A systematic review of team-building interventions in non-acute healthcare settings

Given this state of affairs, we had three goals for this review. First, we aimed to describe the characteristics of team-building interventions that have been applied in non-acute healthcare settings. Second, we aimed to identify the characteristics of empirical studies that have tested these team-building interventions in such settings. Third, we aimed to evaluate empirical results of these team-building interventions in four outcome domains: trainee evaluations, teamwork attitudes/knowledge, team functioning, and patient impact. To our knowledge, this is the first review of team-building interventions to focus specifically on non-acute settings.

For this review we have adopted the definition of team-based healthcare put forth by Mitchell and colleagues in their Institute of Medicine (IOM) discussion paper [1], itself adapted from Naylor and colleagues [17]:

“Team-based health care is the provision of health services to individuals, families, and/or their communities by at least two health providers who work collaboratively with patients and their caregivers—to the extent preferred by each patient—to accomplish shared goals within and across settings to achieve coordinated, high-quality care.” [1] (page 5).

Furthermore, there is diversity in the literature regarding how to label team-building approaches themselves, with some authors using the term “team-building intervention” (e.g. [18]), while others use some variation of “team training” (e.g. [11]), some combination of the two (e.g. [19]), or one of a host of other terms (e.g. [20]). For simplicity we have chosen to adopt the term “team-building intervention” to refer to any systematic approach to improving healthcare team functioning for the purposes of this review (see Methods for details).

We developed a guiding conceptual model of non-acute healthcare team-building based on previous literature (Fig. 1, which we have entitled the Team Effectiveness Pyramid). We propose as a starting point that building effective healthcare teams in non-acute settings requires a baseline level of resources (Pyramid Level 1), including a supportive organizational context [5], basic tangible resources such as staffing [3, 21] and space [22], and psychological resources in the form of civility, mutual respect, and psychological safety [23, 24] for the staff who comprise the team. The model proposes that these preconditions provide fertile ground for team-building interventions (Pyramid Level 2) to lead to enhanced teamwork (Pyramid Level 3). The bullet points at this level are not meant to be comprehensive, but rather to list some of the qualities frequently cited in this domain [5]. Finally, our model posits that good teamwork will in turn lead to improved patient impact in the form of both clinical outcomes and patient satisfaction (Pyramid Level 4) [1, 2].

The four outcomes we chose to investigate for this review align closely with the Team Effectiveness Pyramid. Specifically, as described above, our outcome domains included trainee evaluations (Pyramid Level 2), teamwork attitudes/knowledge (Pyramid Level 2), team functioning (Pyramid Level 3), and patient impact (Pyramid Level 4). While we believe that foundational resources (Pyramid Level 1) are crucial to healthcare team-building, addressing this issue was beyond the scope of this review, as most studies of healthcare team-building provide only general information about the settings in which they are conducted.

Our initial search terms consisted of the following: ((“Patient Care Team”[Mesh]) AND (model[All Fields] AND Review[ptyp])); ((“team training”[tiab] OR “teamwork training”[tiab]) AND review[tiab]); ((“Patient Care Team”[Mesh] OR “patient care team” OR team*[tiab] OR interdisc*[tiab] OR multidisc*[tiab]) AND (model[tiab] OR framework[tiab]) AND review[tiab]). The first author screened all titles resulting from these searches to identify potentially relevant papers for full-text review. Inclusion criteria for these reviews consisted of the following:

We read the manuscript body and reference list of each of the reviews identified in Review Stage 1 above, with a goal of identifying team-building interventions. Inclusion criteria at this stage consisted of the following:

We conducted a series of additional literature searches in Review Stage 3—one for each team-building intervention identified from reviews in Review Stage 2. The goal of these separate searches was to identify empirical studies evaluating the use of each team-building intervention in non-acute healthcare settings. Sources included Google Scholar, PubMed, associated websites (for team-building interventions that are free and/or publicly available), and direct contact with developers of the team-building interventions. Inclusion criteria for empirical support consisted of the following:

The first author and two co-authors independently rated a subset of ten manuscripts (including reviews, team trainings, and empirical support) identified by the search process above, including some manuscripts that the first author determined had met inclusion criteria, and others that the first author determined had not. Fleiss’s kappa for all three raters for this subset of manuscripts was.70, indicating acceptable reliability [27] for our manuscript identification process.

We chose a descriptive approach to achieve our first and second study aims; specifically, we report the characteristics of the team-building interventions and empirical studies identified through our review process. Similar to previous reviews in different healthcare contexts (e.g. [11]) we chose to report the following information for each empirical study: the length of the intervention; the number and types of providers trained; the characteristics of the control condition (if any); whether a pre-training needs analysis was conducted [28]; and whether the intervention was modified from its original version. We also evaluated the quality of the overall body of empirical studies, consistent with criteria on study bias from the Cochrane Collaboration [29]. This involved assessing the risk of selection bias, performance bias, detection bias, attrition bias, and selective reporting in the identified studies.

For our third study goal, the diversity of study designs and outcomes reported in the field made meta-analysis impractical. Instead, we chose to descriptively catalogue the empirical support for each team-building intervention identified in terms of trainee evaluations, teamwork attitudes/knowledge, team functioning, and patient impact. Our approach therefore meets the criteria for a systematic review [30].

Table 3 contains results regarding the four outcome domains. With some exceptions, the 12 empirical studies that collected trainee evaluations reported positive scores among participating staff in this domain (with 68–100% rating their experiences as positive). Only one of the six studies that assessed teamwork attitudes/knowledge [47] found statistically significant improvement in knowledge of teamwork principles as evidenced by the Outcomes elements of the WeLearn framework [48]. Qualitative results from that study also supported increased awareness of teamwork principles. Other studies, however, found no statistically significant differences in attitudes toward teamwork pre- to post-intervention [35, 49], or between the intervention and control group [34].

Eighteen empirical studies also reported results from post-training assessments of team functioning such as the Team Development Measure (TDM [50]) or Workforce Dynamics Questionnaire [51]. Most such studies showed improvement in a few [35, 40, 46, 52] or several teamwork-related domains [33, 41, 53‐55]. However, other studies did not find statistically significant improvement in team functioning post-intervention [20, 34, 38, 39, 56]. This variability also manifested within specific teamwork domains (e.g. some studies reported enhanced communication as a result of the team-building intervention [40, 41, 53] while others did not [39, 44]).

Additionally, six studies investigated clinical outcomes or patient satisfaction for patients treated by clinicians who had participated in the team-building intervention, and, of the studies that did investigate these outcomes, findings were generally mixed. For example, two studies that investigated falls in nursing homes [40] and an orthopedic unit [34] found at least a modest reduction in falls from pre- to post-intervention, but other studies either found no statistically significant changes in clinical outcomes (e.g. [37]) or did not subject such outcomes to statistical testing (e.g. [47]).

Consistent with our guiding conceptual model (Team Effectiveness Pyramid; Fig. 1), we reviewed outcomes in four domains: trainee evaluations; attitudes toward, and knowledge about, teamwork; team functioning; and patient impact. As detailed in Table 3, empirical studies generally reported positive trainee evaluations, although a shortage of credible comparison conditions (e.g. different team-building approaches) made it difficult to determine how meaningful this finding is. Some of the studies we reviewed also found their interventions to be associated with improvements in knowledge of the principles of team-based care or attitudes toward the importance of teamwork—but only one study found such improvements to achieve statistical significance, and several studies either found no significant change in this domain or minimal differences between the intervention and control teams.

Similarly mixed results emerged for team functioning, with some studies finding robust improvements associated with team-building interventions. Others found significant changes for only a small set of team functioning variables, or no differences at all associated with the team-building intervention. In several cases, positive results appeared to be selectively chosen from among many potential subdomains (e.g. focusing on positive findings for one aspect of communication, while downplaying negative findings for other aspects of communication).

Fewer studies investigated clinical outcomes or patient satisfaction for patients treated by clinicians who had participated in team-building. The existing findings in these domains were generally mixed, although two studies that investigated falls in nursing homes [40] and an orthopedic unit [34] respectively found at least a modest reduction in falls from pre- to post-intervention. Clearly, future research should include assessments of patient impact to fully identify the potential benefits of such interventions.

As described in Table 1, many team-building interventions featured a workshop as a central component, with of subset of these including either repeated workshops or use of ongoing (e.g. weekly) team meetings to continue developing teamwork practices. Based on this variation, it is not surprising that the total length of the team trainings ranged widely; median length was about 6 months of active teamwork development. Workshop activities ranged from traditional classroom instruction, to team-building exercises, to case-based learning. In at least one case, trainee evaluations revealed that participants preferred more innovative teaching methods (e.g. case-based discussion [49]), a finding that is consistent with previous reviews [58] and adult learning theory [59].

The teamwork topics included in this review’s team-building interventions were similar to those identified in previous reviews from acute healthcare settings (e.g. [11]). These topics most commonly included communication, leadership, problem-solving, conflict management, and team goal-setting. Many of the team-building interventions we identified also included some clinical training or coverage of administrative issues such as accreditation (e.g. [41]). A pre-training needs analysis is a common component of team-building [11]; 48% of our identified empirical articles (12/25) noted inclusion of such a needs analysis. Our guiding conceptual model (Team Effectiveness Pyramid; Fig. 1) included the capacity for engaging in process improvement as a possible result of enhanced teamwork, but this was rarely mentioned in the identified team-building interventions.

Among the team-building interventions identified via this review, TeamSTEPPS (Team Strategies and Tools to Enhance Performance and Patient Safety) was the only one that has been tested by more than one research team in more than one sample. The core TeamSTEPPS curriculum was initially designed for acute care settings by the Agency for Healthcare Research and Quality (AHRQ [32]), but individual research teams (e.g. [34, 46]) and AHRQ itself have successfully adapted TeamSTEPPS for use in non-acute settings. Strengths of the TeamSTEPPS approach include the ready availability of supporting materials from AHRQ website [32], as well as the incorporation of a pre-training needs analysis to ensure that the specific curriculum implemented matches the needs of the team being trained. Furthermore, for organizations desiring more explicit support in implementing TeamSTEPPS beyond the materials provided by AHRQ, several private entities offer TeamSTEPPS-oriented trainings (e.g. Lifewings [60]).

With the exception of TeamSTEPPS (described above), most of the interventions we identified have only been validated in one study—which was typically published by the developers of the intervention itself. Furthermore, in several cases, multiple empirical studies were published on the same validation sample, and the typical empirical study was conducted with just 6–8 teams within six clinics (for a total of about 100 staff trained in the median study we reviewed). These numbers suggest that few team-building interventions (beyond TeamSTEPPS) have been subjected to exhaustive empirical study in the form of multiple studies conducted by different research teams across multiple non-acute samples.

It was difficult to determine whether individual research teams made systematic modifications to the team-building interventions in the empirical articles we identified. Many such interventions are inherently flexible, making it nearly impossible to differentiate training-consistent from training-inconsistent adaptations based on published literature. However, three articles clearly described either the addition of simulation modules [36] or solicitation of patient goals [37, 38]. Following the conventions of Stirman and colleagues [61], these represent modifying the intervention format as well as adding elements to the training content.

Our study results indicate a high risk of bias in several domains specified by the Cochrane Collaboration [29]. These include selection bias, performance bias, detection bias, attrition bias, and selective reporting bias. The first three of these potential biases were difficult to avoid given the paucity of control conditions (Table 2). Furthermore, the few control conditions we found typically involved the commitment of fewer resources than the interventions being studied. Close inspection of Table 2 reveals that several studies were prone to attrition bias (based on teams dropping from the study before final data collection) or selective reporting bias (by emphasizing significant results and downplaying equivocal results even within teamwork domains). Thus, none of the studies we found could definitively address the question of whether the resources put into team-building would have been better spent on more clinical or administrative staff for the team(s) in question.

Taken together, our results emphasize that research on team-building in non-acute healthcare settings lags behind that in acute settings. Furthermore, we did not find consistent positive results—in terms of improvement in teamwork attitudes/knowledge, team functioning, or clinical outcomes—across the studies we reviewed. Thus, an important next step for the field is to determine the circumstances under which team-building will be most effective in non-acute healthcare settings. For example, teams in some settings may have limited overlap in caseloads among team members, making it especially difficult to establish shared goals within the team or foster enthusiasm for team-building. We have seen this dynamic occur in outpatient mental health teams [62], as staff may find it difficult to commit to team meetings and shared activities if only a small portion of their caseload is treated by other members of the team. In such situations, it may be important to better align team caseloads to maximize the potential for coordination within the team before team-building can begin in earnest.

Similarly, more research is needed regarding what we have labeled foundational resources (Level 1 of our Team Effectiveness Pyramid, Fig. 1). There is broad agreement that such resources are required for teamwork to blossom—and well-developed bodies of literature on the importance of the individual foundational elements listed in the Pyramid (e.g. [23]). However, we are not aware of any concrete methods for determining whether such resources are sufficiently in place for team-building to be indicated. For example, if an outpatient clinic is extremely short-staffed, then taking clinical time offline for team-building may simply result in more stress and burnout on the part of clinic staff. In such circumstances, it is possible that team-building should be postponed until new staff can be hired or patient flow within the clinic can be adjusted to reduce provider burden. Consistent with this, survey results from one empirical study we reviewed indicated that providers were concerned about the amount of time required for team-building [52]. A robust method for determining the minimum levels of foundational resources needed for non-acute healthcare teams to profitably engage in team-building would be valuable contribution to the field. In the meantime, we recommend that any team-building intervention include a thorough pre-training needs analysis [28] that includes an assessment of the resources available to the team. Ideally, such analysis would inform possible adaptations of the team-building intervention itself to match local needs.

Results from this review should be considered in light of several limitations. First, given the breadth of the field, we relied on a multi-stage search process—identifying reviews, then using these reviews to identify team-building interventions, and finally using those reports to identify articles that empirically evaluated each intervention. This leaves the possibility that we missed interventions or studies that have not been included in previous reviews—especially team-building interventions developed too recently to be included in published reviews. However, this type of method has been used before in different healthcare contexts [25], and the use of not just a literature search but also the examination of the reference lists of dozens of review papers leaves us confident in the scope of team-building interventions we identified. Furthermore, our Review Stage 1 resulted in the identification of several team-building interventions directly. We also contacted intervention developers directly to inquire about potential other empirical reports we might have missed; in no cases did this reveal studies that our search process did not. Second, we were limited by the amount of information available in the peer-reviewed articles we reviewed. It is therefore possible that we may have underestimated the extent to which certain elements (such as a pre-training needs analysis or systematic modifications) were used in our reviewed studies. However, we would not expect this to affect our core study findings regarding the outcomes of team-building interventions. Third, given the diversity in the literature, we were unable to conduct a formal meta-analysis. Instead, we endeavored to narratively describe results in four outcome domains that are prominent within the literature and aligned with our Team Effectiveness Pyramid conceptual model (Fig. 1).