently coded in hospitals, and coding guidelines have changed over the past decade.

For costing, we used PbR for most spells, which is a robust source of information on reimbursement rather than cost. However, as PbR does not cover all spells, we supplemented it with cost estimates from the NCC and relevant published literature. Those sources differ in terms of what they capture and how costs are defined, which will introduce discrepancies in the results, especially when disaggregated across the different costing methods within the sample.

We continue to refine the method for costing using the NCC to better handle outlier costs. We anticipate updating the results once this work is completed.

Publication reference: PRN02407_iii