Juggling Ambiguity in Sustained Ignoring Work: The persistent dismissal of warnings at a university hospital

Ignoring work as shifting in its degree and nature of wilfulness

Many studies of ignoring highlight ignoring actors’ clear and stable strategic intentions or political interests in avoiding responsibility or keeping people in the dark (McGoey, 2012; Proctor & Schiebinger, 2008) and emphasize conscious choice and rationality (Hertwig & Engel, 2016). However, recent work also describes ‘less wilful’ and reactive forms of ignoring work, where intentions are less clear-cut and less strategic, and ignoring work is rather driven by silo thinking or the desire to simply muddle through (Alvesson et al., 2022; Essén et al., 2022). We assume that ignoring work may be characterized by shifting degrees of wilfulness and multiple motives for ignoring in prolonged and contested situations.

Ignoring work as managing the ambiguity of the ignored issue

A large share of extant research assumes that ignoring actors know what not to know (Knudsen et al., 2023; Proctor & Schiebinger, 2008). Ignoring is referred to as unseeing and repressing ‘what is known’ or as refusing to acknowledge information that ‘does not correspond with one’s existing cognitive frameworks’ (Roberts, 2022, p. 368). This suggests that there is knowledge from which actors disconnect or redirect. However, the target of ignoring may be fluid and uncertain rather than a matter of clear ‘knowledge’ that can simply be avoided (Alvesson et al., 2022). This is implicitly shown in studies (Kronblad et al., 2024; Schaefer, 2019) illustrating the ignoring of fragments and signs that could in theory be investigated further, although it may be unclear how. Actors also cast doubt on the ignored information, emphasizing its unreliability and incompleteness (Ebbevi et al., 2024; Essén et al., 2022; McGoey, 2012). Hence, the ignored issue can be and can be constructed asambiguous (Alvesson et al., 2022; Maitlis & Sonenshein, 2010) – i.e. characterized by ‘opaqueness and the simultaneous existence of several conflicting views of the same phenomenon’ (Cappellaro et al., 2021, p. 2). ² Hence, we assume that sustained ignoring work may involve not only coping with ambiguity but also its active exploitation (Abdallah & Langley, 2014; Cappellaro et al., 2021).

Ignoring work as selective knowledge avoidance and seeking

The points above challenge the common view of ignoring as knowledge avoidance (Schaefer, 2019, p. 1308). Indeed, persistent forms of ignoring are likely to involve combinations of knowledge and non-knowledge (Gross, 2007), as actors struggle to know what to ignore. Although the phenomenon has been little theorized, a few studies suggest that ignoring can involve seeking some information (e.g. to avoid ‘stepping on land mines’ or appearing totally ignorant about issues) while strategically rendering it unavailable for decision-making (e.g. Heimer, 2012; Knudsen, 2011). Hence, we assume that sustained ignoring work may involve both knowledge avoidance and (partial, selective) knowledge-seeking.

In summary, we assume that sustained ignoring work in dynamic situations is (a) shifting rather than static in nature, (b) not necessarily clearly or only wilful or strategic but often ad hoc and reactive, (c) often involving ambiguity management and (d) a matter of degree (selective). Although there is fragmented support for these assumptions in the emerging literature, their combined manifestation and interaction have not been theorized in previous work. Arguably, these aspects are crucial to understanding how managers sustain ignoring work in complex, contested situations, as in our case.

Stage 1: Collecting documentation and making sense of the evolution of events

We initially compiled and triangulated the extensive available documentation (5,793 pages) about the M case (searching in the two major daily newspapers, the trade press, open online searches and acquiring books written about the case) and created timelines depicting the warnings raised and organizational (non-)responses between 2009 and 2021. We worked with a narrative timeline (about 30 pages) and parallel visualizations to facilitate our sensemaking of the process (Langley, 1999). The documentation confirmed that warnings had been persistently raised over a long period. The documentation however did not explain how responsible managers could, over several years, make sense of things in a way that enabled them to avoid thorough scrutiny of the potential malpractice that the warnings pointed at. Based on the assumption that this non-response was the result of ‘work’ that might be observed or inferred (Knudsen, 2011), we engaged in a second round of data collection and analysis, including interviews.

Stage 2: Conducting and analysing interviews

We constructed multiple versions of a semi-structured interview guide based on our pre-understanding reached in Stage 1. Open-ended questions (adjusted to different individuals) concerned how they and/or others gradually realized that M was engaged in scientific and clinical malpractice and how they and/or others responded to this fact or found reasons not to. In 2021, we (the first author and a research assistant) interviewed (1) ‘critics’ (an external reviewer and certain Karolinska employees) who had expressed concern about M or submitted warnings, (2) managers (who wish to remain anonymous) at five different levels who, according to documents, received warnings (in)directly in the first and/or second phase and had a formal responsibility to act (see supplemental files: Part Two), (3) other Karolinska employees who were present and aware of the evolution of events (and sooner or later were exposed to the warnings) and (4) external experts with unique insight into the case (e.g. journalists who followed the case closely). We conducted 23 interviews with 17 respondents, each lasting 60–120 minutes (including a few follow-up interviews in 2023–25). Some respondents seemed eager for an opportunity to reflect on the experience and turn it into lessons that could prevent the same thing from happening again; others were afraid of becoming scapegoats. We made great efforts to adapt to the requests of the respondents (regarding where, how and when interviews took place) to reduce their anxiety. Many still declined, referring us to documentation instead.

Interviews aimed to capture the interviewees’ sensemaking of the process (Langley & Meziani, 2020). Extensive pre-interview preparation was essential given the complexity of the studied process (numerous related events) and the sensitive nature of the issue. For instance, we prepared for potential ‘defences’ and attempts to manage impressions (e.g. ‘I had no idea’) by referring to timelines that depicted submitted warnings at certain points in time. Communicating that our intention was to understand the unfolding of events rather than blame interviewees was also important, as was ensuring anonymity as far as possible. To reduce pressure, we often started by reaching agreement about the overall timeline of the process and speaking about collective behaviours and what ‘most of the employees’ or ‘others’ did rather than what the specific interviewee had (not) done. This allowed us to gradually delve into more sensitive individual experiences and acts of avoidance that the interviewees now sometimes regretted or themselves still struggled to understand. All interviews except one were recorded and transcribed. Our evolving understanding sometimes resulted in follow-up interviews to elaborate or revise our understanding and/or delve further into alternative explanations for specific issues. (We interviewed some respondents two or three times.)

In parallel, we worked continuously with initial inductive, first-order categorizations of the material that provided cues about why and how the managers might have avoided engaging with the warnings. Data and categories were considered as uncertain cues rather than definitive meanings, calling for consideration in terms of the larger picture, in line with hermeneutic principles (Alvesson & Sköldberg, 2017). We compared the interviewees’ retrospective accounts with managerial arguments in the documentation published during the focal period, resulting in gradually developed interpretations and a growing list of (partly conflicting) examples and explanations.

Stage 3: Using ambiguity management to understand how managers did ignoring work

As we began to perceive saturation (few new event types or explanations seemed to be emerging from interviews), we paused data generation and switched to data analysis (Alvesson & Sköldberg, 2017). We considered numerous different ways to account for managers ignoring the warnings (e.g. they lacked time to engage, genuinely did not understand, or were trapped in an inertial institution). However, several data sources contested such interpretations. Instead, our analysis suggested that the managers actively refrained from engaging with the warnings by interpreting issues in certain ways. In particular, the managers’ way of selectively backgrounding and foregrounding the complexity and opacity – ambiguity – of a number of situational dimensions around M allowed them not to engage with the warnings in a substantive manner, by legitimating such non-knowledge-seeking. We found that previous studies on ambiguity management (Cappellaro et al., 2021; Eisenberg, 1984; Sillince et al., 2012) could help us further explore of the role of ambiguity, and in particular managerial constructions of ambiguity. We treated the ambiguity that was ‘intrinsic’ to the case studied (in terms of the complex and conflicting signals about M’s practices and the complexity of his work ⁴ ) as analytically distinct from the ambiguity that was constructed and exploited by actors in specific situations (Eisenberg, 1984; Sillince et al., 2012). The latter was our focus.

At this stage (spring 2022), we gained access to 44 interviews (audio files of witness hearings) from the court, including interviews with Karolinska managers who had previously declined our interview requests. These interviewees’ answers were likely geared – even more than in our first round of interviews – towards appearing trustworthy and innocent in court (Langley & Meziani, 2020). However, they still provided insight into the managers’ explanations for not engaging with the warnings. Using them in combination with our other empirical sources, we explored the relevance and variations in our tentative concept of ambiguity constructions as a form of ignoring work, writing analytical memos and forming interpretations of underlying meanings and indications.

This led us to identify different kinds of ambiguity constructions among the managers (although they used other words for this), each targeting different aspects of the situation. Some aspects (e.g. M’s practices and the validity of the warnings, which could be seen as at least partially knowable) were constructed (presented) by the managers as incredibly complex, unknowable, or ‘impossible’ to grasp (i.e. highly ambiguous). Other aspects, which – as the warnings suggested – could be seen as far from clear-cut (e.g. M being a genius; who was ‘responsible’ for investigating the situation), were constructed by the managers as very clear (i.e. not ambiguous at all). We identified two broad types of ambiguity constructions that were repeatedly and collectively performed by the managers, i.e. patterned at group level. First, disambiguating acts constructed things as being a certain way rather than any other way – thus backgrounding or cleansing the intrinsic ambiguity of the situation in terms of the existence of multiple, potentially valid yet conflicting views (Cappellaro et al., 2021). Second, ambiguating acts foregrounded ambiguity in terms of the uncertainty, opacity or complexity (Weick, 1995) of broader issues such as truth and the consequences of action.

We understood these constructions of ambiguity as occurring within both the speaker and the recipient (Sillince et al., 2012). More specifically, the constructions were self-directed, as managers reported to us that when they were making sense of the situation alone, they interpreted some things as ‘too complex or uncertain’ to even try to address (i.e. legitimizing their non-engagement with the warnings to themselves). At the same time, the managers’ references to issues as either highly ambiguous or not ambiguous were communicated to others, who partly shared and accepted them. Hence, we approached the ambiguity constructions as ignoring work directed towards self and others, allowing oneself and others to dismiss the warnings.

As further rounds of analysis suggested that the ambiguity constructions facilitated a partial form of knowledge avoidance (non-knowing), which interacted with knowledge-seeking (knowing) efforts, we came to understand ignoring work as ambiguity constructions enabling selective non-knowing and knowing. This was based on our material, which suggested that the managers’ ambiguity constructions happened before (and were a precondition to) their selective knowledge avoidance (although being reproduced later). By telling themselves and others that issues were more or less ambiguous, the managers could make choices about what (not) to know more about. This was evident, for instance, in the handling of negative references received during M’s recruitment in Phase one, and also when the more elaborate warnings were submitted in Phase two, where issues were described (e.g. in emails) as more or less ambiguous before decisions were made to dismiss the warnings. Overall, as the constructions were evident in media-reported statements made by the managers during the focal period and referred to and exemplified by other individuals, we interpreted the constructions as having taken place during the focal period, as opposed to merely post-rationalizations created at the time of our interviews.

While the ambiguity constructions represented active and patterned efforts, they were not an explicit or coordinated strategy performed by the managers. Quite expectedly, the managers did not refer to an intentional exploitation of ambiguity for the purpose of avoiding substantial knowledge-seeking about the warnings. Rather, they communicated to us a frustrating process of coping with a very complex and difficult situation, which they found themselves forced to reactively handle. ‘To be honest, I had no idea what to do, what was the right thing. . .’ as one manager confessed. Interpreting and referring to issues as either unclear or clear was a way of coping with this situation and deciding what knowledge to seek and what not to seek. Having said that, the material also pointed toward a shared unwillingness to engage with the warnings, which was largely driven by strategic interests to protect the potential breakthrough and Karolinska’s or managers’ own reputation. For instance, many referred to a deep antipathy towards engaging with the warnings, and many employees reported managers’ communicating a clear stance of ‘Let’s not go there, it’s not worth the trouble.’ Hence, we increasingly approached the observed ignoring work as both reactive and strategic.

Stage 4: Identifying phases and developing a synthesized conceptualization

We created several visual mappings depicting the chronological evolution of the warnings, the pressures to respond produced by the warnings, the managers’ shifting unwillingness to engage with them, and the related change in the disambiguating and ambiguating acts. This led us to identify two qualitatively distinct process phases (Langley, 1999), as the Figure 1 timeline shows. Each phase was characterized by patterned constructions of the ambiguity of issues as high and low, which we conceptualized as specific configurations of ambiguity-juggling. The transition between the two phases represents a shift in the type of warnings – caused by a lack of response to the warnings, i.e. a ‘setback’ in the process, from the critics’ perspective (Cloutier & Langley, 2020, p. 8). In each phase, ambiguity-juggling constituted a form of ignoring work (process) which justified the managers’ selective (non) knowledge-seeking, and in turn produced a state of accepted organized ignorance among the managers receiving the warnings and their inner circle (Phase 1) and also among the employees in the broader organization (Phase 2). We worked with several alternative visualizations of the process and asked for feedback on them, ultimately arriving at the linear chronological model of the process presented in Figure 3 (Cloutier & Langley, 2020). Finally, we revisited and finalized our data structure (Figure 2) (inspired by Gioia et al., 2012) ⁵ to provide traceability in condensed form in relation to Figure 3.

OPEN IN VIEWER

Figure 2.

Data structure.

OPEN IN VIEWER

Figure 3.

Visual summary of how sustained ignoring work involving shifting ignoring triggers, adaptive ignoring work, and evolving states of organized ignorance manifested in our case.

Ignoring-work trigger

Informal warnings and unwillingness to engage with them to protect the opportunity. The initial period of M’s appointment was characterized by grand hopes tied to his extraordinary clinical and research-related abilities. Many employees described a ‘fairytale feeling’ of being part of his story. Karolinska was struggling to uphold its reputation as one of the world’s leading medical universities and needed a breakthrough, so employees placed great anticipation on M. ‘Finally, someone who could save us,’ one manager said, explaining:

This [M’s practice] was not about incremental advances [. . .] It would undoubtedly lead to something radical, in the history books. The regenerative research would be spread to other areas of transplantation. . . Of course, everyone had different dreams of how [M] would elevate them to glory and fame as individuals. The top managers imagined bringing the next Nobel Prize to Karolinska, more donations and reputation, and the employees expected benefits from having worked with one of the most extraordinary geniuses alive [. . .]. (Manager, interview, referring to recruitment period in 2010)

During this phase, informal warnings about M’s potentially questionable character and practices introduced conflicting interpretations about him. For instance, during the recruitment of M, four of his five referees (former colleagues) noted that while he was a technically outstanding surgeon, he also had a ‘questionable moral compass’ and an inclination to ‘take excessively large risks’ [Asplund, 2016]. Later, four Karolinska employees in non-managerial/professor positions (internal critics) further began voicing concerns about the extreme medical complications that M’s surgical patients suffered during his early employment at Karolinska. His patients never seemed to recover, as the plastic tracheas he implanted failed to integrate with their surrounding tissue, thus continually causing new problems. The internal critics formulated their concerns to their line managers in an open-ended manner: ‘We were careful and communicated the possibility that things seemed unusually problematic and potentially caused by overly immature and risky procedures’ (critic 1, interview).

Although open-ended and communicated from persons with relatively low status, the concerns did pressure the managers to respond somehow. ‘The managers had to at least let the critics know that the information was received. . .’ (employee, interview). At the same time, acknowledging the warnings and the potential problem they indicated risked reducing the potential for a breakthrough or the speed at which it might be achieved, which contributed to a willingness among managers to at least temporarily avoid significant engagement with the warnings. The desire to ‘protect the opportunity’ and see what M could do was strong. As noted by a manager: ‘There was a clear disinclination [. . .] “Oh no, why this, why now?. . . It disturbs this great opportunity. . . Let’s wait with this. . ..”’

As illustrated below, the managers coped with these opposing impetuses – to know and not know – by juggling two complementary constructions of ambiguity: one making (some) things appear very clear and straightforward (disambiguation) and the other making (other) things seem very complex (ambiguation).

Disambiguating individuals’ abilities through role-casting

One clear pattern in the managers’ response to the warnings was their casting of M and the critics into stereotypical roles in the story of a breakthrough attracting jealousy and critique. For instance, faced with the views contradicting or at least muddying the ‘genius’ view of M during his recruitment in 2010, the managers in the recruitment team were initially concerned: ‘We had discussions, of course, and we were surprised [to receive the negative references]’ (manager, interview). However, a shared way of interpreting the critique gradually emerged, in which the concerns were interpreted against what the managers presented as M’s unusual abilities:

Surgeons who have no qualms about excising diseased tracheas from people and replacing them with decellularized necro grafts [. . .] are rare. I do recognize the phenomenon well, having experience with similar people in our business. Sweden’s first kidney and liver transplants were made [with interesting methods], [. . .] contributing greatly to humanity and patients (even if some died in surgery). A specific kind of personality is required to eviscerate people in vivo. The border between success and fiasco is thin, and success is contingent on several conditions and other people who can balance it all. (extract, email from the head of the department to Principal 1 during M’s recruitment, 2010; translated by the authors)

The managers hence dealt with the concerns initially created by the referees’ input by casting M in the mould of trailblazers who achieved significant medical breakthroughs despite initial setbacks. The argument was that high initial mortality was an inevitable corollary of these leaps forward in medicine, which required a certain type of individual: a genius willing to take risks [Asplund, 2016]. Moreover, taking all potential risks seriously in such contexts might prevent anything risky or unconventional from ever being attempted at all. The clear-cut characterization of M as extraordinary implied that he should not be judged by the same criteria as others, and hence the critics’ concerns were not valid, motivating managers to avoid seeking further knowledge about the details. Notably, the recruitment team asked for additional information about M:s character from the Mayor of Tuscany, who was known to be a big admirer of M, thus engaging in a minimal form of knowledge-seeking focused on M’s own person. This letter was very positive towards M. As a result, M was later recruited over critics’ objections:

There obviously will be noise around a great scientist: lingering too long on people having not perceived him as nice [. . .] and other nonconcrete ideas [. . .] is not possible. [. . .] Great breakthrough innovations always generate those kinds of things. [. . .] I think we [involved in the recruitment] needed to reduce the noise and see the overarching question: ‘Could this man be a one-of-a-kind scientist – yes or no?’ The answer was rather straightforward and actionable: ‘yes’. (manager, interview, referring to the recruitment period in 2010)

In the quote above, the narrative of the breakthrough innovation as (‘always’) generating resistance provides an illustrative account of the increasingly agreed-on narrative, with M as its ‘clear’ hero and his critics cast as evil adversaries who cannot possibly understand him.

The construction of M as an extraordinary, unfathomable genius was also reflected in the managers’ response to the four Karolinska employees (internal critics) who later voiced concerns about M’s malpractice early in his employment in 2011. As one critic noted, recalling one of the meetings that were finally arranged (after many fruitless attempts and long silences):

When we first mentioned our concerns about the extreme complications [from M’s surgeries] at an informal meeting with the head, we were met with sad eyes saying, ‘Why don’t you try to really understand what M wants to achieve? Why don’t you take a nuanced view of M? Why are you being so regressive?’ But perhaps above all, ‘How dare you question the skills of this fantastic mind?’ (employee, conversation, referring to the period in 2011)

In retrospect, the managers did admit to feelings of uncertainty in relation to the early warnings. (As one noted, ‘I was shocked. What is this? Surely it can’t be true? At least it’s too early to say?’) However, the above quote and numerous others illustrate how the managers coped with this uncertainty by communicating with each other and the critics in a way that constructed the critics as typical examples of regressive, unimaginative, overly cautious personas. This served to undermine both the credibility and power of the critics’ arguments.

Thus, managers handled their hesitation and concern by making sense of the situation in a way that idealized M (as highly skilled, bold, visionary) and – more reactively than strategically – denigrated the critics (as poorly understanding, cautious, conservative). This made the situation appear relatively unambiguous and made the case for ignoring the warning signals seem fairly unproblematic – i.e. accomplished through ‘ignoring-supportive’ constructions of the situation and people involved. The alternative possibility and storyline – that the critics could be right, and that their warnings in fact suggested that they were able to identify problems tied to M’s risk inclination and practices – could therefore, at least temporarily, be dismissed.

Ambiguating truth through relativization

While the role-casting constructions backgrounded ambiguity by making certain interpretations of M less credible and thus ‘ignorable’, the managers also engaged in parallel sensemaking behaviours that foregrounded the ambiguity of other dimensions of the situation as a response to the warnings. In particular, a salient reasoning among managers was that it was in fact impossible to arrive at a final, ‘complete’ description of M’s accomplishment. The complexity of the situation made it impossible to define the ‘appropriate’ level of risk in relation to the potential. This depended on, and was always relative to, the perspective taken. For instance, in 2013, two of the internal critics articulated their concerns about M in a meeting with the professor who led the ENT unit and the professor who was secretary of the Nobel Assembly, only to be told that the situation was too complex and therefore impossible for the professors to judge: ‘This is too advanced for us to understand’ [Lindqvist, 2018]. Although the critics’ concerns were fairly straightforward – referring to medical consequences among patients, derived from firsthand observations – they were treated as impossible to interpret and act upon because there were likely other sides to the complex story. The managers hence engaged in ambiguation by moving the subject matter beyond the scope of judgment and knowability. This made any thorough attempt to assess the validity of the warnings seem futile:

It’s funny: we are so concerned with evidence and quantitative proof of every little detail when introducing new interventions. But when we’re allowing a new surgeon to operate on patients in real life, it suddenly becomes meaningless and impossible to delve into details about results, to distinguish right from wrong. [. . .] I mean, even the most qualified professors, some of whom were on the Nobel Committee, claimed not to understand the issue [. . .] and suggested that ‘there were probably different ways of looking at this’. This was a reason for not even trying to understand the risks we pointed out. (critic 2, interview, referring to period between 2011–14)

The acts of disambiguation and ambiguation in Phase 1 reinforced each other. The combination of constructed person-ability certainty (ambiguity-cleansed ‘creative genius versus small-minded detractors’ narrative) and epistemological ambiguity – ‘We can’t know’ – made any substantial exploration of the warnings (and thorough attempts to reduce the ambiguity of what was ‘true’ about the situation) seem irrelevant. This legitimized the move from knowledge-seeking about the issue that the warnings pointed at, to a selective, person-centred knowledge-seeking. Worded differently, it enabled a shift from a factual knowledge-seeking logic to a matter of character assessment – moving from ‘what’ to ‘who’:

We were all totally impressed by [M]. . . It was not difficult to dismiss the warnings at this stage. There were many reasons not to stir up a fuss, if I am to be honest. It seemed unlikely that the concerns that were vaguely posed by some individuals on the periphery were worth exploring, given that [M’s] practices were very difficult to understand, and would probably never be fully understood by us even if we tried. [. . .] In combination, it all appeared both meaningless and incredibly tough to even start stepping uphill into the potential mess that the critics pointed at. A mountain to climb that would potentially prove useless in the end. (manager, interview, referring to the period 2011–14)

Enabling a state of contained organized ignorance

Lax ambiguity-juggling worked, in the sense that no-one other than critics questioned managers’ non-action at the end of Phase 2 (end of 2013). Indeed, in addition to justifying the non-engagement with the warnings to themselves, the managers’ reasoning was also communicated at meetings and coffee breaks and accepted by their inner circle of Karolinska colleagues, contributing to a collective view on the subject matter and an accepted passivity in relation to it within this small group (a state of contained organized ignorance). The outer groups of the larger Karolinska organization were however at this time largely unaware of the warnings and the managers’ non-response to them. In parallel, M’s patients continued to experience severe complications.

Ignoring trigger: Targeted alarms and an ambivalent willingness to both know and ignore to minimize damage

In Phase 2, the old and new critics communicated their old and new concerns in an increasingly explicit and formal manner, submitting reports with more evidence, making direct accusations of several forms of severe malpractice, and approaching those formally responsible for dealing with suspicions about clinical and scientific malpractice. The warnings were now delivered not only by M’s previous colleagues and the internal critics but also by external critics with more authority, making it harder to rely on loose references to M as an unquestionable genius in combination with epistemological uncertainty over his practices (as in Phase 1). For instance, from 2014 onwards, several formal reports accusing M of scientific malpractice were submitted to KI’s Principal 2 (who had replaced Principal 1 in 2013). The internal critics submitted a 400-page report describing indications of M’s research malpractice. Further, a foreign researcher submitted an official report of suspected scientific malpractice, arguing that M’s claims of plastic-trachea surgery success were exaggerated in several papers published in The Lancet. The reports called for an investigation into whether M had preceded his human interventions with animal experiments and ethical approval, as required by regulations on medical research, and whether he had followed basic documentation requirements. These requests were combined with reports of M’s patients’ suffering and death, based on reviews of their medical records.

Stronger critical signals thus now collided with the equally strong established view of M as a genius and saviour. This triggered more ambitious ignoring work. Indeed, while the fortified alarms created a stronger pressure to respond, managers’ unwillingness to acknowledge that the warnings might point at a ‘real’ problem also grew stronger, as ‘the more time that passed, the worse it would be to be forced to realize that we had neglected a genuine risk for so long’ (employee, interview, referring to the situation in 2014–15). Interestingly, our material also points towards an increasing willingness to know more about the warnings at this stage. ‘There was more effort to know just enough to be able to dismiss the warnings,’ as one employee put it. Managers also referred to a mix of desires: to understand what had really happened, but without getting lost ‘down the rabbit hole’. If anything, the will to (not) know was multidimensional and ambivalent. Driven by these multiple, conflicting motives, and faced with simultaneous pressures to respond, the managers coped, once again by constructing the ambiguity of some issues around M as very low and around others as very high.

Disambiguating roles through compartmentalization

The warnings delivered in Phase 2 suggested a complicated, problematic situation tied to several dimensions of M’s practice. In particular, the combination of so many question marks was seen in post hoc reviews as unquestionably warranting a comprehensive analysis of the problems suggested by the reports and a step back to look at the totality of the indications that had emerged so far. One unit at Karolinska called a halt to M’s surgical activity. However, the managers at this unit – and, more importantly, Principal 2 and his advisors – located themselves, and smaller parts of the warnings’ subject matter, in formal categories in a way that absolved them of any responsibility to thoroughly investigate the totality and validity of the warnings. This involved compartmentalization by treating parts of the warnings in isolation and tying those parts to ‘formal’, partitioned occupational jurisdictions and categories.

As an example, in response to the report submitted by the foreign researcher in 2014, Principal 2 and his advisors identified functions (other than their own) responsible for investigating certain aspects of the suggested signs of malpractice, subsequently requesting a statement from the KI Ethics Committee. The committee concluded that the research malpractice accusations made by the foreign researcher were unsubstantiated (Svärd Huss, 2022) and referred to the issues raised as matters of ‘philosophy of science’ rather than ‘research ethics’ (Braunerhielm, 2017; Stokstad, 2015). Having accepted that the issues belonged in the ‘philosophy of science’ category, Principal 2 made no attempt to thoroughly scrutinize the warnings or form any judgement on them. Instead, this categorization turned certain parts of the warnings into an issue of scientific debate for the research community to deal with rather than a matter for external investigation. Recalling this announcement, a KI manager reflected that:

There was no hesitation [in the official communication by the top management]: everything was in order. No question marks. It was important to make things [potential malpractice] appear certain, to us and others – although they were, I now of course realize, very grey. (manager, interview, referring to period 2014)

This quote illustrates the backgrounding of ambiguity around M’s practices. In another example in 2015, fragments from M’s patients’ medical records leaked to international websites, prompting a New York Times piece about the accusations. That article, combined with the formal accusation reports, finally prompted Principal 2 to initiate an external review of M’s practices. The external review (Gerdin, 2015a, 2015b), which ran to approximately 1,500 pages, found M guilty of research malpractice. Shortly afterwards, however, Principal 2 announced that KI had cleared M of all allegations of scientific malpractice and did not intend to investigate the issue any further. Justifying this decision, Principal 2 and his advisors referred to M as having acted on conditions belonging to a certain medical category: ‘vital indication’ – that is, life-threatening conditions for which no better treatment alternatives are available, as noted in the so-called Helsinki Declaration:

The surgeries were performed in accordance with §32 of the Declaration of Helsinki at the time: ‘§32. In treating patients where evidence-based preventive, diagnostic, and therapeutic methods are non-existent or have proven ineffective, the physician is free to use informed consent by patients to use untested or novel preventive, diagnostic, or therapeutic methods.’ (extract, author’s translation, Daily Medicine, 2015, https://www.dagensmedicin.se/opinion/debatt/vart-beslut-vilar-pa-saklig-grund/)

The extract illustrates how the Principal referred to the issue as a clear case of vital indication. The critics (e.g. the external reviewers, Karolinska employees, and journalists who had begun following the case by now) questioned this categorization (Asplund, 2016; Braunerhielm, 2017), as did several medical specialists in witness hearings held when the case was brought to court. They claimed that although M’s patients had suffered before his treatment, their conditions could be seen as chronic and/or treatable with alternative methods, at least partly. However, the managers repeated the ‘vital indication’ argument in the press and at subsequent meetings, underlining that they were ‘certain about this, and that there was no need to worry or to act, this was taken care of’ (board member, interview). Once again, any ambiguity around the appropriateness of M’s practices, and whose role it was to actually investigate the problem that the warnings pointed at, was backgrounded – officially, at least – through the focus on small parts of the warnings and categorizing the problems they pointed at in ways that absolved the managers of any responsibility to investigate the totality of the warnings further.

In summary, the managers’ disambiguation of the situation in Phase 2 did not imply a complete disconnect from or avoidance of the targeted alarms; they did initiate investigations that yielded lengthy reports. However, the managers subsequently dismissed the signs of serious problems that these reports confirmed. This partial, initial non-ignoring/knowledge-seeking and subsequent ignoring formed a combination of process-knowing and outcome-ignoring. Boundary-drawing around the problem and procedures here turned issues that could be seen as highly ambiguous (M’s practices, as well as who was responsible for seeking knowledge about them) into something more clear-cut. Paradoxically, this disambiguation of responsibilities through compartmentalization – and the avoidance of engaging in ways that perhaps lay beyond one’s formal or at least everyday tasks – was motivated by the fear that the warnings could, in fact, be partly true. As we illustrate next, this uncomfortable suspicion was followed by compensatory ignoring work constructing the ambiguity around the consequences of acting on the warnings as very high.

Ambiguating consequences of action by amplifying uncertainty

Officially, Karolinska managers continued to support M and insisted that there was no reason to scrutinize him any further. Informally, however, they also expressed views foregrounding uncertainty regarding what might happen if they were to investigate the warnings and thus go beyond their ‘ordinary’ tasks. Behind closed doors, the managers and employees sometimes admitted that the mounting warnings were potentially relevant, albeit perhaps coloured by personal conflicts. Yet the managers still deemed it very risky to delve into the warnings, on the grounds that doing so might lead to ongoing and escalated conflictual views and far-reaching consequences over time. As one manager noted, this raised the question:

‘Why did you not speak up earlier – you are a manager!’ or various other accusations that no one could even imagine. [. . .] You know, people often behave in unexpected ways when crises emerge, and one should not think one can predict this. (manager, interview, referring to the period in 2015)

Some employees perceived this when informally asking managers about the status of the ‘M case’ in this period:

The message was: ‘Put a lid on it. We are not going there. You do not distrust your colleagues. You do not interfere. You are loyal. Full stop.’ [. . .] No explanation, but indirectly saying, ‘This is too dangerous.’ [. . .] This was clearly a no-go zone, as too much was at stake [. . .] the risks [were] too high to Karolinska, its reputation [. . .] in particular, the uncertainty. Perhaps even exaggerated, I would say. It was like: ‘No one can even foresee what will happen if we start investigating this in depth [. . .] [it] might create ripple effects that no one could foresee.’ (employee, interview, referring to the period in 2015)

Hence, the response to questions was a full range of ignoring work constructing the risks tied to distrust, interference and disloyalty as unforeseeable. Faced with such ignorance-supporting engineering of uncertainty regarding the consequences of speaking up and acknowledging potential problems, people tend to respond with passivity and refrain from knowing work. Bystanding managers from more remote departments, who were also aware of the critics’ formal accusations by this time, similarly emphasized the uncertainty and jeopardy communicated informally by managers, suggesting that it was ‘impossible to predict what would happen if we were to start investigating such an extraordinary thing’ (employee, interview). It was impossible to foresee the consequences of trying to seek knowledge about the details of the warnings and thus (dis-)confirming them. More remote managers and colleagues now often complimented the internal critics – ‘I applaud you for explicating these issues’ – but still ended by saying, ‘Personally, I would not take this further’ (employee, interview). This contributed to a perceived uncertainty regarding the degree and nature of the danger involved in discovering more about the M case, and again, a reluctance among both managers and their inner circles, and a growing number of employees in the broader Karolinska organization, to invest time in truly grasping the content of the warnings beyond a very superficial level:

It was more like the diffuse, terrible, uncertain career future that was driving this fear. [. . .] This uncertainty was often mentioned in informal conversation. . . [that]. . . this could be devastating for Karolinska’s possibilities of acquiring donations again, for its reputation, in many unexpected ways. [. . .] Many people worried about exponential effects, if the ball starts rolling [. . .] etcetera [. . .]. Even trying to predict and ‘calculate’ the effects was seen as futile; no one could ever know exactly what would happen if we were to start investigating M’s practices, and God forbid, find that [the critics] were right. [. . .] They did not want to know more; they did not want to understand the details [. . .]. (employee, interview, referring to the situation in 2015)

Statements such as ‘What would happen if. . .’ and ‘No one could ever know exactly what would happen’ signal the foregrounding of a high level of ambiguity tied to the consequences of knowing more about the warnings. Initiating further investigations that could, maybe, go against important stakeholders was perceived as career suicide, although exactly how this would unfold was impossible to foresee. As one internal critic noted:

Who knows what would happen to someone aligning with a troublemaker, or a manager who dared to criticize and scrutinize a star surgeon? [. . .] I think these vague ideas made people simply not want to even get acquainted with the details of the risks. The more you know, the harder it is to extricate yourself from them [. . .]. (critic, interview, referring to period 2014–15)

The acts of ambiguation and disambiguation combined in the tightrope ambiguity-juggling performed in Phase 2 reinforced each other, as the managers’ ramping-up of uncertainty regarding the consequences of action (ambiguation) partly encouraged them to simplify issues by dividing the warnings into smaller parts and assigning them to established procedures tied to someone else’s role (i.e. disambiguation). Since the consequences of responding to warnings were uncertain, sticking to the letter of one’s formal responsibilities seemed the most attractive option. Similarly, the direction of non-action suggested by seemingly clear roles made it easier to accept the ambiguity around the consequences of action and thereby legitimized the very selective, part-focused knowledge-seeking, and the absence of effort to find out what would actually happen should one take steps to further investigate the totality of issues that warnings pointed at.

Enabling a state of distributed organized ignorance

The tight-rope ambiguity constructions legitimized non-action not only among the managers and their inner circle, but also among outer groups of bystanding employees at Karolinska overall, who were at least peripherally encountering and observing the warnings and managerial (non‑)actions. As noted by an employee in an opinion piece published after the fact:

I know that no major clinical decision is taken by one person alone. I know who was at the meeting where they agreed to start operating on plastic tracheas; it is all public. I know who contributed names and titles to the Macchiarini manuscript. I know how many senior managers, board members, and professors knew, appreciated, approved, and encouraged. I know there were hundreds like myself, who knew and perhaps disapproved but chose to remain silent, out of cowardice or laziness or both. (Grönlund, 2021)

We refer to the organized state of ignorance accomplished at the end of Phase 2 as ‘distributed organized ignorance’. At this stage, the dismissal of the warnings was known and accepted, not only by the managers and their inner circle, but also by the wider organization. As noted by an employee, ‘Everyone [top management, heads of clinics, clinical and research groups] knew about the concerning signs,’ but ‘No one wanted to know more about them.’ At the end of the study period, M was still operating for research purposes with full support from Karolinska. (See supplemental files for further details about what happened when the media intervened.)

The ‘why’ of ignoring

Previous work has identified numerous reasons why actors ignore uncomfortable information (Ebbevi et al., 2024; Essén et al., 2022; Heimer, 2012; Klintman, 2019; Knudsen, 2011) but has not explored the nuances of actors’ unwillingness to know and how they may interact with expectations to attend to information. We contribute by conceptualizing the combination of changing warnings, associated perceived pressures to act on them and motives for ignoring them as dynamic triggering mechanisms (shifting ignoring triggers). Our work confirms that periods of ignoring can trigger escalations of warnings (Anvari et al., 2019; Delmestri & Goodrick, 2016; Kenny, 2019), creating heightened pressures to respond. However, depending on how such heightened pressures interact with actors’ unwillingness to engage with the warnings, this may or may not lead to knowledge-seeking responses (Alvesson et al., 2022). As we show, heightened pressures combined with a sustained unwillingness to engage with the warnings may rather trigger new rounds of revised, and rather sophisticated, ignoring work.

In our case, the unwillingness to engage with the uncomfortable information in question became increasingly multifaceted, involving motives such as simply getting on with an issue and accepting a level of uncertainty, but also protecting a specific trajectory and avoiding facing critique and blame. The unwillingness to engage was also increasingly accompanied by a willingness to know (some things). This suggests that motives behind ignoring work can be both coping-oriented and strategic, characterized by ambivalence, and include a will both to ignore and to know. This extends previous portrayals of the motives behind ignoring as relatively stable, unified and one-dimensional (e.g. escaping liability, maximizing profits, preventing disruption, protecting autonomy (Essén et al., 2022; McGoey, 2012; Proctor & Schiebinger, 2008).

The ‘how’ of ignoring

As we show, shifting ignoring triggers exert a distinct influence on the work of ignoring, which often thus becomes more complicated than simple neglect of known or knowable unknowns (Schaefer, 2019). The ignored issue may rather be a moving target, only partially knowable and characterized by conflicting interpretations, i.e. ambiguous. Drawing on previous work suggesting the role of ambiguity and actors’ need to cope with it in contexts of ignoring (Alvesson et al., 2022; Maitlis & Sonenshein, 2010), we here introduce the notion of adaptive acts of ambiguity-juggling as a way for individuals and groups to carry out ignoring work in the face of complex, disturbing and shifting signals or issues.

Acts of ambiguity-juggling are reactive, triggered by the emergence of uncomfortable issues, but also purposeful, driven by actors’ unwillingness to engage with the issues. Hence, the juggling metaphor refers to how uncomfortable issues are encountered, (e.g. warnings received), only to be thrown up ‘in the air’ – again, thus at least temporarily released from one’s responsibility (but returning in adapted form as new warnings – requiring new rounds (configurations) of juggling).

The notion of ambiguity-juggling particularly underscores the role of dynamic constructions in this context, making the ambiguity of issues too great to confront (‘We will never know’ or ‘It is too difficult to try to know’) or non-existent (‘We already know’) and thereby not worth exploring fully. Here, the construction of certain issues as being virtually unknowable (ambiguation), and the resulting uncertainty and open-endedness of action, call for compensatory disambiguation to stabilize and create a sense of social order, motivating pragmatic action. By clear-cutting and truth-doing certain issues, disambiguation in turn contributes to a continued acceptance of letting other issues remain unknowable, thus legitimizing ambiguation. Since actors feel they know enough, and can demonstrate to others that they have done enough to credibly drop or park the issue, they can let other issues (outside the scope of perceived certainty) be. Hence, in combination, these constructions of things as high and low in ambiguity provide self-reinforcing legitimators of ignoring.

These insights confirm the importance of previously outlined ignoring work such as hiding/disconnecting (Heimer, 2012; Kenny, 2023; Kronblad et al., 2024; van Krieken, 2024), redirecting (Knudsen, 2011; Schaefer, 2019) and the justification of non-efforts to know (more) (Ebbevi et al., 2024; Essén et al., 2022), but supplement previous accounts by foregrounding interacting ambiguity constructions as a novel handling and justification mechanism. Indeed, while previous work shows that justifications such as ‘not my job’, ‘it wouldn’t be right’ and ‘compartmentalization’ of issues legitimizes knowledge avoidance, we introduce an ambiguity perspective that shed new light on previous studies. From this perspective, the active construction, combination and playing out of the view that ‘we know enough’ (to act pragmatically) respectively can’t really know means a ‘clarification/ambiguity cocktail’ that facilitates muddling through, at least for some time and with some luck. Further, our emphasis on the adaptive nature of these acts extends previous accounts of stable justifications for ignoring relatively constant sets of information or a consistent narrative (Ebbevi et al., 2024; Essén et al., 2022). As we show, original justifications may lose their legitimizing power over time and call for updates.

The ‘how much’ of ignoring

Our work shows that acts of ambiguity-juggling legitimize selective ignoring work including both knowledge-seeking and knowledge avoidance. Actors may have to do some knowing work to find out what they could or should ignore, and how to communicate ‘credible ignoring’, i.e. use knowledge-seeking to justify further non-knowledge work (Gross, 2007). For instance, in our case, knowledge-seeking about individuals’ ability (minimal, person-centred knowing), enabled the knowledge avoidance regarding the content of the warnings in Phase 1. In Phase 2, knowledge-seeking focused narrowly on those parts of the warnings that lay within an actor’s comfort zone of responsibility (part-focused knowing), enabled knowledge avoidance regarding the totality of warnings. These insights contribute to the literature, which has hinted at combinations of knowledge-seeking and knowledge avoidance in ignoring work (e.g. Gross, 2007; Heimer, 2012; Knudsen, 2011) without providing empirically substantiated illustrations and theorizations of how they may unfold over time. As we show, sustained ignoring work can involve mutually supporting acts of knowledge-seeking (non-ignoring) and avoidance (ignoring), which shift in nature along with changes in external pressures and actors’ willingness to know.

The ‘who’ (and outcome) of ignoring

The ignoring literature generally treats the outcome of ignoring work as the keeping of a secret (or dismissal) by one actor (or a coherent collective actor) – for instance, the pharma industry (Proctor & Schiebinger, 2008), public organizations (Kronblad et al., 2024; Maor et al., 2012), a manager (Schaefer, 2019) or a collective of organizations maintaining unused data systems (Essén et al., 2022). In these works, ignoring is regarded as a stable, enduring act or a unitary state. In contrast, we point out that the outcome of ignoring work is ongoing and takes the form of evolving states of organized ignorance (Knudsen et al., 2023) that vary in terms of the actors involved in keeping the secret (non-acknowledged issue) and in relation to whom they need to legitimate it to. In our case, top management, heads of clinics and clinical and research groups more or less directly involved in making influential decisions come together to form a more complex backdrop than a top manager simply knowing/ignoring or a management team acting in isolation. We show how the ignoring work shifted from a contained version of organized ignorance to a distributed one as more and more people became at least partly aware of the problems, although they refrained from intervening (Thiel & Grabher, 2024). This suggests various degrees of ignoring by multiple actors, contributing to states of organized ignorance in which it is difficult to point to a single actor or group being accountable. In our case, this allowed several factions within the organization to at least tentatively and temporarily protect their interests (to pursue the innovation and, later, minimize damage).

Taken together, our work highlights a type of ignoring work that is persistent but adaptive (there is awareness and some acting/knowledge-seeking) and provides new perspectives on a crucial set of dynamics for understanding ignoring: the ‘why’, ‘how’, ‘how much’ and ‘who’ of ignoring work, and its outcomes. Considering these elements further in research should enrich theory on strategic or wilful ignorance by advancing our theoretical understanding of how sustained forms of ignoring, both inwardly and outwardly directed, are enabled and maintained in organizations.