Background

Patient safety is an important indicator of hospitals’ organizational performance. Approximately 10 % of patients suffer adverse events and half of those are deemed preventable [1]. Vincent defined patient safety as the absence of preventable adverse events – events that are a consequence of healthcare interventions rather than the patients’ condition [2]. Healthcare is predominantly provided by teams – two or more people each with specialized roles and responsibilities whilst interacting with the shared goal of patient care [3]. Consequently, in addition to medical competence, effective teamwork is critical for safe patient care [4–7]. This includes both observable team behaviors and clinicians’ perceptions of interpersonal team processes. For example, several studies have linked better coordination or team psychological safety to fewer medical errors and better patient outcomes such as length of stay [8–10]. Also, specific team behaviors, for example leadership, information sharing or decision making and team properties (e.g., shared mental models) are associated with performance indicators such as decision and execution latency or protocol adherence [5, 11, 12].

Teamwork is also an important predictor of another indicator of hospitals’ organizational performance: the well-being of healthcare providers [13, 14]. Reduced occupational well-being or high psychological strain may develop as an immediate or long-term response to stressors [15] and is highly prevalent in healthcare workers [16, 17]. Teamwork may constitute such a stressor. For instance, dysfunctional inter-professional teamwork predicts increased acute and chronic clinician strain [18, 19]. However, effective teamwork may protect team members from the effects of work stress, since positive perceptions of teamwork are associated with enhanced occupational well-being indicators such as better mental health in nurses and physicians [20, 21].

Lastly, clinicians’ occupational well-being and patient safety are interrelated. Reduced clinician occupational well-being is associated with objective and subjective patient safety indicators such as mortality ratios, clinician-rated safety and reported errors [13, 22, 23]. Highly strained clinicians might thus pose a threat to patient safety since patient safety incidents are stressors that may lead to decreased clinician well-being: clinicians report increased emotional distress following medical error [24].

Studies investigating associations between teamwork, clinician occupational well-being and patient safety originate from very different strands of research – medical, nursing, and psychology. So far, the evidence generated has not been drawn together for systematic evaluation. While this research showed that relationships exist between the independent associations of teamwork, clinician occupational well-being and patient safety, few studies investigated them simultaneously. Moreover, the mechanisms underlying the relationships and causalities between either two – and potentially all three – concepts are largely unknown.

To overcome this research gap, we aimed to provide an overview of the current state of research on relationships between at least two of the three concepts of teamwork, clinician occupational well-being, and patient safety in hospital settings. In a systematic review, we summarized theoretical foundations, sample, methodology, and empirical findings, and evaluated overall study quality. Based on relevant psychological theories and on the findings of the systematic review, we developed a conceptual framework integrating the three concepts. Specifically, we propose theoretically informed causal relationships between the concepts, describe focal points of past research, and identify gaps in the current knowledge. The framework is intended to serve as a blueprint both for future studies intended to benefit clinicians’ occupational well-being and patients’ safety.

Methods

Definition of central concepts

Teams and teamwork

In order to include a diverse array of healthcare teams, we used rather broad definitions of teams and teamwork. A team is defined as a group of two or more people embedded in an organizational system with specialized roles who are interdependent and socially interact with each other in order to reach a common goal [3]. Studies were included if the teams investigated matched these criteria. We based our definition of teamwork on the model by Marks and colleagues, which includes transition (planning, goal formulation), action (coordination, monitoring), and interpersonal processes (conflict management, motivation, or team members’ perceptions thereof, e.g., team climate) [25]. Thus, we excluded studies comparing the effects of team-based work to other forms of work organization. We included leadership if it was clearly directed at the team level, and excluded studies examining dyadic or organizational leadership processes. Lastly, we excluded studies assessing inter-team processes, because we were interested in how working within a team relates to patient safety and clinician well-being.

Clinician occupational well-being

Under occupational well-being, our aim was to identify studies investigating both positive and negative aspects [26–28]. We specifically included studies, based on Lazarus’ stress model, which investigated work-related psychological or physiological strain as an individual’s short- or long-term perception of, or response to, stressors at work, such as burnout [15]. In the case of workplace stressors, these are often referred to as job demands. According to the job demands-resources model, job demands are defined as physical, social, or organizational job characteristics that require increased effort, thereby depleting the individual’s energy and eventually decreasing occupational well-being or increasing strain [29]. We included studies examining mental fatigue (i.e., exhaustion or lack of energy that is not due to physical overexertion) if direct measures of mental fatigue were used rather than being inferred from external indicators such as shift duration [30]. Furthermore, we included general or work-related positive indicators of occupational well-being as an outcome of lack of job demands, or the abundance of job resources, such as work engagement. Job resources are physical, social, or organizational characteristics that help maintain the individual’s energy, thereby increasing occupational well-being or reducing the strain caused by job demands [29]. Our aim was to focus the review on studies examining occupational well-being as the result of appraisal of a stressor or lack thereof. For this reason, we excluded studies examining aspects of occupational well-being in the wider sense, i.e., studies investigating aspects that are the result of a large array of workplace characteristics, such as job satisfaction or organizational commitment. We furthermore excluded studies examining personality traits or psychopathological disorders. Lastly, we excluded long-term chronic somatic disorders such as lower back pain, as it is often unclear whether these conditions are caused by continuous psychological strain or physical activities.

Patient safety

We defined patient safety as “the avoidance, prevention, and amelioration of adverse outcomes or injuries stemming from the process of healthcare” [31]. We included studies covering variables that could directly affect a patient’s health status (i.e., reported or observed errors, key actions not being performed), as well as subjective patient safety ratings and objective morbidity-mortality-data. We excluded studies assessing quality of patient care or using safety climate as a substitute outcome measure.

Search strategy

We searched six databases (Medline, PsycArticles, PsycInfo, Psyndex, ScienceDirect, and Web of Knowledge) to identify relevant literature. Our a priori assumption was that teamwork, clinicians’ occupational well-being and patient safety are related to each other. Thus, we combined two of the three keywords TEAMWORK, PATIENT SAFETY, WELL-BEING with AND. We then combined the results with OR. In order to receive both relevant and manageable results, we applied a number of strategies (e.g., MeSH/thesaurus terms, related terms, alternative spellings, truncations or plural forms, and adjacency terms; the complete search strategy for one database can be viewed in Additional file 1). Further inclusion criteria were: peer-reviewed journal articles, published in English between January 2000 and June 2015, referring to a hospital setting. We included studies sampling practicing nurses or physicians. If multiple publications were based on the same dataset, we either selected the paper that was first published or reported the most extensive data analysis. Finally, we hand-searched reference lists of the selected articles and systematic reviews we identified in our initial search.

Screening and selection procedure

Two raters screened (AW and either MD, SS, or JV) all references independently. We scanned the title and abstract at the first stage and included studies investigating at least two of the three concepts (teamwork, patient safety, clinician well-being) in a hospital setting. At the second stage, we included studies reporting a statistical relationship between at least two of the relevant concepts, which clearly described measurement methods and were published in peer-reviewed journals. Disagreements between raters at the first screening stage led to inclusion, after which we resolved disagreements at the second stage by consensus discussion.

Quality rating

To systematically assess study quality, we combined and slightly adapted existing systems. [32, 33] Ratings were based on a maximum of 19 items (not all items were applicable for all studies) covering topics such as validity of measures or statistical analyses. Items were rated as 0 = major limitations/not applicable/not mentioned, 0.5 = some limitations, or 1 = fulfilled. Two raters (AW and MD) independently evaluated study quality and resolved disagreements through discussion. All quality rating items are available in Additional file 2.

Data extraction

We extracted study setting, study design, method of data collection, data analysis, and study outcomes from the selected studies. If results were described in sufficient detail but effect sizes were not reported, we calculated them according to convention [34, 35] to give an indication of whether a statistically significant relationship was large enough to infer practical significance (see Table 1 for an overview of effect size magnitudes) [36]. In some studies, teamwork, clinician occupational well-being and patient safety may have been analyzed within a larger context (e.g., nurse working environment), however, only relationships between the variables of interest to this review are reported.

Table 1 Overview of effect sizes [34, 35, 147]
Full size table

Framework development

Building on the results of our systematic review, the framework development followed two stages. Based on the assumption that teamwork, clinicians’ occupational well-being and patient safety are correlated, our aim was to provide a framework summarizing the current state of research and exploring the underlying mechanisms and causal directions between the concepts. First, we examined measures, samples, and definitions of teamwork, well-being and patient safety to provide an overview of the evidence, and to detect trends and shortcomings in current research. Second, we drew from the theoretical foundations of the reviewed studies and from psychological theories relevant to the topic to aid interpretation of the findings and formulate hypotheses regarding the causal relationships between teamwork, clinician occupational well-being and patient safety to point out avenues for future research.

Results

The database search from January 2000 to June 2015 yielded 26,870 results. We identified an additional 62 publications through other sources (e.g., hand-searching references lists). After removing duplicates, 21,186 publications remained. Following title and abstract screening, we retrieved the full text of 1697 publications. Examining full-texts and hand-searching reference lists led to the inclusion of 98 publications (see Fig. 1). Of these, 25 (26 %) investigated relationships between teamwork and well-being, 43 (44 %) between teamwork and patient safety, 25 (26 %) between well-being and patient safety, and five (5 %) included all three concepts.

Fig. 1
figure 1

Flow diagram illustrating search method and inclusion/exclusion criteria

Full size image

Quality rating

Selected studies were of medium (49 studies) or high quality (49 studies; see Tables 2, 3, 4 and 5 for individual quality scores). Average study quality was similar across the three concepts; teamwork, well-being and patient safety (i.e., 11.48 for teamwork/well-being [SD = 1.68], 11.03 for teamwork/patient safety [SD = 2.04], 10.92 for well-being/patient safety [SD = 2.013], and 11.20 [SD = 0.75] for teamwork/well-being/patient safety)). We excluded the low quality studies identified in this review at an early stage because the methodological description was insufficient for data extraction and assessment of quality (see Fig. 1).

Table 2 Relationships between teamwork and well-being
Full size table
Table 3 Relationships between teamwork and patient safety
Full size table
Table 4 Relationships between well-being and patient safety
Full size table
Table 5 Relationships between teamwork, well-being and patient safety
Full size table

Relationships between teamwork and clinician occupational well-being

Design & sample

Out of 25 studies examining relationships between teamwork and clinician occupational well-being, 24 (96 %) used cross-sectional self-report designs, with one study adding a pre-post-shift diary design (Table 2 and box A/B in Fig. 2) [18]. One study employed a longitudinal self-report design [37]. Of these 25 studies, 19 (76 %) surveyed only nurses [18–20, 37–52], one (4 %) physicians [21], one (4 %) midwives [53], and four (16 %) included a mixed sample [54–57].

Fig. 2
figure 2

Integrative framework of teamwork, clinician occuptional well-being and patient safety in hospital settings Notes. *as identified in this review. More explanations on the boxes may be found in the results section. Their content is partly based on Tables 1, 2, 3 and 4

Full size image

Measures

Studies operationalized teamwork most often with the nurse-physician-relations subscale of the Nursing Work Index (NWI; 12 studies/48 %) [18, 19, 41, 42, 44, 46, 47, 50, 52, 54, 55, 58]; and clinician occupational well-being with the Maslach Burnout Inventory or short versions thereof (MBI; 11 studies/44 %; see box A/B and box 2 in Fig. 2) [19, 38, 41–47, 55, 59].

Findings

Studies examining relationships between teamwork and well-being focused on interpersonal teamwork aspects (box A/B in Fig. 2). Most authors assumed that teamwork, a variable inherent to the working context, influences individuals’ general occupational well-being, rather than well-being influencing teamwork. Two studies (8 %) focused on acute strain [18, 54] one of which showed that it was negatively associated with team behaviors such as closed-loop communication or backup behavior [54, 60]. The only longitudinal study reported an effect of teamwork at time 1 on well-being at time 2. However, since this study did not conduct comprehensive analyses (i.e., testing for reverse causal relationships), we could not draw definite conclusions regarding causal relationships between teamwork and clinician occupational well-being [37].

Out of 25 studies examining relationships between teamwork and clinician occupational well-being, 19 (76 %) focused on interpersonal team processes in rather stable nursing teams, such as nurses’ perceptions of interprofessional teamwork or team cohesion [18–21, 37, 39, 41–47, 49–53, 57]. Four studies (20 %) did not address specific aspects of teamwork, but measured it on a general level [38, 40, 48, 54, 55]. One study (4 %) included a short questionnaire on all three team processes (i.e., action, transition, and interpersonal) [56].

Some studies examined the larger clinical work context without formulating assumptions about the specific relationships between teamwork and clinician occupational well-being, the respective findings thus being a by-product of the larger study context rather than a focus of investigation (see column ‘primary topic’ in Table 2). Across the 25 studies investigating associations between teamwork and clinician occupational well-being, 48 out of 62 (77 %) relationships reported were significant and matched author’s assumptions. Of these significant relationships, 15 (31 %) showed a positive association between both positive indicators of teamwork and well-being (e.g., work engagement), whereas 33 (69 %) showed a negative association between positive indicators of teamwork and negative indicators of well-being (e.g., burnout). Out of the 14 non-significant associations, six (43 %) were in accordance with hypotheses (i.e., teamwork on the hospital level is not related to individual burnout) [50]. Thus, overall findings indicate that clinicians perceiving higher quality of teamwork also reported higher occupational well-being or less strain. Effect sizes ranged from small (β = −12.85; f2 = 0.13) to medium (r = −0.47, Table 2).

Relationships between teamwork and patient safety

Design & sample

Studies examining relationships between teamwork and patient safety were very diverse regarding study design, construct operationalization, setting, data collection methods and strength of statistical relationships (see Table 3). Of 43 studies, 25 (58 %) employed video- or live-observation of nurses and physicians in real or simulated acute clinical situations (Table 3a) [5, 6, 11, 12, 61–81]. Five studies (12 %) utilized cross-sectional designs with self-report questionnaires (Table 3b and box C in Fig. 2) [8, 60, 82–84]. Another 13 studies (30 %) employed mixed-method designs (e.g., record reviews or observations plus questionnaires) [9, 10, 85–95]. These studies included one intervention (2 %) [88] and three studies (7 %) with longitudinal aspects [8, 88, 89]. Of the studies using questionnaires seven (16 %) surveyed either nurses [60, 82, 89–93] and seven (16 %) surveyed a mixed sample [9, 10, 83, 84, 87, 94, 95]. Observational studies, in contrast, analyzed teams usually consisting of nurses, physicians (and other healthcare professionals) with the exception of four studies (9 %) [11, 70, 75, 81].

Measures

Observational studies most frequently used the Surgical NOTECHS tool (a tool to observe non-technical skills or team behaviors in acute care settings; see box C in Fig. 2) [96] and its adaptations to various clinical settings to assess teamwork (21 %) [6, 62–64, 69, 70, 76, 85, 86]. Studies assessed patient safety using subjective ratings (6 studies/16 %) [8, 60, 82–84, 95], indicators based on hospital records (13 studies/30 %) [9, 74, 75, 85–94] and observational data (22 studies/52 %) [5, 6, 11, 12, 61–70, 73, 77–81, 85, 86]. These observational studies often used execution of key treatment actions (i.e., steps in the care process that are considered indispensable for successful treatment in potentially life threatening situations, such as the administration of magnesium sulfate for eclampsia) as a proxy measure for patient safety (10 studies/23 %) [11, 65, 68, 70–73, 79–81]. Only one study utilized both objective and subjective patient safety indicators [10].

Findings

Overall, findings were rather inconsistent for the relationship between teamwork and patient safety. All authors assumed that teamwork positively influenced patient safety. A longitudinal study confirmed this assumption (box 1 in Fig. 2) [8]. In the 43 studies investigating teamwork and patient safety, authors reported 239 relationships, 105 (44 %) of which were significant. The majority of survey and observational studies (23/53 %) reported positive associations between teamwork and patient safety [5, 6, 9, 12, 60, 61, 63–65, 67, 69–77, 79, 82, 92, 94, 95]. In line with this, the valence of 198 (83 %) of the 239 significant associations matched authors’ anticipations (i.e., a positive correlation between both positive indicators of teamwork and patient safety, such as coordination and clinical performance, or negative correlation between a positive indicator of teamwork and a negative indicator of patient safety, such as errors). However, the valence of 41 associations (17 %) was not in line with assumptions (i.e., a negative correlation between positive indicators of both teamwork and patient safety or a positive correlation between positive indicators of teamwork and negative indicators of patient safety). Thus, eight studies (19 %) contained findings suggesting that better teamwork was seemingly associated with lower patient safety [8, 10, 11, 62, 76, 85–87, 90]. Some of these findings may have been coincidental, but the majority may be explained by study design. In survey studies on medical errors, instead of the number of errors, authors measure participants’ propensity to report errors, which in turn may be fostered by positive interpersonal team relationships. In a similar vein, positive associations between teamwork and unfavorable patient outcomes like complications or operative duration in observational studies may simply reflect the necessity for increased coordinative behaviors in complicated cases (box 3 in Fig. 2). Moreover, studies investigating links between teamwork and objective or observational patient safety indicators were frequently unable to identify significant relationships (Table 3a, b). For example, two studies (5 %) used a sample of clinicians surveyed with a teamwork questionnaire to examine associations with objective and subjective patient safety indicators [82, 91]. While no association between teamwork and preventable adverse events extracted from hospital records was found [91], the effect was significant when using the frequency of these events reported by lead nurses [82].

Studies using observational tools to investigate teamwork in relation to patient safety focused on action and transition processes with nine (33 %) of altogether 27 studies examining just action processes [5, 11, 12, 61, 68, 74, 76, 77, 79], and six (22 % of observational studies) measuring both [62, 66, 67, 71–73]. Eight observational studies (30 %) measured action, transition, and interpersonal processes without clear distinction between these dimensions [6, 63, 64, 69, 70, 78, 85, 86]. Two observational studies (7 %) focused on interpersonal processes only [65, 80]. One study (4 %) examined transition processes [75]; and one study (4 %) did not provide further details on the teamwork measure [81].

Studies using questionnaires to examine teamwork in relation to patient safety were rather diverse with regard to teamwork processes. The largest part examined teamwork in general, with no clear distinction between action, transition, and interpersonal processes (8 studies/44 % of survey studies) [9, 60, 83, 87–90, 95],. followed by a focus on interpersonal processes (e.g., team climate or nurse-physician relations; 5 studies, 27 %) [8, 82, 84, 91, 92]. Two studies examined interpersonal and transition processes (13 %) [10, 94], and one study examined action and transition processes (6 %) (again, with no clear distinction between these dimensions) [93].

Effect sizes ranged from small (r = −0.08) to large (r = −0.66, Tables 3a, b).

Relationships between clinician occupational well-being and patient safety

Design & sample

The majority of the 25 studies examining relationships between clinician occupational well-being and patient safety (Table 4) targeted either nurses (10 studies/40 %) [14, 22, 98–105] or physicians (12 studies/48 %; box D/E in Fig. 2) [23, 97, 106–115], with only three studies (12 %) using a mixed sample [13, 116, 117]. Twenty studies (80 %) employed a cross-sectional design [13, 14, 22, 97–99, 101–107, 109–112, 114, 116, 117] and four (16 % used a design with longitudinal aspects [23, 100, 108, 113]. One study (4 %) combined survey and observational data [115].

Measures

Studies used the MBI [59] most frequently to assess psychological well-being (14 studies/56 %) [13, 22, 97, 102–106, 108, 111, 112, 114, 116, 117]. Studies measured patient safety using a variety of self-report measures (18 studies/72 %) [14, 22, 23, 97–103, 105, 107, 110–114], with 5 studies (24 %) using objective data such as mortality rates [13, 104, 106, 109, 117]. Two studies (8 %) assessed patient safety via observational data [115, 116].

Findings

Authors of the 25 studies examining clinician well-being and patient safety followed two lines of reasoning: Some assumed that committing an error (equaling reduced patient safety) induces (short-term, emotional) distress in clinicians (4 studies/16 %) [13, 97, 100, 103], while the majority of researchers theorized that high (chronic) strain causes employees’ performance to suffer, thus being detrimental to patient safety (20 studies/84 %; box D/E in Fig. 2) [22, 23, 98, 99, 101, 102, 104–113, 115–117]. Overall, results were mixed. Empirical evidence of longitudinal studies lends support to both perspectives [23, 100, 108, 113]. However, due to analytical limitations (i.e., testing for reverse causal relationships), we can draw no definite conclusions [23, 108, 113]. Authors of the 25 studies examining clinician occupational well-being and patient safety reported 123 relationships altogether, of which 64 (52 %) were significant and in line with hypotheses. Of these significant relationships, 42 (66 %) described a positive association between negative indicators of both clinician occupational well-being and patient safety, whereas one (2 %) described a positive association between a positive indicator of clinician occupational well-being and patient safety. Sixteen (25 %) of relationships were negative, describing associations between negative indicators of clinician occupational well-being and positive indicators of patient safety or vice versa. Another five (7 %) associations were unexpected, such as an association between burnout dimension depersonalization and perceived patient safety or heart rate (an indicator of stress) and time spent on cardio-pulmonary resuscitation (an indicator of performance) [102, 109]. However, the latter can be explained by the physically strenuous nature of resuscitation, which is likely to cause an elevated heart rate. Effect sizes ranged from small (OR = 1.09) to large (OR = 8.3, see Table 4).

Relationships between teamwork, clinician occupational well-being and patient safety

Design & sample

Five of the 98 reviewed studies examined teamwork, clinician occupational well-being and patient safety (Table 5), three of which (60 %) sampled nurses only [119, 121, 122]. All studies were cross-sectional self-report studies, with one study (20 %) using risk-adjusted morbidity and mortality rates as objective patient safety indicators.

Measures

Three of the studies (60 %) used the nurse-physician-relations scale of the NWI [58] to assess teamwork, and (parts of) the MBI [59] or its emotional exhaustion subscale to measure well-being [119, 120, 122].

Findings

Studies examining relationships between teamwork, clinician occupational well-being and teamwork focused exclusively on interpersonal team processes. One study (20 %) proposed a model with the teamwork variable psychological safety [123] serving as a mediator between work environment and work engagement, commitment, and patient safety [120]. However, this mediation effect was statistically non-significant. Another study found a partial mediation between nursing work environment (including nurse-physician relations) and adverse events via burnout. Three studies (60 %) covered teamwork, clinician occupational well-being and patient safety amongst other aspects of the (nursing) work environment, but did not analyze the variables simultaneously, and reported mixed results [118, 121, 122]. Altogether, the five studies reported 33 associations between teamwork, clinician occupational well-being and patient safety, 21 (63 %) of which were significant and in line with authors’ assumptions. These 21 associations included five (23 %) negative associations between teamwork and a negative indicator of patient safety, teamwork and a negative indicator of clinician occupational well-being, and clinician occupational well-being and a negative indicator of patient safety. The 16 positive associations (76 %) included relationships between teamwork and patient safety, clinician occupational well-being and patient safety, and between negative indicators of clinician occupational well-being and negative indicators of patient safety.

Effect sizes ranged from small (r = 0.13) to medium (r = 0.39).

Integrative framework

Our aim was to develop a framework applicable to many different healthcare teams in hospital settings. We combined psychological models of team performance and work strain with the findings and theoretical assumptions of this review to formulate specific hypotheses regarding the relationships between teamwork, clinician occupational well-being and patient safety (Fig. 2).

Drawing from the job demands-resources model, we propose that teamwork can be a demand or a resource [29]. This model proposes two parallel processes that influence positive and negative aspects of occupational well-being, such as work engagement and burnout. Job demands deplete the individual’s energy and eventually decrease occupational well-being. Job resources, on the other hand, help employees attain goals, increase occupational well-being or reduce the strain caused by job demands [29].

A team in which actions are not well-coordinated (action team processes), goals are not communicated (transition team processes) and employee’s input to the team is not welcomed by fellow team members (interpersonal team processes) may be demanding for its members and thus directly decrease the team’s ability to provide safe patient care (Fig. 2, arrow C) [10, 11, 25, 120, 123–126]. Simultaneously, ineffective teamwork may lead to decreased clinician occupational well-being: according to the conservation of resources theory, decreased well-being can develop if there is an imbalance between resource investment and resource gain [22, 55, 107, 127]. Ineffective teamwork, as a lack of resource, can lead to a higher individual workload or emotional distress, thereby decreasing well-being [55, 56].

Poor well-being, in turn, may decrease clinicians’ ability to provide safe care (arrow D), because clinicians’ physical and mental resources are depleted [128], cognitive functioning may suffer and they may not be able to exhibit safe working behaviors [129, 130]. The effects of decreased clinician well-being might also be reflected in the team, because distressed team members may not be able to execute relevant team behaviors as effectively (arrow B) [54].

In contrast, if teamwork quality is high, teamwork may act as a resource supporting clinicians to provide safe patient care (e.g., developing shared team mental models, backup behaviors, high psychological safety encouraging clinicians to speak up; or transition, action, and interpersonal team processes; arrow C) [10, 65, 120, 123, 124, 131]. Effective teamwork helps to balance workload, prevent errors, and provide social support in a demanding work environment [126, 132], and may also lead to lower strain levels (arrow A), thereby indirectly supplying clinicians with resources needed for safe patient care (arrow D) [42, 55].

From the reviewed studies, it is not clear whether patient safety influences clinician occupational well-being or vice versa. Clinicians with reduced well-being may not be able to care for patients as safely and effectively due to depletion of resources [23]. Conversely, being involved in an adverse event may lead to guilt and emotional stress potentially compromising psychological well-being in the short- or long-term. [24] Given the existing evidence, we hypothesize that clinician occupational well-being and patient safety are tightly coupled: Tangible patient safety incidents are likely to cause short-term emotional distress [103] and chronic strain in clinicians [24]. Several authors have recognized that, after the patient, the clinician may become the second victim following an adverse event. They may be blamed for errors and have their clinical competence questioned. Sufficient support systems or policies to deal with the effects of error on second victims, such as feelings of anxiety, guilt or shame, do not always exist. [133–135] Chronic strain may also develop due to demanding working conditions which may decrease clinicians’ motivation and efficiency, which could lead to reduced patient safety in the long run (arrows D and E) [23].

Gaps and trends in current research

One aim of this study was to point out current gaps and recommendations to inform future studies addressing the relationships between teamwork, clinician occupational well-being and patient safety. These gaps and recommendations based on the reviewed studies are summarized in Fig. 2. We found that a holistic approach taking account of the complexity of teams in terms of team structure and different teamwork processes in healthcare organizations was missing, especially in survey studies: for instance, in addition to focusing on the individual professions within the team, the entire multi-professional team should be included (e.g., box A/B). Potential multiple team memberships, measures covering transition, action, and interpersonal teamwork processes, and adoption of a temporal rather than static perspective to account for the temporal instability of healthcare teams should be considered (boxes 1 and C in Fig. 2) [136–138]. For example, future studies might employ the team classification developed by Andreatta, which distinguishes between four different team types by classifying team membership and team roles as stable versus variable [139]. Moreover, correlating teamwork behaviors and patient safety indicators over an entire shift is not sufficient to gain an understanding of how they are linked. Instead, changes during the course of a shift or a specific task together with other influencing factors such as disturbances or interruptions need to be taken into account [74, 140].

Future approaches should consider reciprocal relationships between clinician occupational well-being and patient safety, and broaden the assessment of well-being to acute strain, physiological stress indicators or positive outcomes such as work engagement (box 2 in Fig. 2) [141].

With respect to patient safety, there is a clear need to consider how teamwork and well-being interact and impact upon objective safety indicators (boxes D/E and 3 in Fig. 2). This also includes ensuring independence of the objective indicators from other variables. For instance, measuring patient safety via subjective ratings or incident reports may not shed light on a unit’s safety, but rather measure clinicians’ willingness to report errors, which will be higher for clinicians working in a positive team climate [93, 142]. Yet, there seems to be a gap between the need for safety indicators that are feasible and a lack of theoretical discussion of what these indicators actually entail.

We identified several conceptual and methodological issues overarching all three concepts, which could be addressed through more focused study designs (bottom box in Fig. 2). These issues included missing or unclear theoretical foundations, definitions of key concepts, research goals and hypotheses, use of instruments with low validity (despite availability of valid instruments), incomplete description of analyses and reporting of results, mismatch of analyses and research question, and overgeneralization of results.

However, none of the studies suffered from all these drawbacks and many studies investigated the larger work environment so that the comprehensive measurement of teamwork, clinician occupational well-being and patient safety was not within the scope of these studies. Despite the gaps we identified, a large proportion of the reviewed studies were of high methodological quality, using triangulated data, validated instruments and statistical analyses of adequate complexity. Still, validity of results could be greatly improved by supporting pragmatic reasoning with sound theory to define key concepts and formulate clear, measurable research goals and hypotheses. In addition, it will be easier to perform analyses accounting for complexity of both the setting and data (i.e., structural equation or multilevel modeling, longitudinal studies, non-dichotomization of continuous variables).

Altogether, we found the most recent studies seem to address the issues mentioned above, i.e., by employing longitudinal research designs, sampling multi-professional teams or including objective measures of patient safety.

Discussion

This review provides an overview of the current state of research by scrutinizing relationships between teamwork, clinician occupational well-being and patient safety in hospital settings. Overall, ample evidence on associations between combinations of either two of these concepts exists. The volume and diversity of studies highlight the relevance of these concepts and provide a rich source of information for the design of future studies and interventions. Furthermore, the findings of the review in combination with psychological theories served as the foundation for the framework to explain interrelations between the concepts. The framework is intended to aid interpretation of findings, inconsistencies, and gaps in current research, to serve as a blueprint to designing future studies aiming to improve teamwork, clinician psychological well-being and patient safety.

Need to explore mechanisms behind relationships

Based on this review, the fact that some studies found no or only partial support for their hypotheses and reported small effect sizes is mainly due to the aforementioned conceptual and methodological issues, rather than non-existent relationships between concepts. These issues could be addressed by utilizing more stringent study designs. For instance, one may not find a relationship between general perceptions of teamwork and objective patient safety indicators. However, a targeted approach that draws from theory on aspects of teamwork and error types and uses validated measures may show that distorted shared mental models are related to inadequate nursing care.

Five of the 98 studies investigated relationships between all three concepts. These five, rather recent and very diverse studies did not provide a sufficient basis for drawing conclusive conclusions regarding the causal mechanisms between the concepts (e.g., because the entire team was not sampled, contradictory results were found across the studies), but demonstrate that the need for an integrative approach has been recognized.

The next step would be to design coherent studies based on strong theoretical foundations to uncover the mechanisms underlying the well-established relationships between teamwork, clinician occupational well-being and patient safety. Knowledge of these mechanisms may serve as a basis for designing interventions that integrate all three concepts.

Adopting an integrative approach

Teamwork is the predominant form of work organization in healthcare. Clinician occupational well-being and patient safety develop in a teamwork context and are dependent on each other. Consequently, clinician occupational well-being and patient safety should not be viewed as outcomes to be managed separately. They may even seem contradictory - additional policies to ensure patient safety may increase clinician workload and decrease well-being. Our findings suggest that they can be integrated into a comprehensive approach: Teamwork may serve as a means to improve both these central organizational outcomes. Also, team-based interventions may be utilized to benefit from the synergies between teamwork, clinician well-being and patient safety. To achieve this, it is essential to focus on multi-professional teamwork and include nurses, physicians and other healthcare professionals. For example, differences in perceptions of teamwork quality by different professions [143, 144] and different approaches to team tasks may result in interpersonal friction [145] and decreased team effectiveness [5, 12]. Aside from proposing general mechanisms between teamwork, clinician well-being and patient safety, the review and framework provide an overview of the specific aspects (i.e., chronic and acute strain, interpersonal, action and transition team processes) that may help target particular problems.

Outlook

The findings of this review have implications for researchers, and the proposed framework can help to address them in an integrative manner (Fig. 2).

  1. 1.

    Comprehensive approach to teamwork, well-being and patient safety

    There is a clear need to investigate teamwork, clinician occupational well-being and patient safety simultaneously in order to evaluate the complex interrelations between these constructs. Interdisciplinary exchange (e.g., medical, nursing, psychological) during study design would help harvest the full potential of studying these associations. Understanding these relationships may help develop interventions aimed at improving all three concepts.

  2. 2.

    Exploration of causal relationships

    Little is known about the causal associations between teamwork, clinician occupational well-being and patient safety, and their changes over time. Theoretically informed longitudinal studies and practical interventions will shed more light on this issue. Designing and implementing team-based interventions may investigate the simultaneous effect of improved teamwork on clinician occupational well-being and patient safety.

  3. 3.

    Considering the entire healthcare team

    Inter-professional tasks are inherent in healthcare. Thus, only considering nurses and physicians (and other healthcare professionals as appropriate) will provide a comprehensive picture of the complex associations between teamwork, clinician occupational well-being and patient safety. In addition, the complexity of teams in healthcare (i.e., temporal instability) needs to be taken in to account [136–139]. In practice, consideration of the entire healthcare team is likely to increase the impact of team-based interventions on clinician and patient outcomes [146].

Limitations

Although we employed a rigorous search strategy, we may have missed relevant studies. For instance, the lack of consensus between different research approaches concerning terminology for key concepts may have resulted in ambiguous database indexing. However, we compensated for this limitation by including a thorough search of reviews and reference lists. Second, qualitative and interventional studies might have provided additional insights, but – with one exception [88] – were excluded because they did not examine statistical relationships between the concepts that were the focus of this review. Third, study selection, data extraction and rating of study quality were naturally influenced by authors’ reporting style. Nevertheless, the detailed review procedure including structured quality rating proved useful in exploring strengths and weaknesses of the selected studies and thus provided a solid foundation for framework development. Fourth, since disagreements between raters regarding study quality were resolved by consensus discussion, interrater reliability was not calculated. Fifth, we limited this review to acute care hospital contexts, thus, we cannot be sure that our findings are applicable to other (healthcare) settings. However, while other healthcare settings, such as primary care, may differ in terms of team structure or risks to patient safety, we are nevertheless convinced that the overarching issues of this review mentioned in the section above are worth addressing in other contexts. Lastly, as with all reviews, there is always a possibility of publication bias, because non-significant results are often not published.

Conclusion

We identified substantial relationships between combinations of two of the three concepts teamwork, well-being and patient safety, indicating that all three might influence each other. The proposed framework is based on solid research and provides a foundation for overcoming current research gaps and inconsistencies by hypothesizing causal mechanisms between the concepts and investigating relationships between all three concepts simultaneously. In the most recent studies, we identified a trend to address these gaps. Following the three main recommendations (i.e., comprehensive approach to teamwork, clinician well-being and patient safety; consideration of the entire healthcare team and exploration of causal relationships) will generate research that substantially explores and supports the hypothesized links between teamwork, clinician occupational well-being and patient safety. An integrative perspective of the synergies between teamwork, well-being and patient safety will inform future research, and aims to benefit clinicians and patients alike.

Abbreviations

AHRQ, agency for healthcare research and quality; AIC, akaike information criterion; ANOVA, analysis of variance; ANTS, anesthetist’s non-technical skills; ATLS, advanced trauma life support; ATOM, anti-air teamwork observation measure; AW, annalena welp; BIC, bayesian information criterion; CBI, Copenhagen burnout inventory; CD-RISC, Connor-Davidson resilience scale; CFI, comparative fit index; CI, confidence interval; CLABSI, central line associated bloodstream infections; CoMeT–E, coordination system for medical teams - emergency; CPR, Cardio-pulmonary resuscitation; CQS, chirurgisches qualitätssiegel survey; GEE, generalized estimating equations; GHQ-12, general health questionnaire; GLM, generalized linear mixed model; icu, intensive care unit; JV, Johanna Vogt (see Acknowledgments); LQWQ-N, Leiden quality of work questionnaire for nurses; M, mean; MBI, Maslach burnout inventory; MD, Mariel Dardel; MeSH, Medical subject heading; MISSCARE, missed nursing care; NNFI, non-normed fit index; NOTECHS, surgical non-technical skills system (observational instrument); NOTSS, non-technical skills for surgeons; NRP, neonatal resuscitation program; NS, not significant; NWI, nursing work index; NWI-R, nursing work index revised; OCHRA, observation clinical human reliability assessment; OLBI, Oldenburg burnout inventory; OR, odds ratio; OTAS, observational teamwork assessment for surgery; RMSEA, root mean square error of approximation; RR, risk ratio; SAQ, safety attitudes questionnaire; SD, standard deviation; SEM, structural equation modeling; SF-36, short form health survey; SOS, safety organizing scale; SS, Sven Schmutz (see Acknowledgments); SSI, standard shiftwork index; STAI, state-trait anxiety inventory; TCI, team climate inventory; TCT, team check-up tool; TEAM, team emergency assessment measure; TeamSTEPPS, team strategies and tools to enhance performance and patient safety; TLI, Tucker Lewis index; TM, Tanja Manser; TPQ, teamwork perceptions questionnaire; UBOS, Utrecht burnout scale; UWES, Utrecht work engagement scale