Introduction

To estimate age-standardised mortality rates (ASMRs) for sepsis across the 7 NHS regions, we used Office for National Statistics (ONS) death data from 2022 to 2025 accessed via the Secure Data Environment (SDE).

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

Definitions

Deaths were identified where sepsis was recorded as the underlying cause of death, defined using the International Classification of Diseases, 10th Revision (ICD-10) codes, which are used to translate medical diagnoses and causes of death into a standardised format.

For the Sepsis modern service framework, the following codes were used:

• A40
• A41
• A39.2
• A39.3
• A39.4

Methodology

ASMRs provide meaningful comparisons between populations with differing age structures. ONS publishes regional counts of deaths involving sepsis but does not subdivide it into different age bands, so ASMRs cannot be calculated. The National Online Manpower Information System (NOMIS) provides ASMRs by single ICD-10 codes, but these cannot be combined across multiple codes. As we are unable to determine ASMRs from available publications and obtaining a bespoke ASMR from ONS would take time, we produced the ASMRs.

Each death registration records a single underlying cause, so we excluded deaths where sepsis is mentioned but is not the underlying cause. We aggregated counts of deaths by:

• year
• age group
• NHS region

When compared with published ONS death registration statistics (table 3), the regional death counts used in our ASMR calculations were broadly consistent. Regional assignment was informed by the ONS lookup table, and the approach was quality assured to ensure that registration district codes were correctly assigned to NHS regions. Any differences relative to the ONS 2024 publication are likely to reflect changes in the underlying dataset between the time points at which it was accessed and/or minor revisions to the district boundaries, as we mapped to this following the extraction of the data.

The resulting ASMR represents the rate that would have occurred if the observed age-specific rates had applied to a standard population with a fixed age distribution. As per ONS methodology, we used the 2013 European Standard Population (ESP) as the standard population. This is a hypothetical population applied consistently across sexes. Results are presented as annual ASMRs per 100,000 population, which accounts for population size and age structure.

Corresponding 95% confidence intervals are also presented with the ASMR rates (see table 1). These form a confidence interval (CI), which is a measure of the statistical precision of an estimate and shows the range of uncertainty around the estimated rate. Calculations based on small numbers of events are often subject to random fluctuations. While this does not constitute a formal statistical test of differences, instances where there is little or no overlap between a regional confidence interval and a national one suggests that the observed difference is unlikely to be due to random variation alone.

It should also be noted that ASMR accounts solely for differences in population structure (age and sex) and does not account for other risk factors associated with sepsis mortality, such as socio-economic deprivation, smoking status, obesity or underlying comorbidity. Therefore, variation in ASMR observed between region, though important to highlight, cannot be conclusively attributed to the quality of care.

Results

Table 1. Age-standardised mortality rate (ASMR) by region and year (95% CI) per 100,000 population

Table 2. Deaths by NHS region and year (2022 to 2025)

All counts have been rounded to the nearest multiple of 5 in line with disclosure control rules from NHS data export.

Table 3. Comparison of Department of Health and Social Care’s data extraction for total sepsis deaths with ONS published counts by NHS region for 2024

Publication reference: PRN02407_iv