Table 4.

Summary of findings.

No of studies	No of patients	Study design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other	Level of evidence	Findings
Impact of prehospital care upon time to antimicrobial therapy
1	199	RCT	not serious1	none	not serious2	very serious3	very serious4	⊕⊙⊙⊙ very low	[Chamberlain 2009] prehospital antibiotics provided 3.4 ± 2.6 hours sooner (p=0.02).
5	1,927	non-RCT	very serious5	none	not serious2	very serious6	very serious7	⊕⊙⊙⊙ very low	[Band 2011] Median time to antibiotics reduced: 116 minutes (IQR 66–199 minutes) EMS vs 152 minutes (IQR 92–252 minutes) ‘other means’ (p≤0.001). [Studnek 2012] if arriving by EMS vs other means time to antibiotics reduced 111 minutes (EMS) vs 146 minutes (non-EMS); (p=0.001). If EMS recognized and documented sepsis time to antibiotics reduced 70 minutes (documented) vs 122 minutes (not documented) (p=0.003). [Bayer 2013] Median time of administration 19 minutes (IQR 18–24 minutes) after initial emergency call (time of administration estimated as 10 minutes after arriving at scene). [Guerra 2013] No significant reduction in time to antibiotics mean 72.6 minutes (SD 59.3 minutes, pre-alert) vs 98.5 minutes (SD 89.9 minutes, no pre-alert) (p=0.07). [Femling 2014] Time to antibiotics: 87 minutes (EMS, IQR 44–157 minutes) vs 120 minutes (non-EMS, IQR 141–271 minutes), difference 33 minutes (p=0.02).
Impact of prehospital care upon fluid resuscitation
5	2,697	non-RCT	very serious5	none	not serious2	very serious6	very serious8	⊕⊙⊙⊙ very low	[Seymour 2010] patients who received prehospital fluids had shorter time to MAP>65 mm Hg 17/24 (70%, EMS IV fluids) vs 12/26 (44%, no IV fluids), unadjusted RR 1.53 (95% CI [0.9–2.65]), and shorter time to CVP>8 mm H20 15/25 (60%, EMS IV fluids) vs 17/24 (70%, no IV fluids), unadjusted RR 1.2 (95% CI [0.8–1.8]). [Band 2011] Median time to initiation of IVF reduced: 34 minutes (IQR 10–88) EMS vs 68 minutes (IQR 25–121 minutes) ‘other means’ of arrival (p≤0.001). [Bayer 2013] Median time of administration 19 minutes (IQR 18–24 minutes) after initial emergency call (time of administration estimated as 10 minutes after arriving at scene). Patients received 2.5L intravascular fluid (IQR 1.5–3.0L) until admitted to the ED. [Guerra 2013] No significant difference in fluid administration by 6 hours 42.97 cc/kg (SD 33.23cc/kg, pre-alert) vs 35.17cc/kg (SD 26.81 cc/kg, no pre-alert, p=0.30). [Seymour 2014] Median prehospital fluid volume 500mL (IQR 200–1000mL).
Impact of prehospital care upon Early Goal Directed Therapy
6	2,523	non-RCT	very serious5	none	not serious2	very serious6	very serious8	⊕⊙⊙⊙ very low	[Seymour 2010] patients who received prehospital fluids had shorter time to MAP>65 mm Hg 17/24 (70%, EMS IV fluids) vs 12/26 (44%, no IV fluids), unadjusted RR 1.53 (95% CI [0.9–2.65]); shorter time to CVP>8 mm H20 15/25 (60%, EMS IV fluids) vs 17/24 (70%, no IV fluids), unadjusted RR 1.2 (95% CI [0.8–1.8]); and shorter time to SVCO2>70% 13/24 (54%, EMS IV fluids) vs 9/25 (36%, no IV Fluids), unadjusted RR 1.5 (95% CI [0.8–2.9]). [Studnek 2012] if arriving by EMS vs other means time to EGDT reduced 119 minutes (EMS) vs 160 minutes (non-EMS, p=0.005). If EMS recognised and documented sepsis time to EGDT 69 minutes (documented) vs 131 minutes (not documented, p=0.001). [Guerra 2013] No significant reduction in proportion of patients with central venous line placement 62% (pre-alert) vs 68% (no pre-alert, p=0.54). [Femling 2014] Time to central line: 200 minutes (EMS, IQR 89–368 minutes) vs 275 minutes (non-EMS, IQR 122–470 minutes), difference 75 minutes (p<0.01). [Seymour 2014] Prehospital fluids reduced likelihood of increasing organ failures adjusted OR 0.58 (95% CI [0.34–0.98]). [McClelland 2015] Time to ‘sepsis 6’: mean 205 minutes (SD 271 minutes, range 10–720 minutes, EMS identified)* vs 120 minutes (SD 110, 17–450 minutes, not identified). (*Includes outlier where the fluid balance chart was not started for 12 hours, excluding this case mean 76 minutes [SD 95 minutes, range 10–240 minutes]).
Impact of prehospital care upon admission
3	646	non-RCT	very serious5	none	not serious2	very serious6	very serious8	⊕⊙⊙⊙ very low	[Guerra 2013] No significant reduction in length of stay: mean 7.3 days (SD 6.8 days, pre-alert) vs 8.4 days (SD 8.8 days, no pre-alert, p=0.65). [Femling 2014] Length of stay: 15 days (IQR 13–17 days, EMS) vs 14 days (IQR 10–17 days, non-EMS), difference 1 day, not significant. [Seymour 2014] Prehospital vascular access reduced ICU admission adjusted OR 0.41 (95% CI [0.24 – 0.70]). [McClelland 2015] ICU admission: 4% (1/23, EMS identified) vs 13% (3/23, not identified).
Impact of prehospital care upon mortality
5	2,959	non-RCT	very serious5	none	not serious2	very serious6	very serious8	⊕⊙⊙⊙ very low	[Band 2011] No significant difference in mortality was noted: adjusted RR 1.24 (95% CI [0.92 – 1.66, p=0.16). [Guerra 2013] If hospital was ‘pre-alerted’, unadjusted mortality was improved OR 3.19 (95% CI [1.14–8.88], p=0.04). [Femling 2014] No significant difference in mortality was noted 113/378 (30%, EMS) vs 34/107 (31%, non-EMS), difference 1%, not significant. [Seymour 2014] Prehospital vascular access reduced mortality adjusted OR 0.31 (95% CI [0.17 – 0.57], p<0.01). [McClelland 2015] 3 month mortality 21% (5/24, EMS identified) vs 16% (4/25, not identified).
Impact of prehospital antimicrobial therapy on ICU admission
1	199	RCT	not serious1	none	not serious2	very serious3	very serious4	⊕⊙⊙⊙ very low	[Chamberlain 2009] Mean ICU length of stay: reduced 6.8 ± 2.1 days (intervention) vs 11.2 ± 5.2 days (control, p=0.001).
Impact of prehospital antimicrobial therapy on mortality
1	199	RCT	not serious1	none	not serious2	very serious3	very serious4	⊕⊙⊙⊙ very low	[Chamberlain 2009] 28-day mortality reduced: 42.4% (intervention) vs 56.7% (control), OR 0.56 (95% CI [0.32 to 1.00], p=0.049).
Impact of prehospital intravenous fluid therapy on ICU admission
1	1,350	non-RCT	not serious9	none	not serious2	none	none	⊕⊙⊙⊙very low	[Seymour 2014] Prehospital fluids did not reduce likelihood of ICU admission adjusted OR 0.64 (95% CI [0.37–1.10]).
Impact of prehospital intravenous fluid therapy on mortality
1	1,350	non-RCT	not serious9	none	not serious2	none	none	⊕⊙⊙⊙very low	[Seymour 2014] Prehospital fluids reduced hospital mortality adjusted OR 0.46 (95% CI [0.23–0.88], p=0.02).
Risk of bias unclear. Single centre study may limit generalizability. Small study numbers limits precision/accuracy. Published in abstract only, insufficient detail to rule out other bias. Concerns relating to eligibility, exposure, confounding, follow-up Small study numbers limits precision/accuracy, failure to report confidence intervals (Guerra) Abstract only publication (Femling), insufficient detail to rule out other bias, Publication bias likely (Guerra) Publication bias likely (Guerra) Risk of bias unclear

RCT, randomized control trial; EMS, emergency medical services; ED, emergency department; IQR, interquartile range; CI, confidence interval; RR, risk ratio; MAP, mean arterial pressure; CVP, central venous pressure; IVF, intravascular fluid; IV, intravascular; SVC_O2, superior vena cava oxygen; EGDT, early goal directed therapy; OR, odds ratio; ICU, intensive care unit; RCT, non-randomized controlled trial (observational study); SD, standard deviation.