Page 54 - Journal of Special Operations Medicine - Summer 2015

93%–100%) (Table 1). With a windlass, improvised it was 114mmHg (95% CI, 92mmHg–136mmHg; p <
tourniquets failed to stop bleeding 26 times out of 80 0.0001; 95% CI for difference, 42mmHg–92mmHg).
tests (32%; 95% CI, 23%–43%) (Table 1). The differ- Windlass use increased the pressure under the tourniquet
ence in proportions, 99% versus 32%, was statistically compared to no windlass use.
significant (p < .0001). However, neither group was reli-
ably effective: Both with and without a windlass, impro- The mean blood loss volume with no windlass was
vised tourniquets did not achieve 80% effectiveness, the 415mL (95% CI, 383mL–446mL) and with a windlass
minimum threshold of reliable effectiveness. was it was 648mL (95% CI, 596mL–700mL; p < .0001;
95% CI for difference, 172mL–294mL). When blood
Table 1 Hemorrhage Control Results by Windlass or loss was measured as volume per time, the windlass
No Windlass tests bled at an average of 6.7mL/s (95% CI, 6.5mL/
s–6.9mL/s) until bleeding was stopped, while the tests
Failed Hemorrhage Control without a windlass bled at an average of 7.1mL/s (95%
Windlass Adjusted Wald Total CI, 6.9mL/s–7.3mL/s) throughout the test period.
Used Tests, No. (%) 95% CI Tests, No.
No 79 (99) 93%–100% 80 The Role of the User
Yes 26 (32) 23%–43% 80 There were interesting results that varied by user. Even
with the user effect taken into account in the statistical
Note: CI, confidence interval.
methods, there was a very significant windlass use ef-
Pulse results were nearly the same as the hemorrhage fect. For both users, the results of tests with no windlass
control results in that the windlass played a major role were similar in that almost every test failed. However,
in improvised tourniquet performance (Table 2). In tests tests with a windlass varied by user.
with no windlass, attempts to stop the pulse failed every
time (80 of 80 tests, 100%; 95% CI, 96%–100%). With The user with more experience had faster tests (mean
a windlass, however, attempts to stop the pulse failed time, 70 seconds vs. 87 seconds, p < .0001; 95% CI
31% (95% CI, 22%–42%) of the time (25 of 80 tests); for difference, 6–27 seconds). With these shorter times
the difference in proportions was significant (p < .0001). to stop bleeding, the mean blood loss was also less for
The similarity between results of pulse stoppage and the user with more experience (mean volume, 458mL
hemorrhage control indicated that the two phenomena vs. 604mL; p < .0001; 95% CI for difference, 78mL–
were closely related. 213mL). However, the users differed greatly in pres-
sure. The mean pressure applied by the experienced user
Time to Bleeding Cessation, Pressure, was 15mmHg, while the less-experienced user applied
and Blood-Loss Volume Results a mean pressure of 145mmHg (p < .0001; 95% CI for
The mean time to bleeding cessation with no windlass difference, 111mmHg–148mmHg). Based on these re-
was 59 seconds (95% CI, 54–64 seconds), and all but sults, we decided it was necessary to consider the user
one test with no windlass ended in failure (79 of 80 a random effect in the mixed statistical modeling when
tests; 95% CI, 93%–100%). On the other hand, the comparing windlass type for the factors of interest: time
mean time to bleeding cessation with a windlass was to stop bleeding, pressure, blood loss, and blood loss
98 seconds (95% CI, 90–105 seconds), and 32% (26 of per second.
80 tests; 95% CI, 23%–43%) of tests ended in failure.
The difference in mean times was significant (p < .0001; Discussion
95% CI for difference, 28–48).
The first major finding of the present study is that the
Table 2 Pulse Stoppage Results by Windlass or No Windlass performance of improvised tourniquets varied by design,
with the strap-and-windlass method performing sub-
Failed Pulse Cessation stantially better than the strap with no windlass. Tour-
Windlass Adjusted Wald Total niquets with a windlass had higher proportions of tests
Used Tests, No. (%) 95% CI Tests, No. with hemorrhage control, higher proportions of tests
No 80 (100) 96%–100% 80 with suitable pressures, and lower rates of blood loss.
Yes 25 (31) 22%–42% 80 Mean blood loss volumes with a windlass were more
because such tests were reliably effective and lasted lon-
Total 105 160
ger while tests with no windlass were not reliably effec-
Note: CI, confidence interval. tive and ended earlier. Use of a windlass is historically
intended to gain a mechanical advantage in tightening a
The mean pressure applied with no windlass was 46mmHg tourniquet. As a matter of fact, this windlass role is not
(95% CI, 35mmHg–58mmHg), whereas with a windlass, specific to tourniquets but applies to hauling and lifting

44 Journal of Special Operations Medicine Volume 15, Edition 2/Summer 2015