Summary of findings
| No. of studies | No. of patients | Study design | Risk of bias | Inconsistency | Indirectness | Imprecision | Other | Findings | Level of evidence |
|---|---|---|---|---|---|---|---|---|---|
| Derivation of prehospital sepsis screening tools | |||||||||
| 3 | 145, 843 | Non-RCT | None | None | Not serious* | Very serious† | None | Seymour et al29 32 CIS: risk of sepsis 0.76 (95% CI 0.75 to 0.77) | ⊕⊙⊙⊙ very low |
| Polito et al 31 PRESS score: sensitivity 0.85, specificity 0.47, PPV 0.19, NPV 0.96 (95% CI not reported). PRESS score ≥3; sensitivity 0.81; specificity 0.63 | |||||||||
| Bayer et al30 PRESEP score: sensitivity 0.85 (95% CI 0.77 to 0.92), specificity 0.86 (95% CI 0.82 to 0.90), PPV 0.66, NPV 0.95 | |||||||||
| Sepsis recognition by EMS (using a screening tool) | |||||||||
| 2 | 161 | Non-RCT | Very serious‡ | None | Not serious§ | Very serious¶ | Very serious** | Guerra et al12 Screening based on SSC criteria identified 32/67 patients with sepsis (47.8%) (95% CI not reported) | ⊕⊙⊙⊙ very low |
| McClelland and Jones15 Screening using modified Robson tool. Sensitivity and specificity for sepsis 43% (95% CI 28% to 58%) and 14% (95% CI 0% to 40%), respectively. Sensitivity and specificity for severe sepsis 30% (95% CI 12% to 47%) and 77% (95% CI 60% to 95%) | |||||||||
| Retrospective application of EMS data to screening tool by researcher | |||||||||
| 2 | 728 | Non-RCT | Very serious‡ | None | Not serious§ | Very serious¶ | None | Wallgren et al21 Retrospective application of two different screening tools in comparison to clinical judgement. For sepsis, Robson tool: sensitivity 75% (p<0.001), BAS 90-30-90: sensitivity 43% (p<0.001), clinical judgement: 12% accuracy (95% CI not reported). For severe sepsis, Robson tool: sensitivity 93% (p<0.001), BAS 90-30-90: sensitivity 70% (p<0.001), clinical judgement: 17% accuracy (95% CI not reported) | ⊕⊙⊙⊙ very low |
| Bayer et al30 Retrospective application of three different screening tools. (Modified) Robson tool: sensitivity 0.95, specificity 0.43, PPV 0.32, NPV 0.97. BAS 90-30-90: sensitivity 0.62, specificity 0.83, PPV 0.51, NPV 0.89. MEWS ≥4 sensitivity 0.74, specificity 0.75, PPV 0.45, NPV 0.91 (95% CI not reported) | |||||||||
| Sepsis recognition by EMS (use of screening tool not reported) | |||||||||
| 3 | 963 | Non-RCT | Very serious‡ | None | Not serious§ | Very serious¶ | Very serious†† | Erwin et al20 Screening based on SSC criteria. For sepsis: sensitivity 33% (95% CI 18% to 53%), specificity 89% (95% CI 08% to 94%), PPV 50% (95% CI 28% to 72%), NPV 80% (95% CI 70% to 87%). For severe sepsis: sensitivity 20% (95% CI 5% to 51%), specificity 94% (95% CI 87% to 97%), PPV 29% (95% CI 08% to 64%), NPV 91 (95% CI 83% to 95%) | ⊕⊙⊙⊙ very low |
| Shiuh et al27 Screening based on SSC criteria, also stratified by lactate, lactate ≤4 ‘sepsis advisory’ while lactate ≥4 ‘sepsis alert’. 74.2% of ‘Sepsis Advisory’ patients and 76.7% of ‘sepsis alert’ patients received a hospital diagnosis of severe infection or sepsis (95% CI not reported) | |||||||||
| Travers et al28 Screening criteria not defined. Specificity 78.85% (95% CI 75.23 to 82.17), sensitivity 73.4% (95% CI 61.40 to 83.05), PPV 30.59% (95% CI 23.76 to 38.11), NPV 95.86% (95% CI 93.61 to 97.49), accuracy 78% (52 true positives, 440 true negatives) | |||||||||
*Seymour et al CIS not specific to sepsis (CIS intended to identify all cases of critical illness). Polito et al and Bayer et al studies limited to single EMS systems, Bayer et al physician-based EMS.
†Polito et al failed to report CIs, small sample size in Bayer et al study.
‡All studies patient selection/eligibility criteria, exposure/outcome reporting, confounding.
§Guerra et al, Erwin et al and Shiuh et al include lactate measurement (not widely available within EMS). In majority of studies, the population limited to single EMS agency/hospital so limited generalisablity. Bayer et al used physician-based EMS.
¶All included studies have small sample sizes, thus imprecise point estimates. In several studies, CIs are not reported.
**Guerra et al publication bias likely.
††Published in abstract only, unable to reliably critically appraise.
BAS 90-30-90, systolic blood pressure <90 mm Hg; respiratory rate >30 bpm; SpO2 <90%; CIS, critical illness score; EMS, Emergency Medical Services; MEWS, Modified Early Warning Score; modified Robson, Robson tool with addition of SpO2; non-RCT, non-randomised (observational) study; NPV, negative predictive value; PPV, positive predictive value; PRESEP, Prehospital Early Sepsis Detection; SSC, Surviving Sepsis Campaign.