We identified three sub-themes categorising resilient practices for patients hospitalised with suicidal behaviour and HCPs: interactions capturing non-verbal cues, protection through dignity and watchfulness, and personalised approaches to alleviate emotional pressure. All three are grounded in relationships of trust in resilient practices for patients in suicidal crisis. These resilient practices outline a safety-II perspective in inpatient care of suicidal patients, which is based on culturally situated knowledge which cannot be generalised to all inpatient settings (57). However, theoretical conceptualisation can be made regarding the understanding of safety for the suicidal patient, from a resilience perspective (57). Above all, the perspective of resilience provides insight into the system complexity of suicide prevention, and how patients and HCPs cope with challenges and changes (28, 58).
4.1 The complexity of clinical practice for patients hospitalised with suicidal behaviour
The shared understanding of resilient practices offers insight into clinical practice for patients hospitalised with suicidal behaviour. This clinical practice has the characteristics of a CAS (27, 28, 59, 60), which has been previously presented in the first author’s PhD thesis (49).
Interactions capturing non-verbal cues of suicidal inpatients reflect uncertainty in clinical decision-making. The findings correspond with previous studies suggesting that caring for suicidal patients involves uncertainty (61–63) and expand the literature by showing that making sense of suicidal behaviour requires the comprehension of information ‘beyond the spoken word’, using intuition along with other sources of information (e.g. diagnosis, mental state, medical journal, individual risk factors) (64, 65). To attempt anticipating suicide, and to adapt and respond to the suicidal crisis of patients in the wards, HCPs apply multiple sensemaking strategies to improve their situational awareness (66, 67). The complexity and ambiguity of the cues involved in making sense of suicidal behaviour highlight that it is difficult to standardise screening questions or limit a set of cues to look for (proximal risk factors) in the setting of inpatient suicide prevention with high specificity (68). Furthermore, suicide prediction models are imprecise (21), the evidence does not support the use of risk scales in suicide risk assessment (20, 69), and there is no support for the use of clinical intuition as the sole source of suicide prediction, which leaves HCPs with limited support for their clinical decision making. This study supports that ‘work as done’ of suicide risk assessments involves multiple strategies to make sense of uncertainty, which is a characteristic of decision making in complex high-risk and ambiguous work settings (66, 70–72).
Protection through dignity and watchfulness, and personalised approaches to alleviate emotional pressure emphasise that there is no one-size-fits-all approach to the protection and treatment of suicidal patients. Instead, practice relies on interpersonal and context-sensitive adaptations to ensure protection is safe (28, 30, 31, 73). Furthermore, clinical practice for suicidal patients is characterised by multiple conflicting goals which must be resolved by making trade-offs between higher- and lower-level goals (74), e.g., preventing immediate harm and working towards long-term health goals. Dynamic interactions and adaptations to ensure safe work are characteristics of a complex adaptive system (27, 28, 59, 60).
Relationships of trust in resilient practices for patients in suicidal crisis reflect that the complexity in clinical practice is characterised by the establishment of psychological and relational safety, which is only created through personalised and trusted relationships. Personalised relationships are the core characteristic of mental healthcare as a complex adaptive systems (31).
In a CAS there will be unpredictable consequences of standardised patient safety interventions due to the need for adaptations in clinical practice to ensure safe outcomes. For those designing patient safety interventions at the blunt end of the system, deviations from the procedure may be perceived as an error to be corrected (6). However, while errors arise from approximate adjustments of procedures, it is also why everyday work in complex adaptive systems are safe (6). To improve safety in complex adaptive systems, variability should be embraced, not erased (75).
The CAS perspective of clinical practice in suicide prevention has implications for learning. To learn from variability, rather than eliminate variability, it is necessary to move from simple linear models (e.g., root cause analysis) to systemic models (e.g., resilience, CAS). These models focus on why practices vary, succeed or fail at the clinical level, hospital management level and at the health system macro level (9, 30, 73).
4.2 Supporting resilient practices
This study describes resilient practices which outline complexity in the clinical work in suicidal inpatient care. Some generic lessons regarding how to approach patient safety and suicide prevention at the micro level of the system are drawn based on the RHC perspective.
In a CAS, it is impossible for the HCPs to anticipate all of the consequences of adjustments of procedures and trade-offs (76, 77). Feedback systems are needed to acquire knowledge of how to adapt and its outcomes (78). This implies that, systems in mental healthcare need to collect feedback on everyday clinical practice to learn what fosters success under various conditions (Safety-II), not merely from retrospective views of suicide incidents (6, 9, 14). Suicidal patients provide valuable feedback on what makes them feel safe or unsafe, identify conditions for successful treatment, and those that may cause adverse events. All of this corresponds to findings in other domains of healthcare (79, 80). Collecting patient feedback should be considered essential to support resilient practices for suicidal patients.
Furthermore, suicidal patients’ experiences should inform the development of patient safety measures. This study finding corresponds with the literature that identifies trust as fundamental for hospitalised patients’ feelings of safety (81, 82). The relational component of patient safety is the most vital aspect of care from the patients’ perspective, and HCPs can strengthen the sense of safety in those patients by respecting them as human beings, validating their feelings and ensuring that they know that people care about them. A sense of safety is also linked to a sense of control. HCPs may strengthen patients’ sense of control by addressing underlying issues and mental illnesses during hospitalisation and adapting suicide risk assessments and therapeutic approaches to suicidal behavior to meet each patient’s needs. An intervention which may support resilient practices is the collaborative assessment and management of suicide risk (CAMS) that involves the exploration of the suicidal patients’ individual drivers, warnings signs along with addressing the patient’s pain and suffering (83, 84). Furthermore, dialectical behavior therapy, which aims to improve patients’ emotional and practical coping skills are relevant (85, 86).
This study implies that HCPs apply sensemaking strategies to improve their situational awareness of suicidal behavior. These strategies are enacted to gain a fuller meaning of the information they have obtained and a sense of what is going on in complex and ill-defined situations (71, 72, 87). Since obtaining feedback from the healthcare team is essential in the creation of situational awareness (70–72), training in suicide risk detection can benefit from multidisciplinary training involving HCPs who regularly interact as a team to establish a shared vision, values and mental models (88, 89). HCPs need to be able to discuss their clinical judgement in everyday clinical practice with a team of colleagues (71, 72). In dialectical behavior therapy, and evidence-based treatment for borderline personality disorder with a specific focus on suicidal behavior, discussion of clinical assessment with colleagues is an integral part of the intervention (85, 86). Furthermore, the results indicate that opening a collaborative dialogue with the patient may also help HCPs’ sensemaking of suicidal behavior (83). In this regard, sensemaking is concerned with supporting strategies that create a more comprehensible understanding that enables action (90). Future studies are needed to develop complex interventions to improve HCPs’ shared situational awareness, non-verbal communication and relational skills in the patient contact, and may include the development of simulation training interventions for suicide risk assessment (91).
Following the literature, this study highlights the importance of having experienced HCPs (92), who are fully therapeutically engaged with the patient (93, 94) who balance the exertion of control and the building of the therapeutic relationship during observation (95). Observation of patients at suicide risk is a resilient practice that entails watchfulness and sensitivity to cues of a patient’s mental state along with engaging with the patient and establish bonds of trust. The findings are in accordance with research depicting prevention of suicides in constant observation as intertwined with forming connections and regaining hope (93, 96, 97). The finding supports that observation demands resilient strategies from the individual HCPs and required adaptations and the balancing of multiple goals in care (28, 30, 73, 98).
Lastly, this study implies that resilient practice should not rely solely on HCPs’ capacity to adapt without formal support systems. Without reliable sources of trust, the system is brittle. HCPs’ adaptive capacities may be overstretched and threaten the adaptive capacity of the system (99–101). The findings also follow the literature finding that HCPs may distance themselves from suicidal patients’ emotions to protect themselves from emotional discomfort which may erode patients’ trust in their HCPs (102–104). HCPs need to know they are supported, and thus they favour formal arenas for collegial trust, support (61, 105), supervision and training to ensure they can keep working with suicidal patients (62, 93). Support systems are currently not considered a vital part of clinical guidelines for inpatient suicide prevention (10). However, to support resilient practice, HCPs’ capacity to handle daily stress needs to be nurtured after incidents and on a daily basis. Clinical guidelines and patient safety policies related to inpatient suicide prevention need to include psychosocial support system and clinical supervision of HCPs.
4.3 Strengths and limitations
The ability to judge the quality of qualitative research rests on four characteristics: credibility, dependability, confirmability and transferability (106). Triangulation of data sources (patients and professional groups), methods (focus groups, individual interviews and systematic review of literature) and the use of several researchers enhance the credibility of this synthesis: the confidence in the accuracy of the data and ensures that the research investigates what it intended to investigate. Credibility was strengthened by including a sample with sufficient information power (107), that covered significant variations and had relevant experiences with the phenomenon under study (108). A sample size of 18 participants was considered adequate to ensure such information power when studying a heterogenic group of patients with suicidal behaviour (107). A sample size of 18 and 25 HCPs was considered adequate to ensure variability across care settings (locked/open wards), diverse specialities (psychologists, nurses, medical doctors), gender, experience, expertise and patient diagnosis (107).
Dependability is strengthened in this synthesis by the provision of clear, detailed descriptions of all procedures and methods in the first author’s PhD thesis (49), the original studies (41–43), and by providing transparency through the published protocol (48), This allows for appraisal without the need to arrive at the same results (109).
Confirmability is the degree of neutrality and researcher bias (106). Confirmability was strengthened through the sharing of the researchers’ backgrounds, preconceptions and pre-understandings to interpret the data (110). This study adopted an inductive approach to analysis and curiosity about the experienced reality to describe resilient practices in this context, which reduces researcher bias through approaching data with sensitivity and openness (47).
Transferability is the extent to which the findings can be transferred to other settings, context or groups (109). Qualitative data produce culturally situated knowledge which cannot be generalised to all inpatient practice settings (57). The resilient practices described here are part of the processes of clinical practice, not linked to specific outcome measures (e.g., suicidal behaviour, symptoms). We therefore cannot draw conclusions on the effect of such practices related to suicidal behaviour and they should not replace current evidence-based system approaches to suicide prevention (e.g., 11–13, 22, 111, 112–115) However, the findings of this study can be conceptualised at a theoretical level and used to arrive at a deeper insight into the ontology of safety for suicidal patients (116). This study focused on hospitalised patients who survived a suicidal crisis and resilient practices in mental health wards, as experienced by these patients and their HCPs. As such, the study conclusions do not pertain to patients dying from suicide or to patients who were not admitted to hospital wards during their suicidal crisis.