Page 46 - Journal of Special Operations Medicine - Spring 2014

differentiate patients with respect to their survival prob- of prioritizing immediate patients? If yes, when exactly
abilities. To overcome this limitation, the Sacco method should we deviate? While extensive research is needed to
calls for categorizing patients according to their RPM answer these questions confidently, we describe two spe-
(respiratory rate, pulse rate, and motor response) scores, cific methods, called ReSTART and Simple-ReSTART,
which can take one of 13 possible values implying po- mainly to illustrate one way of using the broad guide-
tentially 13 different triage classes. While the question lines depicted by Figure 2 in practice. Both methods are
of whether RPM scoring is superior to START classifica- supported by our mathematical analysis, and simulation
tion is beyond the scope of this article, this more detailed studies show that they are superior to the current stan-
classification comes with a cost: it is difficult to man- dard practice, but they should serve as a starting point
age the whole response effort logistically when there are for more research on this topic and not as a definite
many patients belonging to one of 13 different classes, word on how prioritization decisions should be made.
7
and it is more difficult to incorporate human input (e.g.,
to decide when to override and/or how to modify the pri- To make the ReSTART policies accessible to a wide audi-
oritization policy suggested by the Sacco Triage Method ence, we developed a web-based decision support tool,
in response to changes in the field). available at http://www.restarttriage.com/. The tool al-
lows users to set input parameters such as survival proba-
We take a different approach: Recognizing that it may bilities, travel time, and number of ambulances on sliding
be difficult to reliably estimate all of the parameters in scales. The model output is displayed in a user-friendly
a large model, we model patient prioritization in a way interface that demonstrates the effects of changing these
that works within existing triage frameworks, requires parameters without requiring the use of any mathemati-
relatively simple information, provides a strategy that cal expressions, by displaying a dynamic, interactive
serves as a broad guideline for patient prioritization, chart similar to Figure 2. The chart displays the policy
and encourages providers to incorporate clinical judg- suggested by ReSTART as a function of the number of
ment. The model proposed in this paper was developed delayed patients (on the horizontal axis) and the number
by making parsimonious use of mathematical analysis. of immediate patients (on the vertical axis). The user can
Because of the widespread adoption and acceptance of hover over any point in the chart to see the exact policy.
color-coded triage systems using a small number of triage The tool also provides a proof of concept for the use of
classes, our goal from the beginning was to investigate ReSTART in a practical setting. If medical providers wish
how systems such as START or SALT can be extended to adopt ReSTART, or a similar policy, in the field they
so that priority decisions are made with resource limita- can use the mobile version of the web site, which provides
tions in mind. Having a small number of triage classes to a quick way to calculate S and provides a visualization
deal with is not helpful purely for operational efficiency. of the policy that has been adapted for display on small
The simplicity also makes it possible for triage officers or screens. A mobile app is currently in development so that
emergency responders to have an intuitive understanding on line connectivity will not be required for real time use.
of why the prioritization policy makes sense. It is also
important to point out that for the proposed model and Because the calculations involved in determining the
the methods to work classification based on START is policy are very simple, the web-based implementation
not a must. Any triage protocol meeting the uniform cri- does not require any delay for “solving” the problem,
teria described by Lerner et al. will use the immediate so it can offer instant feedback by showing how the
2
and delayed classes. As long as the survival probability transportation prioritization should change if, for ex-
functions for these classes follow the intuitive structure ample, the classification of one of the patients changes
shown in Figure 1, our model and policies point to a rea- due to retriage or if additional resources arrive.
sonable way of factoring in resource limitations.

Implementation Support Study Limitations
The model depicted in Figure 2 provides broad guid- The method described above is meant to introduce per-
ance for making prioritization decisions. It can serve sonnel making triage decisions to the fact that scarcity
as a template for developing more specific protocols to of transport resources should be taken into account
handle future mass-casualty events, or it can help deci- when triaging patients and that in resource constrained
sion makers who mostly act based on intuition make environments this approach might suggest the need to
more informed decisions in real time in the field. The transport delayed patients before immediate patients
insights provided by this model should be taken as the in order to maximize survival. In a perfect world, we
main contribution of this work. When making decisions would have accurate parameters for T and know the
in the field based on these insights, providers will have shapes of the survival curves. As with Sacco we have
to answer two difficult questions: Are the resources re- had to estimate these numbers. More research is needed
stricted enough to deviate from the standard practice to better define these parameters.

38 Journal of Special Operations Medicine Volume 14, Edition 1/Spring 2014