Study Selection and Data Extraction

Initial database searches resulted in 1992 documents (Figure 1), of which 781 duplicates were removed. The title and abstract screening revealed 31 potential studies meeting eligibility criteria. Of these, 3 studies were not available in full text. Therefore, full-text screening was carried out on the remaining 28 studies, of which 17 were excluded, and 11 papers met the inclusion criteria. One study was included from the citation and grey literature search, resulting in 12 studies for analysis.

Figure 1. Modified PRISMA 2020 flow diagram for new systematic reviews, which included searches of databases and other sources. Abbreviation: PRISMA, Preferred Reporting Items of Systematic Reviews and Meta-Analysis.

Study Characteristics

Study methodologies comprised 6 retrospective observational cohort studies, ^{Reference Sollid, Rimstad and Rehn22–Reference Raux, Carli and Lapostolle27} and 6 case reports (Table 3). ^{Reference O’Keefe, Kheir and Martin28–Reference Leiba, Halpern and Priel33} Two studies were conducted in the USA, ^{Reference Kahn, Schultz and Miller26,Reference O’Keefe, Kheir and Martin28} 2 in Israel, ^{Reference Raiter, Farfel and Lehavi32,Reference Leiba, Halpern and Priel33} 2 in France, ^{Reference Travers, Carfantan and Luft23,Reference Raux, Carli and Lapostolle27} and 2 in the Netherlands, ^{Reference Welling, van Harten and Henny25,Reference Postma, Winkelhagen and Bloemers30} while the remaining were conducted within the UK, ^{Reference McLeod, Hodgetts and Mahoney24} Canada, ^{Reference Mulvey, Shaw and Betzner29} Norway, ^{Reference Sollid, Rimstad and Rehn22} and Taiwan. ^{Reference Lin, Pan and Wen31} Incidents were varying in nature and included terrorist attacks (n = 4) ^{Reference Sollid, Rimstad and Rehn22,Reference Raux, Carli and Lapostolle27,Reference Raiter, Farfel and Lehavi32,Reference Leiba, Halpern and Priel33} and military operations (n = 2) ^{Reference Travers, Carfantan and Luft23,Reference McLeod, Hodgetts and Mahoney24} (Supplemental Figure 1). The remainder focused on single incidents involving a balcony collapse, ^{Reference O’Keefe, Kheir and Martin28} a train crash, ^{Reference Kahn, Schultz and Miller26} a bus rollover, ^{Reference Mulvey, Shaw and Betzner29} and a building fire, ^{Reference Welling, van Harten and Henny25} as well as a flight crash, ^{Reference Postma, Winkelhagen and Bloemers30} and a non-intentional explosion. ^{Reference Lin, Pan and Wen31} Two studies focused on single transportation modes: air ambulances in French military operations, ^{Reference Travers, Carfantan and Luft23} and ground ambulances in Paris attacks, while the rest involved air, and ground ambulances. EMS providers (paramedics, emergency medical technicians EMTs, and physicians) were involved in all studies. Additionally, nurses and emergency physicians responded to 3 MCIs, ^{Reference Sollid, Rimstad and Rehn22,Reference McLeod, Hodgetts and Mahoney24,Reference O’Keefe, Kheir and Martin28} and anesthesiologists responded to the terrorist attacks in Norway. ^{Reference Sollid, Rimstad and Rehn22}

Table 3. Characteristics for the included studies in the systematic review

Abbreviations: IQR, Interquartile range; mins, minutes; N/A, The information was not reported; TA, terrorist attack; 95%CI, 95% confidence interval.

^a Mortality data from a different study that reported the same incident but did not meet eligibility criteria. ^{Reference Yang and Shih34}

^b Patients already dead on hospital arrival (transported via private vehicles).

^c The reviewers calculated the IQR median for any case report that included all patients’ PHT sequentially (per individual patient).

^d PHTs were reported for every patient in two different hospitals.

In total, 2589 patients were included, and overall mortality was 14.7% (n = 269). Most patients were formally triaged at the scene (2237, 86.4%), and “non-triage” occurred where victims were immediately transported or self-presented to the hospital. ^{Reference Sollid, Rimstad and Rehn22,Reference Kahn, Schultz and Miller26} Of the triaged patients, 482 (29.0%) were red (critical, P1), 522 (35.6.5%) were yellow (delayed, P2), 475 (32.4%) were green (minor, P3), and 172 (10.8%) were black (dead-on-scene, P4) (Supplemental Table 2). The START (Simple triage and rapid treatment) system was implemented in 3 mass incidents in North America (USA and Canada), ^{Reference Kahn, Schultz and Miller26,Reference O’Keefe, Kheir and Martin28,Reference Mulvey, Shaw and Betzner29} and the SIEVE system was utilized in 2 European MCIs (UK and The Netherlands). ^{Reference McLeod, Hodgetts and Mahoney24,Reference Postma, Winkelhagen and Bloemers30}

Quality Appraisal

Of the 6 observational studies, 4 papers were rated as moderate quality, ^{Reference Sollid, Rimstad and Rehn22,Reference McLeod, Hodgetts and Mahoney24,Reference Kahn, Schultz and Miller26,Reference Raux, Carli and Lapostolle27} and 2 were deemed high quality. ^{Reference Travers, Carfantan and Luft23,Reference Welling, van Harten and Henny25} For the case reports, 3 were deemed to be of moderate quality, ^{Reference Postma, Winkelhagen and Bloemers30,Reference Lin, Pan and Wen31,Reference Leiba, Halpern and Priel33} while 3 reports were graded as high quality (Table 4). ^{Reference O’Keefe, Kheir and Martin28,Reference Mulvey, Shaw and Betzner29,Reference Raiter, Farfel and Lehavi32}

Table 4. Quality assessment of included studies

Pre-hospital Times in Mass Casualty Incidents

Most studies documented pre-hospital times of 2 hours or longer, ^{Reference Travers, Carfantan and Luft23,Reference McLeod, Hodgetts and Mahoney24,Reference Kahn, Schultz and Miller26,Reference Raux, Carli and Lapostolle27,Reference Mulvey, Shaw and Betzner29–Reference Lin, Pan and Wen31} although PHTs were reported variably as means, medians, and sequences (per individual patient) (Table 3). Three studies focused on reporting the PHTs for individual patients who were of particular note (first and/ or last arrivals). ^{Reference Sollid, Rimstad and Rehn22,Reference Raiter, Farfel and Lehavi32,Reference Leiba, Halpern and Priel33} Two of the studies used sub-grouped medians that were based on triage classifications. ^{Reference Welling, van Harten and Henny25,Reference Kahn, Schultz and Miller26}

In studies where average timings were reported or calculated by the reviewer, the shortest time to arrival at a hospital was a median of 35 minutes, following the balcony collapse in the USA where 16 people were injured (Figure 2). ^{Reference O’Keefe, Kheir and Martin28} The longest reported time was 488 minutes, when 12 critically injured patients were evacuated following the bus rollover in Canada. ^{Reference Mulvey, Shaw and Betzner29} Otherwise, timings ranged from 120 minutes (UK military incident) to 233.5 minutes for the fire incident in the Netherlands (Figure 2). ^{Reference McLeod, Hodgetts and Mahoney24,Reference Welling, van Harten and Henny25}

Figure 2. Bar chart shows pre-hospital time in minutes for mass casualty incidents included in the review. Abbreviations: NA1, 1st incident of Norway attacks (Car bombing) 2011; BC, Balcony collapse in Virginia 1997; SB, Bombing attack in Tel Aviv, Israel 2006; BN, Bombing at a nightclub, Israel 2005; CA, UK military operations, Afghanistan 2006-2007; MC, French military combat, Sahel region 2013-2018; TC, California train crash 2002; PA, Bombing attack and a mass shooting incident, Paris 2015; FC, Flight accident, The Netherlands 2001; DE, Dust explosion, Taiwan 2015; FA, Fire accident, The Netherlands 2001; NA2, 2nd incident of Norway attacks (Mass shooting ) 2011; BR, Bus rollover, Canada 2020.

Of the studies that reported the PHTs of individual patients, ^{Reference Sollid, Rimstad and Rehn22,Reference Raiter, Farfel and Lehavi32,Reference Leiba, Halpern and Priel33} the first victims of a Norwegian car bombing arrived at trauma centers 26 minutes after the first incident (Figure 2). ^{Reference Sollid, Rimstad and Rehn22} Similarly, first patients arrived 20 minutes after a Tel Aviv bomb attack in 2006. ^{Reference Raiter, Farfel and Lehavi32} The most prolonged times for final patient arrival was 118 minutes following a nightclub bombing in 2005, and 246 minutes after the Norwegian mass shooting event. ^{Reference Leiba, Halpern and Priel33} Two civilian MCIs with the largest numbers of casualties reported pre-hospital times of 194 and 232 minutes respectively (Table 3). ^{Reference Raux, Carli and Lapostolle27,Reference Lin, Pan and Wen31} However, this volume-delay association was not observed across all MCIs, and the incident with the most prolonged times involved only 27 patients (Table 3). ^{Reference Mulvey, Shaw and Betzner29} Similarly, there was no obvious relationship between PHTs and proportion of deaths (Figure 2, Table 3), although 3 studies failed to report mortality data. ^{Reference Sollid, Rimstad and Rehn22,Reference McLeod, Hodgetts and Mahoney24,Reference Lin, Pan and Wen31}

Factors associated with delayed pre-hospital time

Difficult access to incident locations

Inaccessibility to the incident site due to geographical location or obstruction was the main reason for patient triage and transportation delays in 3 studies. ^{Reference O’Keefe, Kheir and Martin28–Reference Postma, Winkelhagen and Bloemers30} The Columbia icefield bus rollover occurred in an extreme environment, and despite having access to aeromedical services and physicians in the field, the steep slopes challenged the response teams resulting in a prolonged PHT of more than 4 hours for the first critical patient (Figure 3). ^{Reference Mulvey, Shaw and Betzner29} At the Schiphol air crash, the evacuation and transportation for the first patient was delayed 74 minutes due to extremely muddy conditions. ^{Reference Postma, Winkelhagen and Bloemers30} Similarly, road obstructions following a balcony collapse prevented EMS providers from reaching injured patients, which extended the pre-hospital period to a median of 35 minutes (Figure 3). ^{Reference O’Keefe, Kheir and Martin28}

Figure 3. Scatterplot represents factors associated with delays by pre-hospital times and number of casualties.^a Abbreviations: BC, Balcony collapse in Virginia 1997; BR, Bus rollover, Canada 2020; BN, Bombing at a nightclub, Israel 2005; SB, Bombing attack in Tel Aviv, Israel 2006; FC, Flight accident, The Netherlands 2001; MC, French military combat, Sahel region 2013-2018; FA, Fire accident, The Netherlands 2001; TC, California train crash 2002; DE, Dust explosion, Taiwan 2015; PA, Bombing attack and a mass shooting incident, Paris 2015; CA, UK military operations, Afghanistan 2006-2007.

^aThe terrorist attacks in Norway 2011 were not included in the graph as the number of casualties was not reported.