Silos and Social Identity: The Social Identity Approach as a Framework for Understanding and Overcoming Divisions in Health Care

Abstract

Context

One of health care's foremost challenges is the achievement of integration and collaboration among the groups providing care. Yet this fundamentally group-related issue is typically discussed in terms of interpersonal relations or operational issues, not group processes.

Methods

We conducted a systematic search for literature offering a group-based analysis and examined it through the lens of the social identity approach (SIA). Founded in the insight that group memberships form an important part of the self-concept, the SIA encompasses five dimensions: social identity, social structure, identity content, strength of identification, and context.

Findings

Our search yielded 348 reports, 114 of which cited social identity. However, SIA-citing reports varied in both compatibility with the SIA's metatheoretical paradigm and applied relevance to health care; conversely, some non-SIA-citers offered SIA-congruent analyses. We analyzed the various combinations and interpretations of the five SIA dimensions, identifying ten major conceptual currents. Examining these in the light of the SIA yielded a cohesive, multifaceted picture of (inter)group relations in health care.

Conclusions

The SIA offers a coherent framework for integrating a diverse, far-flung literature on health care groups. Further research should take advantage of the full depth and complexity of the approach, remain sensitive to the unique features of the health care context, and devote particular attention to identity mobilization and context change as key drivers of system transformation. Our article concludes with a set of “guiding questions” to help health care leaders recognize the group dimension of organizational problems, identify mechanisms for change, and move forward by working with and through social identities, not against them.

The achievement of integration and collaboration among different providers of care is one of the foremost challenges facing today's health care system. The theme of overcoming disciplinary, sectoral, and institutional “silos” is echoed in almost every area of health-services research, from primary-care reform to patient safety, chronic-disease management to cost containment (e.g., Clancy 2006; Mann 2005; McDonald et al. 2007). The urgency of the problem has increased in the current policy context, where it can be argued that the success of health reform stands or falls on the ability of delivery system reform to replace fragmentation and waste with coordination and cost-effectiveness. Strategies for achieving this—from the micro level of interprofessional teams and primary care medical homes, to the macro level of accountable care organizations—are fundamentally about collaboration across silos. Indeed, while serving as administrator of the Centers for Medicare and Medicaid Services, Donald Berwick placed the elimination of silos at the top of a list of values needed to improve American health care (Schaeffer 2011). Interest in teamwork and integration, and in the organizational changes that may foster them, is equally strong internationally (Finn, Currie, and Martin 2010; Mann 2005). Yet earlier reviews, albeit extensive (e.g., Oandasan et al. 2006), have not offered a theory of what makes cross-silo relationships flourish or decay. Part of the problem may be that although such relationships clearly constitute an intergroup issue, the theme of group dynamics is frequently ignored. Most of the literature does not focus on the group level, but on the individual and interpersonal (e.g., what traits, skills, and processes facilitate teamwork) or the operational and systemic (e.g., what structures and resources promote integration). To amend this gap, we seek out literature that addresses the neglected group level and examine this body of literature through the lens of a general theory of group processes: the social identity approach (SIA), which comprises social identity theory (SIT; see Tajfel and Turner 1979) and its extension, self-categorization theory (Turner et al. 1987).

There is increasing recognition of the value of theory in guiding health-services and policy research (Grol et al. 2007). However, researchers have also cautioned that a proliferation of theoretical constructs promotes confusion, not understanding (Michie et al. 2005). Whereas one popular approach is to devise frameworks by combining theories, thus risking a fragmented analysis that lacks conceptual coherence, this article explores the integrative potential of a single theory that is sufficiently broad and multifaceted for the task. In doing so, it seeks not to diminish the value of other theoretical approaches, but to highlight the SIA's capacity to synthesize findings and insights from disparate traditions.

The SIA arose from the recognition that group memberships form an important part of the individual's self-concept. Focusing on the nexus between the individual and the group, this approach explores how seeing ourselves and others in terms of social categories affects our perceptions, attitudes, and behavior. The SIA encompasses five key dimensions:

1. Social identity. People categorize themselves and others as members of an in-group (“us”) or an out-group (“them”), because (a) being part of a positively valued group enhances self-esteem and (b) categorization offers a meaningful way to organize the social world (Tajfel and Turner 1979; Turner et al. 1987). People also compare their group with others, often striving to maximize the positive distinctiveness of their own. When social identity is salient, people focus more on the shared attributes uniting group members than on the personal characteristics differentiating them. These perceptions engender group behavior, which, depending on the context, can be either positive (e.g., cooperation) or negative (e.g., discrimination).

2. Social structure. Structural relations among groups, notably intergroup differences in status and power (especially if seen as unstable and illegitimate), can make the difference between conflict and peaceful coexistence (Tajfel and Turner 1979). Groups can alter structural relations when they take collective action on the basis of their social identity.

3. Identity content. The identities we value are defined by specific norms and attributes. Members internalize group norms and use them to guide behavior (including intergroup behavior) and to evaluate other members, including leaders (Haslam 2004; Jetten, Spears, and Manstead 1996). Members also mobilize in support of, or to combat threats to, these shared norms (Ellemers, Spears, and Branscombe 1999).

4. Strength of identification. Individuals typically belong to many groups, but tend to identify strongly with some and weakly with others. Group identification amplifies both positive and negative social behavior (Ellemers, Spears, and Doosje 2002); high identifiers are more likely to take action in the service of group goals and to fight to protect the group from perceived threats to its status, distinctiveness, or norms.

5. Context. The relative prominence of an individual's multiple identities is not fixed but changes with the social context (Turner et al. 1987). This fluidity of social identity—well demonstrated in empirical studies (e.g., Levine and Reicher 1996)—provides hope that by modifying the context, we can change group behavior.

The SIA has three major strengths: (1) as a general theory of group processes and intergroup relations, it enables the transfer of insights across levels (e.g., interprofessional to interorganizational) and domains (e.g., non–health care to health care); (2) as a multidimensional theory, it allows researchers and decision makers to consider multiple influences on (inter)group dynamics; and (3) as a well-established theory that has continued to develop over three decades, it has a strong empirical foundation in both basic and applied research (Ellemers, Spears, and Doosje 2002). Haslam (2004) extensively reviewed the SIA's contribution to organizational psychology, presenting evidence that adumbrates social identity's important role in myriad phenomena, including leadership, employee motivation and commitment, communication, group performance, intergroup negotiation, group decision making, and response to stress. Among the relevant findings: leaders are more likely to win support when they belong to a valued in-group, advance its interests, and endorse its norms—indeed, members tend to prefer leaders whose characteristics differentiate them from an out-group, even when such characteristics are objectively undesirable (Haslam, Reicher, and Platow 2011); organizational identification reliably outperforms job satisfaction as a predictor of an employee's intent to stay (Randsley de Moura et al. 2009); and shared social identity underpins effective communication—not only does information flow less readily across salient intergroup boundaries (Agama 1997), but we are more receptive to influence from in-group members, as we assume that they see the world as we do (Abrams et al. 1990).

We investigated both what the SIA has already contributed and what it can contribute to an understanding of intra- and intergroup relationships among health care providers. Our review questions were (1) To what extent, and how accurately, has the SIA been applied? and (2) How does the overall literature on health care groups (including non-SIA-citing sources) address the five SIA dimensions? The first question required a critical analysis of sources citing social identity; the second demanded a broader exploration of the perspectives from which health care groups have been examined. We describe our approach as a “critical scoping review”—a combination of scoping review (Arksey and O’Malley 2005) and theory-based synthesis.

Method

Although our review was protocol-driven, it allowed for evolution as we became more familiar with the literature. We undertook a systematic literature search for relevant empirical and theoretical sources, using electronic databases (PubMed, Cinahl, Scopus, PsycInfo, Sociological Abstracts, Dissertation Abstracts) and Google Scholar, and also hand-searching relevant journals. We identified additional reports through our own collections and reference lists, using the context in which a reference was cited to assess its probable relevance. Any item we thought might discuss social identity in health care was retrieved; those discussing concepts other than social identity were sampled up to the point of theoretical saturation. The search process captured literature published up to the end of 2010. Two reviewers independently screened each abstract and subsequently assessed each report for inclusion; disagreements were resolved by a tie-breaking vote for abstracts and by consensus for full reports. (Further methodological details and a full list of included and excluded reports are available from the authors upon request.)

To determine how the five SIA dimensions were addressed, we undertook an inductive, iterative analysis of the literature, whereby we derived what we have termed conceptual currents. We did not rely on the authors’ espoused theories as a basis for categorization, as we discovered commonalities across theoretical labels and diversity within them; moreover, many articles were atheoretical. Instead, drawing on the approach of metanarrative synthesis (Greenhalgh et al. 2005), we interrogated each article for its underlying assumptions. We began with simple content analysis, coding sources for the presence of each of the five SIA dimensions; however, since merely documenting a concept's presence provided no information about how it was addressed, we used this as a springboard for thematic analysis. This process involved constant comparison of the data against one another and an emerging list of thematic categories. One reviewer (SK) used the preliminary codes, along with qualitative notes recorded for each report, to sort articles according to the combination and interpretation of SIA dimensions that they reflected; the analysis was then opened to the other team members.

Results

We reviewed 348 reports representing 335 unique sources (166 qualitative, 77 quantitative, and 17 mixed-methods studies; 71 essays, commentaries, or theoretical papers; and 4 reviews). The literature revealed a strong focus on interprofessional and, secondarily, intraprofessional issues (N = 151 and 84, respectively), with some examination of management-staff relations (30), organizational change (29), and other organizational issues (34). As we confined our search to English-language reports, it is not surprising that most of the sources came from English-speaking countries such as the United States (90), United Kingdom (88), Australia (29), and Canada (24).

SIT was referred to in 114 reports, representing 105 unique studies/essays (about a fifth of which addressed it only briefly or used it implicitly by discussing organizational identification). These did not constitute a unified literature; pockets of articles cited one another, while many articles cited no others within the group. Furthermore, some sources were inconsistent with the theory's fundamental assumptions. Not all presented social identity as a group-related issue, instead treating it as an individual-difference variable or an input to interpersonal relationships (e.g., Fuller et al. 2006; Nauta and von Grumbkow 2001). Some authors defined key SIT terms incorrectly (Hallier and Forbes 2005) or evinced awareness of only the earliest or simplest formulations of the theory, not its breadth and depth. For example, the efforts by Mitchell and colleagues (2010) to “integrate” SIT with an individualistic information-processing approach suggested a lack of familiarity with the extensive social identity literature on social cognition and social influence (e.g., Abrams et al. 1990). Furthermore, some of the papers that applied SIT appropriately did not focus on health care, using it merely as a setting in which to test general hypotheses about intergroup or organizational behavior (Bartels et al. 2010; Oaker and Brown 1986). Thus, despite the apparently high number of SIT citations, there was no evidence of a social identity literature on health care groups.

Content analysis suggested that sources featured an average of 2.5 of the 5 SIA dimensions. Whether or not a source cited SIT was not a robust indicator of whether it addressed these dimensions in a manner consistent with the SIA; notably, we found several highly SIA-congruent analyses among non-citers. For this reason, the theory-based synthesis does not privilege SIT-citing sources, but discusses all sources together. In the next section, we group the conceptual currents (in italics) under the SIA dimension that they most strongly reflect. To conserve space, we use citations for illustrative purposes, not exhaustively.

Social Identity

The most basic insight of a group-based approach is that analyses of conflict or cooperation in health care must consider group, not merely interpersonal, dynamics. This insight was powerfully demonstrated in an exposé of virulent intergroup conflict among hospital departments (Hewett et al. 2009). Drawing on SIT, the study revealed that patient charts—supposedly repositories of objective information—were rife with examples of intergroup competition and in-group enhancement. The study also showed how interspecialty competition to “own” (or disown) patients could threaten safe care and may have led to a patient's death. The authors argued that interpersonal-skills training would be an inadequate remedy for physicians’ dysfunctional communication, since the problem resulted not from a lack of skill but from the active expression of group identities.

Such recognition of the importance of groups is echoed in papers that variously invoke SIT (Shute 1997), role theory (Booth and Hewison 2002), evolutionary psychology (Braithwaite, Iedema, and Jorm 2007), and other concepts (Ferlie et al. 2005). However, this body of literature is primarily descriptive. To identify potential solutions, therefore, we must consider social identity in conjunction with other SIA themes.

Social Structure

The health care landscape comprises groups of unequal power and status. Without an understanding of this issue, naive calls for “teamwork” may actually reinforce professional divisions and hierarchies. For example, Finn (2008) documented how various providers appealed to the concept of teamwork to advance their profession-specific norms and interests: to nurses, teamwork meant more equal and respectful relations; to surgeons, that others efficiently followed their orders.

Unequal Partners

A long tradition of literature focuses on structural inequality between doctors and nurses. Many authors stress the gendered nature of the doctor-nurse dynamic, tracing its origins to the subordination of women within the sexual division of labor (Campbell-Heider and Pollock 1987). Others put power first and gender second, emphasizing how the specific unequal relationship between doctors and nurses produces certain patterns of interaction. For example, Tellis-Nayak and Tellis-Nayak (1984) elucidated how structural inequalities are reinforced by social rituals (concerning the differential use of space, time, language, and body language), arguing that any attempt to reduce the power imbalance must address both structural and symbolic factors. Both gender- and power-related role expectations may constrain health care professionals to play the “doctor-nurse game,” in which the nurse must offer any suggestion indirectly so that it seems to be the doctor's idea (Stein 1967). Many structural analyses also touched on identity content by noting how the values, discourses, sources of knowledge, and types of labor associated with high-status groups (men/physicians) are privileged over those of low-status groups (women/nurses) (Wicks 1998).

Roberts (1983) contended that intergroup inequality also taints nurses’ intragroup relations, giving rise to “oppressed group behavior” in which nurses accept the physician-dominated structure as legitimate and inevitable. She explained “horizontal violence” within nursing in terms of nurses’ efforts to repudiate a devalued group identity and align themselves instead with the oppressing group.

While offering a provocative examination of health care hierarchies, this literature has attracted three major criticisms: (1) its perspective tends to be static and deterministic, missing the context-dependent, strategic, and, above all, variable nature of doctor-nurse interactions, and sometimes denying nurses’ agency; (2) its accounts are frequently partisan, reflecting the unquestioned assumption that increased power for a certain group is an absolute good; and (3) its overwhelming focus on doctor-nurse relations obscures other power relations in health care. For example, a UK study (Johns 1992) explored how the division of nurses into primary and associate roles created new power dynamics, including what might be described as a “nurse-nurse game” (i.e., associate nurses confining themselves to indirect suggestions to maintain a facade of harmony).

From an SIA perspective, all three limitations spring from the fact that “unequal partners” analyses reify observed phenomena as immutable social facts, not recognizing them as examples of generalized group processes. According to SIT (Tajfel and Turner 1979), the behavior of unequal-status groups can be understood in terms of two basic responses: individual mobility (attempts to leave one group and enter the other) and social change (attempts to reduce, reverse, or increase the status difference between the groups). Social change strategies include social creativity (creation of ideologies that affirm the in-group's worth) and social competition (competition with the out-group for power or resources). SIT does not treat these strategies as unique to oppressed groups or as dependent on a particular group psychology, but rather as a function of the in-group's structural position relative to the out-group, the features of the structural relationship, and the strength of the member's group identification. The SIA's generality and objectivity allow it to encompass both entrenched intergroup inequalities and scenarios in which power relations are more complex, unexpected, and context-dependent. (For a SIT-informed account of doctor-nurse relations, see Chattopadhyay, Finn, and Ashkanasy 2010.)

The professional strategies tradition, grounded in the sociology of professions (Freidson 1970), explores occupational groups’ struggles for territory and control. Although medical dominance is a frequent focus, Abbott (1988) underscored that the important phenomenon is not the empirical fact of a certain group's dominance (which can change) but the ubiquity of intergroup competition—which occurs both interprofessionally and among intraprofessional subgroups (Currie, Finn, and Martin 2009). Unlike the “unequal partners” literature, this stream tends to regard all groups as self-interested and their ideologies as self-serving. It also offers a less static view, illustrating how the intergroup landscape may change when a strategy succeeds or backfires. In a study of health care reorganization, Daykin and Clarke (2000) illuminated how nurses’ “project” of professionalization was impeded by contradictions between the strategies used to gain territory from medicine and to protect their own. To affirm their status vis-à-vis medicine, nurses claimed a distinct body of knowledge grounded in caring, yet undermined this claim by devaluing the caring labor of health care assistants. The authors argued that, far from enhancing nurses’ professional status, such exclusionary tactics reduced their ability to resist the most pertinent threat: management's Fordist practices of routinization and de-skilling.

The “professional strategies” literature often highlights the role of context; for example, oncologists who worked in different settings used different means to promote their dominance over practitioners of complementary and alternative medicine (CAM) (Broom and Tovey 2007). Hospital physicians used the discourse of science to discredit CAM, proclaiming the superiority of medicine's “scientific mind-set” even in areas where no medical evidence existed. Hospice physicians could not discredit CAM, given its fit with their institution's holistic philosophy; instead, they found subtle ways to subsume CAM within a biomedical paradigm, tightly controlling CAM practice to avoid any challenge to (bio)medical dominance.

With its rich description of both the material and rhetorical strategies that groups use to secure and enhance their professional status, the “professional strategies” literature depicts social competition and social creativity in action. Echoing SIT, it shows how members strategically compare their own group with others, choosing dimensions (May and Fleming 1997), definitions (Norris 2001), and comparison groups (Fournier 2002) that maximize the in-group's positive distinctiveness. It differs from SIT, however, in focusing on groups’ instrumental motivation to gain power and autonomy, seldom recognizing the psychological motivation to maintain a positive social identity. The assumption that groups care only about tangible, not symbolic, benefits can lead to inaccurate predictions (e.g., that health care groups would soon abandon professional ideologies, competing instead over who could meet market needs most cost-effectively) (Light 1988). Skepticism about groups’ self-representations can be taken too far, to the point of dismissing all affirmations of group identity as strategic performances. The SIA recognizes that such performances are not just strategic; they often reflect highly valued identity content.

Identity Content

Group identities are not arbitrary but are defined by certain content: norms, values, and worldviews that are meaningful and important to members. A large literature frames this topic in terms of culture and cultural differences, delineating how professional groups differ in everything from values and attitudes (Degeling, Kennedy, and Hill 2001) to myths and rituals (Dombeck 1997). But bald descriptions of identity content do not elucidate group processes, and inspire few solutions apart from vague calls for cultural sensitivity. Accounts of professional culture may also miss intraprofessional diversity; even explorations of professional subcultures (Leininger 1994) often stop short of asking what the different variants of identity mean. More SIA-consistent studies may explain variability in terms of subgroups with different experiences or structural positions contesting (advancing their own versions of) identity content (e.g., Pratt and Rafaeli 1997).

In contrast to the “cultural differences” discourse, which portrays identity content as static, the SIA takes a dynamic view, examining how different representations of group identity are constructed and mobilized in order to achieve group goals. Several streams of literature reflect the theme of identity content in action. Research on professional socialization considers, among other things, how educational and practice environments encourage or suppress the enactment of a patient-centered identity. Patient-centeredness can be an integral part of physician identity, yet its expression may be actively discouraged through both the formal and informal curricula of medical school (Apker and Eggly 2004). In contrast, nurses are strongly socialized into patient-centeredness—until they enter the working world, where demands for efficiency may imperil this core value, resulting in demoralization and disengagement (Limoges 2007). O’Donohue and Nelson (2007) stressed that professionals’ psychological contract with an organization depends not merely on individual rewards but on the organization's respect for the values central to their collective identity.

Health care professionals are known to resist threats to their identity content (Fiol and O’Connor 2006); however, interventions perceived as identity threatening in one context may be perceived as benign, even identity affirming, in another. Physicians may reject management-imposed system redesign as a threat to the doctor-patient relationship, yet actively support the same redesign when it is seen as contributing to the achievement of a patient-centered medical home (Kreindler 2008). Furthermore, health care professionals invoke different identity representations to combat different identity threats. When the threat involves alternative practitioners’ encroachment on medical territory, medicine is a science (Broom and Tovey 2007)—yet, when it involves managerial pressure to comply with clinical practice guidelines, medicine is an art (McDonald, Waring, and Harrison 2006). At first glance, the finding that such varied, even contradictory, discourses are used to resist change seems to imply that change is impossible. The flexible (re)construction of identity, however, may itself be a key to change. In one study, GPs who became change leaders constructed a hybrid identity that affirmed the primacy of their physician identity while incorporating management skills (redefined as a trivial subset of skills that most physicians possess) (Hotho 2008). They found this new identity more attractive than that of management (still seen as a low-status out-group with objectionable values)—or of rank-and-file GPs (seen as routine oriented and averse to change).

Organizational change efforts have been observed to founder when administrators either ignore social identity (seeking to shape employees’ behavior through individual rewards and sanctions alone) or assume that staff can be rallied behind an imposed organizational identity (Charles-Jones, Latimer, and May 2003; McDonald 2004). Success is more likely when leaders guide change as it grows from the real values of existing groups (Brooks 1996; O’Brien et al. 2004). The fullest elaboration of this strategy occurs within the SIA literature, with the ASPIRe (“Actualizing Social and Personal Identity Resources”) model of organizational development; the ASPIRe process engages employees in building a “mosaic” identity that recognizes both common goals and the distinct contributions of valued subgroups (Haslam, Eggins, and Reynolds 2003). It has also been expressed in terms of “social movement thinking” (Bevan 2008), organic development of a “network community” (Bate 2000), and other concepts. One hospital used the norms of physician culture (concrete, expert-led, decision-oriented discussions) to encourage physician engagement in restructuring (O’Hare and Kudrle 2007). Other studies of positive physician-manager relationships have emphasized the development of a shared identity, facilitated by shared decision making grounded in common values (in particular, service and excellence; see Graham and Steele 2001; Kirkpatrick et al. 2007). Just as social identities can be mobilized to resist change, they also can be mobilized to cope with change, or to achieve it. As the social identity literature has revealed, mobilization is not simply a matter of invoking a ready-made image of group identity, but involves “crafting a sense of us” (Haslam, Reicher, and Platow, 2011) that supports a desired change. Goodrick and Reay (2010) showed how nursing textbooks, using the idea of continuity with the past to foster the reconstruction of nursing identity, highlighted only those aspects of the past consistent with the new vision.

Strength of Identification

Attempts to promote identification with a certain group, to restructure groups, or to mobilize identities must reckon with members’ existing patterns of group identification.

Identities at Work

One stream of literature is concerned primarily with health care employees qua employees, examining organizational identification, its predictors (e.g., good communication up and down the hierarchy, value congruence between organization and employee, the organization's prestige, and the employee's sense of being respected), and its consequences (e.g., cooperation, organizational citizenship) (Dukerich, Golden, and Shortell 2002). Nearly all these studies cite SIT; problematically, though, many are characterized by a management-centric assumption that organizations should, or even can, manipulate employees’ social identities (e.g., Han and Harms 2010). Yet even within this literature, it becomes clear that employee commitment is not so easily manipulable. For example, while participative decision making is among the strongest predictors of organizational identification, opportunities to “participate” are unlikely to foster organizational identification unless employees see them as genuine and relevant (Joensson 2008; Tangirala and Ramanujam 2008).

A second stream of literature recognizes that most health care providers are strongly identified with their profession. Strong professional and organizational identification are not incompatible; indeed, the two are often positively correlated (Bartels et al. 2010). However, most professionals identify more strongly with their profession than with their organization, which typically offers a less distinctive identity and a shorter period of socialization (Callan et al. 2007). Moreover, the combination of high professional and low organizational identification is linked to behavior that, from the organization's perspective, is undesirable. Physicians exhibiting this combination of attitudes have been found to resist influence from management, repay perceived organizational support with reduced compliance, and retaliate against perceived violation of their “psychological contract” with the organization (Hekman et al. 2009a, 2009b).

A limitation of the “identities at work” tradition is that most of the studies, concentrating as they do on drawing general conclusions about organizational psychology, are not firmly grounded in the health care context. For instance, some studies have drawn inferences about “high-status” staff from analyses that excluded physicians or combined them with senior nurses (Callan et al. 2007; O’Brien et al. 2004).

Identity Surprises

Several qualitative studies have amplified the point that group membership does not equate with group identification. Cott (1998) found that higher-status professional groups identified with an “egalitarian” multidisciplinary team. However, direct care nurses and health care assistants, who continued to be excluded from decision making, did not share in this team identity; rather, they felt alienated and unwilling to cooperate with the team's directives.

Physicians are unlikely to identify with an organization whose management they perceive as a devalued out-group—in particular, one that threatens their autonomy in the service of values abhorrent to their profession (Fiol, Pratt, and O’Connor 2009; Hekman et al. 2009b). Moreover, Hoff (1999) found that creating physician-managers failed to bridge the divide between these two groups; on the contrary, the physician-manager population itself became divided. Those who saw themselves primarily as managers embraced their new identity, complete with management jargon, and even defended organizational policies that negatively affected physicians. Those who continued to see themselves primarily as physicians asserted this identity by resisting and sabotaging the activities of the first group, whom they viewed as traitors. These findings invite the application of such SIA concepts as categorization threat (the threat of one's being viewed as a member of a devalued group; see Ellemers, Spears, and Doosje 2002) and the black sheep effect (in which an in-group member who flouts group norms attracts more censure than does an out-group member; see Marques, Yzerbyt, and Leyens 1988). As we have seen, physician-managers can develop hybrid identities (Hotho 2008), but this is not inevitable. Doolin (2002) found that physicians revised their identities to incorporate a new management role only when they perceived the latter as congruent with their physician identity.

Context

Context refers to the external conditions—such as organizational structures, working practices, and physical features of the work environment—that support a particular system of group relations. The SIA holds that changing the context can change the way people view and relate to one another, ultimately altering existing patterns of group interaction.

Adventures in Organizational Development

A handful of articles described successful organizational-development initiatives that put intergroup issues front and center (van de Vliert 1995). Staff have been invited to participate in collective reflection on group dynamics, and the insights derived from this process used to develop or implement new structures and working practices (context changes) that support more positive group relations (Bate 2000; Bleakley et al. 2006).

The creation of multidisciplinary teams is itself a context change. Although the literature abounds with warnings that the mere establishment of a team does not guarantee true teamwork, some studies with a longer follow-up have reached more optimistic conclusions. A British study (Hudson 2002) found that effective teamwork between health and social care providers had begun to develop in practices where providers were co-located and informal working patterns were shifting. A longitudinal study of an interdisciplinary health team traced members’ gradual progression from defensiveness and stereotyping to a shared team culture in which roles were less differentiated (Farrell, Schmitt, and Heinemann 2001). Such findings suggest that what are seen as intractable problems with multidisciplinary teams may actually just be growing pains. There is some evidence, however, that role revision and teamwork are more likely to become realities when social identity is taken into account. A study of an unsuccessful attempt to introduce an enhanced nursing role suggested that the change process failed to engage with the entrenched social identity dynamics associated with a rigid professional hierarchy (Currie, Finn, and Martin 2010). Despite its superficial adoption, the new role was assimilated into the existing social structure and produced little real change. Furthermore, a single organizational intervention may not suffice; context change may entail addressing broader environmental factors that impede collaboration. For example, such contextual factors as organizational integration, resource availability, and various facilitators of long-term working relationships strongly influenced whether the introduction of teams reproduced or transformed professional hierarchies (Finn, Currie, and Martin 2010).

Contexts for Interprofessional Contact

Interprofessional education (IPE) is the only area in which SIA-influenced interventions have been tested through controlled experiments. However, the interventions studied have not mined the depth of the social identity approach. This may be because their foundation is not actually the SIA but the more interpersonally oriented contact hypothesis, which prescribes intergroup contact as a remedy for prejudice (Allport 1954).

Many IPE articles noted that health care professionals’ propensity to categorize themselves in terms of a uniprofessional identity can impede interprofessional collaboration (Lidskog, Lofmark, and Ahlstrom 2008). The literature also reflects an awareness of the various options for reshaping social categorizations (Hean and Dickinson 2005). Two possible approaches are decategorization (encouraging people to see themselves and each other as individuals, not group members) and recategorization (emphasizing a common in-group identity, not subgroup identities). However, both these approaches can provoke identity threat and “backlash” from highly identified health care professionals. Most of the contemporary SIA-influenced literature advocates a third option: acknowledging and valuing both a common superordinate identity and distinct subgroup identities. In contrast to the bottom-up approach of the ASPIRe model (Haslam, Eggins, and Reynolds 2003), the strategies reflected in the IPE literature are top-down: They involve controlling participants’ experiences in order to influence their attitudes in specific, planned ways. Whereas the bottom-up approach makes identity content the driver of change, the top-down approach tends to treat it as incidental.

Proponents of contact-based interventions have maintained that structural issues are beyond their sphere of influence (Carpenter and Hewstone 1996). Unfortunately, the enterprise of creating favorable interprofessional contexts within an inegalitarian social structure is fraught with challenges and contradictions. First, it is not always possible to create equal-status contact between unequal-status groups—as illustrated by a shared education program in which the location and curriculum had been arranged to meet the needs of medical students, thereby marginalizing dental students (Ajjawi et al. 2009). Second, even if an equal-status “bubble” can be created, attitudes developed under such artificial conditions may evaporate when participants return to the real world. This difficulty has led contact theorists to develop increasingly elaborate techniques to encourage the generalization of newly acquired attitudes. This individualistic approach, which locates the problem in personal attitudes and stereotypes, is incompatible with a group-based perspective. The latter, because it views stereotypes as a symptom of a system of group relations that entrenches intergroup conflict, sees context as the necessary target of intervention. Contact theorists’ preoccupation with decontextualized stereotypes has created ironic situations; for example, researchers downplayed an IPE program's successes with team function or client outcomes, lamenting instead that pencil-and-paper tests detected no change in stereotypes (Barnes, Carpenter, and Dickinson 2000; Carpenter et al. 2006).

The “contexts for contact” approach's attentiveness to social categorization represents an advance over approaches that ignore groups or treat them as a nuisance. But without a substantial focus on identity content or social structure, this approach misses a lot of what is social about social identity, and falls short of harnessing the power of social identities to stimulate change.

Contexts for Remedying Intractable Conflict

Fiol, Pratt, and O’Connor (2009) have advanced a sequential, SIA-based model for resolving intractable identity(-based) conflict (IIC) (e.g., between physicians and hospital administrators). In such cases, each in-group draws part of its identity from negating the out-group, making collaboration impossible. The IIC model holds that de-escalation of such conflict must progress through the following stages: (1) readiness (to come to the table), (2) mindfulness (openness to other ways of conceptualizing the situation), (3) positive in-group distinctiveness (ensuring a secure in-group identity without the need to negate the out-group), (4) simultaneous intergroup differentiation and unity (cooperation around specific objectives while maintaining separate, distinct groups), and (5) integrative goals and structures. Organizations may take steps to help a fractious relationship progress from one stage to the next. However, attempts to achieve a higher stage before lower ones have been completed (e.g., making appeals to unity while each group still views the other as a threat) are likely to backfire. The sequence of stages 1 through 4 was borne out in a Canadian study of the relationship between physicians and regional health authorities; the parties might have ultimately progressed to stage 5, but the government dissolved the regional system (Reay and Hinings 2009). In an Australian study, allied health professionals appeared to traverse stages 3 through 5, progressing from unidisciplinary identities to targeted collaboration and finally dual (disciplinary and allied health) identity (Boyce 2006). In its bottom-up approach to the development of a superordinate identity, the IIC model is congruent with the ASPIRe model and other social identity research.

Discussion

Prior research has provided a rich description of how power structures, group norms and values, strength of identification, and contextual factors interact with various social identifications to produce different patterns of group behavior. The social identity approach has the potential to serve as a coherent framework for synthesizing this diverse information and identifying the most promising mechanisms for change. In order to realize this potential, it is important to take advantage of the full depth and complexity of the approach, rather than stop at the basic insight that people engage in social categorization. Further research should also focus on deepening our understanding of the currently neglected group level, rather than assume that a potpourri of group-based and individualistic approaches can offer a better or more complete analysis. It also is essential that research be sensitive to the unique features of the health care context—not because transferability is unimportant, but because overlooking social identifications, structures, or elements of identity content that are highly salient in a particular context can result in shallow or misleading analysis. Finally, since much of the literature has concentrated on the micro level of interprofessional silos and clinical teams, future research might emphasize the macro level of interorganizational silos and system integration. While the same basic mechanisms may apply in both contexts, the specific constellations of social identity variables may be very different. Two directions for future research are (1) in-depth exploration of social identity dynamics during system-integration efforts (e.g., development of accountable care organizations in the United States), with a focus on identity mobilization and context change; and (2) further testing of the ASPIRe and IIC models in health care systems.

A recurring theme in literature from disparate traditions is the importance of identity mobilization and/or context change in driving system transformation. Change seems most likely to occur when both mechanisms are present: Without mobilization of valued identities, attempts to impose context change may provoke identity threat and invite implementation failure; without changes to the real conditions under which people work, identity mobilization may amount to “just another staff development workshop.” The two processes may also reinforce each other cyclically: Mobilization of shared identities can facilitate the adoption of concrete changes (Graham and Steele 2001; Kerfoot 2007), while changes in working arrangements can stimulate the reshaping and reinterpretation of social identities (Farrell, Schmitt, and Heinemann 2001; Hotho 2008). Mobilization may be the natural first step in the process (cf. the ASPIRe model); however, context change can be the impetus for identity reconstruction. In a study of primary care reform, changes at the institutional and organizational levels (capitation, multidisciplinary teams, co-location, etc.) created a context for physicians to reframe teamwork, preventive medicine, and guideline adherence as identity congruent (Chreim, Williams, and Hinings 2007). Whereas managerial attempts to “colonize” staff identities are likely to be resisted or subverted, context changes that can be meaningfully incorporated into existing identities may stimulate constructive engagement (Levay and Waks 2009). In the course of such engagement, providers not only reconstruct their identities to fit the intervention but often reconstruct the intervention to fit their identities. Although such reconstruction may sometimes serve narrow professional interests (McDonald, Harrison, and Checkland 2008), in other cases the result advances the interests of professionals, managers, and, most important, patients (Waring and Currie 2009).

In advancing a social identity perspective on health care silos, we do not mean to imply that silos are wholly a function of social identity dynamics. The SIA recognizes that individual and interpersonal factors remain important, particularly when group identification or social-identity salience is low. Moreover, although practical and operational problems often have a social-identity component, solutions cannot be found by addressing social identity alone. The contribution of the SIA is to articulate when and how various factors will be relevant and to prevent misconstrual of a group-level issue as an interpersonal or purely operational one. We suggest the SIA as a framework, not a replacement, for other group-level theories; for example, while the SIA highlights the ubiquity of politics in organizations, it is not a theory of politics, and other approaches delve more fully into structural relations or power enactment. In short, our claim is not that everything is a social identity problem, but that every problem involving interactions within or among health care groups probably has a social identity dimension, and that understanding this dimension will enable more effective responses.

Because this is not an evidence review, we will not recommend particular interventions. But we do offer the following guiding questions for health care leaders confronting the problem of silos:

1. Who are the relevant groups, and what are their relationships? How might social identity be playing a role in current organizational problems or conflicts?

2. Which groups need to be around the table to develop new ways of working? Can they come together immediately, or does intergroup tension necessitate that they first work separately?

3. Which identities matter most to participants, and how can these be mobilized? What change messages (and messengers) will fit the values and attributes cherished by each group?

4. What is the potential for context change? How might day-to-day factors that reinforce silos and conflict be replaced by others that promote more cooperative and equal interactions?

5. As change proceeds, how can identity threat (to valued groups’ existence, status, distinctiveness, values, etc.) be minimized?

As the more than 300 reports in our review reveal, social identity is a powerful reality in the functioning of the health care system. Rather than attempt to ignore, expunge, or manipulate social identities, we can embrace the opportunity to work with and through them to unite providers around the values that all health care professionals share. Social identity thinking can unlock new options for overcoming silos and bringing about a harmoniously functioning, well-coordinated health care system.