TY - JOUR
AU - MD, Leopoldo C. Cancio,
AB - Abstract Accurate burn estimation affects the use of burn resuscitation formulas and treatment strategies, and thus can affect patient outcomes. The objective of this process-improvement project was to compare the accuracy of a computer-based burn mapping program, WoundFlow (WF), with the widely used hand-mapped Lund–Browder (LB) diagram. Manikins with various burn representations (from 1% to more than 60% TBSA) were used for comparison of the WF system and LB diagrams. Burns were depicted on the manikins using red vinyl adhesive. Healthcare providers responsible for mapping of burn patients were asked to perform burn mapping of the manikins. Providers were randomized to either an LB or a WF group. Differences in the total map area between groups were analyzed. Also, direct measurements of the burn representations were taken and compared with LB and WF results. The results of 100 samples, compared using Bland–Altman analysis, showed no difference between the two methods. WF was as accurate as LB mapping for all burn surface areas. WF may be additionally beneficial in that it can track daily progress until complete wound closure, and can automatically calculate burn size, thus decreasing the chances of mathematical errors. Accurate estimation of the TBSA burned in thermally injured patients is critical for optimal patient care. TBSA influences triage decisions, fluid resuscitation, nutritional requirements, and prognosis. Since the mid-1940s, the Lund–Browder (LB) diagram has been a mainstay of this estimate. This diagram consists of a two-dimensional figure of a human body, front and back. The figure is divided into sections, each of which constitutes a defined percentage of the total body surface. The user draws the injury on the diagram (typically using a colored pencil/marker), estimates the amount of each section that the injury occupies, and manually adds these numbers to obtain the TBSA. The LB diagram allows the user to differentiate between partial- and full-thickness burns and to adjust for patients of different ages.1 However, completion of the LB diagram can be time consuming and may require an experienced provider for accurate use. In one study, the burn size of 18% of the patients transferred to a burn center was overestimated by 100% or more when using the LB approach. Only 42% of patients had accurate initial estimates, and overestimation was more common than underestimation.2 Another approach for burn mapping is the Rule of Nines, developed by Pulaski and Tennison at this institute in 1947.3 It is a faster, but less accurate, method of estimating TBSA than the LB diagram. To address the problem of inaccuracy in burn-size estimates, a previous study used a computer-based mapping program with three-dimensional capability: EPRI's Burn 3D Vision. The program, however, requires a higher level of technical training.4 Another study in Australia focused on the use of an Internet-based computer program for referring facilities to improve burn-size estimates and to decrease discrepancies between referring and accepting facilities.5 Both studies also suggested that computer programs could provide a database for tracking wound progress and the results of treatment. We developed a computer software program (WoundFlow, WF) to more accurately map burn injuries and automatically calculate the patient's TBSA in real time. We deployed this application in our burn center as a process improvement project to optimize burn mapping in a user-friendly manner. The system provides an electronic paint brush to draw the thermal injury onto a two-dimensional figure and automatically calculates the TBSA on the basis of the LB standards. However, WF is more than a computerized LB diagram; it also enables users to modify the diagram at multiple time points, as the patient's status changes and undergoes surgery and/or healing. The objective of this study was to compare the electronic WF system with the traditional paper-based LB diagram. METHODS Software WF is a standalone application written in Java Version 6 and uses MySQL as the database. Current computers having a 2.66 GHz processor and 2 gigabytes of RAM have no difficulties running this program. It is designed to assist providers in mapping burn injuries based on the predefined body section percentages also used in the LB diagram for 15-year-olds to adults (Figure 1). For example, the LB diagram assigns an area of 4.75% to the anterior aspect of each thigh; WF does the same. In WF, wound area is calculated as the number of pixels painted by the user, divided by the total number of pixels within the body part in question. The purpose of WF is 2-fold: 1) to provide a computerized method for accurate initial burn-size estimation, and 2) to enable graphical depiction of subsequent changes in the patient's wounds, as healing and surgery occur. The application can be tied to the electronic medical record, so that users do not need to reenter demographics or other pertinent patient information into WF. Data from WF are stored in a database, which allows users to create multiple maps for a patient during his hospital stay. Injuries are mapped on a window using mouse input, based on a click and drag approach. Users drag the mouse on the two-dimensional body image to draw injury or treatment patterns. Users have the ability to choose from a set of templates representing the various wounds and treatments available. These include partial-thickness burn, full-thickness burn, autograft, temporary coverage, excised and open, debrided, unhealed donor, healed injury, healed donor, and amputation. Each wound or treatment is represented by a different color and hash pattern. Users have the ability to choose different brush shapes (circle, square, or ellipse) and different brush sizes. A set of drawing rules is built into the system, to prevent clinically illogical sequences. For example, the system prevents a user from marking an area with a full-thickness injury that has been previously marked as a donor site and will not let one draw outside the lines of the diagram. Figure 1. View largeDownload slide Example of WoundFlow data entry/drawing page. Figure 1. View largeDownload slide Example of WoundFlow data entry/drawing page. TBSA for the patient is calculated in real time as the injury patterns are drawn on the two-dimensional image. A summary table for each of the body sections and each type of injury/treatment is provided. These numbers are then used to determine the total TBSA without the need for manual calculations. Each map is labeled with the postburn day. Thus, a progress diagram of the burn injury is generated, and the healing rate as a function of time is calculated and displayed. By providing the ability to store and record multiple maps per patient, the history of the patient's healing progression can be examined. In addition, the WF system allows users to upload photographs of the wounds corresponding to each map (Figure 2). Images are uploaded to the WF database system directly from a digital camera. Figure 2. View largeDownload slide Example of digital photographs uploaded in WoundFlow. Figure 2. View largeDownload slide Example of digital photographs uploaded in WoundFlow. Evaluation In the present study, burn patients were simulated by five Rescue Randy Combat Challenge training manikins (Simulaids, Saugerties, NY). The study was completed in two phases, effectively giving us 10 adult male patient representations. Burn wounds were represented as different sizes and in different body areas on the 10 patient representations. Our intent was to compare accuracy of the two tools in estimating TBSA; therefore, distinctions between full- and partial-thickness burns were not made. Participants were burn center providers usually responsible for calculating burn size. This group was made up of providers with a wide range of experience and included experienced staff surgeons, physician assistants, and rotating resident physicians. All participants were given a brief orientation to the WF program as well as to the LB diagram; the time required to achieve proficiency in WF appeared short, but was not measured. All manikins were from the same molds. They had a height of 1.85 m and a weight of 75 kg. Burn victims have varying heights, weights, and genders, but these manikins were used because of the apparent similarity to the figures depicted in the LB and WF diagrams. Preparation for the project started with measurement of the BSA of a manikin in square centimeters. This was achieved using flexible measuring tape placed circumferentially along longitudinal lines for the extremities. Torso and head measurements were obtained in a similar fashion by using anterior and posterior midlines, again measuring to find the total surface area of each section. The BSA was calculated by this method to be 2.024 m2. Each manikin had burns depicted by red adhesive vinyl in varying body areas and sizes. Because of ease of obtaining accurate length and width measurements on the manikin's surface, the flat nature of the burn depictions, and well-defined borders, direct measurements were taken to obtain BSAs. Tracing on square centimeter graphing paper was used for smaller burns with irregular borders.6 These burn measurements were taken after WF and LB mapping, thus effectively blinding participants to the true burn size. The project was divided into two phases with a total of 50 estimates per phase completed by burn center staff. In phase 1, five manikins were prepared by applying red vinyl adhesive to depict burns of various body areas and sizes. The manikins were placed unclothed in hospital beds. Staff members were instructed to complete one estimate for each manikin. The method to be followed was randomly assigned on an instruction sheet. The instruction sheets were made up of 50 sheets of paper with an LB diagram printed on one side. At the right upper hand corner of each sheet was a letter assignment for each corresponding manikin. Twenty-five manikins had instructions, “Lund and Browder use this sheet,” and the other 25 had instructions, “WF use computer.” Five WF sheets and five LB sheets were labeled with corresponding manikin letter. These were then shuffled and placed face down in a folder at the foot of each manikin. The participants were advised to select the top sheet, turn it over, and follow the instructions for each mapping. If the participants received a sheet with instructions “Lund and Browder use this sheet,” they completed the mapping on the sheet using a red pencil. If the participants received a sheet with instructions “WF use computer,” they logged onto a laptop computer set up at the bedside and completed the mapping using the WF program. The laptop used was a Dell Latitude model D630. The operating system was Windows XP with service pack 2. All sheets were then placed in a box kept at the foot of each manikin's bed. In phase 1, burn depictions measured 6.6, 9.5, 15.1, 13.6, and 1.2%. In phase 2, the burn sizes used were 51.8, 14.1, 43.1, 12.4, and 45.1% (Figure 3). Figure 3. View largeDownload slide Mapping done with computer-based WoundFlow as compared with Lund–Browder mapping of same manikin. The actual (directly measured) TBSA was 15.01%. Figure 3. View largeDownload slide Mapping done with computer-based WoundFlow as compared with Lund–Browder mapping of same manikin. The actual (directly measured) TBSA was 15.01%. Statistical Analysis For each manikin burn, the LB and the WF results were compared with the actual TBSA using a one-sample Student's t-test. Then the LB and WF results were compared with each other using a two-sample Student's t-test. An F-test comparing the SDs for each burn size was then performed. P values were adjusted for multiple corrections using Bonferroni's method. To determine whether there was any overall bias or trending bias, LB and WF results were compared using a Bland–Altman analysis. RESULTS Figure 4 shows each burn estimate plotted against the true burn size for each manikin's burn. Table 1 shows the mean differences between the actual burn size (TBSA), and either the WF or LB estimates, for each manikin. Note that three manikins (B2, C1, and E1) featured significant differences between either the WF or the LB estimate, and the actual burn size. These were manikins with laterally placed burns on the thighs and/or flank. Manikins B2 and C1 had 14.1 and 15% burns, respectively; the estimates that were significantly different were 12.6 and 16.8%—well within clinical significance. However, the LB estimate for the smallest burn of 2.6% was more than double the actual TBSA (1.2%). Other studies show LB with a coefficient of variation (CV) around 10 to 20%, depending on the size and location of the burn.7,8 This is consistent with the results here and with the CV values for WF (Table 1), except again for the smallest burn (1.2%), where the %CV for WF was 42%. Table 1. Comparison of either Lund–Browder (LB) or WoundFlow (WF) estimates, on one hand, to directly measured actual burn size, on the other hand, for each of the 10 manikins View Large Table 1. Comparison of either Lund–Browder (LB) or WoundFlow (WF) estimates, on one hand, to directly measured actual burn size, on the other hand, for each of the 10 manikins View Large Figure 4. View largeDownload slide Burn estimates plotted against the true burn size for each manikin. Figure 4. View largeDownload slide Burn estimates plotted against the true burn size for each manikin. Comparing the LB results with the WF results, the two-sided t-test showed no significant difference. The F-test comparing the SDs of the two methods also showed no significant differences for any of the manikin burns. Finally, a Bland–Altman analysis to measure reproducibility was performed. No bias or trend in bias was detected in the analysis.9Figure 5 shows the Bland−Altman plot along with the fitted regression line. Figure 5. View largeDownload slide Bland–Altman plot along with the fitted regression line for Lund–Browder (LB) vs WoundFlow (WF) estimates. No bias or trend in bias was detected in the analysis. Figure 5. View largeDownload slide Bland–Altman plot along with the fitted regression line for Lund–Browder (LB) vs WoundFlow (WF) estimates. No bias or trend in bias was detected in the analysis. DISCUSSION The principal findings in this study were that both the paper LB method and the new computerized WF method enabled accurate quantification of the total burn size simulated on manikins. Statistically significant difference did exist between LB and WF, which was a 1 to 2% variation on a single small laterally placed burn. The variation was deemed clinically insignificant because of the total size of the lateral burn of 1.2%. In the mid-1980s, Wachtel et al reviewed an electronic medical data system developed by Hewlett Packard that integrated a patient BSA calculator within their proprietary Patient Data Management System program. The study compared accuracy between hand-calculated and computer-calculated burn surface areas. They concluded that the computer-assisted calculation was more accurate and less variable. The program functioned more like a data input system rather than a calculate as you draw utility. The user filled in numeric percentages in designated fields and the program then ascertained TBSA, resuscitation, nutritional needs, and prognosis. Following the development of the HP Patient Data Management System, other publications compared newly developed computerized tools. The newer versions evolved into body-diagram–based utilities.10 One such program is Surface Area Graphic Evaluator (SAGE). The program was developed by Philip Parshley and supported by the Oregon Burn Center. The software is Internet- or PDA-based, and has been accessed thousands of times worldwide for determining burn size. This utility has algorithms for weight, height, and gender. SAGE is based on the areas of two-dimensional diagrams.11 In the mid-1990s, Dr. Raphael Lee, the Electric Power Research Institute (EPRI), and the Chicago Hospital Burn Center began publishing data on burn mapping using a three-dimensional computer program. By early 2000, EPRI's 3D Burn Vision was available to clinicians. This program is capable of rotating and manipulating the body and extremities within the diagram. The program also uses various height, weight, and gender algorithms to improve accuracy (Table 2).4,11 Table 2. WoundFlow (WF) comparison adapted from Neuwalder J M, Arch M, Sampson C, et al. A review of computer-aided BSA determination: SAGE II and EPRI's 3D burn vision. J Burn Care Rehab. 2002 View Large Table 2. WoundFlow (WF) comparison adapted from Neuwalder J M, Arch M, Sampson C, et al. A review of computer-aided BSA determination: SAGE II and EPRI's 3D burn vision. J Burn Care Rehab. 2002 View Large Current burn-size methods are typically based on several traditional methods of estimation, most commonly LB and the Rule of Nines. These methods can lead to variability among providers, and do not allow for effective modification while the burn injury is treated during hospital stay. Although LB has been proven to be less variable when compared with Rule of Nines, it still requires manual addition, a potential source of error.7 Given the absence of significant variability between the LB and WF burn mapping tools, we can conclude based on the present study that WF mapping is as accurate as LB. The importance of burn mapping accuracy is underscored by the fact that burn-size estimates are often factored into triage decisions and into resuscitation protocols like the American Burn Association consensus formula.12 Limitations of current mapping strategies such as LB, Rule of Nines, palmar comparisons, and serial halving have been further disputed as body types, weight differences, and surface areas are inherently variable.3,7,10,13,–18 As noted by Nichter et al computer-based three-dimensional mapping utilities required specific training.19 In WF, the computer mouse is used for drawing in a simple manner, such that many first-time users have described it as easy and intuitive. However, more work is required to determine whether WF can successfully be used by nonburn providers, and to quantify the amount of time required to achieve proficiency. Although users can depict and account for amputations at various levels, WF at present cannot be customized according to patient's body habitus or gender. Future improvements may address this limitation.20 In 2011, a call was made to develop a standardized and widely accepted computerized Lund and Browder chart, citing excessive inconsistencies in burn-size estimates. The need for a program that can be readily updated and that can be tied into patient's electronic medical record are additional reasons supporting a movement beyond the paper LB diagram.21 Indeed, WF is an easy-to-use computer-based utility that can also provide data useful for performance improvement and for research. This project demonstrated that WF is as accurate as LB in determining the burn size. The WF program also has the added ability to track the progress of wound healing and to map wound healing trajectory from initial injury to healed burn wound. References 1. Lund CC, Browder NC. The estimation of areas of burns. Surg Gynecol Obstet. 1944;79:352–358. 2. Hammond JS, Ward CG. Transfers from emergency room to burn center: errors in burn size estimate. J Trauma. 1987;27:1161–5. 3. Knaysi GA, Crikelair GF, Cosman B. The role of nines: its history and accuracy. Plast Reconstr Surg. 1968;41:560–3. 4. Tuch DS, Lee RC. Three-dimensional wound surface area calculations with a CAD surface element model. IEEE Trans Biomed Eng. 1998;45:1397–400. 5. Ahn CS, Oneill SP, Maitz PK. Improving accuracy of burn referrals through the use of an internet-based burns chart. Eur J Plastic Surg. 2011;5:331–5. 6. Plassmann P. Measuring wounds. J Wound Care. 1995;4:269–72. 7. Miller SF, Finley RK, Waltman M, Lincks J. Burn size estimate reliability: a study. J Burn Care Rehabil. 1991;12:546–59. 8. Wachtel TL, Berry CC, Wachtel EE, Frank HA. The inter-rater reliability of estimating the size of burns from various burn area chart drawings. Burns. 2000;26:156–70. 9. Altman DG, Bland JM. Measurement in medicine: the analysis of method comparison studies. Statistician. 1983;32:307–17. 10. Wachtel TL, Brimm JE, Knight MA. Computer assisted estimate of the area and depth of burn. J Burn Care Res. 1983;4:255–9. 11. Neuwalder JM, Sampson C, Breuing KH, Orgill DP. A review of computer-aided body surface area determination: SAGE II and EPRI's 3D Burn Vision. J Burn Care Rehabil. 2002;23:55–9; discussion 54. 12. Lindford AJ, Lim P, Klass B, Mackey S, Dheansa BS, Gilbert PM. Resuscitation tables: a useful tool in calculating pre-burns unit fluid requirements. Emerg Med J. 2009;26:245–9. 13. Nichter LS, Williams J, Bryant CA, Edlich RF. Improving the accuracy of burn-surface estimation. Plast Reconstr Surg. 1985;76:428–33. 14. Hussain S, Ferguson C. Best evidence topic reports. Bet 1: Assessing the size of burns: which method works best?. Emerg Med J. 2009;26:664–6. 15. Rossiter ND, Chapman P, Haywood IA. How big is a hand?. Burns. 1996;22:230–1. 16. Yu CY, Lin CH, Yang YH. Human body surface area database and estimation formula. Burns. 2010;36:616–29. 17. Smith JJ, Malyon AD, Scerri GV, Burge TS. A comparison of serial halving and the rule of nines as a pre-hospital assessment tool in burns. Br J Plast Surg. 2005;58:957–67. 18. Sakson JA. Simplified chart for estimating burn areas. Am J Surg. 1959;98:693–4. 19. Nichter L, Bryant C, Edlich R. Efficacy of burned surface area estimates calculated from charts – the need for a computer-based model. J Trauma. 1985;25:477–81. 20. Neaman KC, Andres LA, McClure AM, Burton ME, Kemmeter PR, Ford RD. A new method for estimation of involved BSAs for obese and normal-weight patients with burn injury. J Burn Care Res. 2011;32:421–8. 21. Jeng JC. By Fiat: an enlightened approach to The American Burn Association's sensible quest for a universally accepted electronic Burn Diagram (Lund-Browder Diagram). J Burn Care Res. 2011;32:e157. Footnotes ‡ Presented at the 43rd Annual Meeting of the American Burn Association, Chicago, IL, March 29 to April 1, 2011. Copyright © 2012 by the American Burn Association
TI - Comparison of Traditional Burn Wound Mapping With a Computerized Program
JF - Journal of Burn Care & Research
DO - 10.1097/BCR.0b013e3182676e07
DA - 2013-01-01
UR - https://www.deepdyve.com/lp/oxford-university-press/comparison-of-traditional-burn-wound-mapping-with-a-computerized-9IHzG4Np3S
SP - e29
EP - e35
VL - 34
IS - 1
DP - DeepDyve
ER -