Towards An Efficient Method for Studying Collaborative Practices Of Emergency Care Teams Aleksandra Sarcevic, Michael E. Lesk, Ivan Marsic

Randall S. Burd

Rutgers University 4 Huntington Street New Brunswick, NJ 08901

Childrenʼs National Medical Center 111 Michigan Ave NW Washington, DC 20010

{aleksarc, lesk}@rutgers.edu, [email protected]

[email protected]

ABSTRACT

obtaining access and permission to videotape, gaining cooperation with medical staff, overcoming medico-legal obstacles, and securing patient privacy. This paper summarizes our experiences of using video recording in trauma resuscitation, a safety-critical domain in which medical teams use complex work processes while treating severely injured patients early after injury. Our ability to use this method, however, was limited since human subject protections required that we erase video recordings within 96 hours and rely solely on anonymized transcripts for subsequent analysis. This requirement has led us to develop a method for formally representing and analyzing the collaborative processes during trauma resuscitation without relying on actual video recordings. We describe this method and present our future steps in studying trauma resuscitation domain.

The paper presents our approach to studying collaborative practices of trauma teams during trauma resuscitation, a fastpaced and information-laden process of evaluating critically injured patients early after injury. We conducted a series of studies in a Level 1 trauma center using field observations and video recording. Our ability to use video recording was limited since human subject protections required destruction of video records within 96 hours. We describe our experiences of using video recording and summarize a method that we have developed to overcome this research limitation.

Categories and Subject Descriptors H.5.3 [Group and Organization Interfaces]: Evaluation / Methodology; J.3 [Life and Medical Sciences]: Health

2. RESEARCH OVERVIEW The goal of our research is to understand the nature and extent of team errors in complex work settings, and suggest technological or procedural means of improving teamwork. Over the past three years, we conducted a series of observational studies at the Robert Wood Johnson University Hospital (RWJUH), a 600-bed academic medical center and the principal hospital of UMDNJRobert Wood Johnson Medical School in New Brunswick, NJ. The hospital is also a verified Level 1 Trauma Center, one of three in New Jersey. The center receives about 1200 patient per year. Patients treated at the trauma center have sustained major injuries in motor vehicle accidents, interpersonal violence (e.g., gunshot or stab wounds), or falls.

General Terms Design, Human Factors.

Keywords Healthcare, Trauma resuscitation, Methods, Video recording.

1. INTRODUCTION Video recording is a valuable tool for studying human performance in complex, dynamic environments such as those found in emergency care clinical settings. It provides a rich, reusable record of events for repeated scrutiny and analysis of collaborative work. Yet, a brief overview of past CSCW research in clinical settings revealed rare use of this method. Researchers have mostly applied ethnographic approaches such as field observations, shadowing, interviews, and focus groups, e.g.,[1], [3], [8], [15]. An exception are “workplace studies” [4], but they have mainly focused on less dynamic medical settings such as operating rooms and anesthesia, e.g., [5], [14].

To gain the requisite knowledge and understanding of how trauma teams work, we employed a mix of approaches for studying the domain. To begin with, we assembled a research team comprised of researchers, engineers, computer scientists, trauma surgeons, and trauma nurses. This diverse group of experts resulted from our desire to holistically address the barriers to effective introduction of technology in the trauma bay. More important, close collaboration with three trauma surgeons and a nurse throughout the studies has been essential. The surgeons have been involved as research partners from the beginning of the project and have played a crucial role in securing access to the setting and obtaining approval to videotape live trauma resuscitations. Second, the non-medical members of our research team have been trained in the processes of trauma care in a one-day didactic course modified from the Advanced Trauma Life Support (ATLS) course to give them the needed skills to understand and analyze the conduct of trauma resuscitations. The ATLS course is a critical tool in guiding patient care during trauma resuscitation. It presents a concise approach to assessing and managing patient injuries and is a required course for physicians involved in trauma

Despite many advantages, the use of video recording in emergency care clinical settings imposes a number of challenges that are often difficult to overcome. The challenges include

1

Figure 1. Summary of the research approach. care [2]. Third, we used both inductive (bottom-up) and deductive (top-down) approaches to collect and analyze the data (Figure 1). Our top-down approach was driven by two research frameworks that are concerned with the analysis, design and evaluation of complex socio-technical systems: cognitive work analysis (CWA) [16], and skills-rules-knowledge framework, also known as the SRK framework [7]. The use of CWA enabled analysis of the goals, work practices, role assignments and tasks of trauma resuscitation teams. In conjunction with the SRK framework, CWA also served as an overarching framework for developing a coding scheme for tagging activities of trauma team members in video recordings. Bottom-up approaches such as ethnographic data collection methods (e.g., field observations, interviews, video recording, focus groups) and grounded theory approach [13] enabled analysis of communication, information needs and teamwork inefficiencies during resuscitations. The combined use of top-down and bottom-up approaches helped with understanding different aspects of trauma care, as well as with identifying and describing the causes of team errors.

our research protocol to the Institutional Review Board (IRB) at the hospital. To circumvent the risks involved in videotaping live resuscitations, such as patient privacy and medico-legal concerns, we explicitly stated that (a) our study did not focus on the patient; thus, no personal patient data would be recorded, and (b) any written records produced during the study would exclude resuscitation dates, times, and any personal or other information that could permit identification of a patient, a specific resuscitation, or a trauma team member. As a result, IRB approval for videotaping live resuscitations was secured, but mandated informed consent from all team members and destruction of video records within 96 hours of videotaping. Because of the large number of individuals participating in trauma resuscitation, the research purpose and consent process had been presented to select large groups (e.g., surgery residents and faculty, emergency nurses and technicians) before studies began. This was done for two reasons. First, by obtaining consent in advance, we avoided a risk of not having the necessary time to distribute and discuss consent forms before resuscitations. Second, the team participating in a given resuscitation may be drawn from a pool of over 200 hospital personnel. Obtaining consent before studies began assured that all participating team members had been informed about the research and videotaping procedure. Based on the signed consent forms, we estimate that our studies involved about 250 health care providers, including physicians, residents, nurses, technicians, anesthesiologists, pharmacists, EMS paramedics and medical students.

Although we collected data through field observations, interviews and focus groups, we primarily relied on video recordings of real trauma resuscitations to examine collaborative practices and team errors. We were fortunate to gain access and permission to videotape live trauma resuscitations but human subject protections limited the kinds of studies we could do. In the following section, we describe our experiences of using video recording, first detailing the process of gaining permission to videotape, and then presenting a method that we have developed to overcome this research limitation.

3.2 Analysis of Video Records The 96 hours time limit posed several constraints on data capture and analysis. We needed an efficient yet detailed transcription scheme that would facilitate quick data transfer from a video- to a paper record. Our transcription was thus based on the parallel columnar transcription scheme, which is commonly used in interaction analysis [6]. Each discernable action of each team member was transcribed in a separate row. For each action, we

3. VIDEO RECORDING AND ANALYSIS 3.1 Obtaining Permission to Videotape As mentioned already, the involvement of trauma surgeons in the project has been essential. In addition to securing access to the setting, the surgeons provided support in drafting and submitting

2

MAIN CATEGORIES Observing Information acquisition and retrieval

Memory recall & info retrieval

Verbal communication Communication & Intervention

Nonverbal procedures Skill-based behaviors

Decision-making

Rule-based behaviors Knowledge-based behaviors

SUBCATEGORIES

CODE

Physical examination (auscultation, palpation) Manual measurement Simple sensing: sight, sound, touch, time Instrument reading Knowledge recall Situation recall Info retrieval from artifacts (trauma flowsheet, notes, x-ray workstation…) Directives (Task assignment / Instruction / Command) Report (about patient status or team member activity) Inquiry / Request for information Response to an inquiry or request for information Clarification (Request for retransmission of information) Relay Acknowledgement Summons Therapeutic intervention / Treatment Setting-up instrument/equipment Recording information (e.g., on a trauma flowsheet) Handing / Receiving an object Task coordination Solo decision making Strategic planning Judgment: Approval vs. Disapproval vs. Praising Takeover / Handover of leadership role Group decision making Coaching / Educating

EXM MM SEN IR LTM STM MEXT DIR RP Q RS CL RLY ACK SM INT SET RC HRO TC SDM SP JU (A/D/P) TO GDM ED

Table 1: Coding scheme for categorizing control tasks and communication in trauma resuscitation. included a brief description of the action and utterance, and identified who performed a task or spoke based on their role (e.g., patient, team leader, paramedic). A domain expert (trauma surgeon or nurse) verified the accuracy of transcripts while the video record was still available. Transcribing the videos was a difficult and tedious process. Although resuscitation events lasted between 20 to 30 minutes, they were extremely fast-paced and information-laden. At times, there were several conversations happening simultaneously, with trauma team members performing many parallel tasks. On average, it took about 20 hours to transcribe one event that typically included >400 discrete tasks or communications. Because of the amount of time required in producing and verifying the transcripts, we were able to videotape, transcribe and analyze only 30 percent of observed resuscitations (18 out of 60).

information retrieval” as the fifth major task category. These five categories provided an overarching framework for identifying other collaborative practices during trauma resuscitation. Using the grounded theory approach, we identified 26 activities corresponding to the above five task categories (Table 1). For example, observation activities include patient physical examination (EXM), such as chest auscultation; manual measurement of a patient state variable (MM), e.g., blood pressure; simple sensing of a patient state variable (SEN), e.g., pulse monitor alarm; and, instrument reading to obtain the value of a patient state variable (IR), e.g., pulse rate. The subcategories emerged from our pilot observations, as well as from observations of two simulated trauma resuscitations. The medical-task coding scheme was developed with the assistance of the trauma surgeons on our research team. Medical tasks were derived from the goals of the team activities driven by the ATLS evaluation protocol and provided semantic context for the control tasks. For example, code C2, which is a matching code for pulse rate, was put next to the technician’s verbal report about the patient’s pulse rate.

To enable analysis of collaborative activities based on anonymized transcripts, we developed two coding schemes: (1) control-task coding scheme, which represents the behavioral aspects of the action (Table 1), and (2) medical-task coding scheme, which represents the medical goals of the action (not shown due to the space restrictions). The control-task coding scheme was based on a simple information-processing model of complex teamwork, adapted from cognitive work analysis [16]. According to this model, tasks in trauma resuscitation can be grouped into four information-processing activities: (1) observe, i.e., information gathering; (2) communicate, i.e., information sharing; (3) decide, i.e., information interpretation; and, (4) intervene, i.e., decision execution. Pilot data also indicated that trauma team members rely on their memory and various artifacts (e.g., trauma flowsheet, handwritten notes) to retrieve situationspecific and medical information when performing their tasks. To account for these activities, we added “memory recall and

Coding of the transcripts was done after the videos had been deleted. Each row was assigned one or more control-task- and medical-task codes in the appropriate columns in the transcript. The first author coded the transcripts twice, with 6 to 11 months passing between the first and the second coding. Cohen’s Kappa was used to verify the intra-coder reliability of the control-task coding scheme. Separate kappa values were calculated for the high-level categories. “Communication and intervention” and “Information acquisition and retrieval” categories obtained “Almost perfect” (kappa value of .924) and “Substantial” (kappa value of .697) levels of agreement, respectively. These kappa

3

values are considered to be an acceptable indicator of agreement above chance level. The “Decision making” category attained “Moderate” level of agreement (kappa value of .582). Although limited, the above described technique for transcribing and coding activities during trauma resuscitation allowed us to create detailed representations of collaborative practices of trauma teams and to conduct analyses after the video record was deleted. To date, we successfully applied this method to examine different aspects of trauma care, including division of labor and trauma resuscitation tasks [11], [12], information needs of trauma team members [10], and information handover between EMS paramedics and trauma teams [9]. In addition, use of this method enabled quantifying interactions in the trauma bay, which is one of the novelties of this research.

importance of collaboration with and support of medical personnel involved in research; and, (c) effective use of video records for the purposes of analyzing collaborative practices that are of interest to CSCW community.

5. REFERENCES [1] Abraham, J., and Reddy, M. 2008. Moving patients around: A field study of coordination between clinical and nonclinical staff in hospitals. Proc. CSCW’08, 225-228. [2] American College of Surgeons. 2005. Advanced Trauma Life Support® (ATLS®), 7th Edition, Chicago, IL. [3] Bardram, J. E. 2000. Temporal coordination: On time and coordination of collaborative activities at a surgical department. Computer Supported Cooperative Work, 9, 2, 157-187.

4. CONCLUSION AND FUTURE WORK The paper presents our approach to studying collaborative practices of trauma teams during trauma resuscitation. We described our experiences of using video recording and a method that we have developed to overcome research limitations inherent to emergency care clinical settings. Although we gained permission to videotape live trauma resuscitations, the amount of time we had for video analysis limited the kinds of studies we could do. For example, we were not able to perform microanalysis of interactions among trauma team members, such as those found in workplace studies.

[4] Heath, C., and Hindmarsh, J. 2002. Analysing interaction: Video, ethnography and situated conduct. In Qualitative Research in Practice, T. May, Ed. London: Sage, 99-121. [5] Hindmarsh, J., and Pilnick, A. 2002. The tacit order to teamwork: Collaboration and embodied conduct in anesthesia. Sociological Quart. 43, 2, 139-164. [6] Jordan, B., and Henderson, A. 1995. Interaction analysis: Foundations and practice. J. Learning Sciences, 4, 1, 39-103. [7] Rasmussen, J. 1983. Skills, rules, and knowledge: Signals, signs, symbols and other distinctions in human performance models. IEEE Trans. on Systems, Man, and Cybernetics, 13, 3, 257-266.

We are currently extending our research to an additional medical setting and we hope to be able to perform the types of analyses we could not do previously. Our new setting is a trauma center at the Children’s National Medical Center (CNMC), in Washington, DC. CNMC is the main pediatric teaching hospital of George Washington University Medical School and is a Level 1 trauma center with over 1000 trauma admissions per year. The center already uses video recording of live trauma resuscitations for the purposes of improving team performance. We have permission to use these video records for extended period of time and to participate in video review sessions with trauma team members. In addition, we have access to the state-of-the-art simulation lab, as well as to the communication exchanges between the hospital’s Emergency Communication and Information Center and EMS paramedics in the field. Our plan is to extend current research framework and perform detailed analysis of video records and of other types of data collected through fieldwork. We also plan to conduct a series of participatory workshops with trauma team members to generate design concepts and evaluate them in a simulation setting. The inclusion of the second site will allow us to (a) perform microanalysis of collaborative activities of highly dynamic patient care teams that are of interest to CSCW community; (b) compare the conduct of resuscitation and teamwork at different institutions; and, (c) improve the generalizability of our findings across trauma centers and patient types (children and adults).

[8] Reddy, M., and Dourish, P. 2002. A finger on the pulse: temporal rhythms and information seeking in medical work. Proc. CSCW ‘02, 344-353. [9] Sarcevic, A., and Burd, R.S. 2009. Information handover in time-critical work. Proc. GROUP ‘09, 301-310. [10] Sarcevic, A., and Burd, R.S. 2008. What’s the story? Information needs of trauma teams. Proc. AMIA ‘08, 641645. [11] Sarcevic, A., Lesk, M.E., Marsic, I., and Burd R.S. 2008. Quantifying adaptation parameters for information support of trauma teams. Ext. Abstracts CHI ‘08, 3303-3308. [12] Sarcevic, A., Marsic, I., Lesk, M.E., and Burd, R.S. 2008. Transactive memory in trauma resuscitation. Proc. CSCW ‘08, 215-224. [13] Strauss, A., and Corbin, J. 1998. Basics of Qualitative Research: Techniques and Procedures for Developing Grounded Theory (2nd ed.). Thousand Oaks, CA: Sage. [14] Svensson, M.S., Heath, C., and Luff, P. 2007. Instrumental action: The timely exchange of implements during surgical operations. Proc. ECSCW ‘07, 41-60. [15] Tang, C., and Carpendale, S. 2009. Supporting nurses’ information flow by integrating paper and digital charting. Proc. ECSCW ’09, 43-62.

Our studies raised several questions about conducting research in emergency care clinical settings. Although we addressed some of the challenges we faced, we would like to further elaborate on these issues with workshop participants. We are particularly interested in discussing (a) different approaches one can take to overcome limitations of an emergency care clinical setting; (b) the

[16] Vicente, K.J. 1999. Cognitive Work Analysis: Toward Safe, Productive, and Healthy Computer-based Work. Mahwah, NJ: Lawrence Erlbaum Associates.

4

Towards An Efficient Method for Studying Collaborative ...

emergency care clinical settings imposes a number of challenges that are often difficult .... account for these activities, we added “memory recall and information ...

315KB Sizes 0 Downloads 216 Views

Recommend Documents

Particle Swarm Optimization: An Efficient Method for Tracing Periodic ...
[email protected] e [email protected] ..... http://www.adaptiveview.com/articles/ipsop1.html, 2003. [10] J. F. Schutte ... email:[email protected]

DART: An Efficient Method for Direction-aware ... - ISLAB - KAIST
direction with respect to his/her movement or sight, and the direction can be easily obtained by a mobile device with GPS and a compass sensor [18]. However,.

Towards an Efficient Public Key Cryptosystem
EC-KCDSA Elliptic Curve Korean Certificate-based Digital Signature Algorithm ... Chapter 6 presents the implementation and analysis results assessment of the ... using some secret data (cryptographic key), this operation is called encryption. ... met

Differential Evolution: An Efficient Method in ... - Semantic Scholar
[email protected] e [email protected] .... too much control, we add the input value rin(t)'s squire in ..... http://www.engin.umich.edu/group/ctm /PID/PID.html, 2005.

Differential Evolution: An Efficient Method in ... - Semantic Scholar
[email protected] e [email protected] .... too much control, we add the input value rin(t)'s squire in ..... http://www.engin.umich.edu/group/ctm /PID/PID.html, 2005.

Frequency interleaving towards spectrally efficient ...
redesigned by frequency interleaving of two adjacent OSSB + C formatted ..... and passed through the electrical OFDM receiver to recover transmitted data bits.

A Highly Efficient Recombineering-Based Method for ...
Mouse Cancer Genetics Program, Center for Cancer Research, National Cancer Institute, Frederick, Maryland ..... earized gap repair plasmid or from uncut DNA (data not ...... Arriola, E.L., Liu, H., Price, H.P., Gesk, S., Steinemann, D., et al.

ServiceFinder: A Method Towards Enhancing Service Portals
17. ServiceFinder: A Method Towards Enhancing. Service Portals. XIAO FANG. University of Toledo. OLIVIA R. LIU SHENG. University of Utah and. MICHAEL CHAU. The University ... tured in the homepage of a service portal such that users can be directed t

Simple and efficient method for carbon nanotube ...
Cystic Fibrosis Center, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599. R. Superfine, M. R. ... introduced, tip-side down, onto the submerged platform. The ... the CNT cause the tubes that come into contact with the.