Hacking Teamwork in Healthcare: Addressing Adverse Effects of Ad Hoc Team Composition in Critical Care Medicine

Abstract

The continued need for improved teamwork in all areas of healthcare is widely recognized. The present article reports on the application of a hackathon to the teamwork problems specifically associated with ad hoc team formation in rapid-response teams. Hackathons -- problem-solving events pioneered in computer science – are on the rise in healthcare management. The focus of these events tends to be on medical technologies however, with calls for improvements in management practices as general recommendations. The hackathon reported here contributes to healthcare management practice by addressing improvements in teamwork as the focal problem. We describe briefly the event, present summaries of the solutions proposed by the participating teams, and discuss implications of the proposals for teamwork practice in healthcare and of using the hackathon approach.

Issue

The literature on team dynamics is replete with evidence highlighting the importance of careful team formation and composition processes (Bedwell, Ramsay, & Salas, 2012). In teams with stable membership, these processes contribute over time to the development of effective communication and coordination structures, familiarity with each other’s knowledge, and mutual trust. Such productive team characteristics are nowhere more vital than for resuscitation teams -- charged with responding immediately to the often-unpredictable and complex emergency medical situations that arise concerning critically ill patients. For a variety of reasons (e.g., staffing levels), these resuscitation or rapid response teams are composed chiefly by propinquity -- that is, composed of care givers who happen to be physically near or can reach the patient very quickly. As a result, the composition of these teams is fundamentally ad hoc: the nurses, technicians, residents, fellows, attending physicians and other clinicians who rush to a bedside may not know each other or the patient, may have never worked together before, and may be unaware of each other’s experience and specialization. Further, the composition of these teams is characterized by high temporal fluidity in that members arrive and leave -- often unexpectedly -- at different times throughout an emergency event. Like teams with more stable membership, rapid response teams need efficient structures and mutual knowledge awareness and trust, but they also need to achieve these factors within extremely short time frames (minutes or seconds) and in the face of rapidly changing, circumstances.

This paper describes a hackathon focused on the challenges of ad hoc team formation in critical care medicine. Originating from computer science, hackathons have been found to be an effective organizational practice for stimulating innovation in a wide variety of fields; increasingly this includes healthcare (Chowdhury, 2012). During hackathons, which typically last from just a few hours up to several days, teams work during intensive face-to-face sessions in competition against each other to develop the most innovative high-quality solutions to a specific problem. Diversity in membership with respect to expertise, discipline, and stakeholder identity is a hallmark of these teams.

We believed a hackathon would be a promising approach to the challenge of ad hoc team formation in critical care medicine for several reasons. First, despite the clear empirical evidence for the positive impact of effective teamwork on patient outcomes and organizational processes (Hughes et al., 2016), the rate of problems attributable to communication failures and other problems of teamwork remains unacceptably high (Mitchell, Parker, Giles, & Boyle, 2014; Sorbero, Farley, Mattke, & Lovejoy, 2008) and therefore presents an ongoing need for innovative solutions. Whereas some healthcare professionals may perceive teamwork as merely common sense, a considerable body of research across the communication, management, and psychology literatures has demonstrated otherwise (Weller, Boyd, & Cumin, 2014). Most importantly, teamwork skills can be learned and improved. Therefore, the second reason for our interest in a hackathon was to bring researchers and practicing health care providers together face-to-face to work intensively on this process challenge. A final reason we turned to a hackathon was that we believed our experience with this event would make an important contribution to the growing literature on the specific topic of ad hoc healthcare team composition (Bedwell et al., 2012), and to healthcare team effectiveness more generally. By making the process of teamwork the target challenge, our event extends the traditional hackathon focus from tools and technology to improving interaction processes. We offer our experience as a template for adapting this method to a variety of teamwork and organizational challenges.

Theoretical Analysis

Ad Hoc Team Formation in Critical Care Medicine

Ad hoc teams, such as resuscitation teams, are short-lived teams formed to address a specific problem; these types of teams are not limited to critical care medicine (White, Eklund, McNeal, Hochhalter, & Arroliga, 2018). Other high-stakes healthcare settings (such as the emergency department or operating room) provide services to critically ill patients who are the most likely to require ad hoc response teams. Estimates show that the largest proportion of critically ill patients are treated in critical care units, whereas less than 10% of emergency department patients are characterized as critically ill ( Nguyen et al., 2000) and the vast majority of in-hospital cardiac events occur within critical care units (Peberdy et al., 2008). Moreover, teams in critical care settings that respond to sudden deteriorations in patients’ conditions are nearly always ad hoc. Special attention to the unique nature of ad hoc teaming in critical care is therefore warranted.

When a patient in the intensive care unit (ICU) suffers a cardiac arrest, for example, those physically closest to the patient are expected to respond. The members comprising the responding team must be ready to assume roles and responsibilities necessary to treat patients with whom they have had little, if any, familiarity, and must be further prepared for their roles to change over the course of a resuscitation event as other providers arrive and depart. One analysis identified at least 16 different roles (e.g., RN Team Leader, Recorder RN, CPR RN) that need to be filled on an ad hoc basis during such events (Su et al., 2014). The responsibility for who is doing what needs to be constantly reassessed. Role confusion combined with elevated emotional intensity may negatively affect performance in ad hoc resuscitation teams (Hunziker et al., 2011).

The synchronization of members’ behaviors is essential to effectiveness of any team (Driskell, Salas, & Driskell, 2018), but this is especially difficult in ad hoc teams (White et al., 2018) because of the unpredictable mix of specialties and temporal shifts in membership. Therefore, achieving effective role and task coordination quickly is a key challenge for ad hoc teams. Critical care teams rarely have time for discussions to become familiar with each other (Rosen et al., 2018). As a result, team members may have difficulty assessing each other’s expertise and experience and the extent to which each will act according to expectations. Although knowledge of role expectations facilitate the functioning of critical care teams (Ervin, Kahn, Cohen, & Weingart, 2018), to the extent that ad hoc team members are not familiar with one another, the shared knowledge can only represent a prototypical situation; it cannot encompass the specific experience, skills, and weaknesses of other team members, nor person-specific styles and preferences. Further, roles can nevertheless be enacted quite differently across people. For example, in an investigation of trauma teams, Raley et al., (2017) found inter-individual differences in issues like, “offering assistance, emotion controlling, working in an organized manner, and being responsive” (p. 176). As described by Faraj and Xiao, 2006, p. 1166), “the combination of expertise specialization, overlapping interdependencies, and a rapid tempo requires…members to enact new coordinative responses.” Thus, effective coordination in ad hoc resuscitation teams requires individual and team adaptability (Bedwell et al., 2012).

Despite the extensiveness of the literature on teamwork training (Hughes et al., 2016), the experiences of the practicing healthcare professionals who took part in our hackathon suggested that insights from this literature are generally not seen in practice. This is due in part to persistent gaps in translating research to practice, and in part to there being relatively few training programs designed for the unique challenges of critical care and emergency medicine settings. Rosenman et al. (2018) note for example, that much healthcare teamwork training is designed around “prolonged shared experiences” (p. 133), which helps teams develop strong affective structures like team identity and efficacy, and strong cognitive structures like shared mental models. Such long-term designs are impracticable for the ad hoc teams seen in critical care and emergency medicine, however. Rosenman et al. instead recommend that training should center on adaptive approaches like “generalizable protocols [and] communication strategies” (p. 133). In a similar vein, Bedwell et al. (2012) proposed that the training of teams with fluid membership should focus on generalizable skills and how to be strategic in communicating information. Other training approaches include providing ad hoc teams with small-scale structures (i.e., “scaffolds”: Valentine & Edmondson, 2014) as guides for effective interpersonal processes, and simulations of emergency team processes (Rosenman et al., 2018). Ultimately, anecdotal evidence from practicing healthcare professionals, committed to solving this issue, was a key motivation for our turning to a hackathon as a new source of innovative ideas.

Hackathons in Healthcare

Hackathons are increasing in number and prominence in the healthcare industry, as evidenced for example by MIT’s “Hacking Medicine Initiative” (DePasse et al., 2014). As described by Angelidis et al., (2016, p. 393), a “hackathon integrates collaboration, idea generation and group learning by bringing together different stakeholders in a mutually supportive setting.” Examples of the wide range of healthcare problems that have been tackled by hackathons include issues in rehabilitative medicine (Silver, Binder, Zubcevik, & Zafonte, 2016), healthcare access for underserved populations in low and middle-income countries (Angelidis et al., 2016), and technological solutions to support user-driven self-care (Day, Humphrey, & Cockcroft, 2017). Proponents of healthcare hackathons have argued that these events allow disruptive innovation, which can complement more deliberate long-term innovation approaches (Chowdhury, 2012; Day et al., 2017; DePasse et al., 2014).

Although there is not general agreement on a single definition or model of a hackathon (Day et al., 2017; Silver et al., 2016), there is clear consensus in the literature on the importance of some features of healthcare hackathons. Chief among these is interdisciplinary team composition. There is universal agreement about the value of bringing healthcare practitioners in direct face-to-face contact with other stakeholders. Walker and Ko (2016, p. 98), for example report “we have seen firsthand how cross-discipline partnerships can provide incomparable opportunities, particularly for physicians.” Chowdhury (2012) argues that the primary goal of hackathons should be to reduce barriers between people from different disciplines and that reducing those barriers begins with placing them in the same physical space. The cross-disciplinary contact also is seen as important for exposing participants to ways of thinking and practices not available to them in their everyday work environment (Chowdhury, 2012; Day et al., 2017).

Another common feature is the time-limited intensive nature of these events. Although in some cases the work sessions might be spread over several weeks (Angelidis et al., 2016; Chowdhury, 2012), most occur within the span of a few days or hours (Olson et al., 2017). In all cases the sessions involve highly focused attention on the target problem. Competition contributes to the intense nature of the setting. Hackathons are characterized by strong norms of “cooperative competition” and openness (Silver et al., 2016). The typical physical set-up is open spaces in which teams’ activities are in full view of everyone, and teams are usually encouraged to consult with each other and to share resources. Most important, teams share the superordinate goal of wanting effective solutions to the target problem. Based on research showing that moderate levels of intergroup competition spurs within-group creativity (Baer, Leenders, Oldham, & Vadera, 2010), we argue that this competitive yet supportive atmosphere results in a level of competition optimal for team creativity.

Most of the healthcare hackathons described in the literature typically focus on development of innovative technologies or the innovative application of technologies to healthcare problems. Common goals are increased quality, safety, accessibility and affordability of medical care (Angelidis et al., 2016; Chowdhury, 2012). Reports of these events describe successes along multiple dimensions including development of viable clinical products and establishment of healthcare start-up companies (Angelidis et al., 2016; DePasse et al., 2014), the participation of hard-to-reach populations (Angelidis, et al., 2016), improvement in educational practices (Silver et al., 2016), and longevity in collaborations established during the hackathon (Olson et al., 2017).

Although medical technology innovation is often the primary outcome of hackathons in healthcare, many of the greatest challenges facing the healthcare industry are human in nature. Improvement in the processes of teamwork is frequently included among problem solutions; however, the mention of teamwork rarely moves beyond being a recommendation. A major contribution therefore of our hackathon was to place solving teamwork problems as the starting point. In other words, we “hacked” healthcare teamwork.

The Hackathon Event

The setting for the event was the 2017 annual conference of the Interdisciplinary Network for Group Research (INGRoup), an organization which “(a) promotes communication about group research across fields and nations, (b) advances understanding about group dynamics through research, (c) advances theory and methods for understanding groups, and (d) promotes interdisciplinary research” (www.ingroup.net). Recent years have seen a steady increase in the amount of programming at this conference devoted to healthcare teamwork, attendance by healthcare practitioner-researchers from across the globe, and interdisciplinary research collaborations around this topic among conference attendees. The specific teamwork problem we chose -- mitigating adverse effects of ad hoc team formation and composition in critical care medicine -- emerged from one of these collaborations: it was a fitting problem because of its complexity, high practical importance and relevance to healthcare management, reliance on multiple team and management theories, and inherent interdisciplinary nature.

The focus on ad hoc team formation reflects largely unsolved process challenges experienced by teams across all areas of healthcare. Like many medical teams, signature characteristics of rapid response teams include unpredictable changes in membership and in the specific problems encountered (Bedwell et al., 2012; White et al., 2018), and the presence of multiple disciplines or professions (DiazGranados, Dow, Appelbaum, Mazmanian, & Retchin, 2018). Given that teamwork quality has been shown to affect rapid response team practices (Hunziker et al., 2009), common in this literature is the call for continued investigation and innovation in ways to increase effectiveness of healthcare teamwork. In tackling teamwork problems directly, our hackathon responded to this call.

The authors were members of three six-member teams (comprised of practicing physicians and other healthcare professionals and academics from disciplines of communication, psychology, and organization sciences) formed prior to the conference based on criteria of (1) maximizing breadth of disciplinary background, and (2) inclusion of at least one student member to align with the INGRoup organization’s strategic goals related to professional development of group and team scholars.

The hackathon work took place in several discrete sessions scheduled over two and a half days during the INGRoup conference during which team members reviewed background materials to develop a better understanding of the challenges facing teams in critical care before providing input based on individual areas of expertise and developing their ideas and solutions. To model the short intensive nature of hackathons, we established a rule that teams could work on their hackathon projects only during the specified work sessions. Teams presented their solutions in a plenary session during which the audience voted on which solution best met the criteria of originality, effectiveness, feasibility, and interdisciplinarity.

The Solutions

We highlight how each team defined unsolved problems of ad hoc teamwork in healthcare and how their proposal addressed that problem. Like other reports of hackathons (e.g., Angelidis et al., 2016; Silver et al., 2016), the solutions are presented at only a high level of detail because we wanted to illustrate what was possible with a hackathon focused on interaction processes among resuscitation team members rather than on specific products or technical solutions. Following the description of the solutions we discuss common themes and implications for managing healthcare teamwork.

1. JANUS: Wearable sensors to facilitate RRT composition and coordination

The team’s solution integrated team science with computer science to propose a wearable device interface they named JANUS (Judging Ad-hoc Needs and Uniting Specialties). Their solution comprised a three-pronged approach: (a) monitor patients to proactively prevent events requiring a rapid response team (RRT), (b) enhance the effectiveness of RRT formation, and (c) provide continuous feedback for improvement of RRT processes. Thus, the focus was to develop technology assistance to improve the interaction processes of the resuscitation team members.

Sensors in providers’ wristwatches and in patients’ locations collect data that can be used to identify patients at risk for clinical deterioration and alert providers if rounds on a patient are overdue. This preventative information can help decrease the occurrence of rapid response events. When a rapid response event does occur, JANUS would integrate data to populate a team formation algorithm based on the core factors of (1) physical proximity to the patient, (2) familiarity with the patient, and experience with other providers in proximity to the patient, (3) expertise and experience with the condition of the patient or the code event, and (4) appropriate number of personnel. The algorithm would be optimized to select the ideal RRT member composition (given the circumstances), identify each member’s role, and notify members of the location of the emergency event. A monitor in the patient’s room would display the RRT members and their roles to facilitate role clarity and responsibilities, given that team composition and role confusion tend to be major challenges in a code event. JANUS would be capable of tracking arrivals, activating requests for additional support, and scaling back the number of team members in the room (often too many “helpers” arrive only to further complicate teamwork matters).

Finally, through machine learning principles, the system would use RRT providers’ effectiveness ratings and patient outcome results to improve its algorithm, in real time. In sum, the JANUS solution provides an innovative, practical, scalable, and sustainable solution that offers improvement in how resuscitation members interact and integration their medical and patient knowledge.

2. MRC: Medical Responsibility Clarification

This team focused on the problem of poor role clarity during rapid response events (Castelao, Russo, Riethmüller, & Boos, 2013). Contributors to this problem include temporal factors (e.g., RRTs tend to be formed sequentially; senior personnel tend to arrive later), that RRT members may come from outside a patient’s immediate care team and RRT members may or may not be familiar with each other, and the repeated transfer of leadership and responsibility. These factors are amplified by a context of high pressure and cognitive load, multiple professional cultures, and rapid switching to the low-frequency occurrence of emergency codes from ongoing routine work.

This team’s solution centered on a set of rules as a protocol, specifying what each member should do when joining the RRT. The person in charge (a) states or requests arriving medical staff to identify themselves and describe what they are prepared to do; (b) is clear and explicit in assigning, confirming, or resolving conflicting team members’ roles; and (c) allow leadership to be explicitly transferred accordingly as members arrive and depart. Their solution included careful analysis of barriers to implementing these rules, including medical culture, that the proposed rules violate common communication norms (e.g., avoid redundancy), the role of stress, and individual differences in ability to follow these rules. Their recommendations for overcoming these barriers include tailoring rules for local contexts, monitoring and providing feedback of resuscitation team performance, and incorporating such rules into existing protocols and medical education and training, thus lowering the risk for mistakes common in such situations (Ervin et al., 2018).

3. Tackling the image of teamwork by healthcare professionals: A focus on the science

This team’s solution centered on the specific challenge that healthcare professionals generally overestimate their ability to learn and execute team skills and overly depend on a skilled leader (Cooper & Wakelam, 1999). Thus, it is possible that the integration of teamwork as a process has not been as well integrated with medical training as many believe. They point to the gap between extensive data and well-grounded teamwork theory and application of this knowledge in practice in medical settings. They proposed a three-point solution focused on team training that integrates research in ways valuable to practitioners.

The first point was rebranding team training with the goal of getting practitioners to recognize the complexity of effective teamwork for which effective models exist. An example is TeamSTEPPS (Baker, Battles, & King, 2017), the dominant model in American healthcare developed by the Agency for Healthcare Research and Quality, American Research Institute and the Department of Defense. Although this model has been shown to improve teamwork and provides an excellent foundation and common team-oriented language, it may not address more advanced teamworking skills that are commonly needed in emerging situations. This hackathon team’s specific suggestions include highlighting the science of teams and team training (Salas & Cannon-Bowers, 2001), incorporating teamwork requirements into professional review processes, focusing on measurable benefits of effective teamwork, setting policies for teamwork training requirements, and increasing communication of the science of teams research implications to key healthcare organizations and professional associations.

Their second point was, to develop clear tactics for implementing team training that will further help professionals to appreciate and adopt effective practices. Tactics include prioritizing training needs, designing the training to be at an accessible scale (e.g., frequent, iterative and brief), and to mirror real conditions (e.g., field-based, asynchronous, videos that illustrate specific teamwork principles). Their third component focused on using big data. This involved both making teamwork science available to medical providers and collecting data on teamwork practices in the field. For example, elements such as “team leader identified” and “the number of resuscitation members who received team training in the past year” could be collected on the American Heart Association-Get With The Guidelines (GWTG) resuscitation form. Adding this kind of data to a clinically oriented form would emphasize that effective team performance is as important as ensuring that defibrillation occurred in a timely manner, and provide data on key teamwork factors (e.g., identified leader and teamwork training) largely understudied in non-simulated settings. These teamwork-related queries could then be tracked in electronic health databases, like the AHA’s CPR database, which could not only provide a rich source of data on best teamwork practices but also could addresses the challenge of the expense and resource demands required to directly observe details of teamwork in the field. Other methods like wearable devices that collect sociometric data (e.g., who is in proximity to whom at a given time) could give useful real time information about teamwork during practice. Data such as these would improve understanding of what real healthcare teamwork practice looks like and in collaboration with team researchers the field could better identify management practices and training to support improved patient care.

Discussion and Implications for Practice

The three solutions were distinct, yet complementary; each with the potential to influence teamwork processes in different ways. They can also be arrayed along a rough continuum of general to specific. Team #3’s approach might be considered the most general as it offers systemic solutions for improving teamwork effectiveness focused on attitudes, policy, and training (Marsilio, Torbica, & Villa, 2017). Targeting these fundamental factors would increase the healthcare system’s capacity to improve teamwork and receptivity to appropriate interventions, as well as contribute to building a culture of high-performance teamwork. As such, their recommendations could be applied beyond critical care and emergency medicine; it also dovetails well with the more focused solutions offered by Teams #1 and #2.

The technology application offered by Team #1 might be considered the next most general. This team offered a holistic technology-based learning approach--ranging from prevention to continuous system improvement – which complements the systemic approach seen in Team #3’s solution. The core of Team #1’s more specific focus on rapid response settings was attention to team composition – getting the right mix and number of people to a patient’s side with the least delay possible. It is also notable that this team’s technology-based solution was the one most closely aligned with the classic hackathon roots in information and computer science, and as such encompassed probably the broadest interdisciplinary reach of the three solutions. This solution integrates for example literatures on team composition (Bedwell et al., 2012) organizational continuous improvement (Murray & Chapman, 2003), and machine learning (Wojtusiak, 2014).

The proposal offered by Team #2 might be considered the most specific with its focus on rules for role clarity. The importance of role clarity has long been recognized in the healthcare management literature (Suter et al., 2009). As implied in this team’s solution, clarity of the leadership role is particularly important in the rapid response setting (Delaloye et al., 2016). Reducing role ambiguity is also a goal of the solution proposed by Team #1 and the technology they propose could provide the means to automate some of the process suggestions offered here by Team #2. Team #2’s approach incorporates a temporal perspective (Kusunoki & Sarcevic, 2015) by focusing on the dynamics of a rapid response situation and identifying the key moments when role ambiguity is both most likely to occur and to cause most damage, namely when RRT members arrive and depart and poor management of the resultant hand-offs. The core idea of their solution – explicit announcements of identity and responsibilities – offers a straightforward way of managing these problems. The systemic reforms in attitudes, policies and training suggested by Team #3’s proposal would provide the enabling conditions to support practices like those suggested here by Team #2.

Hackathon Impacts

A key issue in organizing a hackathon is defining success (Angelidis et al., 2016). The definition depends on an organization’s goals for the event, and metrics must include the perspective of participants and the external impacts. Our goals were several. First, we each are deeply interested in the problems created by poor teamwork in healthcare and are concerned that the advances we see in academic research barely touch frontline practice. The INGRoup conference provided the rare opportunity for researchers and practitioners to work together, face-to-face in real time, on a major healthcare teamwork issue that requires true joint problem-solving. The solutions that emerged from our hackathon were backed by both field-based experience and solid research evidence. We do not claim that these solutions are wholly unique or new. Other solutions that might complement those we proposed could include, for example “structural changes, such as the scaffolding described by Valentine and Edmondson, (2014). The true innovation of our hackathon was the joint application of practical and theoretical knowledge to interpersonal communication and coordination problems. The complexity of such problems -- arising from factors such as changes in medical knowledge and practices and differences across settings and kinds of teamwork – calls for multifaceted solutions and continuous innovation. Our event demonstrated that hackathons can offer a high leverage spur to innovation through the ability to bring together key stakeholders (Chowdhury, 2012).

Second, the event was tied to strategic goals of the INGRoup organization to: “translate groups research for impact,” “build interdisciplinary relationships,” and “serve developmental function for groups scholars” (www.ingroup.net). We saw the direct collaboration between researchers and practitioners as the opportunity for two-way translation whereby research findings could be brought to bear on practice and practice could inform future research. The goal of interdisciplinary relationship building was addressed by the practice of pre-forming the teams with an eye to maximizing the disciplinary diversity and providing development opportunities for young scholars. Evidence from other hackathons, for example, has demonstrated that students benefit from participating in these events in terms of learning content and skills, and gaining self-confidence (Day et al., 2017), especially when they are on teams with professionals (Lyndon et al., 2018).

Based on the participants’ feedback immediately following the event and 18-months later, we judge the hackathon to have successfully met our goals. All three teams proposed practical process solutions that were well-informed by empirical research. Although the ideas varied in how quickly or easily they could be developed, they were all implementable. Moreover, many of the ideas are applicable to settings outside of critical care. Participant feedback suggested that we also met our goals of creating interdisciplinary connections between people. The following two comments reflect the general feedback received from all the participants: “Working with individuals I’d never met for our project was the best networking experience I had in the entire conference,” and it “sparks meaningful conversations, networking, and ideas.”

We offer several examples of the lasting effects of the hackathon. Aligned with the suggestions offered by Team #3 to look for ways of connecting team science to ongoing practice, a physician member of that team has presented the three solutions at “Resuscitation in Motion,” an annual healthcare event for researchers and clinicians to optimize outcomes of cardiac arrest. This physician also co-hosted the 1^st annual team science and team communication symposium at Stanford University, which brought organizational science scholars together with healthcare faculty and educators. Three participants who are co-authoring a medical textbook chapter about ECMO¹ teams realized as a result of the hackathon experience that the chapter needed to provide better access to team science concepts to healthcare workers.

The lasting impacts were also reflected in participants’ reports of both deepening pre-existing collaborations and developing new ones with colleagues they had met for the first time – collaborations all focused on healthcare teamwork and informed by the hackathon experience. For example, two participants restructured their current research team on organizational science in healthcare to expand the interdisciplinary representation among the members. They reported that as a result of this restructuring, organizational science researchers were appointed to several transformative healthcare initiative committees in the healthcare system with which they worked. For another set of participants this experience led them to collaborate on submitting a foundation grant.

Participants also reported changes in individual practices ranging from new ways of approaching their writing to changes in teaching methods. One participant, for example, admitted to sometimes resorting to “creative blue-skying” when it came to writing the practical implication section of articles. Several months following the hackathon this researcher reflected that, “engaging on a practical problem from the outset…was a nice way to start working and thinking from a different direction…how literature and knowledge could be brought to bear on a very specific problem…that was a new way of considering and working, not quite as ‘pure’ academics, not quite as ‘pure’ practitioners, and not quite as ‘pure’ consultants.” Mirroring the spirit of this comment, all participants reported that the experience helped them better appreciate and think more deeply about the value and implications of interdisciplinary connections.

Lessons Learned

Overall, our approach was well-aligned with the “best practices for healthcare hackathons” presented by Silver et al., (2016) -- we defined a problem, set time parameters, engaged interdisciplinary teams to think through innovative solutions, invited presentations of the solutions, and assessed the success of the event. We took lessons from the spread of hackathons from computer science to other disciplines and extended them further by showing that the approach can help improve interpersonal processes in addition to identifying technological solutions. Moreover, the ability to run the event within the larger structure of our annual conference, similar to Silver et al., (2016), meant that we could leverage the presence of scholars and practitioners from across the globe. As a result, the hackathon was low-risk and virtually of no financial cost to the INGRoup organization. A hackathon motivated by addressing teamwork challenges from a social science perspective represented the most effective integration of the dedication of the organization to interdisciplinary group research and the creative potential of its members. We thus demonstrated a hackathon model which might lend itself to other low-tech or process-oriented problems where an interdisciplinary approach to generating innovative science-based solutions would be of interest.

We also learned several lessons to pass on to others considering a healthcare hackathon that departs from a focus on medical technologies. First, the degree of topic specificity is important to consider. The challenge is balancing between giving teams sufficient guidance to keep them from floundering and providing so many strictures as to stifle creativity. Next is pre-event planning and communication. We found it was important for team members to familiarize themselves ahead of time with background information. Relatedly, we found that forming teams ahead of time, rather than during the event as is typical in traditional hackathons, facilitated teambuilding. We believe this was especially important given the short amount of time teams worked together during the event. Pre-forming the teams also gave us control over maximizing the disciplinary diversity in the teams. On the other hand, one problem with this team formation approach was that our participants spent a significant portion of their time on simply identifying viable solutions. Having teams form around potential problem solutions, an approach frequently used in technology-focused hackathons that has the advantage of giving teams a clear focus at the start. Thus, it is important to consider how the advantages and disadvantages of various pre-planning methods meet an organization’s goal priorities.

We found that despite the advantages of having the hackathon be an integral part of the INGRoup conference, this was also participants’ greatest criticism. They all shared dissatisfaction with how the schedule crimped their participation in the larger conference; direct time conflicts and the difficulty of switching focus between the hackathon sessions and attending or presenting at paper sessions added to the exhaustion and stress already attached to the event. This issue nevertheless did not dampen their unanimous and resounding enthusiasm for the event. Rather, the lesson we take from this is the need for flexibility and innovation in event planning to balance the multiple and sometimes competing goals for the staging of something like this. An additional suggestion for future implementation is to consider creating structures to help teams to reflect on their solutions after the event concludes, like requesting reports of post-event progress. Such follow-up may better facilitate solution implementation, continued participant team collaboration, and the creation of products or additional scholarship stemming from hackathon solutions (Chowdhury, et al., 2012; Day, et al. 2017).

Finally, our experience reinforced the value of the “cooperative competition” norm we imported from classic hackathons. Teams in our event found the competitive aspect to be exciting and fun. The prospect of having an audience of respected peers judge their solution was highly motivating to the teams. In addition to giving “bragging rights,” participation in the competition was seen as a non-trivial addition to a C.V. At the same time, everyone shared the goal of finding solutions to the specific problem of ad hoc team formation in critical care and of improving teamwork in healthcare more generally. Cooperative norms were reinforced through putting the teams in a common workspace and explicitly encouraging them to communicate and share resources across teams. We found that these norms achieved the intended goal of stimulating both learning and achievement, as reflected in the following comment regarding all the solutions from one of the participants: “Amazing presentations. I really loved that hour of the conference…I couldn’t have been more pleased with the results.”

Conclusion

Integrating across the points of intersection among the three hackathon proposals yields the clear insight that multi-pronged solutions for emergency-oriented teamwork are needed. Any single tool or practice will have limited effectiveness if a healthcare organization’s culture and policies do not communicate strong norms about the integral role of teamwork process to the quality of care. Teamwork training will likewise be limited without sophisticated technologies -- in the form of procedures as well as software – to facilitate daily communication and coordination practices. Our hackathon on the challenges facing teams in critical care medicine with participation from teamwork experts in multiple disciplines highlights the scale of collaboration and effort necessary to tackle the many complexities in healthcare that significantly impact outcomes for providers, patients, and health organizations. As perhaps the first hackathon to foreground teamwork, our event also points to a promising new avenue of applied research on teamwork innovation.