Page 25 - Journal of Special Operations Medicine - Winter 2016

Five representative graphs (Figures 1–5) are presented Figure 4 One parameter as a difference between two users
to represent the spectrum of results. The first graph is with one tourniquet model. A difference (white dots) between
one user’s learning curve of one parameter with one two users was graphed for the number of windlass turns. The
tourniquet model (Figure 1). The second graph depicts trend line with this parameter was virtually flat, indicating
the three continuous parameters (time, pressure, blood no learning.
loss) for one user with one tourniquet model (Figure 2).
The third graph includes data of one parameter for one
user with three tourniquet models (Figure 3). The fourth
chart depicts one parameter as a difference between two
users with the same tourniquet model (Figure 4). The
fifth graph depicts a second parameter as a difference
between two users with one model (Figure 5).

Figure 1 One user’s learning curve of one parameter with
one tourniquet model. Time to effectiveness (white dots)
by use number. The trend line (black line) is a power curve, Figure 5 One parameter as a difference between two users
a classically shaped learning curve showing moderately with one tourniquet model. A difference between two
increased user speed over time. users was graphed for time to effectiveness. The trend line
crossed zero at 101 uses, indicating that the less experienced
tourniquet user then performed equally to the expert user; it
took 101 to become expert.

Figure 2 Three parameters for one user with one tourniquet
model. Three curves were made (time: solid line; blood loss:
short dash; pressure: long dash). Pressure data are hidden for
clarity of other data (i.e., time [gray open circles]; blood loss
[gray open squares]). Time to effectiveness and blood loss
were classically shaped learning curves showing improved
user performance over time as user experience was accrued.
However, pressure was virtually flat, indicating no learning
with this parameter selected.
Statistical methods include descriptive statistics used to
summarize results. Graphic charts were used to aid hy-
pothesis generation for future study. Measures of learn-
ing were extracted from graphic displays (Microsoft
Excel 2010, Redmond, WA) including best-fit trend lines
2
of the data. R-squared (R ; coefficient of determination)
values were calculated using Microsoft Excel and pro-
vided information on how well data fit the calculated
model; they also represent the amount of variability in
Figure 3 One parameter for one user with three tourniquet the dependent (y) factor that is accounted for by vari-
models. Three curves were made (C-A-T: solid line; ability in the independent (x) factor.
SOFTT-W: short dash; RMT: long dash). Data points are
hidden to see the trend lines more clearly. All three trend
lines showed similar learning in classic power curves; the user
became faster over time as more experience was accrued. Results
Results of One Parameter for
One User With One Tourniquet Model
We first examined the learning curve for one user with
the standard-issue military tourniquet, the C-A-T, and
evaluated time to effectiveness. The response curve of
the less experienced user (trend line in Figure 1) was
approximately parabolic, wherein the trend line for the

Learning Curves of Emergency Tourniquet Use 9