Criteria for SIRS, sepsis and septic shock

Clinical background

The criteria for SIRS, sepsis and septic shock were produced at a consensus conference organised by the American College of Chest Physicians and the Society of Critical Care Medicine in Northbrook in August 1991 [1]. The aim was to create a uniform terminology for a syndrome that had until then been described under a range of different and incompatible labels. The conference produced a stepwise model in which systemic inflammation, infection, organ dysfunction and circulatory failure were defined as a continuum: SIRS, sepsis, severe sepsis and septic shock.

The instrument serves two purposes. First, it provides a rapid bedside assessment requiring no resources beyond a thermometer, a pulse count, a respiratory rate and a differential white cell count. Second, it has historically been the dominant inclusion criterion in sepsis studies and quality registries. The decision the instrument serves is really twofold: to classify a patient into the right severity stage and to set a threshold for when an infection has become a systemic illness.

The model has since 2016 been formally replaced by the Sepsis-3 definitions, in which sepsis is defined by organ dysfunction (SOFA) rather than inflammation (SIRS) [2]. This calculator reflects the classic criteria, which still appear in clinical routines, electronic record systems and older study protocols.

Applying the SIRS criteria

SIRS is defined as the presence of at least two of four criteria:

$$\text{SIRS} \geq 2 \text{ of:}$$ $$\text{Temperature} > 38,^\circ\text{C} \text{ or } < 36,^\circ\text{C}$$ $$\text{Heart rate} > 90\text{/min}$$ $$\text{Respiratory rate} > 20\text{/min} \text{ or } P_{a}\text{CO}_2 < 32\text{ mmHg}$$ $$\text{White cell count} > 12 \text{ or } < 4 \times 10^9\text{/L, or } > 10,$$

The staging then rests on the accumulation of conditions:

$$\text{Sepsis} = \text{SIRS} + \text{suspected or confirmed infection}$$ $$\text{Severe sepsis} = \text{Sepsis} + \text{organ dysfunction or hypoperfusion}$$ $$\text{Septic shock} = \text{Sepsis} + \text{persistent hypotension despite adequate fluid resuscitation}$$

The derivation is an expert consensus, not an empirical cohort study [1]. The criteria were chosen not by statistical modelling of a patient population but by expert agreement on physiological parameters considered to reflect systemic inflammation. This is central to understanding the instrument's limitations: the thresholds (90 beats/min, 20 breaths/min, 12 or 4 ×10⁹/L) are not optimised against any outcome, and there is no derivation cohort whose composition limits their application. A revision in 2001 (Sepsis-2) expanded the list of signs of sepsis but kept the SIRS criteria unchanged at its core.

Interpretation in practice

The calculator gives a stepwise classification in which each step implies an escalation in the level of monitoring and the intensity of treatment.

It is important to note that the SIRS criteria are not in themselves prognostic. A patient with two SIRS criteria and a urinary tract infection does not necessarily have sepsis in the modern sense, and a patient with life-threatening organ dysfunction may not meet two SIRS criteria.

Validation and performance

Since the SIRS criteria were derived by consensus and not by empirical modelling, there is no derivation cohort to compare against. External validation instead concerns how well the criteria perform as a screening tool and prognostic marker in real patient populations.

In a large retrospective cohort of 10,981 adults with suspected infection at a US acute hospital (2011 to 2017), SIRS was the weakest of four scores compared for predicting in-hospital mortality, with an AUC of 0.79 (95% CI 0.78 to 0.81), compared with SOFA 0.90, NEWS 0.85 and qSOFA 0.84 [3]. In the intensive care unit the difference was even more marked: SIRS AUC 0.68 (95% CI 0.66 to 0.70) versus SOFA 0.82. In non-intensive care settings, SIRS was marginally worse than SOFA, NEWS and qSOFA alike.

Another cohort of 30,677 patients with suspected infection outside intensive care (University of Chicago, 2008 to 2016) confirms the picture [4]. SIRS had an AUC of 0.65 (95% CI 0.63 to 0.66) for in-hospital mortality, compared with NEWS 0.77, MEWS 0.73 and qSOFA 0.69. At the conventional threshold of ≥2 SIRS criteria, sensitivity for the composite outcome of death or transfer to intensive care was 91%, but specificity only 13%. This means that in practice SIRS flags almost all sick patients but also most of those who are not sick.

In a smaller prospective cohort of 124 patients with acute hepatobiliary infection in Singapore, the pattern was confirmed: SIRS had high sensitivity (60% for high-dependency care, 62.5% for intensive care) but low specificity (49.5% and 53.3% respectively), while qSOFA showed the reverse pattern [5].

The Sepsis-3 committee summarised the criticism of SIRS explicitly: the criteria have poor discriminant validity because they are common in inpatients without infection, and 1 in 8 patients admitted to intensive care units in Australia and New Zealand with infection and new organ dysfunction did not meet the requirement of ≥2 SIRS criteria [2].

Limitations

The most important limitation is that SIRS-based definitions have formally been superseded. Sepsis-3 (2016) defines sepsis as life-threatening organ dysfunction caused by a dysregulated host response to infection, operationalised as an increase in SOFA of ≥2 points [2]. The term "severe sepsis" has been discarded as redundant. Septic shock is now defined as sepsis requiring a vasopressor to maintain a MAP ≥65 mmHg together with a lactate >2 mmol/L after adequate fluid resuscitation, with an in-hospital mortality above 40%.

The SIRS criteria are furthermore not specific to infection. They are triggered by trauma, pancreatitis, haemorrhage, surgery, ischaemia and a range of other conditions unrelated to sepsis. This was a deliberate design, since SIRS was originally intended to describe systemic inflammation whatever the cause, but it becomes a weakness when the criteria are used to screen for sepsis.

The thresholds are moreover static and take no account of age, comorbidity or medication. Beta blockers can mask tachycardia, and older patients may have sepsis without fever or leucocytosis. Nor do the criteria have a prognostic dimension: they state that a patient has an inflammatory response, not how ill the patient is or what risk of death or organ dysfunction exists.

One concrete misuse is to use the absence of SIRS criteria to exclude sepsis. Given that up to 1 in 8 patients with infection and organ dysfunction do not meet ≥2 SIRS criteria [2], this can lead to patients with genuine sepsis being misclassified as not being ill.

References

1. Bone RC et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. Chest 1992. PMID: 1303622
2. Singer M et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016. PMID: 26903338
3. Kovach CP et al. Comparative prognostic accuracy of sepsis scores for hospital mortality in adults with suspected infection in non-ICU and ICU at an academic public hospital. PLoS One 2019. PMID: 31525224
4. Churpek MM et al. Quick Sepsis-related Organ Failure Assessment, Systemic Inflammatory Response Syndrome, and Early Warning Scores for Detecting Clinical Deterioration in Infected Patients outside the Intensive Care Unit. Am J Respir Crit Care Med 2017. PMID: 27649072
5. Mak MHW et al. A prospective validation of Sepsis-3 guidelines in acute hepatobiliary sepsis: qSOFA lacks sensitivity and SIRS criteria lacks specificity. Int J Surg 2019. PMID: 31678690