American Journal of Emergency Medicine (2009) 27, 10851090

www.elsevier.com/locate/ajem

Brief Report

External blood loss estimation using the MAR Method

Mark A. Merlin DO, EMT-P a,, Scott M. Alter BS, NREMT-P b , Brian Raffel BS a , Peter W. Pryor II MD, MPH a
University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School Department of Emergency Medicine, New Brunswick, NJ 08901, USA b University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, NJ 08854, USA
Received 9 July 2008; revised 28 July 2008; accepted 29 July 2008
a

Abstract Background: An element lacking in medical education is training to estimate blood volumes. Therefore, health care workers currently use visual estimation as their only means of determining blood volumes, which has shown to be highly inaccurate. This study proposes and tests a new method using one's fist to determine external blood loss. Methods: Increments of human whole blood were measured and used to compare fist size to surface area of blood present. A formula was created averaging blood per fist, hereafter known as the MAR Method. Two scenarios were staged using set quantities of blood (75 and 750 mL). Participants estimated blood volumes before and after being taught the MAR Method in a 1-minute session. Errors in estimation before and after using the MAR Method were compared. Results: The MAR Method was created using a fist to cover a surface area of blood that equals 20 mL. A total of 74 participants had errors of 120% and 73% for visualization of the small and large pools, respectively. For the smaller volume, the average error from the mean decreased by 76% (P b .0001), and the interquartile range of errors decreased by 60%. For the larger volume, the average error from the mean reduced by 40% (P b .0001), and the interquartile range of errors reduced by 45%. Conclusion: Use of the MAR Method improves blood volume estimations. After less than 1 minute of instruction, participants were able to determine blood volumes with improved accuracy and precision. 2009 Elsevier Inc. All rights reserved.

1. Introduction
Estimation of external blood loss is a routine occurrence at the scene of trauma as well as within the hospital setting, yet most health care providers do not receive training in determining blood volumes. Many people rely on visual

approximation to estimate external blood loss; however, until now, no method has been successful using a method to confirm estimates are close to accurate.

1.1. Background
Previous studies have shown that both prehospital and in-hospital medical personnel are unable to accurately estimate blood loss in simulated scenarios [1,2]. Additional experiments attempted to create various educational sessions to improve blood volume estimations. One method

Corresponding author. Robert Wood Johnson Place, MEB 104, New Brunswick, NJ 08901, USA. Tel.: +1 732 235 8717; fax: +1 732 235 7379. E-mail address: merlinma@umdnj.edu (M.A. Merlin). 0735-6757/$ see front matter 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2008.07.039

1086 showed improvements in accuracy but not precision after using a 20-minute slide presentation and visual demonstration [3]. These authors concluded that their method should not be used. Another study reported slight decreases in mean percent errors by demonstrating known volumes, either by actual pools of blood or pictures in a slideshow, for memorization and use as a future reference [4]. Each of these previous studies had several limitations. First, none of the scenarios used whole blood. One study used laundry detergent with food coloring [2] and others used packed red blood cells reconstituted with saline [1,4]. The viscosity of these substances was not tested and compared to that of whole blood, so the pattern of dispersion might not mimic actual scenarios health care providers would encounter. A major shortcoming of the previous studies relates to their probabilities of long-term success. Specifically, they relied on participants' ability to recall and equate visualizations of known volumes to the currently presented scenario of blood loss. Over long periods, it is unlikely for health care providers to keep these visualizations in their memories, rendering the training useless.

M.A. Merlin et al. established with our University Hospital. In addition, use of the sewage system for disposal of blood was approved by the county hazardous material team coordinator.

2.1. Method determination

Two experimenters, both male and approximately 68 in tall, developed a method to estimate blood volume, using the fist as a measuring device. Units of recently expired whole blood were obtained from the transfusion services department of the hospital. Using a 60-mL syringe, a pool of blood was created in increments of 50 mL directly on the floor of the room (vinyl). Blood was applied to the ground such that it would pool with negligible splatter. After every increment, the experimenters took turns determining how many fists would cover the pool. Holding the fist dorsum up approximately 5 cm from the floor, the experimenters moved their closed fists over the blood, counting the number of fists required to cover the surface area of the pool without any spacing. Additional trials were conducted using increments of 20, 25, and 60 mL, creating a pool initially to a maximum of 450 mL. To achieve a pool of larger volumes, further trials were conducted using 60-mL increments on sheets of acrylic glass (polymethyl methacrylate), which were laid over the floor and leveled. This expanded the available usable surface area, and trials up to 1200 mL were conducted. The results from the different trials were analyzed and compared. Through this comparison, a value was assigned in milliliters to the amount of surface area 1 fist can cover. This formula of milliliters per fist is known as the MAR Method, an acronym of the authors' names.

1.2. Importance
Reliable determination of blood loss is an important estimation that may influence patient triage and treatment. Knowing the quantity of blood lost can aid health care providers in choosing the proper fluids and volumes for infusion to achieve homeostasis [5]. In the prehospital setting, the amount of visualized blood may factor in the decision to transport a patient to a trauma center. In addition, estimated blood loss is often a required item to report for both trauma and surgery, yet there has been no prior effective method for external blood loss determination.

2.2. Study design

The study design is an unblinded crossover trial. Participants served as their own control because the same participants were studied before and immediately after the intervention.

1.3. Goals of this investigation

To overcome the aforementioned problems of previous studies, the present research devises and tests a new method to determine external blood loss that uses human whole blood and does not depend on visualization. The proposed method equates volumes of blood to the anterior surface area of a clenched fist, similar to the palmar method of burn estimation, where the palm is estimated to equal 1% of total burn surface area [6]. The fist has been chosen instead of the palm to avoid confusion as to what area of the hand to use [7,8]. The goal of this study is 3-foldfirst, to devise such a method; second, to educate health care providers in its use; and third, to validate its reliability and usefulness in achieving improved accuracy and precision of blood volume estimation.

2.3. Setting
The study was conducted in the decontamination room of an academic emergency department (ED), which has a volume of 80 000 patients per year and is a Level I Trauma Center. The room was separated by dividers into 2 sections, such that one could not see into the other half of the room. On each side of the room, 1 sheet of acrylic glass was placed on the ground and leveled with 4 4s to adjust for the sloping floor. Human whole blood was measured with syringes and pooled on the acrylic glass. A small pool was created with 75 mL of blood, and a large pool with 750 mL.

2.4. Selection of participants

2. Methods
The protocol is approved by the institutional review board of our medical school, and a collaborative agreement is A convenience sample of health care professionals was recruited from the adjacent ED. There were no specific inclusion or exclusion criteria, and verbal informed consent

Blood volume estimation using MAR Method was obtained after a consent form was read to subjects. Participants included registered nurses, physician's assistants, medical students, and physicians of various training from within the hospital, as well as basic emergency medical technicians and paramedics who transported patients to the hospital.

1087 forms by the experimenters. Responses were confirmed by the experimenter repeating back the recorded values to the subjects. Items recorded for each participant included type of training, combined years of experience and training, first blood pool visualized (small or large), and the estimates of blood volumes for each pool, before and after the training session. At the conclusion of each data collection day, all information was entered into a database.

2.5. Interventions
Before entering the decontamination room, participants were asked about their type of training (ie, registered nurse, physician, medical student, etc) and years of combined medical training and experience. Participants were then randomly assigned to visualize either the small or the large pool first. Individually, participants entered the room and were given up to 1 minute to estimate the volume of blood in the first pool, using milliliters. Subsequently, participants moved to the other side of the room to estimate the volume in the second scenario within 1 minute. Only 1 pool could be observed at a time, and responses were not allowed to be changed once the second pool was visualized. Next, participants were brought outside of the room and were instructed in use of the MAR Method. This session consisted of an experimenter reading a script to the participants, which included, This method utilizes your fist to estimate blood volumes. The volume of a blood pool depends on how many fists it would take to cover the pool. One fist equals 20 mL of blood. So, if it takes 3 fists to cover a blood pool, 1, 2, 3, you would have 3 fists times 20 mL, which equals 60 mL of blood. The last sentence was said as the experimenter demonstrated proper use of the method on the ground with blood present. This intervention was conducted outside of the room so that the participants could not visualize the actual pools while learning the method. The same experimenter led the session for all participants, and the entire training session lasted less than 1 minute. Finally, participants were brought back into the room and given a second opportunity to estimate the volumes of each blood pool using the MAR Method. Blood pools were visualized in the reverse order of initial random assignment, and participants were again given 1 minute to estimate each pool volume. Participants then resumed their normal job duties without being told the actual volumes of each pool.

2.8. Outcome measures

Besides using the raw estimations given as a measure, errors were computed by calculating the absolute value of the difference between the guesses and the actual volume of the blood pools. In addition, percent errors were calculated by dividing the error by the actual volume. Using the absolute errors and percent errors allow for direct comparison and identification of improvement in estimation, which could not be done with simple averages of raw estimations.

2.9. Primary data analysis

Analyses were performed using SAS 9.1.3 (SAS Institute, NC) and SPSS 15.0 (SPSS Inc, IL) statistical software. To determine whether participants' errors in estimation were smaller before or after using the MAR Method, the differences between estimations were analyzed using the Wilcoxon signed rank tests, as the distributions of variables were not normal. Because the Wilcoxon signed rank test does not assume normality, exact 95% confidence intervals (CIs) are unlikely to be obtained, and the preceding intervals have confidence levels slightly greater than 95%. Comparison of average errors in estimation allows for the accuracy of the MAR Method to be tested versus the control of uneducated guesses. The precision of the MAR Method can be demonstrated by comparing the standard deviation and interquartile range (IQR) of errors for the pre- and postintervention values.

2.6. Methods and measurements

Participants' estimations of blood volumes were recorded in milliliters immediately as they were completing the procedure. Some participants who gave a range of volumes were asked to choose 1 number. Participants' volume estimations using the MAR Method were always 20 times the number of fists counted, to prevent any learning effects bias.

3. Results
3.1. Method determination
Number of fists for each known volume was averaged between the different experimenters and floor surfaces, which produced a linear response. Using least squares means from a 2-way analysis of variance, the average milliliters per fist was 19.71. There was no significant difference in volume per fist between the vinyl and acrylic glass surfaces (M = 19.44 vs 19.95 mL). The Pearson product-moment correlation coefficient between the 2 experimenters was r = 0.99.

2.7. Data collection and processing

The experimental sessions were administered by 2 medical students. Participants' responses were recorded on

1088

M.A. Merlin et al.

Table 2 Characterizations of errors in blood volume estimations by scenario for each guess in milliliters Volume estimations Visualization Small pool Mean (SD) Median (IQR) Percent error Mean (SD) Median (IQR) Percent error 90 (112) 50 (25-106) MAR Method 33 (27) 25 (13-45) b.0001 P

3.2. Characteristics of study subjects

Seventy-four subjects with various types of training and years of experience estimated blood volumes on 3 separate days. Participants included basic emergency medical technicians (36%), paramedics (7%), registered nurses (32%), physician assistants (4%), resident physicians (4%), attending physicians (11%), medical students (7%), and other training (7%). Many participants had multiple levels of training, so percentages add up to more than 100%. Participants averaged 10.41 years of combined training and experience (SD = 8.21), with a median of 9 years.

Large pool

120 44 549 (553) 244 (130) b.0001 500 (250-613) 250 (150-350) 73 33

4. Main results
To test for normality of data, histograms, Q-Q plots, and the Shapiro-Wilk W test were used to examine blood volume estimations by visualization and by the MAR Method for the small and large pools. For all variables, the graphs showed nonnormal distributions. In addition, the Shapiro-Wilk W test showed that normality is rejected at a significance level of 0.05 for MAR Method estimation of the large pool and at a significance level of 0.01 for the remaining variables. Therefore, nonparametric tests were used to compare volume means. Volumes estimated before and after receiving education in the MAR Method were compared using both the raw guesses (Table 1) and the errors from the actual volume (Table 2). For the small pool, the median error of initial visual estimations was 50 mL (95% CI, 40-55), and the IQR of errors was 81 mL (25 to 106). The percent error from the actual volume of the pool was 120%. After using the MAR Method, the median error of estimations decreased to 25 mL (95% CI, 25-45), and the IQR of errors was 32 mL (12 to 45). The percent error of the second estimation was 44%. The average error from the mean between the first and second estimations reduced significantly by 76% (P b .0001), and the IQR of errors reduced by 60%. The estimations of the large pool also yielded reductions in errors. The median error of initial guesses was 500 mL (95% CI, 350-550), and IQR of errors was 363 mL (250 to 613). The percent error of the initial guess from the actual volume of the pool was 73%. Second estimations using the MAR Method had a median error of 250 mL (95% CI, 210-250) and
Table 1 Characterizations of raw blood volume estimations by scenario for each guess in milliliters Volume estimations Visualization Small pool Large pool Mean Median Mean Median (SD) (IQR) (SD) (IQR) 127 85 623 400 (135) (30-181) (771) (200-750) MAR Method 104 100 598 540 (31) (80-120) (232) (435-725)

an IQR of errors of 200 mL (150 to 350). The percent error of the second estimation reduced to 33%. The average error from the mean between the first and second estimations reduced significantly by 40% (P b .0001), and the IQR of errors declined by 45%. Analysis of the first pool participants viewed (small vs large) by Wilcoxon rank sum tests did not make a difference in any volume estimations or errors. After using the MAR Method, 76% of all participants estimated volumes equal to or closer to the actual value of the small pool, and 83% estimated volumes equal to or closer to the actual value of the large pool. Correlations using Spearman were performed between participants' years of experience and errors in estimation of blood pool volumes. Between years of experience and the error in visual estimation of the small pool, a weak correlation of 0.379 was significant at the P b .001 level. No other correlations were found to be significant.

5. Limitations
A limitation of this study is the manner in which the MAR Method was derived. The 2 experimenters who created the method were unblinded to the volume of the pools throughout this process and were fully aware of each other's responsesleading to an unusually high correlation coefficient. Another reason for such a high correlation is that the 2 experimenters had approximately the same size fists. Within the population of health care providers, fist sizes may vary; however, the MAR Method can still be used to achieve a more accurate estimation of blood volumes than by visualization alone. Another limitation was that the time participants took to provide estimations of blood volumes was not recorded. When participants were shown each scenario, they were allowed 60 seconds to estimate a volume. Although most participants required less than this maximal permitted duration, the exact amount of time is unknown. Participants also spent a significant amount of time visually trying to estimate the blood pools. Instinctively, one may think using

Blood volume estimation using MAR Method the MAR Method requires more time than visual estimation. However, the time spent actively using the MAR Method may not be much, if any longer than the time spent looking at a pool while trying to visually estimate a volume. Further research is needed to make this determination. The MAR Method itself also has a limitation in that it has a narrow applicability. Currently, the method has only received extensive testing on a flat surface. The method should also be limited to nonporous materials. Although other surfaces besides vinyl and acrylic glass were considered, they were ultimately rejected from this initial study. For example, when testing blood application to carpeting, there was only a small visible spot of blood, as most of the volume seeped through to the underlying floor. Thus, using any method on carpeting would not be effective without being able to visualize the underlying floor. A cement surface was also considered; however, there was no suitable physical space available for use. Because of a relatively small blood supply, larger volumes of blood could not be used. For each data collection session, only 2 units of whole blood were available, totaling approximately 900 mL. Because a unit of blood varies in quantity, 900 mL was not always guaranteed by 2 units. To have 2 pools of blood, one small and one large, the volumes were decided to be 75 and 750 mL. The volume of 750 mL was chosen because it is the cutoff between class I and class II hemorrhagic shock [6]. The volume of 75 mL was then chosen as a smaller, noneven quantity such that the total amount of blood required was less than 850 mL, leaving room for variations in the available blood supply. Although the MAR Method is expected to work at any volume, it was only tested up to 1200 mL by the experimenters and 750 mL by the participants. The method must be tested with additional size pools for validation. Repeat testing is currently underway yet remains limited due to available blood in our institution.

1089 of use with rapid results. In determining the method, there was a linear response between volume and fists, and the average volume per fist was calculated between trials and experimenters. Although the experimentation actually produced a volume of 19.71 mL per fist, this number was rounded to 20 mL, as a whole number for ease of use and memorization. The purpose of using the MAR Method is not to calculate an exact volume to the milliliter. Its use is to determine blood volumes with increased accuracy and precision over visual estimation alone. Therefore, it is not necessary to have an exact volume but rather a close approximation. As consistent with prior findings, participants' initial visual determination of volumes (small and large) produced large variations in approximations over a considerable range of estimations. Participants who had more years of experience had slightly better visual estimations of the small volume but not the large volume. After using the MAR Method, participants were able to more accurately estimate the blood volumes present, indicated by reductions in errors from the actual volumes. In addition, the IQRs and standard deviations of the errors also reduced, showing an increase in precision. An even greater accuracy by use of the MAR Method could be obtained by providing the subjects with additional education. After learning the method in less than 1 minute, many participants were observed to be using the method improperly. One problem was that some participants used the wrong aspect of the fist, counting with the metacarpal phalyngeal joints pointing downward (as if punching the ground) instead of upward. Another misuse of the method was having spacing between fiststhe proper usage does not have any space between fists. Some participants counted with their fist high above the surface of the pool, causing a fist to appear to cover more area than if it were closer to the ground. The correct way to use the method is to get as close to the blood pool as possible when counting fists. These misapplications of the MAR Method can be attributed to lack of explicit mention during the educational session. A solution to these problems would be an expanded educational session. With less than 60 seconds of training, participants' accuracy of estimations improved drastically. The details of which aspect of the fist to use, distance from the pool, and spacing between fists were not mentioned in the educational session. Participants only obtained this information from watching the experimenter demonstrate on a fictitious pool. Specifically mentioning these points and increasing the time dedicated to teaching the method should reduce misuse and further increase the accuracy and precision of the MAR Method. Future research on estimating blood volumes should focus on how blood behaves on other surfaces and materials. In addition, the effects of coagulation on size of a blood pool should be studied. Although the MAR Method is currently limited to pools of blood, the technique would still be useful for health care providers who encounter patients with traumatic injuries. A brief in-service for medical personnel on use of the MAR Method is all that is required to learn a quick, simple, and

6. Discussion
Medical personnel currently do not receive any formal training in estimating blood volumes, although this can be an important piece of information used for patient treatment. Previous studies have shown that health care providers cannot accurately estimate blood volumes, yet no steps have been taken for improvement. This study has corroborated that medical personnel cannot accurately estimate amounts of pooled blood and additionally created a method to better determine blood volumes. Use of the technique, known as the MAR Method, showed an improvement in accuracy and precision of blood volume estimations. The MAR Method was designed to be a simple way to equate a volume of blood to its surface area. The anterior surface of a fist was chosen as the reference point for determining surface area due to the minimal variability in the average-sized person's fist. It was also chosen because of ease

1090 easy way to measure external blood loss. Because blood loss is routinely documented, its proper estimation based on sound methods cannot be overemphasized.

M.A. Merlin et al.

[3] Duong B, Gilbert G, Norris R, et al. An educational presentation does not improve visual estimation of blood loss by paramedics [Abstract]. Prehosp Emerg Care 2007;11:113. [4] Moscati R, Billittier AJ, Marshall B, et al. Blood loss estimation by outof-hospital emergency care providers. Prehosp Emerg Care 1999;3: 239-42. [5] Shafi S, Kauder DR. Fluid resuscitation and blood replacement in patients with polytrauma. Clin Orthop Relat Res 2004;422:37-42. [6] American College of Surgeons. Advanced Trauma Life Support for Doctors. 7th ed. Chicago: American College of Surgeons; 2004. [7] Rossiter ND, Chapman P, Haywood IA. How big is a hand? Burns 1996; 22:230-1. [8] Jose RM, Roy DK, Vidyadharan R, et al. Burns area estimationan error perpetuated. Burns 2004;30:481-2.

References
[1] Tall G, Wise D, Grove P, et al. The accuracy of external blood loss estimation by ambulance and hospital personnel. Emerg Med (Fremantle) 2003;15:318-21. [2] Patton K, Funk DL, McErlean M, et al. Accuracy of estimation of external blood loss by EMS personnel. J Trauma 2001;50:914-6.