Page 53 - Journal of Special Operations Medicine - Summer 2015

of the US Army Institute of Surgical Research. A labora- included a smartphone-like touchpad. Software (version
tory experiment was designed to compare the function 1.9, CHI Systems Inc.; www.chisystems.com/index.php)
of improvised tourniquets with and without windlass integral to the thigh allowed the manikin to stand alone
use. The study design was based on first responder ac- and be operated by user input through finger touch on
tions in Boston. the pad. The thigh had no blood, but bleeding was rep-
resented by red lights that transilluminated the wound.
The study group was a set of tests of an improvised tour- The number of lights illuminated represented the rate
niquet design with a windlass. The strap-and-windlass of bleeding—all 26 lights on meant maximal bleeding;
design included a strap that was a cotton tee shirt; the no lights on meant bleeding had stopped. Users tight-
shirt was used for each test. For this experiment, a set ened tourniquets until they perceived that simulated
of standard bamboo chopsticks was used as the wind- bleeding stopped or until efforts proved futile. Arterial
lass mechanism; we used six chopsticks taped together pulses were palpable in the popliteal artery area behind
into a functionally single windlass. The windlass reli- the knee. The time for hemorrhage control was that in-
ably kept itself in a bundle. The windlass, after insertion terval from iteration initiation until cessation of bleed-
into the tourniquet knot, was twisted by the user in 180° ing, as evidenced by the absence of lights. Effectiveness
turns, thereby tightening the tourniquet strap. The con- was defined as cessation of blood loss. When hemostasis
trol group was constituted similarly to the study group was achieved, users stopped turning the windlasses. The
except there was no windlass used. manikin settings included a constant hemorrhage rate
(635mL/min); the resulting bleed-out time in this sce-
There were two tourniquet users—one experienced and nario was 4 minutes—240 seconds in which to success-
one inexperienced in tourniquet use. The experienced fully apply the tourniquet. The system reported blood
user always preceded the inexperienced user, and the loss volume as calculated from the product of hemor-
control group was tested by each user before the study rhage rate and time until hemorrhage control. The casu-
group was tested. There were 40 tests per group per alty had a medium build and the setting was Care Under
user; hence, each user had 80 tests (40 tests times two Fire, a setting resembling emergency care when under
groups times two users), or 160 tests altogether for the gunfire.
experiment.
The critical outcome was effectiveness (hemorrhage con-
A black tee shirt (lightly worn, cotton, short sleeve, trolled: yes or no). An important outcome was absence
large men’s size; Lands End, Inc.; www.landsend.com) of palpable pulse distal to the tourniquet (yes or no).
was used in the trials. The shirt was folded into a strap Minor outcomes included time to cessation of bleeding
to encircle the limb. The line of folding was diagonal (seconds), pressure applied to the skin by the tourniquet
from one sleeve to the opposite waist to maximize cir- (mmHg), and the volume of blood loss (mL). Effective-
cumferential length around the thigh. Users wrapped ness, time to stop bleeding, and pressure were measured
the shirt around the manikin at the proximal thigh, ty- by the manikin, while pulse stoppage was measured by
ing a half-knot and pulling tightly to maintain tension the user. Historically, a threshold has been used as a
in the strap and create pressure on the underlying skin. rough guide to tourniquet effectiveness such that when
The user terminated the test when one of three condi- 80% or more of uses are successfully effective, then the
tions existed: (1) hemorrhage was controlled; (2) there tourniquet has reached a minimal level of reliability—
was futility after repeated efforts to generate sufficient a so-called 80% solution. Descriptive statistics were
8
tension (repeated efforts led only to unceasing failure); used to analyze results. For categorical variables, a chi-
or (3) when unsafe use occurred (e.g., lacerated skin of squared test was used and the likelihood ratio p values
the manikin). were reported (SAS Institute Inc.; www.sas.com). For
continuous variables, a mixed model was used with user
When a windlass was tested, the same procedure was as a random effect, as there was a clear user difference in
used, except the user put the windlass atop the half-knot the results. Confidence limits were adjusted Wald 95%
8
and then tied another half-knot atop the windlass before confidence intervals (CIs). Significance for results was
twisting it to wind the knot and supposedly create more established when p values were less than .05.
strap tension.

The tourniquets were tested on a manikin in the labora- Results
tory. The investigators used a HapMed Leg Tourniquet
™
Trainer (CHI Systems Inc.; www.chisystems.com/index The Role of the Windlass in
.php)—a simulated right-thigh (leg number 000F) with Improvised Strap-and-Windlass Tourniquet Use
an above-knee amputation injury was the testing appa- Without a windlass, improvised tourniquets failed to
ratus. The medial hip had an embedded computer that stop bleeding 79 times out of 80 tests (99%; 95% CI,
6, 7

Windlass in Improvised Tourniquet Use 43