A qualitative exploration of mental health service user and carer perspectives on safety issues in UK mental health services

Abstract Background Service user and carer perspectives on safety issues in mental health services are not well known and may be important in preventing and reducing harm. The development of the Yorkshire Contributory Factors Framework—Mental Health (YCFF‐MH) provides a broad structure within which to explore these perspectives. Objective To explore what service users of mental health services and their carers consider to be safety issues. Design, setting and participants Qualitative interviews with 13 service users and 7 carers in the UK. Participants were asked about their experiences and perceptions of safety within mental health services. Perceived safety issues were identified using framework analysis, guided by the YCFF‐MH. Results Service users and carers identified a broad range of safety issues. These were categorized under ‘safety culture’ and included psychological concepts of safety and raising concerns; ‘social environment’ involved threatened violence and sexual abuse; ‘individual service user and staff factors’ dominated by not being listened to; ‘management of staff and staffing levels’ resulting in poor continuity of care; and ‘service process’ typified by difficulty accessing services during a crisis. Several examples of ‘active failures’ were also described. Discussion and conclusions Safety issues appear broader than those recorded and reported by health services and inspectorates. Many safety issues have also been identified in other care settings supporting the notion that there are overlaps between service users and carers’ perspectives of safety in mental health services and those of users in other settings. Areas for further research are suggested.


| BACKG ROU N D
Patients can be harmed while receiving health care, 1

and evidence
suggests that between 3% and 36% of admissions to hospital result in an adverse event (eg a medication error) and up to half of these events are thought to be preventable. 2 The majority of patient safety research has focussed on general hospital settings but has begun to include community and primary care. [3][4][5] However, it is widely acknowledged that patient safety in a mental health context has received less research attention. 6,7 Over the last decade, evidence suggests that systems can be designed in general hospitals that collect feedback from patients on the safety of their care. 8,9 This research has shown that patients are willing and able to provide information about the safety of their care.
This can help to identify issues not necessarily recognized by services or regulators and has the potential to inform interventions to improve safety and prevent harms and adverse events. [10][11][12] Research has also explored the potential for patient involvement in patient safety within primary care 13 and mental health services. 14,15 Given that improving safety is a priority in the delivery of mental health services, including the views of service users and their family and carers about the safety of their care is paramount. 15 Mental health services following regulatory scrutiny by the Care Quality Commission (CQC) often need to improve the safety of care.
In the UK in 2017, the CQC considered over a third of mental health services deficient in terms of safety. 16 Of particular concern were sexual safety 17 and the use of restrictive practices such as restraint and seclusion. 18 Mental health services span both hospital and community settings and have significant differences to general care; for example, staff-completed incident reports are dominated by violence and self-harm. 19 Harms associated with mental health services might not necessarily be caused by treatment error but by iatrogenic harms caused by medication side-effects, the use of restraint, seclusion, forced medication or even diagnoses leading to exclusion from services. 20

| Theoretical framework
The Yorkshire Contributory Factors Framework (YCFF) was systematically developed to account for factors contributing to safety incidents in general hospital care and was subsequently amended for mental health services. 21,24 The resulting Yorkshire Contributory  Figure 1). Active failures include mistakes, slips, lapses and violations. 25 The YCFF-MH is intended for use to support the development of interventions to promote safety within mental health services.
The aim of this study was to explore service user and carer experiences of safety issues across mental health services using the YCFF-MH as a theoretical basis in order to inform the development of interventions to improve the safety of mental health services.

| ME THODS
This study followed on from a survey study which resulted in the adapted YCFF-MH. 21 Given the limitations of a survey approach, we devised a qualitative study to generate a richer picture of the key safety issues and further contextualize the adapted framework. A qualitative study permitted exploration of the topic of safety. We sought to conduct semi-structured interviews with a convenience sample of UK mental health service users and carers, recruited via social media (reported elsewhere 21 ). Social media enabled recruitment of those with a breadth of experiences across many different mental health services and organizations. Service users, carers and health professionals aged over 18 with experience of contact with UK mental health services in the past two years were invited to take part in one telephone interview from their homes to discuss safety issues in services. Interviewees were assumed to be alone during the interview although as they could not be seen this may not have been the case. Due to the rich data generated, we present solely the service and carer findings in this paper. Consent forms and information sheets were emailed to people who expressed an interest. The aim of the study was reiterated prior to the interview, and verbal consent was taken and audio-recorded. Interviewees received a £10 shopping voucher as a token of appreciation for taking part. Forty-seven people expressed an interest to take part, 24 responded to invitation with 20 interviews conducted (13 service users, 7 carers-see

| Data collection
A broad interview guide was developed informed by an earlier study on the same topic. 21 Additions were made to the guide based on early interviews, for example the issue of service user articulacy was raised by one interviewee and subsequently was raised with other

| Data analysis
Initial data analysis was concurrent with data collection. Interview data were analysed using framework analysis 26 which is recognized as a useful approach when multiple researchers are working on a project and for managing large data sets which aim to generate a descriptive overview. 27 We used a combined deductive and inductive approach. Deductive in so far as the exploration of how the data mapped to the YCFF-MH framework, but also inductive as additional codes and sub-codes were generated for data which did not 'fit' within the framework, these data were coded openly. The inductive approach allowed us to further understand a particular contributory factor in the context of mental health services at a much deeper level. For example, within the 'safety culture' coding we were able to tease out more specific coding around psychological and physical safety and raising concerns.
The analysis was supported by NVivo. 28 Authors KB, GL and AA familiarized themselves with the transcripts prior to coding, and data could be coded onto more than one YCFF-MH factor. Summaries of main points of each transcript were shared and discussed amongst the team before detailed coding was carried out on all transcripts by KB, with 20% of transcripts also coded by GL and AA. Any discrepancies were resolved through discussion. Data coded to the most populated factors were further assigned to subcategories to give more understanding and context (see Figure 2). Data were explored in-depth by the three researchers during intense analysis meetings 29 ; at this point, it was decided to combine the carer and service user data as analysis showed there was consensus amongst most of the issues discussed. During analysis, although we experienced 'code saturation' with no new codes being added, we did not consider 'meaning saturation' to have occurred due to the broad nature of the questioning and the diversity of participants experience. 30 All the researchers were experienced academic researchers educated to PhD level, KB (female) and JB (male) research mental health services and GL and AA (both female) research patient safety.

| Patient involvement in the design and conduct of the study
The focus for this research arose from general social media discussions involving authors JB and KB, who participate in social media debate with a broad range of people in their networks, including (ex) service users, family members, carers and professionals in a range of roles about improving and understanding key safety issues in mental health services. Participants were invited to consent to contact for this study as part of a larger survey; this is detailed elsewhere. 21 For the purpose of member checking, a draft of the findings was sent to interviewees to give them the opportunity to comment. Several interviewees responded to say that the findings represented their views accurately. Some interviewees provided further comments that extended beyond the scope of this paper but were useful in terms of future research.

| FINDING S
Data about safety issues related to 13 of the 20 YCFF-MH factors.
The subthemes of most frequently coded YCFF-MH factors are shown in Figure 2. In the following sections, we provide a descriptive account of the nature of these most prominent factors and illustrative excerpts. Other factors did have some coding, for example communication systems, but are not reported.

| Safety culture
The definition of positive safety culture within the YCFF is one where organizational values, beliefs and practices support the management of safety and learning from error. 24 Learning from error is affected by how comfortable service users and carers feel about raising safety concerns. This section presents data which were coded according to the original YCFF definition.

| Psychological vs physical safety
Safety was conceptualized differently between service users and carers. Carers were primarily concerned with services ensuring physical safety, to prevent self-harm and suicide. Service users described safety in services from two distinct perspectivesː physical safety (including seeking help with managing self-harm or protection from other service users) and psychological safety (experiences within services leading to fear and distress). The two were intertwined as treatment intending to prevent physical harm, that is prevent self-harm When a concern had been raised, there was often no satisfactory response: I complained on the ward at the time… it was very defensive. I never heard anything about it, it didn't get followed up, as far as I heard.

Male carer ID05
Others felt there might be repercussions as a result of raising a concern: My advice would be don't complain because it will make things worse, and actually unless it's actively really harming you, you're better off to just not mention it, you're better off to just suffer it.
Female service user ID13

| Active failures
Active failures include mistakes, lapses and violations. Several active failures were described although they had few common themes.
Examples included sending a confidential letter to a former address of a service user who had left there due to domestic abuse; warnings from carers with regard to antecedents to absconding and self-harm being ignored; a Mental Health Act detention not been completed following procedure; and home visits not occurring as arranged.
Experiences of violations ranged from ward staff speaking to service users in a disrespectful or taunting manner to sexual assaults:

| Situational factors
Situational factors include those relating to individual staff and service users, as well as the social environment on a ward.

| Local working conditions
Local working conditions include supervision and leadership, lines of responsibility and management of staff and staffing levels. It was the latter that was most salient to interviewees.

| Latent organizational factors
Latent organizational factors included service process: the way in which people gain access to, transition between or, are discharged from services. Participants talked almost entirely in terms of gaining access to services.

| Service process
The majority of discussion about service process related to access-  There has never been a state-level recognition of harm caused by mental health services. Commentators have advocated the use of truth and reconciliation processes as a means of addressing this deficit 33 ; however, the lack of state sponsorship, engagement of professionals most in need of change and retributive justice and legal reforms make them unlikely to be successful. 34 Nevertheless, restorative justice models have been successfully applied in health services 35 and even if not adopted systemically, have features (eg independent facilitation, acknowledgement of harm caused, reparation) that could potentially improve safety culture. 36 Not defining safety for participants allowed them to relate it to their experiences of feeling unsafe as a result of contact with services. There were consistencies between our findings and previous safety research about mental health 21 and primary care services. 3,4 Some interviewees would avoid particular services, for instance a crisis service, to protect their psychological safety and because they feared a hostile or dismissive response. 3 In other care settings, patients have the right to choose between care providers and can avoid certain individuals or services. 3,4 For mental health service users however, this autonomy often cannot be exercised due to lack of treatment providers and threat of compulsory treatment, thus making it even more important that if autonomy is overridden, safety should be assured. The trauma of experiencing restrictive practices was graphically described by participants. These practices have been found to persist, despite being legally determined as a last resort. 16 The barriers faced by service users and carers to raising concerns, having them taken seriously and responded to, echoed previous research 14 and continue to be a thread running through health-care service failures. 23,37 Some of the active failures described by interviewees may not have been recorded by staff on incident reporting systems.

| D ISCUSS I ON
Procedural errors, assault by other patients and communication failures might be recorded by staff; however, experiences of verbal abuse from staff would require a complaint to be made, and this has been shown to be difficult. 14 Being spoken rudely has been found to cause psychological harm to people in primary care. 4  There was a considerable degree of overlap between the issues identified in this data and those previously identified in primary care service users' perceptions of safety. 4 These parallels reinforce that safety in mental health services should not be treated as a separate domain but, as has been previously advocated, as an integrated part of the discipline of patient safety. 6

| Recommendations for practice
Services should explore the use of the YCFF-MH in collecting and reporting quality and safety data.
Service user and carer feedback should be proactively sought, and it should be made easier for service users to raise concerns and make complaints.
Mental health services must acknowledge when they have caused harm to service users and carers; lessons can be learned from restorative justice.

| Limitations
This was a small-scale, exploratory study although the findings overlap with those reported from primary care. Our codes reached saturation but we did not consider the study to have achieved 'meaning saturation', that is more data, particularly from carers would enrich some factors that did not have many codes associated with them. 25 Interviewees were recruited from an online survey about safety issues, and this may mean that those who had negative experiences were more likely to participate. However, people who have not had these experiences may have less to contribute. 43 There are many people in contact with mental health services who would not have been included in this research; using social media to publicize the study limited recruitment to users of such platforms. Different response may have emerged by recruiting directly from NHS mental health services or via alternative recruitment mechanisms which did not rely on social media.

| Implications for future research
This study highlights the following key factors from the service user and carer perspective that are particularly important in terms

| CON CLUS ION
This study shows that service users and carers consider there to be a broad range of safety issues associated with mental health services.
Interviewees described predominantly psychological harm caused not only by treatment but the behaviour of other service users, within the context of services that are understaffed and difficult to access. Patient safety has been defined as 'the prevention of harm to patients', and the discipline of patient safety has been defined as 'the coordinated efforts to prevent harm, caused by the process of health care itself, from occurring to patients.' 30 Our findings reinforce that in mental health services the definition could usefully be expanded to include harm caused when trying to access services and self-harm provoked by contact with, or rejection from services. Efforts to improve safety in mental health services from policy level downwards, particularly accessing help in a crisis, should be underpinned by research evidence reporting the harms experienced by service user and carers; their concerns need to be central to ensure that the narrow general health service conception of safety does not continue to dominate.

E TH I C A L A PPROVA L
Ethical approval (reference:12/LO/1588) was obtained from the University of [redacted] School of Healthcare Ethics Committee.

ACK N OWLED G EM ENTS
We would like to thank all the people who participated in this study by sharing their views and experiences.

CO N FLI C T O F I NTE R E S T
The authors declare they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared due to privacy or ethical restrictions.