Included Studies

After duplicate removal, 3,116 studies were identified. Title and abstracts were screened for eligibility, and the remaining 185 studies were screened for eligibility through full-text review. Finally, 112 studies were excluded. A total of 344 on-going studies were identified at clinicaltrials.gov, WHO ICTRP, and PROSPERO. This resulted in 73 eligible studies. Another four studies were included after reviewing the reference lists of the systematic reviews and literature reviews. Two of the studies were not found in the original search, one because it was published before 2000¹¹ and the other because tourniquet was not mentioned in the title or abstract. ^{Reference Kotwal, Montgomery and Kotwal12} Two more studies were originally excluded from the title and abstract review, but were included after reading the full-text from references: one case report ^{Reference Hauschild, Noldge-Schomburg and Hoitz13} and one literature review. ^{Reference Fox, Patel and Clouse14} After excluding all non-primary literature such as literature reviews ^{Reference Fox, Patel and Clouse14–Reference Niven and Castle29} and systematic reviews, ^{Reference Mawhinney and Kirk30–Reference Bulger, Snyder and Schoelles35} a total of 55 studies were included in this systematic review. The process from search result to inclusion is illustrated in Figure 1.

Figure 1. PRISMA Flow Diagram of the Selection of Included Studies.

Quality of Evidence

All included studies were published from January 1, 2000 through January 10, 2019 and consisted of 15 civilian retrospective observational studies, ^{Reference Kue, Temin and Weiner36–Reference Passos, Dingley and Smith50} two civilian case series, ^{Reference Rtshiladze, Andersen and Nguyen51,Reference Callaway, Robertson and Sztajnkrycer52} 12 civilian case reports, ^{Reference Hauschild, Noldge-Schomburg and Hoitz13,Reference Gouk, Pasricha and Lingathas53–Reference Kippnich, Jelting and Kraus63} six prospective observational military studies, ^{Reference Kragh, Walters and Baer64–Reference Kragh, Beebe and O’Neill69} 16 military retrospective observational studies, ^{Reference Pillgram-Larsen11,Reference Kotwal, Montgomery and Kotwal12,Reference Staudt, Gurney and Valdez-Delgado70–Reference Shlaifer, Yitzhak and Baruch83} two military case series, ^{Reference Dayan, Zinmann and Stahl84,Reference Nelson, Clark and Stedje-Larsen85} one military case study, ^{Reference Mucciarone, Llewellyn and Wightman86} and one military case report. ^{Reference Kragh, Baer and Walters87} The included studies are summarized in Table 1. None of the studies reported on all of the criteria, with 36%-87% of the criteria being met. No published high-quality studies were found, all published studies had very low evidence according to GRADE, mostly due to their observational character and small sample size. ^{Reference Guyatt, Oxman and Vist10} A detailed quality appraisal is available as Appendix III (available online only).

Table 1. An Overview and Summary of the Included Studies, Excluding Case Reports

Abbreviations: CS, case series; CSt, case study; OP, observational prospective; OR, observational retrospective; TQ, tourniquet.

Some of the studies included tourniquets applied in the emergency department alongside tourniquets applied in the prehospital setting. When the two groups were clearly differentiated, only the report on the prehospital tourniquet application group was reported.

Mortality

Fifty studies reported mortality in patients treated with tourniquet. Twenty-seven of these were either case reports or had less than 20 participants and were not included in assessing the primary outcome. ^{Reference Pillgram-Larsen11,Reference Hauschild, Noldge-Schomburg and Hoitz13,Reference Kalish, Burke and Feldman37,Reference Ballas, Saetta and Peuchot44,Reference Duignan, Lamb and DiFiori46,Reference Passos, Dingley and Smith50–Reference Kippnich, Jelting and Kraus63,Reference Tien, Jung and Rizoli65,Reference Holcomb, McMullin and Pearse71,Reference Gerhardt, Berry and Mabry77,Reference Cheng, Zhao and Li78,Reference Dayan, Zinmann and Stahl84–Reference Kragh, Baer and Walters87} Twenty-three studies were included to assess mortality. ^{Reference Kotwal, Montgomery and Kotwal12,Reference Kue, Temin and Weiner36,Reference Inaba, Siboni and Resnick38–Reference Scerbo, Mumm and Gates43,Reference Scerbo, Holcomb and Taub45,Reference Teixeira, Brown and Emigh47–Reference King, Larentzakis and Ramly49,Reference Kragh, Walters and Baer64,Reference Kragh, Walters and Baer66,Reference Kragh, Littrel and Jones67,Reference Kragh, Beebe and O’Neill69,Reference Brodie, Hodgetts and Ollerton73,Reference Beekley, Sebesta and Blackbourne74,Reference Kragh, Dubick and Aden79–Reference Shlaifer, Yitzhak and Baruch83} Some of the studies reported data from the same population/database. Figure 2 shows data extracted from the latest published study to report mortality rate, consisting of ten civilian and seven military studies. ^{Reference Kotwal, Montgomery and Kotwal12,Reference Kue, Temin and Weiner36,Reference Inaba, Siboni and Resnick38–Reference Leonard, Zietlow and Morris42,Reference Scerbo, Holcomb and Taub45,Reference Teixeira, Brown and Emigh47–Reference King, Larentzakis and Ramly49,Reference Kragh, Beebe and O’Neill69,Reference Brodie, Hodgetts and Ollerton73,Reference Beekley, Sebesta and Blackbourne74,Reference Kragh, Dubick and Aden79,Reference Dunn, Fares and Kusnezov81,Reference Shlaifer, Yitzhak and Baruch83} The figure shows the reported survival, publication year, and sample size in each of these included studies. The studies reported a survival rate between 87%-100%. None of the studies were randomized controlled studies, and studies with comparison groups were hampered by unclear indications for tourniquet placement and bias due to a risk of more severe injuries in the tourniquet group.

Figure 2. Studies Reporting on Mortality, Excluding Case Reports and Studies with Less Than 20 Participants.

Note: Circle size indicates the number of patients treated with tourniquet. The center of the circle indicates the survival rate. Each study is represented by its reference number.

Overall, the military studies reported similar mortality rates in patients treated with tourniquets and patients who were not, despite the patients in the tourniquet group being more severely injured. Early application before the onset of shock was strongly associated with increased survival. One study ^{Reference Kragh, Littrel and Jones67} found that 96% of patients survived when the tourniquet was placed before the onset of shock compared to four percent of patients who received a tourniquet later. Patients treated with prehospital tourniquet (89%) also had higher survival rates than those who received a tourniquet upon arrival at the hospital (76%-78%). ^{Reference Kragh, Walters and Baer66,Reference Kragh, Littrel and Jones67}

Overall, studies indicated a survival benefit for patients treated with tourniquets in a civilian setting. Patients treated with prehospital tourniquet had a lower incidence of shock when arriving at the hospital compared to those who were not treated with tourniquet. ^{Reference Smith, Ochoa and Wong48} One study reported that death from hemorrhagic shock was more frequent in patients who received a tourniquet in-hospital (14%) compared to those who received a prehospital tourniquet (three percent). ^{Reference Scerbo, Holcomb and Taub45} Most studies reported an average tourniquet application time less than two hours.

Blood Transfusions

Seventeen studies reported on blood transfusions in patients treated with tourniquets: nine civilian observational studies, ^{Reference Inaba, Siboni and Resnick38,Reference Schroll, Smith and McSwain40,Reference Leonard, Zietlow and Morris42,Reference Scerbo, Mumm and Gates43,Reference Scerbo, Holcomb and Taub45,Reference Teixeira, Brown and Emigh47,Reference Smith, Ochoa and Wong48,Reference Passos, Dingley and Smith50,Reference Callaway, Robertson and Sztajnkrycer52} three military observational studies, ^{Reference Beekley, Sebesta and Blackbourne74,Reference Kragh, Nam and Berry80,Reference Nelson, Clark and Stedje-Larsen85} and five civilian case reports. ^{Reference Gouk, Pasricha and Lingathas53,Reference Fludger and Bell54,Reference Robertson, McCahill and Riddle59,Reference Malo, Bernardin and Nemeth62,Reference Kippnich, Jelting and Kraus63} In the majority of the civilian studies with a control group, patients with a prehospital tourniquet received fewer blood products than patients not treated with tourniquets or if the tourniquets were placed in-hospital. In the military studies, the tourniquet patients received more blood products than their respective control groups. In one study, ^{Reference Kragh, Nam and Berry80} the difference between the study groups was significant, but most of the patients in the tourniquet group also had non-extremity injuries, which may have affected the data.

Other Complications

Thirty-five of the studies reported on complications associated with the use of tourniquets, 21 in the civilian setting and 14 in the military setting; 10 of these were case reports. Fifteen studies reported on nerve palsy associated with the use of tourniquet. ^{Reference Kue, Temin and Weiner36,Reference Schroll, Smith and McSwain40,Reference Leonard, Zietlow and Morris42,Reference Scerbo, Mumm and Gates43,Reference Smith, Ochoa and Wong48,Reference Callaway, Robertson and Sztajnkrycer52,Reference Kragh, Walters and Baer64,Reference Kragh, Walters and Baer66–Reference Kragh, O’Neill and Walters68,Reference Lakstein, Blumenfeld and Sokolov72–Reference Beekley, Sebesta and Blackbourne74,Reference Shlaifer, Yitzhak and Baruch83,Reference Dayan, Zinmann and Stahl84} Six civilian studies reported on nerve palsies attributed to the use of tourniquet in 18 of 465 patients. One study found that tourniquets were not associated with nerve palsies when comparing two groups of patients (127 treated with tourniquet versus 77 patients not treated with tourniquet). ^{Reference Smith, Ochoa and Wong48} Neurological complications seemed to be few and most were transient.

Nine studies reported on amputations as a complication of the use of a tourniquet, one of which was a case report. Four of the studies were civilian ^{Reference Inaba, Siboni and Resnick38,Reference Scerbo, Mumm and Gates43,Reference Callaway, Robertson and Sztajnkrycer52,Reference Malo, Bernardin and Nemeth62} and five were military. ^{Reference Kragh, Walters and Baer64,Reference Kragh, Walters and Baer66,Reference Beekley, Sebesta and Blackbourne74,Reference Dayan, Zinmann and Stahl84,Reference Mucciarone, Llewellyn and Wightman86} To summarize, very few amputations were solely due to the use of tourniquets, but the few casuistic amputations related to tourniquet use per se were mainly related to improvised tourniquets and prolonged tourniquet application time.

A need for fasciotomy and/or compartment syndrome in patients treated with tourniquet was reported in 17 studies, three of which were case reports. ^{Reference Kalish, Burke and Feldman37,Reference Inaba, Siboni and Resnick38,Reference Schroll, Smith and McSwain40,Reference Leonard, Zietlow and Morris42,Reference Scerbo, Mumm and Gates43,Reference Scerbo, Holcomb and Taub45,Reference Smith, Ochoa and Wong48,Reference Passos, Dingley and Smith50,Reference Callaway, Robertson and Sztajnkrycer52,Reference Mullins and Harrahill55,Reference Eikermann, Velmahos and Abbara58,Reference Kragh, Walters and Baer64,Reference Kragh, O’Neill and Walters68,Reference Brodie, Hodgetts and Ollerton73,Reference Shlaifer, Yitzhak and Baruch83,Reference Dayan, Zinmann and Stahl84,Reference Kragh, Baer and Walters87} Fasciotomy seemed to be a frequent treatment modality after the use of tourniquets, and compartment syndrome was not infrequent if fasciotomy was not performed. Tourniquet application time greater than two hours seemed to increase fasciotomy rates. ^{Reference Kragh, O’Neill and Walters68}

Seven studies reported on infections associated with tourniquet use: five civilian ^{Reference Inaba, Siboni and Resnick38,Reference Schroll, Smith and McSwain40,Reference Leonard, Zietlow and Morris42,Reference Teixeira, Brown and Emigh47,Reference Smith, Ochoa and Wong48} and two military studies. ^{Reference Clasper, Brown and Hill75,Reference Brown, Murray and Clasper76} The civilian studies had a total of 764 patients with tourniquets, 67 of which were reported to have a wound infection (nine percent). Tourniquets were not associated with a difference in secondary infection rates compared to the non-tourniquet groups.