The growing integration of digital technologies in the National Health Service (NHS) has enabled the transformation of traditional clinical protocols into digital tools, with sepsis screening standing as a critical area of innovation. Sepsis, a life-threatening response to infection, necessitates early identification and treatment to improve patient outcomes. A multi-site natural experiment examined the effects of introducing digital and paper-based sepsis screening tools across four NHS Trusts. By analysing data from over 700,000 adult inpatients between 2010 and 2020, the study assessed how different screening strategies influenced in-hospital 30-day mortality, especially when adjusted for casemix and temporal trends.
Comparative Impact of Screening Interventions
The study evaluated Trusts A, B and C, which implemented sepsis screening tools—paper-based in Trust A and digital in Trusts B and C—while Trust D served as a control, introducing only electronic patient records (EPR) without sepsis-specific alerts. Across all Trusts, mortality rates decreased following the introduction of screening tools, although the impact varied after adjustments for pre-existing mortality trends and patient characteristics. Trust C demonstrated the most pronounced reduction in mortality, with a 14% decrease in odds of death, despite a prior upward trend. In contrast, Trust B exhibited a pre-existing downward trend, and no significant effect of the alert was observed in adjusted models. Trust A's paper-based approach showed age-specific benefits, particularly for older patients, but lacked a consistent overall effect once adjusted for casemix.
Trust D's introduction of an EPR without a sepsis tool was associated with a higher risk of mortality, possibly reflecting changes in data recording or broader systemic factors. This contrast highlighted that EPR adoption alone, without targeted digital screening interventions, does not necessarily yield improved clinical outcomes.
Patient-Specific Outcomes and Demographic Trends
The impact of digital screening tools differed among patient subgroups, reinforcing the necessity of tailored clinical strategies. Age emerged as a significant modifier: older patients benefited most from screening tools in both Trusts A and B. No significant interactions were found between comorbidities and tool effectiveness. Moreover, the study illuminated disparities based on ethnicity and time of admission. Across Trusts, mortality risks rose with age, higher comorbidity scores, winter admissions and non-morning admissions. Gender also played a role, with female patients generally experiencing lower mortality.
Must Read: Reducing Sepsis Mortality: Strategies for Success in Hospitals
Ethnic differences were less consistent. In Trust A, Asian patients had higher odds of death, while in Trust B, they had lower odds. In Trust C, no ethnicity-specific differences were observed. The presence of missing or non-stated ethnicity data further complicated interpretation, with indications that coding practices varied across Trusts. Additionally, limited data on deprivation showed that patients without a recorded deprivation score had the highest odds of death, suggesting that some of the most vulnerable populations—possibly those without stable housing—may be under-identified in routine datasets.
Evaluation Challenges and Future Directions
Assessing the effectiveness of sepsis screening tools is complicated by inconsistencies in sepsis coding and the lack of a diagnostic gold standard. The study used a broader “suspicion of sepsis” (SoS) cohort defined by infection codes rather than relying solely on sepsis-specific ICD-10 codes, which can fluctuate based on clinical and policy-driven changes. Sensitivity analyses confirmed that increases in sepsis diagnoses aligned with the introduction of screening tools and national coding policy changes. However, these shifts did not always correlate with improved mortality outcomes.
The inclusion of a control cohort—patients with gastrointestinal bleeds—provided further evidence that observed mortality reductions were specific to infection-related admissions. In this cohort, mortality trends remained unchanged following the introduction of sepsis tools, affirming the specificity of the tools' impact.
While digital alerts are integrated into most NHS EPR systems, the study found that current tools often replicate paper-based logic and underutilise the potential of real-time EPR data and advanced analytics. More sophisticated models incorporating patient-specific data, machine learning and dynamic alerting may improve sensitivity and specificity, enabling better triaging and more appropriate interventions. Parallel qualitative research also indicated a demand for more nuanced, context-aware tools that bridge primary, secondary and emergency care settings.
The multi-site evaluation revealed that digital sepsis screening tools can reduce 30-day mortality, particularly among older adults, but their impact varies significantly across different NHS Trusts and patient demographics. Adjustments for patient casemix and pre-existing mortality trends underscore the need for cautious interpretation of raw outcome data. These findings support recent NICE recommendations advocating for more personalised, context-aware screening strategies. Leveraging the full capabilities of EPRs, aligning screening logic with local population needs and incorporating clinician feedback could enhance tool effectiveness. Ultimately, digital innovation in sepsis care holds promise, but requires rigorous evaluation, equity-aware design and collaborative implementation to realise its full potential.
Source: BMJ Health & Care Informatics
Image Credit: iStock
