The Characteristics of Medical Teams: A Quantitative and Empirical Study of Teamness in Healthcare

Introduction

The delivery of healthcare is a highly complicated system, partially due to the interrelatedness of its many components (Kannampallil et al., 2011), and this complexity can trigger adverse patient outcomes. One of the prevailing solutions is the building of “resilient” healthcare systems, and a key feature of resilient systems is the use of coordinated multidisciplinary teams to adapt and overcome errors in a highly variable workplace (Anderson et al., 2020). Effective team-based care has been shown to improve patient outcomes with higher quality of care, lower utilization of acute care, and lower healthcare costs (Reiss-Brennan et al., 2016). Not surprisingly, effective healthcare teams also improve provider outcomes by creating a safer and more engaging workplace with lower levels of burnout and better mental health. These provider outcomes are crucial in an industry like healthcare, which is characterized by high rates of burnout, depression, and suicide (Rosen et al., 2018). In this study, we consider “healthcare teams” to include teams within large hospital systems that provide medical care to patients. This includes surgical teams, nursing teams, and emergency medicine teams, for example, but it does not include teams that work in positions unrelated to care delivery (e.g., administrative teams, facilities teams, and regulatory/quality teams).

Despite the many benefits of effective teams in healthcare, the characterization of an effective healthcare team remains opaque. In 2024, Cooke et al. (2024) presented the construct of “teamness,” according to which teams can be characterized and compared to one another “by qualifying the differences and similarities between and within instances of teaming.” While teamness is conceptually sound, the specific dimensions within the healthcare context remain unclear. To make this construct of teamness more directly applicable for the characterization of healthcare teams by hospital administration, systems engineers, and patient safety experts, the underlying dimensions should be specific and identified on the basis of empirical evidence and quantitative analysis. The objective of this work is to uncover the underlying dimensions of teamness in healthcare teams through empirical data and quantitative analysis of academic and non-academic text sources utilizing content analysis and factor analysis.

Background

In 1992, Salas and colleagues presented a generic definition of teams as being interdependent groups of people who interact dynamically and adaptively to pursue a common goal (Salas et al., 1992). This was an important step forward in teaming research, but this definition does not provide a clear set of dimensions along which a team can be characterized. In 1990, Sundstrom and colleagues proposed nine dimensions of teams based on an interpretive literature review (Sundstrom et al., 1990). In 2012, Hollenbeck and colleagues performed similar work and proposed three dimensions along which generic teams could be characterized based on their literature review (Hollenbeck et al., 2012). While these works were important efforts in identifying a set of dimensions to characterize teams, in both studies, the authors presented dimensions based solely on their interpretation of the literature. Both works based their dimensions entirely on academic articles, ignoring the insights of non-academic sources, which may have presented different perspectives, and there was no empirical foundation upon which three (in the case of Hollenbeck) or nine (in the case of Sundstrom) dimensions were chosen. Furthermore, neither Hollenbeck’s nor Sundstrom’s work was tuned to healthcare, a unique and complex setting where teams perform distinctly differently from generic teams.

Several researchers have started to study some specific aspects by which medical teams can be classified, but they have all focused on a single dimension or aspect of teaming, like expertise (Garrett et al., 2009), collaboration (Fox et al., 2024), or task complexity (Molleman et al., 2010). Much like the works of Sundstrom and Hollenbeck, the dimensions presented in each of these studies are solely based on the authors’ interpretation of the academic literature, and they all lack an empirical basis for the definition and number of presented dimensions. Furthermore, they each only focus on a particular aspect of teaming, which is valuable in achieving a deeper view of the studied aspect but does not offer a comprehensive understanding of the multidimensional nature of healthcare team characterization.

When Cooke et al. (2024) presented the multidimensional construct of teamness, they proposed five “potential dimensions” based on their review of teaming literature and research questions raised by students in a seminar course. In another article published shortly thereafter, the authors commented on their prior findings stating, “The concept of teamness needs further refinement to address the needs of team researchers; for instance, there are likely many teamness dimensions beyond those currently identified by Cooke et al. (2024)” (Cohen et al., 2024). This illustrates a clear need for empirical data-based quantitative analysis to uncover the underlying dimensions of teamness. Moreover, when the concept of teamness was presented, the authors stated that the “dimensions of teamness vary within and between contexts” (Cooke et al., 2024). This points to a need for a medical domain-specific analysis of teamness. In this research, we endeavor to uncover the dimensions of teamness that are specifically relevant in the medical field by analyzing both academic and non-academic sources. Our work aims to further the healthcare research practice and equip practitioners with the tools they need to characterize healthcare teams. While this study is not an evaluation of healthcare team effectiveness, the dimensions presented by this study could further enable team effectiveness research.

Approach

This study used content analysis and factor analysis to empirically and quantitatively investigate and extract the underlying dimensions of teamness in a medical context. Content analysis (Drisko & Maschi, 2016) is the quantitative analysis of text materials, and the outputs of content analysis are used as inputs for conducting factor analysis: a quantitative method to extract a small number of unobservable constructs or dimensions that underlie a large number of observable variables and data (Kline, 2014). Before conducting content analysis, we collected several dozen academic and non-academic text sources related to teaming in the medical field. We used various search terms to collect these texts, including but not limited to “healthcare teams,” “teamwork in medicine,” and “collaborative healthcare.” The distinction between academic and non-academic sources was based solely on the texts’ features and origin. For academic sources, we limited our search to peer-reviewed articles published in the last 30 years, which were exclusively identified through the Google Scholar search engine. Non-academic sources consist of blogs and videos published in the last 15 years collected via the Google search engine, and all the sources in video format were converted to a transcript before content analysis.

Non-academic articles do not have a standard indexing method nor dedicated databases, making a systematic search through the impossibly large body of work infeasible and inappropriate. Instead, we adopted a theoretical saturation approach (Saunders et al., 2018) by which our search ceased once newly included texts no longer introduced new concepts or applications to the already-identified sources. We mimicked this approach for the collection of academic articles to ensure consistency. Sources discussing a diverse set of sub-disciplines were included (e.g., nursing, surgery, emergency medicine, and primary care) to ensure broad coverage of the source material and relevance to healthcare teams. All text sources were taken from North American or European journals/websites, thereby primarily presenting data from a Western cultural perspective.

Following the collection of sources, we proceeded with content analysis. The text sources were entered into the content analysis software MAXQDA (VERBI Software, 2024) to perform frequency analysis, which outputted the most common words and phrases used in the text sources. Frequency analysis is a transparent and efficient way to uncover the central themes in a body of literature, and it is the most commonly used content analysis method. Frequency analysis is supported by the lexical hypothesis (Saucier & Goldberg, 2001), which posits that the importance of traits is reflected by the number of terms we have for them, so the most frequently used words/phrases often reflect the most important concepts. The list of common words and phrases was filtered by a researcher to remove unequivocally irrelevant, nonsensical, and redundant words/phrases. Redundancy was defined as words that have the same root and reflect the same information (e.g., “coordinate,” “coordination,” and “coordinating” all imply the team is coordinated), and only the most apparently nonsensical terms were removed (e.g., “not,” “how,” and “there”).

This set of common words/phrases was used with the keyword-in-context feature of MAXQDA to ensure that the final items for factor analysis were worded according to the text from which they were extracted. The keyword-in-context feature contextualized the most common words/phrases identified in the frequency analysis, and an archetypal phrase from the source materials was used as the basis for item generation. For example, the words “social,” “interpersonal,” and “relationship,” were linked to the sentences “Other doctors indicated that social relations could have a positive effect on teamwork and that team stability,” “Interpersonal incompatibilities, have been shown to detract from effective team functioning,” and “They also state that good social relationships between team members as well as group cohesion are important ingredients of teamwork.” The first sentence appeared to be representative of the idea that positive social relationships and compatibility are related to teaming outcomes, and it inspired the item “Team members have friendly social relationships.” Each of the items was written to be precise and universally understandable while still being rooted in the text sources. Other examples of items include “Team members work within an environment of innovation,” “The team works in a high-stress environment,” and “The team has a distinct leader.”

After the extraction of many distinct items, these items were arranged into a relationship matrix for similarity analysis. Individually and independently, three researchers rated the similarity of each item pair from very dissimilar (−1) to very similar (+1). Fluency in English and some familiarity with teams were the only qualifications needed for the researchers to give similarity ratings. Because we aimed to characterize human teams and humans are experts in their lived experience, there was no specific expertise required on the part of the raters aside from being on (or observing) a team at some point. For example, most humans would consider “Teams are made up of professionals with different roles” and “People with different expertise work complementary to each other” to be fairly similar, while “Team members collaborate on tasks” and “Team members operate as individuals” are fairly different. All three researchers used a set of practice items to ensure the clarity of the rating procedure.

Ratings provided by the researchers produced three rating matrices which were averaged to generate a similarity matrix—the input for factor analysis. The factor analysis was run in R using varimax orthogonal rotation and a minimum residual factoring method. The outputted potential factors with eigenvalues > 1 were plotted on a scree plot, which was used to determine the number of factors to retain. The factor loading scores were used to determine the meanings of the factors. Factor analysis was used in this study because it is a statistical methodology that describes the relationship among many observed variables in terms of a few underlying and unobservable factors (dimensions).

Outcome

Our search for text sources generated a total of 29 academic articles, 13 blogs, and 7 videos (a sample of the text sources is cited in the references). Based on these texts, our content analysis yielded 50 distinct items for evaluation during the factor analysis. As a result of the factor analysis, we found 17 potential factors with eigenvalues > 1, and following the standard method of inspecting the generated scree plot, 8 factors are retained.

The conceptual label and associated meaning of each factor were determined based on the factor loading scores of the various items. A summary of the retained factors is included in Table 1, and they are listed as follows. (1) Synergistic versus antagonistic: The extent to which team members communicate, collaborate, and act politely. This factor explains the most variance (largest eigenvalue), and it appears to capture a wide variety of elements. It was highly correlated (|λ| > .3) with more than 20 of the 50 items presented for factor analysis, including “Information is shared between team members,” “Team members collaborate on tasks,” and “Team members have friendly social relationships.” From these highly correlated items, it is clear that this factor captures a wide range of elements, including interpersonal relationships, cooperation, and information sharing, making it essentially a catch-all dimension for positive social interaction. (2) Heterogeneous versus homogeneous: The extent to which all team members are similar in skill and perspective. Despite being the factor that accounts for the second-highest amount of variance, it is highly correlated with fewer than 10 of the 50 items. This factor is positively correlated with items like “Team members have different backgrounds” and “Miscommunication is common in the team,” and it is negatively correlated with factors like “Team members interpret tasks similarly” and “Team members share the same task assumptions.” This factor appears to capture elements of team composition, including role assignment/diversity, frequency of misunderstandings as a result of that diverse composition, and shared mental modeling as affected by the team composition. (3) Hierarchical versus self-managed: The extent to which the team operates in a rigid hierarchical structure. This factor was highly correlated with items like “The team has a distinct leader,” “Team tasks are determined by a manager,” and “The team includes trainees who learn from supervisors.” This factor captures elements associated with leadership and leadership structure, including trainee-trainer relationships, which are common in healthcare. This factor was also highly correlated with the item “There is a clear hierarchy where subordinates support a head physician.” This suggests that this factor also captures elements of mandated (or hierarchical) support as opposed to the more mutual support captured by the synergistic versus antagonistic dimension. (4) Goal-oriented versus process-oriented: The extent to which the team is driven toward a specific goal. This factor was very positively correlated with the items “Team members work towards common goals” and “The team accomplishes their goals.” This factor appears to capture elements related to both goal pursuit and achievement, setting goal-oriented teams apart from teams more focused on process work, like ideation and innovation teams. (5) Supportive versus reserved: The extent to which team members provide each other with emotional support. This factor was positively correlated with items like “The team is an emotionally safe environment,” “Team members have a social support network,” and “Team members support each other, limiting exposure to stressors.” This factor appears to capture elements based on deep emotional support in the face of trauma (as commonly experienced in the healthcare field). Not to be confused with synergistic versus antagonistic, this factor is related to emotional support in the face of trauma, while synergistic versus antagonistic is more concerned with support in a broad sense, as being related to typical work functions like coordination and information sharing. (6) Complexity versus simplicity: The extent to which the team works in complex environments. This factor was positively correlated with items like “The team works in complex, highly variable environments,” “Teams adapt and evolve in response to the environment,” and “The team works in a high-stress environment.” This factor appears to capture several environmental elements associated with complexity, including complexity (directly), variability, and stress. (7) Technological versus traditional: The extent to which the team is accepting of cutting-edge technology. This factor was positively correlated with items like “Robotic team members assist human team members with simple tasks” and “Robotic team members proactively support the decisions of human team members.” Based on these item correlations, it seems like the factor captures elements of robotic integration and acceptance. However, many medical teams do not use robots specifically, like doctors treating patients with mental health conditions, so we propose a slightly broader conceptual definition to include acceptance of all cutting-edge technology (which oftentimes happens to be robotics). (8) Specialist versus generalist: The extent to which the team works on/in specific tasks/environments. This factor is positively correlated with items like “Teams are formed to respond to the demands of a specific environment” and “The team is formed for a specific task,” while being negatively correlated with items like “Teams are formed spontaneously for emergent tasks.” This factor appears to capture the elements of team structure that are tied to task specificity. This is not to be confused with the complexity versus simplicity factor, which is meant to differentiate teams that work in complex versus simple environments.

OPEN IN VIEWER

Table 1.

Identified Factors and Their Associated Meanings.

Conceptual label	Associated meaning
Synergistic versus antagonistic	To what extent do team members communicate, collaborate, and act politely?
Heterogeneous versus homogeneous	To what extent are all team members similar in skills and perspective?
Hierarchical versus self-managed	To what extent does the team operate in a rigid hierarchical structure?
Goal-oriented versus process-oriented	To what extent is the team driven toward a specific goal?
Supportive versus reserved	To what extent do team members provide each other with emotional support?
Complexity versus simplicity	To what extent does the team work in complex environments?
Technological versus traditional	To what extent is the team accepting of cutting-edge technology?
Specialist versus generalist	To what extent does the team work on/in specific tasks/environments?

Conclusion

This research utilized the quantitative and empirical approaches of content analysis and factor analysis to uncover the underlying dimensions of teamness in the medical setting. The results provide strong evidence for eight dimensions by which medical teams should be characterized. The dimensions presented in this study have some key differences and similarities when compared to findings from the literature that were not based on quantitative analysis and empirical evidence. Based on the authors’ interpretation of the literature (as summarized in Hollenbeck et al., 2012), Sundstrom et al. (1990) proposed nine factors: industry, organizational hierarchy, education level of the team, scope of team activity, member autonomy, activity routinization, amount of time members work together, degree of skill differentiation, and difficulty of performance evaluation. Some of these factors are consistent with the factors we present here: skill differentiation is similar to heterogeneous versus homogeneous; member autonomy is related to hierarchical versus self-managed; and difficulty of performance evaluation is related to complexity versus simplicity. The dimensions presented in our study also have similarities and differences when compared with the three dimensions presented by Hollenbeck and colleagues: skill differentiation, authority differentiation, and temporal stability. Two of these dimensions are addressed by our empirically based dimensions on a 1:1 basis: skill differentiation by heterogeneous versus homogeneous and authority differentiation by hierarchical versus self-managed. Interestingly, the dimensions presented by the works of Sundstrom et al. (1990) and Hollenbeck et al. (2012) do not address the factor that accounted for the most variance in our study: synergistic versus antagonistic. The synergistic versus antagonistic dimension was one of the most interpersonal-focused dimensions presented in our study (along with supportive versus reserved), and it is possible that the works of Sundstrom and Hollenbeck overlooked more interpersonal factors. This could be due to differences in analysis methodology, or this difference could signal the healthcare specificity of our presented dimensions.

Furthermore, it is notable that the healthcare-specific dimensions we present here do not cover all the dimensions presented by either Sundstrom et al. (1990) or Hollenbeck et al. (2012). This could be due to the healthcare-specific nature of our dimensions as compared to the more generalized dimensions presented in earlier works. This is likely the case for why we do not present dimensions that address Sundstrom’s dimensions of industry or organizational hierarchy: all healthcare teams function within the same industry, and most major hospital systems have similar hierarchical structures, eliminating the need for those two dimensions. Furthermore, the dimensions presented here do not account for Sundstrom’s education level dimension, likely because most medical practitioners have incredibly high levels of education, and they do not account for Hollenbeck’s temporal stability dimension because most healthcare teams are frequently changing. This is evidence to suggest that the dimensions presented in this article are uniquely applicable to the healthcare domain, and they may not be as applicable elsewhere. Further emphasizing this point, we can see that previously presented single-aspect healthcare team research studies on expertise (Garrett et al., 2009), collaboration (Fox et al., 2024), or task complexity (Molleman et al., 2010) are all addressed by our dimensions. Expertise is covered by a combination of specialist versus generalist and heterogeneous versus homogeneous, collaboration is covered by synergistic versus antagonistic, and complexity is covered by complexity versus simplicity. While some of the dimensions we present in this work appear to reflect the characterization of both healthcare-specific and more general teams, some of our dimensions appear to be more healthcare-specific (synergistic versus antagonistic, supportive versus reserved, and technological versus traditional). Both healthcare-exclusive dimensions and dimensions that also apply to more general teams can collectively make up a set of dimensions specifically for the characterization of healthcare teams.

This is an important first step in defining a set of dimensions that can be used to characterize healthcare teams and uncover team characteristics that promote effective care delivery. A clearly defined set of teamness dimensions, specific to the healthcare setting and based on empirical data and statistical analysis, can allow for better characterization of healthcare teams, thereby improving the research practice. Moreover, once these dimensions are associated with patient and provider outcomes, interventions can be implemented to address factor scores associated with negative outcomes like provider burnout and patient harm.

Most of the academic sources we obtained were about studies conducted at major teaching hospitals, and most of the non-academic sources referenced healthcare teams at large hospitals. Therefore, these results are generalizable to healthcare teams in large hospitals, including surgical teams, nursing teams, and emergency medicine teams. Future research will aim to enhance the validity of these dimensions by exploring a more diverse mixed-methods analysis methodology which could deepen our understanding of the identified dimensions. Other qualitative and quantitative methodologies, like thematic analysis or code mapping, could complement our existing analysis. Finally, we seek to study these dimensions in healthcare settings to evaluate practicing healthcare teams and explore the relationship between factor scores and patient care outcomes.