Page 32 - JSOM Winter 2017

persons participated one at a time to apply tourniquets: a cli- Wilcoxon method or Tukey least significant difference (LSD)
nician-scientist, an associate researcher, a military cadet, and method was used, whichever was most appropriate. The num-
a research scientist, numbered users 1 through 4, respectively. ber of pairwise comparisons for the nine glove groups was
36 ([{9 – 1} × 9]/2); the number of pairwise comparisons for
The limb tourniquet model was the generation 6 Combat Ap- the four individual users and six individual tourniquet devices
plication Tourniquet (C-A-T; C•A•T Resources, www.combat was six and 15, respectively. LSD and Dunnett corrections
tourniquet.com). Tourniquets were already unwrapped and were used for pairwise comparisons. Pairwise comparisons
reconfigured for one-handed use. The user donned the gloves of group means were then put into levels based on statistical
by glove group in a randomized order. After starting, the user significance.
unrouted the tourniquet band by removing it from its course
through a slit in the buckle, and then the user applied the tour- A mixed-model ANOVA was also used, which included the
niquet to the manikin. If wear or tear of a tourniquet device user as a random effect in the model. User effects were pre-
occurred, it was replaced upon completion of the test. Each sented as a percentage of the overall variance component
user started with an unused tourniquet. Two devices sustained based on the restricted maximal likelihood variance method.
wear or tear and were replaced, making a total of six devices R was reported as the percentage of the response variable
2
used in the study. variation that is explained by a linear model. Significance for
results was established when p < .05. All statistical analyses
Because the intervention required the glove group, the tour- were conducted with SAS software (JMP version 12.0; SAS
niquet, and the user altogether, there was no capacity to sta- Institute, http://www.sas.com) and MS Excel 2003 (Microsoft;
tistically differentiate between them for an individual test. www.microsoft.com).
20
However, looking at all tests allowed analysis of effects of us-
ers, gloves, and devices. The number of tests was 180. The Results
four users performed five tests of the nine glove groups each,
for a total of 45 tests each. Effectiveness in Bleeding Control
For users, 179 of the 180 tests (99%) ended with effectiveness,
The manikin on which we tested the performance of tourniquet and interuser differences were not statistically significant (p =
use was reported previously. Briefly, a HapMed Leg Tour- .4252). User 2 had the ineffective test.
22
™
niquet Trainer (CHI Systems, http://www.chisystems.com)
simulated a right thigh with an amputation injury. Run-time For gloves, 99% of the tests resulted with effectiveness, and
feedback was off. Metrics included effectiveness in bleeding intergroup differences were not significant (p = .8154). The
control as a yes or no result, time in seconds to determination ineffective test was with cold gloves layered under mittens.
of bleeding control, trial time (sum of time to determination
of bleeding control and time to remove the tourniquet from For devices, six individual tourniquets were used. Users 1 and
its wrapper), distal pulse stoppage (yes or no), patient status 3 had one each; users 2 and 4 had two each. Of 180 tests, 179
(stable, bleeding, or dead), tourniquet placement (none [no resulted in effectiveness, and the ineffective test was with de-
tourniquet detected], good [location], or incorrect [location]), vice U2-2 (user 2’s second device). Interdevice differences were
tourniquet pressure (amplitude in mmHg and categorized as not significant (p = .2965).
loose, good, or tight), blood loss in milliliters, and test status.
Test status was a composite binary result (i.e., satisfactory or Pulse Stoppage
unsatisfactory result) indicating whether the test was, in ag- For users, gloves, and devices, 99% of tests resulted in pulse
gregate, satisfactory for all the following metrics: patient sta- stoppage, and interuser, interglove, and interdevice differ-
tus was stable, tourniquet placement was good, and pressure ences were not significant (p = .4252, .8154, and .2965, re-
was good. The manikin reported satisfactory test status as a spectively). User 2 had the test without stoppage, which was
‘go’. Pulse stoppage was determined by the user; the manikin performed with cold gloves layered under mittens and device
determined the rest. U2-2.

All data were generated during the present study except for Patient Status
time to unwrap, which came from the prior study. Unwrap- For users, 99% of tests resulted with the patient as stable, and
11
ping times were specific to the user, glove group, and model of interuser differences were not significant (p = .4252). User 2’s
C-A-T tourniquet, but there was only one set of nine times for final test ended with bleeding.
each user because there were nine glove groups. To simulate a
realistic time it takes to use a tourniquet, an additional metric For gloves, results were similar: 99% of tests ended with the
was generated by summing the time to unwrap and the trial patient stable; one test, in which cold gloves were layered un-
time; this was used as an overall time. No bleeding occurred der mittens, ended with bleeding (p = .8154).
while unwrapping.
For devices, 99% of tests resulted with the patient as stable.
Descriptive statistics were used to portray results. Categorical The test with bleeding was with device U2-2. Interdevice dif-
data were analyzed by contingency tables, and likelihood ra- ferences were not significant (p = .2965).
tios were calculated. Continuous data (e.g., time to determina-
tion of bleeding control) were summarized by median (range) Tourniquet Placement: Incorrect Placement or
or mean (standard deviation), which were analyzed using Good Placement at the Correct Site
analysis of variance (ANOVA) to see differences. Fixed-effect For users, gloves, and devices, 99% of tests resulted with good
tests were made by glove group and by tourniquet device. tourniquet placement, and interuser, interglove, and inter-
For pairwise comparisons of group means, a nonparametric device differences were not significant (p = .4252, .8154, and

30 | JSOM Volume 17, Edition 4/Winter 2017