Page 37 - Journal of Special Operations Medicine - Summer 2016

student was relatively inexperienced in tourniquet use, The manikin settings also included a constant simu-
whereas the clinician-scientist was a tourniquet expert lated hemorrhage rate (635mL/min). With such a
and had tourniquet experience in trauma care. The clini- hemorrhage rate, the resulting bleed-out time was 4
cian-scientist trained himself on the new tourniquets and minutes (240 seconds); thus, in the absence of any hem-
had formal military training on the standard-issue tourni- orrhage control, simulated death would occur at 240
quet. The clinician-scientist trained the student. Training seconds. If partial hemorrhage control occurred, then
included reading the instructions for use, handling the de- longer survival would result. The touchpad reported
vice, and supervised practice with each tourniquet model simulated blood loss volume as calculated from arte-
in one or two uses on the manikin before testing began. rial flow and time; although the bleeding volume per
pulse dropped over time, the pulse rate increased such
The clinician-scientist tested before the student; the that the bleeding rate, if mechanically unimpeded, was
control group was tested by each user before the study linear. Tourniquets, users, tests, and outcomes were
group. Testing of models of tourniquet was done in runs uniquely identified.
of ten tests in a row per model. The order of models
tested in the study group was TPT3, TPT2, and EMT. Results were summarized by outcome and by tourniquet
models. The critical, or primary, outcome was effective-
There were 10 tests per tourniquet model per user; ness (yes-no, hemorrhage control). Another important
hence, both users had 40 tests. The overall number of outcome was absence of palpable pulse distal to the
tests was 80 replicates for the experiment. Devices were tourniquet (yes–no). Secondary outcomes included time
examined throughout testing for structural and func- to cessation of bleeding (seconds); pressure (mmHg) ap-
tional integrity. plied to the skin (piezoelectric transducers were under
the silicone skin) by the tourniquet in an attempt to
The tourniquets were tested on a laboratory manikin achieve hemorrhage control, and the volume of simu-
that was designed to train users by providing feedback lated blood loss (mL). Effectiveness and pressure were
on user performance. The investigators used a HapMed measured by the manikin, whereas breakage and pulse
™
Leg Tourniquet Trainer (CHI Systems, http://www.chi stoppage were determined by the user. Categorical data
systems.com); a simulated right thigh with an above- such as effectiveness were summarized using percent-
knee amputation injury was the testing apparatus. 20–24 ages and compared using the chi-square test or Fisher
The medial hip had an embedded computer including a exact test, whichever was most appropriate. Continu-
smartphone-like touchpad. Software (version 1.9; CHI ous data such as time to stop bleeding, pressure applied,
Systems) integral to the thigh allowed the manikin to and blood loss volume were summarized using box-and-
stand alone and be operated by user input through fin- whisker plots, with a solid line representing the median,
ger touch on the pad. The thigh was placed on a labo- the box containing the interquartile range (IQR), and
ratory benchtop and was operated in accordance with the whickers representing the upper and lower bounds
the manufacturer’s instructions. The thigh did not bleed, excluding outliers based on the 1.5 IQR criteria (i.e.,
but bleeding was represented by red lights that transil- data points more than [(quartile 1 – IQR) × 1.5] were
luminated the wound. The number of lights illuminated considered outliers). Analysis of variance (ANOVA)
25
represented the bleeding rate: all 26 lights illuminated with a Tukey adjustment for pairwise comparison was
indicated uncontrolled bleeding; few lights blinking in- used to compare the tourniquets. Tiers separating the
dicated intermediate control; and no lights illuminated tourniquet performance were based on the Tukey ad-
indicated bleeding had stopped. Arterial pulses were pal- justed p-value being <.05. All analyses were performed
pable in the popliteal area. Touchpad readouts for each using JMP version 10.0 (SAS, http://www.sas.com).
iteration included hemorrhage control, the time to stop
bleeding, the pressure exerted under the tourniquet, and
the simulated blood loss volume. The measure of time Results
to stop bleeding extended from the start of the iteration
until the manikin detected that no more blood was lost. Bleeding Control and Pulse Stoppage
Effectiveness was determined by the cessation of blood Results of bleeding control (yes-no) by model of tour-
loss (i.e., hemorrhage control). Iterations began with a niquet showed no significant differences: all four tour-
tourniquet laid out flat and undone on the benchtop. niquet models tested were 100% effective. Similarly,
Iterations ended when the user touched the touchpad results of pulse stoppage distal to the tourniquet (yes-
button, assessing that the hemorrhage was stopped. no) were 100% effective for all four models.
Users tightened tourniquets until they perceived that
simulated bleeding stopped. The casualty had a medium Safety-Related Events
build and the setting was “Care Under Fire,” a setting No tourniquet broke, no injury occurred to the mani-
resembling emergency care when under gunfire. kin, and no tourniquet applier sustained any injury.

Comparison of Pneumatic Tourniquet Models 23