Defining and measuring suspicion of sepsis: an analysis of routine data.

OBJECTIVES: To define the target population of patients who have suspicion of sepsis (SOS) and to provide a basis for assessing the burden of SOS, and the evaluation of sepsis guidelines and improvement programmes. DESIGN: Retrospective analysis of routinely collected hospital administrative data. SETTING: Secondary care, eight National Health Service (NHS) Acute Trusts. PARTICIPANTS: Hospital Episode Statistics data for 2013-2014 was used to identify all admissions with a primary diagnosis listed in the 'suspicion of sepsis' (SOS) coding set. The SOS coding set consists of all bacterial infective diagnoses. RESULTS: We identified 47 475 admissions with SOS, equivalent to a rate of 17 admissions per 1000 adults in a given year. The mortality for this group was 7.2% during their acute hospital admission. Urinary tract infection was the most common diagnosis and lobar pneumonia was associated with the most deaths. A short list of 10 diagnoses can account for 85% of the deaths. CONCLUSIONS: Patients with SOS can be identified in routine administrative data. It is these patients who should be screened for sepsis and are the target of programmes to improve the detection and treatment of sepsis. The effectiveness of such programmes can be evaluated by examining the outcomes of patients with SOS.