1 INTRODUCTION

Protecting patients from harm is a key activity for healthcare professionals. Adverse event is defined as an injury and is a result of unsafe care related to “medical management, in contrast to complications of disease” (World Health Organization [WHO], 2005, p. 8). This includes all aspects of care and could thus occur in nearly any interaction within the healthcare system (Jha et al., 2010). Adverse events are associated with significant morbidity and mortality globally during hospitalization (Jha et al., 2013). Learning from failures has been acknowledged as one important strategy to avoid reoccurrence of adverse events (WHO, 2005). Common subjects of patient safety analysis in healthcare are different types, prevalence, and causative factors of adverse events (Griffey et al., 2019; Howell et al., 2015; Rafter et al., 2015; Westbrook et al., 2015). Further, it has been identified that there are variabilities in definitions and perspectives on adverse events (Hågensen et al., 2018; Rafter et al., 2015). The decisions as to what patient safety actions need to be taken are entrusted to healthcare professionals often based on normative rules within healthcare (Hågensen et al., 2018; Schwappach & Gehring, 2014; Vennik et al., 2016). We draw upon the assumption that healthcare professionals' roles and engagement in patient safety depend on how adverse events are being understood and talked about. Therefore, rather than focusing on different types, prevalence, and causes of adverse events, we focus on how descriptions of adverse events and involved healthcare professionals are constructed.

1.1 Patient safety in acute care setting

In acute care settings, healthcare professionals are expected to provide safe care based on (often fast) interpretations, assessments, prioritizations, and decision-making in various contexts and situations. Traditionally, individual decisions were made by physicians, who had sole responsibility for the choice of accurate treatment based on patients' narratives and what was visible to the eye (Ekman et al., 2014; Foucault, 1989; Mathews & Pronovost, 2008). Today, decision-making processes in acute care rely on a spectrum of diverse competencies and specialized knowledge of multiple professionals, and advanced technology (Ödegård & Wallgren, 2007). Thus, decisions can be made by actors involved in various situations of healthcare. Guidelines are also available in which decisions have already been made by expert groups, even before a patient enters the hospital. In addition, patients and relatives are given increasing responsibility for patient safety (Berger et al., 2014). Overall, the distribution of responsibility among individual healthcare professionals, patients, relatives, organizational systems, and technology has become a complex issue in modern healthcare settings. It might not appear as a surprise that cross-cutting problems (e.g., teamwork, communication, and organizational culture) and variation in the use of language have been reported as contributors to adverse events (Runciman et al., 2009; Shojania & Thomas, 2013).

1.2 Gaps in the literature

The literature on patient safety continues to grow. For example, valuable knowledge about healthcare environment (Singh et al., 2015), work environment, workflow (Baernholdt et al., 2020), detection of early warning signs (Spångfors et al., 2016), or choice of materials (Hahnel et al., 2020) help healthcare professionals engage in patient safety actions. Furthermore, patient safety management has, to great extent, come to focus on the identification of technical and procedural errors (Donabedian, 1988; Ödegård, 2019; Reason, 2000)—an approach that has generated an increasingly standardized healthcare with policies and regulations (Braithwaite et al., 2015; Hollnagel, 2013). Patient safety has in some ways become a unifying matter across interprofessional teams, though finding a common language has proven to be complex, and different professionals have been reported to do patient safety differently (Rowland & Kitto, 2014).

Several barriers for patient safety actions and communication have been pointed out. For one, to act according to patient safety policies has shown to be challenging because of unsupportive healthcare systems, power relationships, and healthcare professionals' view of their own and others' role in patient safety actions (Rowland & Kitto, 2014; Skagerström et al., 2017; Ullström et al., 2014). Furthermore, some matters have been reported to be more likely to be talked about than others, such as medical safety issues rather than issues related to hygiene and isolation (Schwappach & Gehring, 2014). Reporting systems for adverse events, which are established today in many countries, were designed with the intention to support a blame-free cycle of improvement (Howell et al., 2017). Nevertheless, a variety of perceived barriers hindering healthcare professionals from communicating adverse events has been reported. Fear of blame, legal consequences, having one's own competency questioned, unsupportive colleagues, and concerns that the reporting would affect others or the hospital, are examples of barriers that have effects on communicating adverse events (Pfeiffer et al., 2010). Depending on the context, narratives about adverse events have been found to be constructed and reconstructed based on historical and structural rules such as moral norms, distant policy agendas, and contradictory aims (Nicolini et al., 2011; Waring, 2009). Through an ethnographic study of a method called Root Cause Analysis, structural investigations of adverse events in healthcare were analyzed by Nicolini et al. (2011). The focus of the method was on the identification of adverse events, investigation of facts, and the search for causes behind adverse events to identify threats and solutions to patient safety. The authors however, encountered multiple challenges with this process. One major challenge was the investigation of facts. Facts were found to be based on information from multiple sources, such as different computer systems containing little specific documentation, and through interviews with involved healthcare professionals who often felt they had to defend their position. This collection of parts of information about adverse events was found to be constructed into pieces of evidence that became the foundation for further risk management (Nicolini et al., 2011). Waring (2009) also exemplifies how narratives about adverse events were constructed differently in different stages throughout the process of a reporting system in a teaching hospital. Findings showed that everyday talk about adverse events was constructed in one way and that narratives of the same events were constructed in another way in written reports. The everyday talk focused on understanding what had happened and the written reports mostly focused on an effort to explain the event, often with recast responsibility. Moreover, the processing of the written reports by risk managers involved further reconstruction of the narratives through a process of coding and recoding from a set of predefined measures and categories narrowing the range of issues (Waring, 2009). What is being viewed as fact is consequently a product of how meaning is being constructed. Thus, patient safety can be viewed as a social construction, which has an effect on the knowledge production of patient safety.

Overall, the construction of patient safety has an impact on the understanding of how to protect patients from harm, and consequently affect healthcare professionals' interpretations, assessments, prioritizations, and decision-making at the sharp end of care. However, to our knowledge, little is known about how a certain meaning is made possible in adverse events and the mechanisms that govern the inclusion and exclusion of discourses about healthcare professionals in the context of acute care. With the assumption that different discourses point out different directions of possible actions, we made use of discourse analysis to explore discursive rules and truth regimes that bring power to these discourses. We thus explored the discourses about healthcare professionals as constructed in written reports of adverse events as reported by patients, relatives, and healthcare professionals.

2 METHOD 2.1 Study design

A discourse analysis was conducted with a Foucauldian perspective on discursive rules and regimes of truth. This was supported by Potter's (1996) action-oriented point of view on text and how it can be organized so that it appears trustworthy with a status of truth. Through discourse analysis, regularities in statements were explored to reveal rules that enabled this fact production. By focusing on how statements were presented as true representations of reality, what is taken for granted, what is not possible, and who is allowed to speak, rules that set limits to and make discourse of healthcare professionals possible were analyzed. The study was conducted in a university hospital in Sweden.

2.2 Discourse as conceptual underpinning

Discourse refers to the way we understand something through the use of language (Potter, 1996; Winther Jørgensen et al., 2000). The conceptual underpinning for this study draws from Foucault's definitions of discourse that practices “systematically form the objects about which they speak” (Dreyfus & Rabinow, 1983, p. 62). Different discursive formations help us to see what is true, though, simultaneously set limits to what is conceivable (Foucault, 1980). What is considered to be true and false are also the result of productive power systems that constitute knowledge and discourse. Power exists in many forms such as economy, law, administration, and work through the installation of techniques that always refer to different scientific truths. Knowledge is thus tightly connected to power in constructing our social world. Consequently, regimes of truth are produced through discourse and are dependent on socially constructed system of rules forming the identity of an object or a subject and social relationships (Howarth, 2000). With “truth” viewed as a discursive construction, the analysis of this study focuses on how regimes of truth are being constructed, that is, the rules for what can be said and what cannot. However, we do not view discourse as something that operates on its own. In Foucault's later genealogy, he introduces dispositif as a conception—an apparatus comprising both discursive and nondiscursive elements, power, and knowledge (Dreyfus & Rabinow, 1983; Howarth, 2000). An institution such as a hospital would be a nondiscursive element together with connected regulatory decisions, policies, laws, material objects, architectural forms, and techniques (Foucault, 1980; Howarth, 2000). Within a hospital, with its strategic purpose and need, different discourses and nondiscursive elements are connected in a certain way in response to these needs. Further, the needs and strategies are supported by the relationships between power and knowledge. From the perspective of dispositif, discourse is connected to systems of power, knowledge, and nondiscursive elements driven by a purpose.

2.3 Sample and procedure

As specified in the Swedish National Board of Health and Welfare Regulations and General Advice about Management System for Systematic Quality Work (SOSFS, 2011), the university hospital in which this study was conducted has an electronic reporting system for adverse events. All healthcare professionals at the hospital can register an adverse event and are obligated under the Swedish Patient Safety Act (Patientsäkerhetslag, 2010) to report adverse events when identified as such. However, patients and relatives have to report a complaint (oral or written) to a healthcare professional at the hospital, who then registers their complaints in the report system. These complaints are later assessed as to whether the complaint reflects an adverse event. When complaints are reported orally, the registering healthcare professional summarizes the patient's description. This means that the written complaint may be affected by the healthcare professional's interpretation of the patient's or a relative's reported experiences. It is also possible for patients and relatives to report complaints to an external authority independent from the hospital.

This study included registered adverse events reported by patients, relatives, and healthcare professionals between June 2014 and June 2015. All included reports were based on experiences from eight wards with different medical care specialties. The reports were recorded as closed in the system to be included in this study. This ensured that the whole process was completed in the reporting system. During the period between June 2014 and June 2015, 843 adverse events had been registered: 72% had been reported by registered nurses, 20% by other healthcare professionals, and 8% by patients and relatives. Of these, 134 were registered as closed cases. In this study, 29 reported adverse events were included. We used purposeful selection to ensure that the reports covered variations in healthcare professionals, various types of adverse events, and included patients and relatives. This way, discourses that include all healthcare professions and not only one or only the patients, could be analyzed. The length of the written reports varied between a few lines and several pages. Overall, we analyzed seven reports made by patients and relatives and 22 reports registered by healthcare professionals (registered nurses, n = 8; assistant nurses, n = 7; physicians, n = 6; physiotherapists, n = 1), totaling 29 anonymized reported adverse events. The following types of adverse events were included: adverse drug effect, bladder overfill, burn, fall accident, healthcare associated infection, humane violation, pain, and pressure ulcer.

2.4 Data analysis

Each of the 29 written reports was initially read several times and reread during the analysis process. First, the text was analyzed with a focus on the fact construction—what appeared as normal, taken for granted and true, and from the perspective of whom. These claims of truth were viewed as discursive persuasions, as described by Potter (1996). The analysis of the process of fact construction was thus focused on the following: (1) the use of resources that worked on the identity of the reporter, which increases the reporter's trustworthiness as a reliable source and (2) the use of resources that helped put the reporter in an independent position from the description. Discursive persuasions in the text were recognized in different ways.

A hierarchy model of modalization helped identify how something was expressed to sound trustworthy, real, and even unnegotiable. For example, a description with low modality may sound suspicious (e.g., “the fall accident probably happened very fast since the nurse did not see the patient”) or trustworthy (e.g., “ECG [electrocardiogram] clearly and classically shows Wellens' sign”). The perspective of the author of the written reports was further explored through the analysis of expressions of emotion and words of comparison, revealing the perspective of the reporter that either strengthened or weakened the credibility of the description. Furthermore, category entitlement, that is the credibility of the author, could be captured by the presentation of an adverse event from an individual, a group, or collective, or an institutional perspective. For example, by designating oneself or another as a physician, a mother, or a coworker, the credibility of the description was strengthened or weakened. After this first analysis, an overall analysis was performed to identify the regularities of different resources and the contexts in which they were used. The focus of the analysis at this stage was on rules, discursive and nondiscursive elements, supporting a certain way of talking while excluding another.

2.5 Ethical considerations

All the accounts of written reports were anonymized. To avoid identification of any individual, shortened extracts of the described situations have been cited. The Regional Ethical Review Board in Gothenburg approved this study (Dnr 447-15), and all procedures complied with the Declaration of Helsinki (World Medical Association, 2013).

3 FINDINGS 3.1 Healthcare professionals as experts

The analysis of the reported adverse events showed that healthcare professionals were talked about in a way that made them appear as experts. The category expert is, in this context, viewed in relation to the category layman—where the expert has specific knowledge and where a layman in contrast is understood as illiterate. The expert discourse refers not only to expert knowledge connected to medicine, but also to organizational arrangements including routines, work procedures, and knowledge about when and where healthcare activities take place. In the written reports of adverse events, patient harm was dominated by descriptions of physical harm, which could be observed and measured with the help of techniques and instruments. Further, patient harm was expected to be prevented by following routines of which healthcare professionals had solely insight. Patients' and relatives' part as co-constructors of patient safety, accordingly, appeared to be marginalized by these expectations.

3.2 Healthcare professionals as experts governed by science of medicine The expert discourse was found to be governed by medical science, which appeared to reinforce power to professional objectivity and thus the trustworthiness of the description. The reported adverse events, mostly written by the healthcare professionals, contained technical details of physical signs verifying medical conditions. With the regime of truth that what you can observe is what is considered to be real, healthcare professionals appeared as experts by the use of a repertoire with several discursive resources of distanced objectivity written with a factual modality (e.g., “The patient is bleeding” [registered nurse]). These were commonly supported by precise instrumental variables, dates, times, and quantities, most often in chronological order, leading to a final conclusion.

The patient was admitted 1st of February and had a pacemaker the 3rd of February. The 4th of February, the patient developed a cough and had temperature and rising CRP [C-reactive protein, a marker of inflammation], antibiotic is prescribed. The 9th February, sample for the flu was tested and the result [on] the 10th February was received as positive. Accordingly, the patient caught the flu at the department, which is a hospital-acquired condition. (registered nurse)

Observations of patients' physical conditions were described in detail using medical terminology, which contributed to increasing the credibility of the descriptions and sometimes in knowledge positioning of the person reporting the adverse event. Healthcare professionals' descriptions most often lacked the use of first-person personal pronouns in this context and were written from an external perspective. This made these adverse events appear as if they had just happened without the interference of any person. In using these resources to describe patients' conditions or the progression of adverse events, and presenting medical objective facts as fundamental for assessments, healthcare professionals appeared as sole experts.

Without a sufficient base for making accurate assessments, the modality of the descriptions dropped to a nonfactual modality (e.g., “The patient has probably walked from the table with the walker and lost balance” [assistant nurse]). However, by describing details of the environment as it was presented when the adverse event was discovered with a factual modality, credibility was brought into account and strengthened the assessments.

The patient was found on the floor around 2:00 a.m. It seems like the patient has fallen/stumbled, brought the table with him with the glass of water, which had fallen and soaked the floor. (registered nurse)

The connected suggestions for improvement in the reports reflected this discursive trend of healthcare professionals as experts. These suggestions mostly contained recommendations to support the early detection of risks that aimed to prevent or reduce a problem by identifying abnormal physical signs (e.g., by more “bounteous screening” [registered nurse]), better observation opportunities (e.g., “higher staffing” [assistant nurse]), or improving healthcare professionals' competency (e.g., “how to analyze the ECG” [physician]), all of which contributed to objective interpretations and assessments dedicated to healthcare professionals.

3.3 Healthcare professionals as experts governed by organizational structures The analysis of the written reports showed that the position of healthcare professionals as experts was supported by the organization itself with its routines and work procedures of how, when, and where care was delivered. By organization, we refer to an entity with people working for a common purpose steered by rules governing activities within it, often based on written documents, and of which hierarchical structures allocate accountability (Cooke & Philpin, 2008). Routines and work procedures were constructed as key for healthcare professionals to create continuity of care between work shifts and different healthcare professionals. When workflow was broken or patients harmed, reports focused on measures (or lack thereof) taken by other healthcare professionals. Reports that identified a lack of performance of adequate measures within the clinic, were followed by markers of requirements (e.g., “Report to the next work shift if one don't have time, so that one can help [the patient] immediately” [assistant nurse]). These requirements were more commonly directed to a specific professional, such as a registered nurse or physician, with the amplification of their responsibility in the matter. The critique of coworkers within the clinic was sometimes written with a tone of disapproval or frustration; although, concurrently, healthcare professionals' descriptions revealed carefulness. The following adverse event concerns a woman with pathological levels of troponin-t (a marker for heart muscle injury) in the blood, which was not signed and lacked documentation of assessment and measures taken by the next work shift.

Incoming pathological blood samples should be remediated immediately even if the physician who ordered the sample no longer is on duty. In case the responsible physician assesses the sample as pathologic, but does not change the management, the physician ought to sign the sample and document the result in the patient's medical record together with the assessment and eventual measures. (physician)

The suggestion for improvement was described with a nonfactual modality; that is, as a recommendation. Additionally, the adverse event turned hypothetical, strengthening the intention to be cautious with any admonishment. The situations of shared responsibility to maintain continuity in care contained corrections directed to coworkers that were sometimes sharp and sometimes less pointed. Overall, the analysis showed that reporters of adverse events distanced themselves as contributors to the events in different ways. They (i.e., patients, relatives, and healthcare professionals) were either those affected by others' substandard workmanship or those discovering the adverse events. This suggests that routines are highly desired and expected to maintain patient safety. Not following routines might equal patient harm, a failure to the purpose of the organization, and the ethical code of healthcare.

Another inclusion mechanism of the expert discourse included the healthcare professionals' close attention to time. This was evident in the following citation from a patient's report of an adverse event at the emergency department.

I walk to the reception and said that I have been waiting for three hours and ask if they have forgotten about me. With a little raised voice, another registered nurse turns to me and says, you are lying, you have been waiting for two hours. (patient)

The patient's focus on time appeared to be connected to the perception of being forgotten; that is, from an individual perspective. Conversely, the focus of the registered nurse was on time, and appeared to be based on the accurate assessments connected to life threatening conditions or aspects such as lead times; that is, from an organizational perspective.

3.4 Exclusion mechanisms of the healthcare professionals as experts The exclusion mechanisms of the discourse concerned procedures of prohibition, such as privileges to specific areas in the hospital, rituals for specific occasions, and sharing of opinions by certain people. As in the example above, the registered nurse's perspective of time was presented as a nonnegotiable fact, together with an unfavorable accusation about the patient, eliminating the patients' possibility to express her or his perception of a situation. A too-rigid hold on work routines, for example, made healthcare professionals seem distanced to patients and relatives. This appeared to limit patients' role in the co-construction of patient safety. The exclusion mechanisms were present in different situations, such as in care environment contexts, communication events, or behavioral circumstances, giving power to healthcare professionals as experts. Overall, these situations contributed to healthcare professionals' “distance”—where patients and relatives perceived they were in no position to bring about encounters based on their needs.

We tried to get in contact with the staff; but they sat at the nursing station with closed doors. We had to go home to call the healthcare unit to at all come in contact with anyone…Also, be aware that it does not look good sitting inside the nursing station with a closed door while using one' s private cell phones! It shows! (relative)

These adverse events were presented with factual modality (e.g., “He did not introduce himself” [patient]). Markers of comparison were also commonly used. In this account, the comparison marker in the sentence does not look good showed the value perspective of disapproval, which was strengthened by contradictory attributes to the category staff, such as sitting down and using private cell phones. Staff credibility as healthcare professionals was further diminished by them appearing to be oblivious to their own behavior being a problem. Concurrently, the writer's subjective perspective showed through the use of a first-person personal pronoun (“we”), referring to a group, in this case the family, which added power to the victim of not being the only witness. However, returning home to get in contact with healthcare professionals can be viewed as a response, or appellation to the position of being a subject (a relative) not having access to just any space in the hospital or to the healthcare professionals' attention when needed. A specific aspect of these descriptions was that they were constructed with a discursive repertoire connected to reactions to a need for support that was not met. Healthcare professionals were described as dismissive, anonymous, unprepared, or unengaged—all of which created a distance to patients and relatives. Symbolically, the closed door described by the relative above reoccurred as a phenomenon in the descriptions and took different shapes.

After (a first encounter and assessment), the registered nurse asked me to go outside again, to the waiting room outside the emergency department and not to the other patients inside the emergency department. Why was I different from other care-seeking patients? (patient)

The exact rooms that were defined as part of the inside or outside of the ED may be irrelevant, although the perception of not being included in care was significant. Healthcare professionals' presence was described as important and presented as something more than physical attendance. It was about being present with a name, being prepared for encounters (e.g., being familiar with test results), communicating information about assessments and care plans, and making room for encounters despite routine obligations.

4 DISCUSSION

The findings presented a discourse about healthcare professionals as experts in written adverse events as reported by patients, relatives, and healthcare professionals. The unnegotiability of organizational and medical truth regimes was found to bring power to the expert discourse and at times seemed to contribute to inflexibility and depersonalization of patients, relatives, and healthcare professionals. We posit that expert discourse, as constructed in adverse events, suppressed alternative ways to talk about adverse events and, in turn, how patient safety can be understood and achieved. At the same time, the discourse seemed to limit the patients' and relatives' potential to act as experts and subsequently their roles as co-constructors of patient safety.

The focus on adverse events as something dominantly physical has similarly been presented by other researchers and has been shown to be a unifying matter across interprofessional teams (Rowland & Kitto, 2014). Objective assessments of the body based on instrumental variables made by healthcare professionals has been described as “the sovereignty of the gaze” (Foucault, 1989, p. 89) to know, decide, and govern, and is a knowledge production that has been reproduced over time. The rationality of what is knowable and true has been described to characterize today's science—not least the science of medicine—which offers related techniques that help fulfill the purpose of hospitals (Howarth, 2000; Wright, 2000). Without knowledge of the accuracy in assessing physical signs or the significance of time, the chances for survival, for example, after cardiac arrest, would rapidly drop (Hessulf et al., 2020). Healthcare professionals' knowledge of using these techniques, for example, how to monitor the heart and analyze the ECG or control the blood pressure and how to understand these numbers, is thus crucial for patient safety and is a knowledge production that was evident in the analysis of adverse events. Abidance to routines and work procedures was also found to be a part of the knowledge production. Such a dominant system perspective within healthcare and the trust in it as a key to patient safety—for example, by well-functioning work procedures and the healthcare professionals' vigilance of these—has earlier been discussed in patient safety literature. A benefit of routinization might, for example, be that accumulated knowledge within an organization can be exploited by healthcare professionals who do not necessarily have knowledge of all the underpinnings of different procedures (Hollnagel et al., 2013). A modern technicalized system with its professional expertise as this, can be understood as an expert system. According to Giddens (1990), the expert system, in a fundamental way, builds on both the ones representing the expert system and the laymen's trust in abstract capabilities. In this case, the trust in that the healthcare is organized and works in a way that ensures patients' safety.

Based on our findings, we would like to stress that social construction of the meaning of patient safety goes beyond standardization, technicalities, and intellectualization. Waring (2009) found that healthcare professionals tried to establish unambiguous and authoritative explanations through often linear and technical narratives of professional roles that reinforced professional competence in written reports of adverse events. We acknowledge that narratives about adverse events intended to be processed in reporting systems are reconstructed in ways that deflect notions of blame and reinforce boundaries of professional responsibilities, as argued by Waring (2009). However, we also posit the power of the expert discourse risks excluding alternative ways to understand and think about possible patient safety actions. This assumption is strengthened by Hågensen et al. (2018), who argued that professional knowledge and hospital environment gave discursive power to healthcare professionals to define any patient situation and to decide on what actions to take when patients suffered from adverse events. In accordance with our findings, Hor et al. (2013) showed that the ignorance of patients' contribution rather provoked, for example, frustration, anger, pain, and acute illness. The resulting distance between the healthcare professionals and the patients and relatives, might, to some extent, be explained by Giddens's (1990) view of the expert system's disembedding mechanisms of social relationships, for example, by impersonal instruments and objectiveness and the pervaded trust in the system. Access points, that is, the encounters of engagement between laymen and experts, are described as important to preserve trust (Giddens, 1990). Healthcare professionals' unavailability or anonymity, as found in the reported adverse events, appeared to diminish these important access points and consequently the trust, both between individuals and to the expert system. Concurrently, the power imbalances that the expert discourse created from time to time breached the norms of politeness when the engagement in patients' needs dropped. For patients, relatives, and healthcare professionals to have the opportunity to co-construct safe care in alternative ways, there needs to be a shift in power toward the recognition of lay peoples' experiential knowledge and the significance of mutual trust through engagement encounters.

4.1 Limitations

The findings of this study were based on expressions of how different interpretations and perspectives and ways of thinking were shaped and structured a part of social reality. Therefore, these findings should not be misinterpreted as a mirror of reality transferable to any context. Similar to other discourses, the meaning of specific discourses connected to patient safety is, from a Foucauldian perspective (Foucault, 2010), dependent on the social practice where specific discourses are legitimized in a certain way and change over time depending on historical context. The present analysis, in which the repertoire of resources for making a statement or expression seems trustworthy and real, can be viewed as having a stronger connection to action than to language in comparison to Foucault (Herz & Johansson, 2013). From an action-oriented perspective, where actions can be viewed as being performed with intention or willingly (e.g., to solve a problem), it is not possible to determine whether the identified resources through language were used in a conscious way.

From the perspective that the reported adverse events were communicative events produced and consumed in a discursive practice, where discourses constituted and constituted by these communicative events, as described by Fairclough (2010), this study might have benefited from data drawn from the whole process. For example, complementary interviews to reveal how the patient, relative, or healthcare professional initially recognized an event as an adverse event, or how reported adverse events were understood and converted into patient safety actions. Such information might have shed light on the whole process of the production and consumption of discourses.

Another aspect is that these findings reflected the context of unsafe care where events were identified to harm the patient. From the perspective of how adverse events are currently defined (WHO, 2005), it is important to clarify that most patients will not experience adverse events during hospitalization. Although adverse events do not affect most patients, adverse events may occur at any time in healthcare; and more research is needed that focuses on preconditions for healthcare professionals to provide safe care, together with the patients and relatives as co-constructors.

5 CONCLUSION

We posit that the apparatus of the hospital supported and brought power to the discourse of healthcare professionals as experts. This knowledge production through reports of the adverse events appeared to be limited to the understanding of patient harm as something that needed to be verified through objective physical signs—a regime of truth that healthcare professionals were expected to have knowledge of. The aspiration to rely on physical signs and comply with routines and work procedures appeared to create a distance to patients and relatives. This sometimes unintendedly and overly bureaucratic approach contributed to inflexibility and depersonalization of patients, relatives, and healthcare professionals. Consequently, the discourse of healthcare professionals as experts seemed to limit patients' and relatives' potential to act as experts and subsequently hindered their roles as co-constructors of patient safety. The desire to standardize healthcare, follow routines, and focus on physical signs might seem to be a neutral and unproblematic way toward patient safety, bringing even more objectivity into care and a possibility to treat all patients equally. We suggest that this regime of truth might also suppress other ways to understand and talk about adverse events and actions toward patient safety. Social change is a slow process, however, and, if patients and their families are to be enabled as co-constructors of patient safety, one has to consider how existing structures and processes support or suppress their roles. The terms for patient engagement can be changed by giving meaning to adverse events as something more than merely a physical harm, acknowledging that they constitute a process involving meaning construction and actions. Viewing both patients', relatives', and healthcare professionals' experiences, knowledge about themselves and their situation as equally important, the knowledge production of patient safety might take new directions.

ACKNOWLEDGMENTS

We are grateful to The Swedish Health Care Facilities Network and the foundation of Hjalmar Svensson who have supported this study by grants. We also thank Wiley Editing Services (https://wileyeditingservices.com/en/) for editing a draft of this manuscript.

CONFLICT OF INTERESTS

The authors declare that there are no conflict of interests.

AUTHOR CONTRIBUTIONS

All designated as authors met the criteria for authorship. Anna Gyberg, Ingela Henoch, and Kerstin Ulin initiated the study and contributed to the research design. Anna Gyberg contributed to the data collection and the majority of the data analysis and drafting of the manuscript, to which all authors contributed with interpretations and critical revision of important content. All authors approved the final version to be published and agreed to be accountable for all aspects of the work. Kerstin Ulin was the study supervisor and responsible for critical revision and supervision.