The use and impact of surveillance-based technology initiatives in inpatient and acute mental health settings: a systematic review

Abstract

Background

The use of surveillance technologies is becoming increasingly common in inpatient mental health settings, commonly justified as efforts to improve safety and cost-effectiveness. However, their use has been questioned in light of limited research conducted and the sensitivities, ethical concerns and potential harms of surveillance. This systematic review aims to (1) map how surveillance technologies have been employed in inpatient mental health settings, (2) explore how they are experienced by patients, staff and carers and (3) examine evidence regarding their impact.

Methods

We searched five academic databases (Embase, MEDLINE, PsycInfo, PubMed and Scopus), one grey literature database (HMIC) and two pre-print servers (medRxiv and PsyArXiv) to identify relevant papers published up to 19/09/2024. We also conducted backwards and forwards citation tracking and contacted experts to identify relevant literature. The Mixed Methods Appraisal Tool assessed quality. Data were synthesised narratively.

Results

Thirty-two studies met the inclusion criteria. They reported on CCTV/video monitoring (n = 13), Vision-Based Patient Monitoring and Management (n = 9), body-worn cameras (n = 6), GPS electronic monitoring (n = 2) and wearable sensors (n = 2). Sixteen papers (50.0%) were low quality, five (15.6%) medium quality and eleven (34.4%) high quality. Nine studies (28.1%) declared a conflict of interest. Qualitative findings indicate patient, staff and carer views of surveillance technologies are mixed and complex. Quantitative findings regarding the impact of surveillance on outcomes such as self-harm, violence, aggression, care quality and cost-effectiveness were inconsistent or weak.

Conclusions

There is currently insufficient evidence to suggest that surveillance technologies in inpatient mental health settings are achieving their intended outcomes, such as improving safety and reducing costs. The studies were generally of low methodological quality, lacked lived experience involvement, and a substantial proportion (28.1%) declared conflicts of interest. Further independent coproduced research is needed to more comprehensively evaluate the impact of surveillance technologies in inpatient settings. If they are to be implemented, all key stakeholders should be engaged in the development of policies, procedures and best practice guidance to regulate their use, prioritising patients’ perspectives.

Background

Inpatient mental health settings are challenging environments, both for those receiving and those delivering mental healthcare. The core purpose of inpatient wards is to provide a physically and psychologically safe place for people experiencing acute mental health difficulties to recover and receive care; however, both patients and staff have reported feeling unsafe on wards [1,2,3]. Inpatient mental health patients report (re)traumatising experiences including abuse, coercion, aggression and violence on wards from staff and/or other patients [4,5,6,7,8]. Staff also report abuse and violence on the wards from patients [9, 10], as well as having to risk-assess for and respond to incidents of self-harm and suicide attempts, which are prevalent in these settings [11].

In this context, some mental health service providers in the UK are increasing their use of surveillance-based technologies in inpatient settings [12]. Such surveillance technologies include closed circuit television (CCTV), body-worn cameras (BWCs) and remote monitoring devices (such as smart watches, Global Positioning System (GPS) trackers and infrared cameras). Use of these technologies is justified on the basis that they may be able to detect or prevent aggressive and violent incidents, reduce self-harm incidents and suicide attempts, improve staff and patient safety, change patient behaviour and staff conduct, provide accurate records to help resolve complaints and to contribute to legal cases, and reduce staffing costs [13,14,15,16,17]. Conflict and providing adequate staffing on wards are costly [18] but interrelated [18, 19]. Reducing cost is a driving force for many service providers; therefore, surveillance technologies may appear to offer a cost-effective solution.

The use of video technologies implemented with the stated purpose of improving security is becoming increasingly common. For example, in the UK, BWCs are now used by the police [20], emergency healthcare workers including paramedics [21,22,23], and retail staff [24,25,26]. However, the use of some of these technologies on inpatient wards is controversial [27, 28]. Patient and service user groups, as well as advocates and disability rights activists, have consistently called for scrutiny of these technologies regarding potential risks of iatrogenic harm and ethical concerns [29, 30]. For example, issues raised by the Stop Oxevision campaign [31] include (i) ethical considerations around use of surveillance technologies and obtaining informed consent (for example, concerns about the ability of services to provide adequate information for informed consent, potential consequences for patients not providing or withdrawing consent, and whether consent can reasonably be given to being filmed or recorded whilst acutely unwell on an inpatient ward), (ii) concerns about data access, storage, security and human rights violations, (iii) distress caused by being recorded or monitored, or the exacerbation of existing paranoia, trauma or distress [14,15,16,17, 32] and (iv) fears that it could result in reductions in staffing and one-to-one contact between staff and patients on wards.

In order to plan effective and safe mental health service delivery, it is important to determine whether evidence supports the use of surveillance technologies. Both potential benefits and harms should be considered, including impacts on outcomes such as safety, care quality, mental health and treatment satisfaction. However, a comprehensive review of the evidence underpinning the use of surveillance technologies in inpatient settings has not yet been undertaken. Therefore, we conducted, to our knowledge, the first systematic review of a range of surveillance technologies in inpatient mental health settings.

Both quantitative and qualitative evidence is synthesised to answer the following overarching research question: how are surveillance-based technology initiatives being used and implemented in inpatient mental healthcare settings, and what is their impact? Our specific three research objectives were as follows:

• (1) How are surveillance-based technologies in inpatient mental health settings being implemented and what are the related implementation outcomes?

• (2) How are surveillance-based technologies in inpatient mental health settings experienced (e.g. by patients, staff, carers, visitors)?

• (3) What is the effect, including benefits, harms and unintended consequences, of surveillance-based technologies in inpatient mental health settings for outcomes such as patient and staff safety and patient clinical improvement?

Methods

We conducted a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [33]. The PRISMA checklist can be found in Appendix A [see Additional File 1]. The protocol for our review was registered with PROSPERO (CRD42023463993). Amendments to this protocol are described in Appendix B, with justifications [see Additional File 1].

This review was conducted by the National Institute for Health and Care Research (NIHR) Policy Research Unit in Mental Health (MHPRU) based at King’s College London and University College London, which conducts research in response to policymaker need (e.g. in the Department of Health and Social Care or NHS England). Our working group met weekly, and included academic and lived experience researchers, and clinicians.

Lived experience involvement

The working group included five lived experience researchers, who took part in all stages of the research from design, screening and extraction to analysis and write-up. The lived experience researchers included people with experience of inpatient care; conducting patient-led ward inspections; peer advocacy and support; being a carer; and direct experience of surveillance technologies during admission to inpatient mental health services. Some of the lived experience researchers were in liaison with service user groups and patients with experience of surveillance technologies. Due to the sensitive nature of the topic and related experiences, some lived experience researchers in the group have chosen to remain anonymous. Two experts by experience who had direct experience of surveillance technologies, including in an inpatient mental health setting, contributed to the lived experience commentary.

Search strategy

We searched five electronic databases (Embase, MEDLINE, PsycInfo, PubMed and Scopus) for peer-reviewed literature relevant to our research objectives. We searched for grey literature relevant to research objective 2 on a grey literature database (the Health Management Information Consortium) and two pre-print servers (medRxiv and PsyArXiv). Database searches were initially conducted between 17/09/2023 and 18/09/2023, with no date or language restrictions and were updated on 19/09/2024. Screening of non-English language papers was conducted using Google Translate; extraction and quality appraisal of full texts was conducted by someone with knowledge of the language. We contacted experts (including from NHS England, the Care Quality Commission and research experts internationally) to request additional literature we may not have identified. Our lived experience networks supported the identification of additional grey literature. We also reference list screened and citation tracked included studies and relevant systematic reviews. Our search strategy included key terms relating to surveillance and inpatient mental health settings, as detailed in Appendix C [see Additional File 1].

Screening

Title and abstract and full text screening were conducted in Rayyan [34]. Title and abstract screening was conducted by seven researchers (KS, UF, JG, AG, CR and two NIHR MHPRU Lived Experience Researchers). One hundred percent of titles and abstracts were independently double screened. Full text screening was conducted by nine researchers (KS, UF, JG, AG, CR, RC and three NIHR MHPRU Lived Experience Researchers). One hundred percent of full texts were independently double screened. Any disagreements were resolved by discussion between KS, UF, JG and AG.

Eligibility criteria

Participants

Participants include mental health patients (of any age, sex or gender), staff, carers and visitors to inpatient mental health services, Section 136 suites and places of safety.

Intervention

Surveillance-based technology initiatives include CCTV, remote monitoring initiatives, smart watches and body-worn cameras. We excluded studies which focused solely on door locking, door security, or key card access practices and policies, without explicit reference to surveillance technologies.

Comparators/controls

Studies with or without any type of comparator or control group were eligible for inclusion.

Outcomes

For research objective 1 we included studies which mapped where, when, how, how often and by whom such surveillance initiatives are used and who they are used on. Information related to lived experience involvement in the development, implementation, use and evaluation of the intervention was also included, as were implementation outcomes including appropriateness, adoption, feasibility, fidelity, sustainability, penetration and costs.

For research objective 2, we included studies which reported qualitative data on patient, staff and family/carer pre-implementation perceptions and post-implementation experiences of surveillance technologies.

For research objective 3, we included quantitative data on outcomes including safety of patients, staff, carers and visitors, use of restrictive practices and other containment measures, cost-effectiveness, care quality outcomes, clinical mental health outcomes, wellbeing and satisfaction of patients, staff, carers and visitors.

Setting

Inpatient mental health/psychiatric hospitals, Section 136 suites and places of safety were included. Inpatient mental health services admit patients, either voluntarily or involuntarily, for at least one night to receive treatment for mental health difficulties. Inpatient mental health services with any length of stay were eligible for inclusion, ranging from acute inpatient mental health wards offering shorter-term support to longer-term rehabilitation wards and forensic wards. We included general inpatient mental health wards, as well as those for any specific psychiatric diagnoses. A “place of safety” is a designated location, such as a hospital or police station, where individuals detained under mental health legislation are taken by police for mental health assessment. Section 136 suites are a specific type of place of safety, typically a dedicated area within a hospital. We excluded studies based in emergency departments, dementia-specific wards, care/nursing homes, outpatient and drop-in crisis services.

Design

We included all study designs reporting quantitative, qualitative and mixed methods data. We excluded conference proceedings, abstracts without an associated full text, books, PhD/MSc/BSc theses, opinion pieces, reviews, blog posts and social media content. For grey literature to be eligible for inclusion in relation to research objective 2, the sources had to, at least briefly, describe their methodological approach. No language or location restrictions were imposed during our searches or screening.

Data extraction

A data extraction sheet was designed in Microsoft Excel and revised based on feedback from the working group and piloting on an eligible paper by JG. The final data extraction sheet template can be seen in Additional File 2. Data extraction was conducted by eight researchers (KS, JG, UF, AG, CR, RC and two NIHR MHPRU Lived Experience Researchers). Data were independently double extracted for 4/32 (12.5%) of the included papers and an expert quantitative researcher (ARG) checked the accuracy and interpretation of all quantitative data extracted.

Quality appraisal

As the included studies varied in design, we used the Mixed Methods Appraisal Tool (MMAT) to assess quality [35]. We also noted any additional ethical issues, the degree of lived experience involvement in the studies, and conflicts of interest reported in the papers, such as author affiliations with surveillance technology companies or funding received from them. Potential undisclosed conflicts of interest were also investigated through online searches of authors using search engines. Quality appraisal was conducted by nine researchers (KS, JG, UF, AG, CR, RC, PB and two NIHR MHPRU Lived Experience Researchers). Independent double quality appraisal was conducted for 9/32 (28.1%) of the included papers. Any discrepancies in quality appraisal ratings were resolved by discussion and involvement of a third review author where necessary.

Evidence synthesis

Evidence synthesis was led by JG and UF. The interpretation of data and synthesis of results was supported by KS and the working group. Data were synthesised by research objective, and study characteristics were tabulated. Where possible, results were reported separately by type of surveillance technology.

Synthesis methods by research objective:

• (1) Implementation mapping and outcomes: We mapped the way the surveillance-based technologies were used in our settings of interest by technology type, including details (where available) on where, when, how often and by whom surveillance-based technologies are used and who was being surveilled. We tabulated and narratively described [36] implementation outcomes including appropriateness, adoption, feasibility, fidelity, costs, penetration and sustainability [37].

• (2) Perceptions and experiences: Quantitative and qualitative data documenting perceptions and experiences of surveillance technologies were narratively synthesised [36]. We synthesised data separately according to whether perceptions and experiences were reported pre- or post-implementation of surveillance technologies. Findings were grouped into benefits and potential uses, concerns and potential harms, or neutral views, and then by respondent (e.g. patients, staff, family/carers) where possible.

• (3) Quantitative measures of effect: Quantitative outcome data were tabulated and summarised narratively [37]. This included reporting original measures of effect (e.g. risk ratios, odds ratios, or risk differences for dichotomous outcomes, and mean differences or standardised mean differences for continuous outcomes) and p-values, where available. Results were grouped according to surveillance technology type. We were unable to perform a meta-analysis due to heterogeneity across the types of outcomes, measures of effect, populations and length of follow-up.

Results

Figure 1 presents the PRISMA flow diagram [33] of the screening and selection process. We identified 32 studies for inclusion. Excluded full texts and their reasons for exclusion can be seen in Appendix D [see Additional File 1]. Approximately two-fifths of included studies reported on CCTV/video monitoring (n = 13), other studies reported on VBPMM (n = 9), BWCs (n = 6), GPS electronic monitoring (n = 2) or wearable sensors (n = 2). Most studies were conducted in the UK (n = 23), with two conducted in Germany, one multi-country study and one each conducted in Ireland, Malaysia, Finland, Australia, Israel and USA. Fifteen (46.9%) studies were quantitative in design, eight (25.0%) qualitative and nine (28.1%) mixed methods. Most studies reported data from a mix of ward types (n = 11), followed by acute wards (n = 7), low/medium secure wards (n = 5), forensic wards (n = 5) and psychiatric intensive care units (PICUs) (n = 4). Only two studies specified that they included wards with inpatients under the age of 18 [38, 39]. The remaining studies either exclusively focused on inpatient wards for adults or did not specify the age of the inpatient populations.

Fig. 1

PRISMA flow diagram

Sixteen papers (50.0%) were rated as low quality, five studies (15.6%) were rated as medium quality and eleven studies (34.4%) were rated as high quality. For full details on MMAT ratings, see Additional File 3. Nine papers (28.1%) disclosed conflicts of interest. One report was produced by a surveillance technology company [39], whilst other conflicts of interest included the project being funded by a surveillance technology company [40,41,42,43], authors’ time or expenses being funded by a technology company [44, 45], technology companies contributing to the conduct of the study [40,41,42,43, 46] or authors working for a surveillance technology company [40, 45,46,47]. In one of these papers, authors were listed as being affiliated with a technology company, but this was not reported as a conflict of interest in the declarations section [47]. Out of the 32 studies included in this review, we also identified a potential undeclared conflict of interest in one study, involving an author receiving prior funding from organisations involved in the use of surveillance technologies. Study characteristics, including quality ratings, are summarised in Table 1. A more detailed version of this table is provided in Appendix E [see Additional File 1].

Table 1 Table of study characteristics

Research objective 1: how are surveillance-based technologies in inpatient mental health settings being implemented and what are the related implementation outcomes?

Below we have summarised how surveillance technologies have been implemented, and reported implementation outcomes, by type of surveillance technology. Full details on implementation process, setting, informed consent procedures and lived experience involvement can be found in Appendix F [see Additional File 1] whilst implementation outcomes can be found in Table 2.

Table 2 Summary of implementation outcomes (appropriateness, feasibility, fidelity, adoption, sustainability, penetration, costs) across the surveillance technologies

Vision-Based Patient Monitoring and Management (VBPMM)

Description of implementation

Nine studies explored VBPMM [32, 39,40,41,42,43, 45,46,47]. All were UK-based and utilised Oxevision (a VBPMM device by Oxehealth). Eight of the nine studies reported conflicts of interest [39,40,41,42,43, 45,46,47]. All studies were rated as low quality except one which was rated high quality [32]. This high-quality study was the only VBPMM paper which did not report any conflicts of interest [32]. Inpatient settings included acute wards [41, 43, 45, 47], older adult inpatient mental health wards [41], psychiatric intensive care units [40, 42, 46] and a high secure forensic inpatient service [32]. VBPMM was used in patients’ bedrooms in all but one study, where it was used in a seclusion room [40].

VBPMM involves an anti-ligature, wall-mounted system equipped with an infrared-sensitive camera (a Class IIa medical device), also referred to as an “optical sensor”, which remotely monitors patients’ pulse and breathing rate at regular intervals [41, 45]. It also tracks patients’ movements, generating location and activity-based alerts [43]. Video can be viewed by staff for up to 15 s when taking vital sign measurements or responding to an alert [41, 46]. In the latter case, only blurred video is available [41, 46]. Dewa et al. [32] states that de-pixellated video can “only be viewed with express permission in exceptional circumstances” (e.g. if there is potential risk to the patient), though it did not state who provides permission. The VBPMM system can be accessed via monitors in the nurses’ station and portable tablets [43]. It differs from CCTV in that it has additional physical health monitoring functions and video stream viewing is intermittent “on-demand” rather than continuous observation [45, 46].

Ndebele et al. [45] described how consent for VBPMM use was sought from patients, or from a suitable consultee, such as their carer or the ward’s consultant psychiatrist, in cases where patients lacked capacity to consent. If consent was not given, the system remained switched off in the patient’s bedroom for the duration of their stay. If patients who lacked capacity initially later regained capacity, consent was then sought from them. Malcolm et al. [42] also stated that patients can request for VBPMM to be turned off. The remaining papers did not describe patients being able to opt-in or out of VBPMM use.

Stated aims of the technology

Reported aims of VBPMM include as follows: reducing staff disturbance to patients by enabling less intrusive observations [32, 47], allowing staff to respond to patient needs more quickly and efficiently [42, 43], improving safety and quality of care [41, 42], aiding monitoring of self-harm risks [45], monitoring early signs of agitation to enable early intervention for aggression [46], preventing incidents [32, 46], supporting care-planning [39], supporting compassionate and dignified care [39, 43] and reducing NHS mental health care costs. VBPMM is reportedly intended as an adjunct to usual care, not as a replacement for therapeutic interactions or physical care [39, 43, 47]. However, it is unclear how this adjunctive role is envisioned alongside the stated aim of cost reduction.

Lived experience involvement in implementation

Three out of nine papers reported lived experience involvement in VBPMM implementation [39, 40, 47]. This included a pre-implementation patient focus group [40] and meetings with former patients, relatives and nursing staff [47]. The Oxehealth report [39] stated that as an organisation, they have continuous patient and caregiver involvement throughout the implementation process. These descriptions of lived experience involvement lacked methodological detail.

Implementation outcomes

Five studies reported VBPMM implementation outcomes [41,42,43, 45, 47]. Barrera et al. [47] reported high fidelity, with no significant gaps in VBPPM use and staff observations being conducted as required, and high penetration, stating that the sensors appeared to be embedded in the ward’s day-to-day clinical practice. In terms of feasibility, Ndebele et al. [45] reported VBPMM consent rates of 67 and 76% on a female and male acute ward, respectively. It was not clear whether any consenting patients later withdrew consent, and whether these figures capture those individuals.

Three studies [41,42,43] used a cost calculator approach to estimate the costs of implementing VBPMM in addition to standard care, compared to standard care alone. Malcolm et al. [42] calculated that if VBPMM were implemented in addition to standard care for adults admitted to PICUs across England, the total costs per year would be as follows: £10,926 (GBP) per patient, £228 per occupied bed day, £897,907 per average sized ward and £68,839,567 per year to the NHS in total. These calculations were based on estimates that VBPMM would cost £319 per patient (including annual licence fees, installation and cabling and staff training costs), and that the costs per patient would be £158 for night-time observational hours, £9943 for one-to-one observational hours, £167 for assaults and £338 for rapid tranquillisations. A full breakdown of costs and how they were calculated is available in Table 2.

Malcolm et al. [43] estimated that if VBPMM were implemented in addition to standard care for adults admitted to acute mental health wards across England, the total costs per year would be £1879 per patient, £39 per occupied bed day, £205,849 per average sized ward and £153,872,139 per year to the NHS in total. This was based this on estimates that VBPMM would cost £242 per patient (including licence fees, annualised installation and cabling and staff training costs) and that it would cost per patient £63 for night-time observations, £1444 for one-to-one observations and £129 for self-harm incidents. A full breakdown of these costs and how they were calculated is available in Table 2.

From figures provided by Buckley et al. [41], it can be calculated that they estimated the annual cost of an average-sized acute adult mental health inpatient ward with VBPMM in England (with 90% occupancy) to be approximately £374,054 from an NHS and Personal Social Services (PSS) perspective, and £368,799 from an NHS mental health trust perspective. They estimated the annual cost of an average-sized older adult mental health inpatient ward in England (with 90% occupancy) with VBPMM to be approximately £441,116 (NHS & PSS perspective) or £293,613 (NHS mental health trust perspective). They estimated VBPMM costs to be £29,457 on an average-sized acute adult inpatient mental health ward, and £29,472 on an average-sized older adult inpatient mental health ward, including annual licencing fees, annualised installation and cabling fees and staff training costs. A full breakdown of the costs included in calculations in this study is available in Table 2.

Closed circuit television (CCTV)/video surveillance

Description of implementation

Thirteen studies explored CCTV/video surveillance [38, 48,49,50, 57, 59,60,61,62,63, 66, 67, 69]. No studies declared conflicts of interest, seven studies were rated as high quality [49, 50, 57, 60,61,62, 69], three were rated medium quality [38, 63, 67] and three low quality [48, 59, 66]. These studies were based in the UK (n = 5), Germany (n = 2), Australia (n = 1), Ireland (n = 1), Finland (n = 1), USA (n = 1), Malaysia (n = 1) and one study recruited experts from a range of countries. CCTV/video surveillance had been implemented in acute wards [49, 61], PICUs [38] and low-secure [67], medium-secure [60] and high-secure forensic wards [63]. Curtis et al. [49] described the setting as an inpatient psychiatric facility with beds for acute psychiatric conditions, geriatric conditions, learning difficulties and forensic cases. In six papers, the type of inpatient ward was not specified [48, 50, 59, 62, 66, 69]. Within wards, CCTV was described as being implemented in communal areas (e.g. ward corridors, exit doors), patients’ bedrooms [67] and seclusion rooms [49, 50, 57, 66]. Some specified that it was not used in private areas such as patient bedrooms [49, 50] or bathrooms [50].

Stated aims of the technology

The functions of CCTV/video surveillance described in the papers included monitoring patient behaviour [49, 57, 62, 69] and staff behaviour [49], monitoring who is leaving the ward [57], monitoring safety during mechanical restraint [62], reducing institutional incidents [63] and preventing violence [62].

Lived experience involvement in implementation

None reported in the included papers.

Implementation outcomes

Five papers reported CCTV/video surveillance implementation outcomes [38, 57, 59, 62, 63]. Krieger et al. [38] reported that only 44% of patients understood why they were under surveillance at the time, and only 56% understood 4–5 days after surveillance ended. Adoption rates varied between studies (from 15.9 to 100% in different locations across the USA, UK and Germany) [57, 59, 62]. In terms of penetration, Krieger et al. [38] reported that 9.4% patients in three PICUs in Germany had been monitored via video, though it was unclear whether all the PICUs had video surveillance technology and its location on the wards.

In terms of appropriateness, Tapp et al. [63] conducted a Delphi expert consensus study to try to reach consensus on the elements of high-security hospital services that would be essential for the rehabilitation of forensic patients. During round one, 82% of staff and academic experts agreed that “CCTV monitoring should be implemented in the secure environment to reduce institutional incidents”, which met the 80% threshold for consensus. In round three, 62.2% of experts rated CCTV as “Important – the element of care is desirable, but its absence would not have a direct effect on the described outcome [institutional incidents]”. This did not meet the threshold for consensus, which the authors concluded meant that CCTV should not be applied prescriptively in high-secure hospital inpatient services.

Body-worn cameras (BWCs)

Description of implementation

BWCs were investigated in six UK-based studies, two of which were rated high quality [51, 68], one medium quality [54] and three low quality [44, 52, 55]. Conflicts of interest were reported in one study [44]. In four studies, the brand was Calla [44, 51, 52, 55]. Brands were not specified in the other two studies. Inpatient mental health settings included acute wards, low-secure, medium and medium enhanced forensic wards, recovery wards, PICUs and a health-based place of safety room at a psychiatric hospital. Hakimzada et al. [54] explored staff perceptions of BWCs in inpatient settings where BWCs had not been implemented, including acute wards, secure wards and a PICU.

BWCs are recording devices worn by trained staff in inpatient settings to document interactions between staff and patients via audio and video recordings [52]. They are manually activated by staff at their discretion. This may generally be signalled by a red flashing light and audible beep, with staff advised to inform patients before recording [54]. In Hardy et al. [55], staff were trained to explain to staff and patients that the camera was for safety, to narrate their actions and intentions to the camera, and inform patients if they stop recording due to it exacerbating the situation. Staff could turn the camera around to record sound only if necessary [55].

BWC footage access was protected by a PIN to prevent data retrieval if the camera was misplaced [55]. In Hardy et al.’s [55] study, BWCs were docked, recharged and data uploaded to a secure cloud via computer in the reception area at the end of each shift. This secure cloud was provided and administered by the BWC manufacturer. Footage is kept for a fixed length of time before being automatically deleted, unless required for a specific purpose, e.g. internal investigation [44].

Methods of informing patients of BWCs were reported in one study and included displaying information posters in high visibility areas on wards, providing written information, and by staff verbally informing patients about them on admission, in morning meetings, patient experience groups and community meetings [55].

Hardy et al. [55] stated that preparing for BWC implementation involved establishing the necessary policies, IT infrastructure and information governance compliance—e.g. completing a full privacy impact assessment and self-assessment tool from the surveillance camera commissioner. Patients and visitors were informed, and training was delivered to staff by the BWC supplier, which was then cascaded to other ward staff. Certain staff members also received specific training to become administrators [55].

Foye et al. [52] described how, before implementing BWCs, the NHS trust piloting them carried out extensive preparatory work, including directorate and senior management-level planning, consultations with individuals with lived experience, and the development of relevant policies and protocols which incorporated a human rights assessment and legal review. BWC training was provided to staff by the BWC manufacturer.

Stated aims of the technology

The aims of BWCs were described as increasing transparency; resolving incidents and complaints by providing accurate incident records; improving staff performance by providing footage for training and monitoring; improving staff conduct and patient behaviour; preventing incidents of violence and aggression; improving safety and to “counter false allegations” [44, 52, 54, 55, 68].

Lived experience involvement in implementation

Foye et al. [52] described how before implementing BWCs, the NHS trust piloting them conducted extensive preparatory work, including lived experience involvement and consultation. There was no lived experience involvement in implementation of BWCs described in the other studies.

Implementation outcomes

Two studies, both rated low-quality, reported BWC implementation outcomes [52, 55]. In Hardy et al. [55] most staff reported no operational or practical difficulties with the BWCs. Where difficulties were reported, most were minor and easily resolved. Only 68% of surveyed patients reported that they had been made aware that some nurses were wearing BWCs [55]. Hardy et al. [55] reported the following purchase costs: camera and software (£6,540), accessories (£1,109) and storage (£569) though these were provided free by the BWC manufacturer for the study. It also provided a breakdown of staff requirements (e.g. to deliver and attend training, create policies, provide IT support, to upload and review recordings and sort out problems with cameras) but did not report the associated costs.

In Foye et al.’s study [52], BWC use was reported in 50/543 completed shift reports, indicating that they were used on less than 10% of shifts during the 6-month pilot. Seventy eight percent of uses were on the acute adult inpatient ward, and 22% were on the adult PICU. Most deployments occurred with a warning (n = 30, 60% activations). 19/50 (38%) of uses the BWC was deployed with a warning, but not activated, and there was only one occasion (2%) where a BWC was activated without warning. There were moderate positive correlations between BWC use and verbal aggression (r = 0.37, p < 0.001) and physical restraint (r = 0.31, p < 0.001).

Global Positioning System (GPS) electronic monitoring

Description of implementation

Two low-quality papers reported on GPS electronic monitoring [56, 65]. Neither reported conflicts of interest. One study used the brand Buddi Tracker [65]; the brand was unspecified in the other. Both studies were set in UK-based medium-secure inpatient mental health services.

In both studies, GPS electronic monitoring devices were attached to patients’ ankles when they went on leave. They were only used with consenting patients, with the exception of high-risk patients requiring urgent hospital or court transfer. It was unclear whether the use of GPS electronic monitoring in these instances was court-ordered or the result of a clinical decision. Consent rates were not provided in either study. Clinical decisions about the appropriateness of GPS electronic monitoring were made on an individual basis following a specific risk assessment protocol in Murphy et al. [56]. Tully et al. [65] described how it was primarily intended to be used with patients in the early stages of leave, when risk of leave violation is highest.

The “Buddi Tracker” device [65] employs secure straps with anti-tamper features, transmitting location via GPS signals to monitoring software via a mobile phone network. Geographical parameters (“geo-fences”) can be set, allowing inclusion and exclusion zones to be created. If a patient breaches a geo-fence, an alarm goes off which causes the device to vibrate and an alert to be sent through the in-built monitoring software. Information from each device is monitored by a security company. Breaches of agreed terms and conditions trigger a predetermined alert to relevant parties and a risk management plan [65].

Stated aims of the technology

GPS electronic monitoring tracks patients on leave, with the aim of preventing leave violations such as absconding or failing to return [56]. It was hypothesised to reduce leave violations, increase overall leave and increase the proportion of unescorted leave [56].

Lived experience involvement in implementation

Tully et al. [65] states that the introduction of GPS electronic monitoring was discussed with patients and legal advisors, and consent and information forms were developed. However, there is no methodological detail for patient consultation provided. No lived experience involvement in implementation was reported in Murphy et al. [56].

Implementation outcomes

Two papers reported GPS electronic monitoring implementation outcomes [56, 65]. Though Tully et al. [65] did not directly discuss feasibility, the authors did state that the technology was still in use at the time of publication, suggesting evidence of feasibility. Tully et al. [65] also reported high fidelity; the authors claimed it was mostly used in the early stages of patients being granted leave or transitioning from escorted to unescorted leave and was only used immediately before discharge in a minority of cases. However, data were not provided to evidence this claim [65]. Murphy et al. [56] calculated that the total cost of GPS electronic monitoring over the 3-month study period was £34,653, equating to an average cost of £286 per patient. They estimated that the total cost of implementing GPS electronic monitoring over the 3-month study period, taking into account the costs of escorting staff (£59 per hour), technology costs (£114,336 for up to 70 devices, and £199 for each additional device) and leave violation costs (cost breakdown not provided), was an average of £1617 per patient. Tully et al. [65] simply reported that each GPS electronic monitoring device used in their study cost £133.

Wearable sensors

Description of implementation

Two papers, one rated as low quality [64] and one as high quality [53] examined wearable sensors. Neither reported conflicts of interest.

Tron et al. [64] evaluated the use of GeneActiv smart watches for monitoring movement in patients with psychosis at a psychiatric inpatient facility in Israel. These smart watches were equipped with accelerometers, light and temperature sensors. Medical staff managed their placement and removal and uploaded data from the memory card in the device to a central storage location for analysis.

Greer et al. [53] explored staff’s perceptions of using two different remote monitoring devices to conduct real-time monitoring of patients’ psychophysiological signals to predict aggression. One device (E4, Empatica Srl) is worn around the wrist, and the other (Everion, Biovotion Ltd) is worn around the upper arm. Staff were recruited from a medium-secure forensic inpatient service in the UK and did not have prior experience with these devices.

Stated aims of the technology

Tron et al. [64] aimed to use the GeneActiv smartwatch to monitor patient movements and correlate them with mental states to better evaluate schizophrenia symptom severity, characterise schizophrenia subtypes and causes, and personalise treatments. In Greer et al. [53], the aim of the devices was to monitor patients’ physical indicators to predict aggression.

Lived experience involvement in implementation

Greer et al. [53] stated that the interview topic guide was informed by consultation with two service user–caregiver advisory groups. No lived experience involvement was reported in Tron et al. [64].

Implementation outcomes

Tron et al. [64] reported that movement features detected by smartwatches during the “free time” window (4–5 pm) were the most effective in explaining variance in patients’ scores on factors of the clinician-administered Positive and Negative Syndrome Scale (PANSS). Combining data from all time windows throughout the day resulted in substantially higher explained variance on all PANSS factors. They also reported a case where a patient’s step count increased during a period where their medication dosage changed. They argue that this evidence suggests the potential of using smartwatches for continuous tracking of schizophrenia-related symptoms and patient states in hospital settings.

Research objective 2: pre-implementation: how are surveillance-based technologies in inpatient mental health settings perceived pre-implementation?

Vision-Based Patient Monitoring and Management (VBPMM)

One study explored pre-implementation perceptions of VBPMM [40] (see Table 3 for full results). It reported a conflict of interest and was rated low quality. It reported overall positive pre-implementation staff views of VBPMM and mixed patient views. No papers reported carer views.

Table 3 Staff, patient and carer pre-implementation perceptions of Vision-Based Patient Monitoring and Management (VBPMM)

Staff

Staff were reported to largely feel that VBPMM could be a positive addition to seclusion rooms, as it could facilitate vital sign monitoring [40].

Patients

Some patients felt that VBPMM could improve safety and reduce disrupted sleep, whereas others feared that it would reduce human interaction in seclusion, or that the cameras could control or harm them [40].

Closed circuit television (CCTV)/video surveillance

Two papers explored pre-implementation perceptions of CCTV/video surveillance [49, 69] (see Table 4 for full results). Neither paper reported any conflicts of interest, and both were rated high quality. Patient views were mixed, and staff and carer views were negative.

Table 4 Staff, patient and carer pre-implementation perceptions of closed circuit television (CCTV)/video surveillance

Patients

Whilst some patients felt comfortable with the idea of being video monitored, others felt that it would cause them stress and disrupt their daily routines. Privacy concerns led some patients to prefer cameras to be positioned in communal rather than private areas. Patient preferences varied regarding camera visibility and who should be able to view the footage [69].

Mixed sample (patients, staff, carers)

Curtis et al. [49] reported apprehension towards the use of CCTV in communal ward spaces amongst a mixed sample of staff, patients and carers.

Body-worn cameras (BWCs)

Four studies explored pre-implementation perceptions of BWCs [44, 51, 52, 54] (see Table 5 for full results). One reported a conflict of interest [44]. One paper was rated high quality [51], one medium quality [54] and two low quality [44, 52]. Staff views were mixed. No studies reported pre-implementation carer views.

Table 5 Staff, patient and carer pre-implementation perceptions of body-worn cameras (BWCs)

Staff

There were mixed views amongst nursing staff about whether they would feel comfortable wearing a BWC, whether it would deter them from working, modify staff behaviour or put their minds at ease. Some felt that BWCs could be a useful training and reflection tool, helping to reinforce good practice and identify faults in staff behaviour. However, others thought they may make staff less willing to get involved in incidents, or that staff and patients may “act” for the camera. Some staff felt that footage from BWCs could provide accurate, unbiased documentation of incidents, and most felt that they would reduce “false patient accusations”. However, some questioned the transparency of BWC recordings, highlighting concerns about bias and power dynamics due to the cameras being controlled by staff.

Some staff believed that BWCs could improve staff and patient safety and help reduce and de-escalate conflict and violent incidents, and so reduce constraints on patients. However, others thought they could increase and exacerbate violent and aggressive situations. Staff expressed concerns about the cameras intimidating patients, and increasing paranoia, particularly for patients experiencing psychosis or with previous negative experiences of being filmed. A minority thought BWCs might enhance staff-patient relationships, whilst others believed they could harm therapeutic relationships, increase power differentials between patients and staff and undermine patient trust. Furthermore, ethical concerns were raised by some staff regarding the difficulty of obtaining informed consent, and the potential of BWCs to violate patients’ privacy and confidentiality.

On a practical level, some staff believed that having similar protocols to other safety measures, such as personal alarms, could make BWC implementation easier. However, concerns were also raised about BWCs increasing workloads, staff forgetting to wear, activate and charge the cameras, and fears that BWCs could be broken and used as weapons by patients. Staffing issues, such as reliance on bank and agency staff, were considered a barrier to effective BWC implementation. Staff emphasised the need for continued BWC training, rather than a one-off training session during induction, to ensure changes in personnel, practice and shift patterns are part of the implementation process [44, 51, 52, 54].

Patients

Some patients believed BWCs could hold staff accountable and improve care, though worried that this effect would wear off after a few months as people become accustomed to the cameras. Concerns were raised about the transparency of BWC recordings, as the cameras are controlled by staff, creating issues of bias and power. Some patients felt that BWCs would not reduce incidents due to some patients being too acutely unwell to have insight into their actions or those of staff activating the cameras. Others doubted their effectiveness in reducing incidents since they do not address the wider systemic factors contributing to violence and aggression. Some patients viewed BWCs as being an intervention primarily for staff safety and benefit. They viewed them as increasing a culture of “them and us”, leading to patient resistance towards BWC use, especially amongst people with mistrust towards staff or negative experiences of BWC use by police. Like staff, patients viewed staffing issues and burnout as barriers to effective BWC implementation [52].

Global Positioning System (GPS) electronic monitoring

No papers reported on staff, patient or carer pre-implementation perceptions of GPS electronic monitoring.

Wearable sensors

One paper explored pre-implementation perceptions of wearable sensors [53] (see Table 6 for full results). It did not report any conflicts of interest and was rated high quality. Staff views of wearable sensors were mixed. No studies reported patient or family/carer views.

Table 6 Staff, patient and carer pre-implementation perceptions of wearable sensors

Staff

Staff recognised wearable sensors’ potential for facilitating less obtrusive monitoring, increasing patients’ self-awareness and providing information that may not otherwise be shared with staff. Some also felt that they could aid risk-monitoring, reduce violent incidents and prevent situations from escalating. However, concerns included patients misusing them as ligatures or weapons, exacerbating patient paranoia, data security and patient confidentiality issues, fluctuating patient willingness to use them and increased workload for staff.

Research objective 2: post-implementation: How are surveillance-based technologies in inpatient mental health settings experienced post-implementation?

Vision-Based Patient Monitoring and Management (VBPMM)

Six papers explored post-implementation experiences of VBPMM [32, 39, 40, 45,46,47] (see Table 7 for full results). Five of these studies reported conflicts of interest [39, 40, 45,46,47]. Five were rated low quality [39, 40, 45,46,47], one was rated high quality [32]. Experiences of patients, staff and carers were mixed.

Table 7 Staff, patient and carer post-implementation experiences of Vision-Based Patient Monitoring and Management (VBPMM)

Staff

Benefits of VBPMM perceived by staff included less intrusive monitoring, which could improve patients’ sleep, and enhanced safety for staff and patients, through better physical health monitoring, reduced patient aggression and more effective prevention and mitigation of incidents. Some staff also believed that VBPMM gave patients a greater sense of agency over their own safety, could reduce use of restrictive practices and contributed to improved staff-patient relationships. There were mixed perspectives on its impact on patients’ privacy. Staff also flagged concerns about technological issues (e.g. poor Wi-Fi), incorrect use of the technology, insufficient staff training and doubts about its accuracy. Some felt VBPMM should not replace standard care and physical observations [32, 40, 45,46,47].

Patients

Some patients also felt that VBPMM improved patient safety and sleep. Other benefits reported by patients included increased independence from staff and a greater sense of connection in seclusion. However, patients also raised ethical concerns about VBPMM’s negative impact on their privacy, dignity and human rights. They cautioned about how being monitored can cause distress, exacerbate power imbalances and damage trust between patients and staff. Concerns were also raised about a lack of patient choice, and inadequate or inaccurate communication from staff regarding VBPMM [32, 39, 40].

Carers

One paper reported that carers had mostly positive perceptions of VBPMM, but some had concerns about a negative impact on care quality [40].

Closed circuit television (CCTV)/video surveillance

Five papers explored post-implementation experiences of CCTV/video surveillance [49, 50, 60, 66, 67] (see Table 8 for full results). None of these studies reported any conflicts of interest. Three were rated high quality [49, 50, 60], one medium quality [67] and one low quality [66]. Three studies explored experiences of CCTV/video surveillance in communal ward areas [49, 50, 66], one in a seclusion room [60] and one in patients’ bedrooms [67].

Table 8 Staff, patient and carer post-implementation experiences of closed circuit television (CCTV)/video surveillance

Staff

Staff’s experiences of CCTV in communal spaces varied [49, 50, 66]. Some identified benefits including improved staff and patient safety, monitoring of self-harm, violence and absconding. However, others doubted its ability to control behaviour or prevent incidents. Some saw value in using CCTV to provide evidence to investigate incidents and allegations and felt it could be used to scrutinise staff behaviour. Ethical concerns were raised about its impact on patients’ privacy, on dignity and human rights and on therapeutic engagement. Some staff felt CCTV should not be used as a substitute for in-person care [49].

Staff’s views of CCTV use in patients’ bedrooms at night were also mixed [49, 50, 67]. Perceived benefits included improved monitoring of patients, enhanced staff safety and reduced disruption of patients’ sleep compared to physical checks. Some staff felt they could rely on CCTV for patient observation, whereas others emphasised the importance of still conducing physical checks. Some staff raised concerns about negative impacts of CCTV in patients’ bedrooms on privacy, increased patient distress and paranoia and reduced opportunities for therapeutic engagement. There were also reports of staff feeling uncertain about how to use the technology, using it incorrectly, finding it unreliable and it producing low-quality images [67].

Patients

Patients had mixed views on CCTV monitoring in communal areas. Some felt it enhanced staff and patient safety, whilst others considered it an invasion of privacy. CCTV use in communal areas did not appear to affect patients’ use of these spaces [50]. In seclusion rooms, some patients believed CCTV could aid staff observations, prevent self-harm, help recognise emergencies and foster a sense of safety. However, concerns included a lack of control, privacy issues and security concerns, worsened by poor communication about the technology [60].

Regarding CCTV use in patients’ bedrooms at night, some patients found it enhanced their sense of safety, for example by deterring other patients from rule-breaking or stealing property. Some considered it less invasive and disruptive to sleep than physical checks since it reduced staff movement and the frequency of staff entering bedrooms for checks. However, others felt it was intrusive, impeded relaxation, negatively impacted therapeutic relationships with staff, and feared that it could result in traditional observations being neglected. Misunderstandings amongst patients about how and when CCTV was being used were reported, and there were also instances where patients were video monitored in their bedrooms outside of designated times or without consent [67].

Carers

One study reported that carers had concerns staff would not always be monitoring CCTV and so may miss things [49].

Body-worn cameras (BWCs)

Four papers explored post-implementation experiences of BWCs [51, 52, 55, 68] (see Table 9 for full results). None reported any conflicts of interest. Two were rated high quality [51, 68] and two low quality [52, 55]. Staff and patient experiences were mixed, no carer experiences were reported.

Table 9 Staff, patient and carer post-implementation experiences of body-worn cameras (BWCs)

Staff

Benefits of BWCs perceived by staff included reduced violence, aggression and restrictive practices. Some staff felt that they improved safety by improving staff awareness and reflexive practice, rather than changing patient behaviour. Staff identified various uses for BWC footage including: providing evidence to aid incident and complaint resolution (including “false allegations” against staff) and prosecutions; documenting interventions (e.g. physical restraints); and facilitating debriefing and staff training. However, concerns were raised about potential bias, as BWCs remain in staff control and only capture footage from the time of arrival, not the preceding events. Some staff also expressed concern about BWCs being used to monitor staff behaviour.

Some staff were sceptical about the effectiveness of BWCs in reducing violence and aggression. Some staff felt this is because they do not address underlying systemic causes, and because some patients are too acutely unwell to have insight into their actions and those of staff using cameras. Others viewed violence and aggression as just “part of the job”. Some staff even felt that BWCs could instigate incidents and increase patient distress and aggression. BWCs were viewed by some as a punitive measure, contributing to patients’ feelings of criminalization and intimidation. Some staff felt that BWCs could negatively impact therapeutic relationships and contribute to a “them and us” culture.

Staff had concerns about the impact of BWC use on patients’ privacy and dignity, with some staff feeling there needed to be limits around their use (e.g. a rule to not use them in private spaces such as patient bedrooms and bathrooms). However, it was acknowledged that these boundaries could be difficult to adhere to in practice. Staff emphasised the need for flexibility and to be able to decide not to use cameras in situations where they could cause harm, enabling a person-centred approach.

Whilst flexibility was felt to be important, staff also identified inconsistency in BWC implementation across trusts as an issue, with some feeling that national guidelines are needed. Staff said that guidelines and policies need to be developed considering how, why, what, where and when to use BWCs, with leadership and management involvement, and consultation with patients and staff, from the beginning of the implementation process. Feedback from staff and patients was viewed as important for ensuring continued improvement of BWC implementation. They noted that policies should include robust governance and data management plans, supported by good IT infrastructure and support, to help assure patients and carers of the safety of their data.

Staff also raised ethical and legal concerns around patient consent and the potential for BWCs to be used as a substitute for good care and safe staffing. They felt it was important to inform patients about how, why and when BWCs are used, and their rights in relation to them, to ensure transparency and hold staff to account to use them correctly. Staff recognised the importance of providing this information continually, since patients’ health and capacity can fluctuate during their inpatient stay.

Some staff found BWCs small, easy to use, user-friendly and viewed them as having little direct impact on their daily routines. However, others expressed concerns about useability, especially for staff less confident with technology. Practical challenges included overheating causing skin discomfort, cameras pulling on clothing, infection control concerns, charging and storage issues and the potential for BWCs to be lost or damaged. There were issues with staff forgetting to wear, activate or recharge cameras. The dynamic and reactive ward environment means that staff often react to incidents quickly, which can result in them forgetting or failing to use BWCs. There were mixed views on whether reviewing BWC footage saved time or added to workloads, with varying procedures across services for footage management.

There was a lack of consistency in staff understanding of BWC uses and procedures, and in BWC training received, with some staff receiving no or very limited training. BWC training was widely thought to be insufficient, with staff expressing a need for more comprehensive training, delivered on a rolling basis. They felt this should cover not just operational aspects, but, for example, ethical considerations, decision-making around BWC use, communication techniques, use in sensitive situations and how BWC use could impact on therapeutic relationships, treatment and recovery.

Some staff saw BWCs as top-down directives from senior management and policy stakeholders with limited frontline experience, contributing to poor staff buy-in and concerns about misspent funds. Some staff felt that de-escalation skills, communication and positive relationships need to be the primary tools for violence reduction, before BWCs, and that there needs to be greater investment in staff training and support, rather than in new technologies [51, 52, 55, 68].

Patients

Some patients reported feeling safer with BWCs due to them providing evidence to support their claims and protect them against staff misconduct. However, others felt BWCs did not improve safety and considered the footage biased since the cameras are controlled by staff and only capture incidents when staff are present. Some patients expressed concerns that BWCs negatively impacted their recovery, privacy and dignity, and could act as a “trigger”, particularly for people experiencing paranoia, psychosis or who had previous negative experiences of being filmed. Many patients felt less comfortable being filmed in private spaces, such as bedrooms and bathrooms, compared to communal ward spaces, and so felt that BWC use in these areas should be avoided unless “really necessary”.

Similar to staff, some patients felt that BWCs fail to address the systemic causes of violence and aggression, and that any improvements in safety are due to increased staff awareness and reflexivity, rather than changes in patient behaviour. Some patients viewed BWCs as punitive, contributing to feelings of criminalization and exacerbating power imbalances between patients and staff. Patients also raised ethical and legal concerns about the role of patient consent in BWC use, and reported receiving a lack of information about how and why the cameras were used and their rights.

Patients noted inconsistency in how BWCs are implemented and identified a need for policies and procedures to be developed around their use, including clear boundaries, to ensure a consistent culture and understanding. Some called for a national directive around BWC use to ensure greater consistency across services. They emphasised the importance of consulting with patients throughout the implementation process, to ensure that policies and procedures are sensitive to their needs and concerns. People who are autistic, experiencing psychosis or who have a history of trauma, were identified as particularly important to involve given the potential negative impact of BWCs on these groups [51, 52, 55, 68].

GPS electronic monitoring and wearable sensors

None of the included studies explored staff, patient or family/carer post-implementation experiences of GPS electronic monitoring or wearable sensors.

Research objective 3: what is the effect, including unintended consequences, harms and benefits, of surveillance-based technologies in inpatient mental health settings for outcomes such as patient and staff safety and patient clinical improvement?

Fifteen studies reported outcomes on the effectiveness of surveillance strategies in inpatient mental health settings [40,41,42,43,44,45,46,47, 52, 55, 56, 61, 65,66,67]. Overall, the findings were limited and mixed. The findings below are reported by type of surveillance and tabulated in Table 10.

Table 10 Quantitative evidence for the impact of surveillance technologies in inpatient mental health settings

Vision-Based Patient Monitoring and Management (VBPMM)

Seven studies reported on the effect of VBPMM [40,41,42,43, 45,46,47]. All studies reported on Oxevision by Oxehealth. All studies were rated low quality, and all declared conflicts of interest [40,41,42,43, 45,46,47]. Study designs included a mixed methods non-randomised controlled pre-post evaluation within a pilot study, which compared two intervention wards with VBPMM to two control wards without VBPMM [45], three economic analysis studies utilising cost-calculator approaches [41,42,43], two uncontrolled pre-post designs [40, 46] and a pre-post study with a concurrent control period [47].

Self-harm and ligature incidents

One study investigated VBPMM’s effect on self-harm and ligature incidents; it reported a significant relative reduction in self-harm and ligature incidents in bedrooms on the VBPMM wards compared to the control wards. However, when considering the VBPMM wards alone, there was a significant decrease in ligature incident rates, but not in self-harm rates, after introducing VBPMM [45].

Assaults

One study evaluated the impact of VBPMM on assaults; it reported a significant overall reduction in assaults on a PICU ward following its implementation. However, there was no significant change when specifically considering assaults in patient bedrooms. An increase in patient-to-patient assaults in bedrooms was found, but the statistical significance of this change was not reported [46].

Restrictive practices

Three studies reported on VBPMM’s effect on restrictive practices [40, 46, 47]. Barrera et al. [47] reported no significant effect on rapid tranquillisation frequency, whilst Ndebele et al. [46] reported a significant decrease in rapid tranquillisation events related to assaults after introducing VBPMM. Clark et al. [40] reported no significant impact upon seclusion session frequency or duration.

Clinical outcomes

One study investigated VBPMM’s effect on clinical outcomes [47]. It reported that insomnia severity significantly decreased the longer patients slept in a bedroom with VBPMM. There was a significant positive correlation between nights in rooms with VBPMM and hospital length of stay, although there was no significant difference in patients’ average hospital admission duration post-VBPMM and the average admission duration for all patients admitted to the ward in the 12 months before VBPMM introduction. There was also no significant difference in the use of hypnotic and anxiolytic medication [47].

Care quality

One study reported VBPMM’s effect on care quality-related outcomes [47]. It reported a 100% match of vital sign reports between observations with and without sensors.

Cost-effectiveness

Three studies investigated the cost-effectiveness of VBPMM [41,42,43], all utilising cost calculator approaches. Malcolm et al. estimated that VBPMM in addition to standard care, compared to standard care alone, reduced costs on a PICU by £72,286 per ward per year, equivalent to £880 per patient per year, or £18 per occupied bed day [42]. They estimated that if rolled-out to all adult PICUs in England, VBPMM would lead to an estimated cost saving to the NHS per year of £5,541,924. These calculations were based on estimates that VBPMM would cost £319 per patient (including annual licence fees, installation and cabling and staff training costs) and that it would reduce costs per patient by £109 for night-time observation hours, £806 for one-to-one observation hours, £60 for assaults and £223 for rapid tranquillisation events. Full breakdowns of these costs and calculations are available in Table 10. The key driver of these savings was 36 h of staff time saved per patient per year, primarily driven by a decrease in one-to-one observation hours. Scenario analyses showed that these results were robust to statistically significant changes in input parameters [42].

Malcolm et al. [43] estimated that VBPMM in addition to standard care, compared to standard care alone, reduces costs on acute adult mental health wards by £29,827 per average sized ward per year, equivalent to £272 per patient per year, or £6 per occupied bed day. They estimated that if rolled-out to all acute adult mental health wards in England, VBPMM would lead to an estimated cost saving to the NHS per year of £22,295,434. This was based this on estimates that VBPMM would cost £242 per patient (including licence fees, annualised installation and cabling and staff training costs) and that it would reduce costs per patient by £44 for night-time observations, £369 for one-to-one observations and £101 for self-harm incidents. The main driver of cost savings was the reduction in one-to-one observations. They reported that threshold analysis showed a 33% reduction in one-to-one observations is needed in an average-sized adult acute mental health ward to generate cash savings. Scenario analyses showed that if one-to-one observations were removed from the economic model, VBPMM no longer provided cash savings [43]. Full breakdowns of these costs and calculations are available in Table 10.

Buckley et al. [41] calculated the cost impact of implementing VBPMM on an average-sized acute adult inpatient mental health ward, and an average-sized older adult inpatient mental health ward, in England, from the perspective of NHS and Personal Social Services (PSS) and from the perspective of an NHS mental health Trust. They calculated that per average-sized adult acute inpatient mental health ward, per year, VBPMM would result in cost savings of £93,433 (NHS & PSS) or £89,305 (NHS mental health trust). This equates to a Return on Investment (ROI) of 3.17 (NHS & PSS) or 3.03 (NHS mental health trust). These calculations were based on estimates that a VBPMM system would cost £29,457 per ward per year, and that introducing VBPMM would reduce costs per ward per year by £7380 for night-time observations, £75,895 for one-to-one observations (cash-releasing), £27,038 for one-to-one observations (opportunity cost saving) and £12,578 (NHS & PSS) or £8449 (NHS mental health trust) for self-harm incidents. They identified a reduction in one-to-one observations as the main driver of these cost savings. They predicted that if VBPMM was implemented across all NHS England acute mental health services, this could lead to approximately £69 million in cost savings, assuming 90% average occupancy, of which around a half would be cash-releasing. Full breakdowns of these costs and calculations are available in Table 10.

Buckley et al. [41] calculated that implementing VBPMM on an average-sized older adult inpatient mental health ward in England would result in cost savings of £413,651 (NHS & PSS) or £265,376 (NHS mental health trust) per year. This equates to a ROI of 14.04 (NHS & PSS) or 9 (NHS mental health trust). These calculations are based on estimates that a VBPMM system would cost £29,472 per ward per year and that introducing VBPMM would reduce costs per ward per year by £29,969 for night-time observations, £107,101 for one-to-one observations (cash releasing), £140,335 for one-to-one observations (opportunity cost saving), £142,948 (NHS & PSS) or £17,443 (NHS mental health trust) for bedroom falls at night, £18,308 (NHS & PSS) or £0 (NHS mental health trust) for A&E visits resulting from bedroom falls at night, and £4461 (NHS & PSS) for emergency services visits resulting from bedroom falls at night. A reduction in one-to-one observations was again identified as the primary driver of cost savings. They calculated that if VBPMM was implemented across all NHS England older adult mental health services, this could lead to cost savings of approximately £89 million, of which one-third would be cash-releasing. Full breakdowns of these costs and calculations are available in Table 10.

Complaints and damage

One study reported on VBPMM’s effect on complaints [47]; it reported that during the study period, no incidents related to VBPMM were recorded on the Trust’s online incident reporting system. During the first four nights of the new observation protocol (where VBPMM was used to conduct most observations of patients at night, instead of physical checks), eleven patients who completed questionnaires each night expressed no negative comments about the system. Details were not provided about how these patients were selected, or the format or content of the questionnaire.

Closed circuit television (CCTV)/video monitoring

Three studies [61, 66, 67] reported the effect of CCTV. One was rated high quality [61], one medium quality [67] and one low quality [66]. All three reported no conflicts of interest. One study had a cross-sectional design [61], one was mixed methods with a cross-sectional quantitative component [67] and one was a pre-post evaluation [66].

Violence and aggression

Two studies reported on CCTV’s effect on violence and aggression [66, 67]. It is unclear whether Warr et al. [67], who investigated the impact of CCTV use in patients’ bedrooms at night on the frequency and nature of incidents, conducted any statistical significance testing. However, they reported that there were fewer incidents at night compared to during the day, but that there was no difference in the nature of the incidents. They also stated that there was no evidence of any association between the nature of incidents and the presence or use of CCTV, or the choice of the patient to be observed using CCTV or not. Vartiainen & Hakola [66] did not conduct any statistical significance testing but reported that violent acts reduced on the CCTV-monitored wards.

Clinical outcomes

Two studies reported on CCTV’s effect on clinical outcomes [61, 66]. Simpson et al. [61] reported that CCTV (at exit) had no significant impact on substance or alcohol use on the ward. Vartiainen & Hakola [61] reported no significant changes in subjective mental health or paranoid traits, or ward atmosphere, on any of the wards (with or without CCTV).

Complaints and damage

One study reported on the impact of CCTV on damages [66]; it reported that no damage had occurred to cameras in two years of TV monitoring.

Body-worn cameras (BWCs)

Three studies reported the effect of BWCs [44, 52, 55]. All were rated low quality and one declared a conflict of interest [44]. One had a quasi-experimental repeated-measures pilot study design [44], one had a mixed methods uncontrolled pre-post pilot study design [55] and the other was a mixed methods repeated measures study [52].

Self-harm

One study examined the effect of BWCs on self-harm [52]. It reported no significant differences in self-harm before, during and after the BWC pilot phase on the acute ward, but significantly higher self-harm during the BWC trial period compared to pre- and post-trial periods on the PICU [52].

Violence and aggression

All three studies reported mixed results on the impact of BWCs on violence and aggression [44, 52, 55]. Ellis et al. [44] reported no significant changes in the overall numbers of violent and aggressive incidents. They reported a significant reduction in incident seriousness on two of the wards (“local services admissions” wards) but no significant changes on the remaining five wards. Hardy et al. [55] did not conduct statistical significance testing but stated that violence decreased on three wards and increased on two wards. They also noted an increase in verbal abuse on three wards, a decrease on one, and no change on another.

Foye et al. [52] found no significant differences in types of incidents (violence against people, violence against objects, verbal aggression, self-harm or conflict) before, during and after implementing BWCS on the acute ward. However, on the PICU, verbal aggression was significantly higher in the post-trial period compared to the pre-trial and trial periods but there were no significant differences in violence against people, violence against objects or conflict behaviour. They also reported no significant differences in incident severity or police involvement on the PICU across the trial periods, but on the acute ward, incident severity was significantly higher during the BWC trial period and post-trail periods compared to the pre-trial period, whilst police involvement was significantly lower in the post-trial period compared to the pre-trial and trial periods [52].

Restrictive practices

All three studies reported on restrictive practices [44, 52, 55]. Ellis et al. [44] reported no significant change in levels of incidents requiring restraint or rapid tranquillisation overall across the wards. They did report a significant decrease in rapid tranquillisation on the two local services admissions wards, but not on the five remaining wards. Hardy et al. [55] did not conduct significance testing but reported an increase in low-level restraint on two wards, a decrease on two, and no change on one. Hardy et al. [55] also noted a reduction in emergency restraint on three wards and an increase on two. Foye et al. [52] reported significant reduction in the use of restrictive practices on the acute ward in the post-BWC trial period compared to the pre-trial and trial periods, and a significant decrease in restrictive practices during the trial and post-trial periods on the PICU.

Complaints and damage

One study reported on BWCs’ effect on complaints and damage [55]. No statistical significance testing was conducted but they reported that three complaints were made during the BWC pilot period, none of which were related to a particular incident or restraint. They stated that this was lower than in the comparison period the previous year before BWC implementation, where eight complaints were made, two of which had related to an instance of restraint.

Global Positioning System (GPS) electronic monitoring

Two studies reported the effect of GPS electronic monitoring technology [56, 65]. Neither reported any conflicts of interest. One was rated as medium quality [56] and one low quality [65]. Both had date-matched pre-post study designs.

Absconding, leave violations, and leave type

Both studies reported on absconding and leave violations with GPS electronic monitoring [56, 65]. Tully et al. [65] reported that following the introduction of GPS monitoring, leave episodes were significantly more likely to be unescorted. There was no significant change in the odds of a leave episode resulting in leave violation during the initial follow-up (1 year later). However, during the subsequent follow-up (another year later), leave episodes were significantly less likely to lead to an incident of leave violation. Murphy et al. [56] reported no changes in the overall number of leave violations after implementing GPS electronic monitoring.

Complaints and damage

One study reported on complaints relating to GPS electronic monitoring [65]; it reported two events of patients challenging the use of GPS electronic monitoring. It did not report the number of patients involved and number of opportunities to challenge the use of GPS electronic monitoring.

Cost-effectiveness

One study reported on the cost-effectiveness of GPS electronic monitoring [65, 70]; it reported a no significant change in the average total cost per patient following the introduction of GPS electronic monitoring.

Discussion

Key findings

Our paper has summarised the use of surveillance technologies on inpatient wards internationally, how these technologies are being implemented, staff, patients’ and carers’ views and experiences of them, and the impact these technologies have on quantitative outcomes such as restraint, seclusion, self-harm, violence and aggression and absconding. There were no randomised controlled trials identified, and very few studies with control groups, meaning that causal inferences regarding the impacts of surveillance technologies cannot be drawn. Overall, there is currently insufficient evidence to suggest that surveillance technologies in inpatient mental health settings are achieving the outcomes they have been employed to achieve.

Key findings regarding implementation included a particular lack of research on certain types of surveillance technologies, such wearable sensors and GPS electronic monitoring, reflecting the novelty of these technologies in inpatient settings. Only two studies specified that they included wards with patients under the age of 18. There was more evidence of implementation of surveillance technologies in the UK than any other individual country. All of the studies on VBPMM and BWCs were UK-based, indicating an increasing adoption of these technologies in the UK [70]. All of the studies declaring conflicts of interest were examining these technologies, with 8/9 (88.9%) VBPMM studies and 1/6 (16.7%) BWC studies reporting conflicts of interest. Therefore, various factors may be driving research into VBPMM and BWCs in the UK, including national strategic directions promoting the integration of digital technologies in mental healthcare, pressures to explore potential healthcare cost-saving strategies, concerns about staff recruitment and retention, showcasing by regulatory bodies, and the financial interests of technology companies [71,72,73,74,75].

Our lived experience researchers highlighted discrepancies between the way surveillance technologies were described as being implemented in the literature and their use in practice. For example, they noted that in their experience, staff can decide to view multiple segments of VBPMM video feed instead of it only being viewable when vital sign measurements are made. This underlines the fact that this review only captures how surveillance technologies are described as being implemented in the included papers, and so does not capture the variety of ways in which they may be implemented in practice. Furthermore, it is important to consider that the implementation of surveillance technologies is dynamic, varying across contexts and evolving over time in response to technological innovations and developments in policies and practices.

There have been numerous efforts to develop an understanding of what “best practice” could look like given these technologies are already being implemented. Such guidelines have been established by healthcare regulatory bodies, professional associations and charities, as well as internal protocols by specific healthcare providers. This includes guidance around the use of surveillance technologies in general [76,77,78], as well as guidelines and recommendations for specific technologies such as BWCs [12], VBPMM [79, 80] and CCTV [81,82,83]. Given the growing use of differing surveillance techniques, further research to explore these guidelines and understand their commonalities and differences (e.g. how best practice may differ across cultures and countries) could provide a better position for developing a more robust message to those institutions implementing them.

Evidence from the included papers related to experiences should inform how best practice guidance is developed. Prominent themes in qualitative results were patients’ and staff’s ethical concerns about privacy invasion, data protection, patient confidentiality and informed consent, in-line with previous literature [15,16,17, 84]. These were reinforced by some quantitative evidence indicating that a substantial proportion of patients did not consent to the use of VBPMM [45] or understand the reasons for being monitored via video [38]. Only two studies specified that they included wards with patients under the age of 18; therefore, the literature fails to account for the unique ethical considerations when using surveillance technologies within children and young people’s care settings. These findings highlight the danger of surveillance technology use infringing upon patients’ human rights, choice and autonomy. If surveillance technologies are to be implemented in inpatient settings, establishing best practice guidance could potentially help to regulate their use and mitigate some of these adverse effects. However, additional oversight by regulatory bodies to ensure audits of standards and adherence would be required as simply developing and implementing best practice guidelines, standards and policies does not necessarily mean that they will be adhered to in practice. This was exemplified by Warr et al. [67] who highlighted instances in their study where patients were subject to surveillance via CCTV against protocol, at times it was not meant to be in use and with patients who had not consented to its use. Similar concerns are being articulated in lived experience literature [31].

Staff, patient and carer experiences of and attitudes towards surveillance technologies on inpatient wards in the included papers were complex, with variation both within and between these groups. This mirrors findings elsewhere on surveillance technologies [16, 17, 85]. Qualitative literature in this review revealed some perceptions that surveillance technology could reduce violence, aggression and self-harm in inpatient settings. However, quantitative papers examining these outcomes presented inconsistent or weak results. This finding is consistent with previous systematic reviews reporting a poor and inconsistent evidence base for the use of BWCs in public sector services, including law enforcement, physical and mental healthcare settings [14, 20, 86,87,88]. This dissonance between qualitative perceptions of surveillance technology in inpatient settings and quantitative evidence is noteworthy; it is unclear whether it is a result of poor-quality evidence, the limitations of the surveillance methods being employed or the complexity of the issues being addressed through surveillance and the context within which such endeavours take place. It is important to consider that perceptions of surveillance technologies are influenced not only by their effectiveness in practice, but also other external factors. These include, for example, how they are marketed by technology companies and described to people by staff, and broader societal attitudes towards surveillance, particularly amongst those more vulnerable and sensitive to close observation.

A notable discrepancy between the stated aims and the evidence base lies in assertions that surveillance technologies reduce costs [89]. Only four studies in this review explored the cost-effectiveness of surveillance technologies. One found that GPS electronic monitoring use in a forensic inpatient setting did not significantly decrease costs [56], whilst the other three reported potential cost savings associated with VBPMM use in PICU settings [42], acute mental health ward settings [41, 43] and older adult inpatient mental health services [41]. These economic analyses had notable limitations, such as most of the parameter values in the models being based on data from small numbers of wards within single services or NHS Trusts, and the models not considering costs such as maintenance, upgrades, wear and tear, ongoing staff training and data compliance administration. Downstream costs incurred from the impact of surveillance technologies upon outcomes such as length of inpatient stay, readmission rates and post-discharge service use were also not accounted for. Furthermore, the accuracy of some of the estimates used in the models has been called into question. For example, the economic models used in two studies [41, 43] assume VBPMM reduces self-harm in patient bedrooms by 44%. This may be an over-estimation because, while the only published study investigating VBPMM’s impact on self-harm reported a 44% relative reduction in self-harm rates in patients' bedrooms on two VBPMM wards compared to two control wards without VBPMM, the actual reduction in self-harm rates on the VBPMM wards alone was only 22% [45]. Additionally, these models calculated Accident and Emergency self-harm treatment costs using the weighted average of fracture codes, which risks over-estimating cost savings. Given these methodological limitations, the full ongoing costs of implementing surveillance technologies in inpatient mental health settings remains unknown, meaning that claims about their cost-effectiveness are not currently robustly substantiated by the evidence base.

In the three studies examining the cost-effectiveness of VBPMM, the main driver of identified potential cost savings was a reduction in one-to-one staff observations [41,42,43]. Qualitative evidence suggested that both staff and patients agreed that surveillance technologies should not replace or reduce human interaction. Indeed, research suggests that human contact, trust, support and empowerment form integral elements of therapeutic inpatient care, including during episodes of containment such as seclusion and restraint [15, 18]. Malcolm et al. [43] argue that a reduction in one-to-one staff observations with VBPMM could potentially free-up resources which could be used on other, more therapeutically beneficial activities. However, in practice, there is no guarantee that this freed-up staff time will be used for these purposes, posing a risk of reducing therapeutic interactions between staff and patients [90]. Over-reliance on technology and increased staff workloads (e.g. due to additional documentation requirements or inappropriate use of technology to justify assigning more patients to each member of frontline staff) could also lead to less vigilant and proactive care. As a result, surveillance technologies could promote more passive practice, where staff observe patients from a distance instead of engaging in therapeutic interactions with them. There is therefore a risk that the use of surveillance technologies to reduce staffing costs could result in decreased human interaction and so quality of care in inpatient settings.

Qualitative findings revealed that staff, patient and carer perceptions and experiences were mixed across the surveillance technology types. Some of the perceived benefits of surveillance technologies included improved staff and patient safety, enhanced monitoring and prevention of incidents (e.g. absconding, self-harm, violence and aggression) and facilitation of less intrusive observations of patients. Providing evidence to help investigate incidents and complaints was another perceived benefit, although some noted that surveillance technologies do not necessarily capture the entirety of events (e.g. due to some being turned on and off at the discretion of staff, and because they may not capture all of the events leading up to an incident). Concerns were also expressed by staff and patients that surveillance technology use could have wide-ranging negative effects, including negatively impacting patients’ recovery, privacy and dignity, decreasing feelings of safety, exacerbating distress and paranoia, reducing quality of care, damaging therapeutic relationships with staff and exacerbating power imbalances between patients and staff. Indeed, patient and service user groups, along with advocates and disability activists, have consistently voiced concerns about the potential iatrogenic harms associated with the use of surveillance technology in inpatient mental health settings [29, 91]. These harms have been the subject of media attention [29, 92], including news reports that at a recent inquest, concerns were raised about “alert fatigue” associated with surveillance technology use, suggesting that it can even have fatal consequences [93].

However, many of the included studies did not comprehensively investigate potential impacts, including unintended consequences, quantitatively. For example, very few quantitatively investigated surveillance technology’s impact upon patients’ mental health, absconding rates, self-harm or care quality. Further, even when these outcomes were investigated, there may have been limitations in how they were measured. For example, Ndebele et al. [45] only measured self-harm frequency in bedrooms and bathrooms, and so they did not capture any possible impact of VBPMM on rates of self-harm in communal ward spaces or on self-harm severity. This is a concern, given reports from patients that VBPMM use can worsen self-harm [31]. Many possible effects were not investigated at all in any of the included studies, such as the impact of surveillance technologies on therapeutic alliances, treatment satisfaction, staff and patient well-being, patient quality of life, recovery, engagement with services and longer-term outcomes such as readmission rates and post-discharge mental health and service use. Therefore, this review shows that our understanding of the impact of surveillance technologies in inpatient mental health settings, including their full range of potential harms and risks, remains incomplete.

Methodological quality of the included studies

There were significant methodological limitations in half (50.0%) of the included studies. Furthermore, there were declared conflicts of interest in over a quarter of studies (28.1%), all in studies examining VBPMM and BWCs. We also identified an additional potential undisclosed conflict of interest, and another VBPMM paper which was retracted from an academic journal due to an undeclared conflict of interest by the authors [94]. We noted that several of the studies with positive findings had declared conflicts of interest relating to the technology of interest, for example, studies being funded or conducted by the technology company themselves. This may not be surprising given their drive to demonstrate the efficacy of their technology. Many of these studies were also rated as low quality. Results therefore need to be interpreted with caution.

There was often a lack of information about how participants were recruited, and how surveys and interviews or focus groups were conducted, making it difficult to assess potential biases (e.g. risk of cherry-picking participants, excluding the most unwell patients, power imbalances inhibiting sharing of criticisms of technology by patients and staff). Consequently, the literature may underrepresent the perspectives of populations facing greater barriers to research participation (e.g. patients lacking capacity to consent, people with concerns about confidentiality, distrust towards research or facing language barriers). The lack of transparency in methodologies, e.g. no pre-registration of studies, makes it difficult to ascertain how reported outcomes were chosen, and raises questions around whether negative outcomes (such as harms, verbal aggression and property destruction) were purposefully omitted. Methodologically, no randomised controlled trials were identified, and few studies had control groups, with mainly before and after comparisons. Many papers did not adequately consider the complexity of the issues and variables surrounding surveillance, for instance, the role of confounding or contextual factors in interpreting results.

There was in general a significant lack of lived experience involvement in the implementation and evaluation of surveillance technologies, and a lack of lived experience involvement in the studies themselves. Even when it was reported, it was often poorly described, for example, lacking detail about numbers of people involved, their demographics, recruitment methods and how (and to what degree) they were involved in the research process. Furthermore, in some studies there lacked a clear distinction between the involvement of individuals with lived experience in the research process versus participation in the study by patients.

Strengths and limitations

Our review is a comprehensive, systematic synthesis of the available literature on the implementation, experience and impact of surveillance technologies in inpatient mental health settings. We reported information on lived experience involvement in the study design and the implementation of the surveillance, exposing significant gaps which should be addressed and prioritised. We also reported information on declared conflicts of interest and funding in the included papers, which have enhanced our ability to assess the validity and independence of the evidence presented.

We sought to identify both academic and grey literature in our review, although, due to time constraints, grey literature was only included in relation to research objective 2 (exploring patient, staff and carer perceptions and experiences of the technology) and was limited to studies which included a description of their methodology. We acknowledge that there may be perspectives which are therefore underrepresented in our synthesis, including perspectives from those with lived experience of surveillance on inpatient wards. Another limitation is that we did not contact authors of included studies for missing information. Had this information been obtained, it may have impacted our quality appraisal ratings. There is a risk of publication bias (i.e. studies showing positive outcomes being more likely to be published) given the number of included studies which declared conflicts of interest, although we were unable to investigate and confirm this.

Implications for policy and practice

The findings of this review suggest that the current evidence base does not support the use of surveillance technologies as a means of improving safety, care quality or reducing costs in inpatient mental health settings.

More independent, coproduced research is needed thoroughly evaluating the impact of surveillance technologies, including their full range of potential harms, in inpatient settings. As is best practice with the implementation of any new intervention, they should only be deployed if the resulting evidence supports their use.

However, the current reality is that surveillance technologies are already being implemented across a variety of inpatient services across the globe, and it is unlikely that this will come to a complete halt. If these technologies continue to be implemented, there will be an urgent need to develop trauma-informed policies, procedures and guidelines for their use, centred around the perspectives of patients. This could contribute to developing more acceptable ways of using surveillance technologies and help maximise their potential benefits and mitigate their harms. These guidelines and policies would need to be accompanied by comprehensive and ongoing training for staff, ideally coproduced with patients, and systematic monitoring and auditing of services’ adherence to them to help ensure compliance.

These policies and guidelines should comprehensively address the tensions and ethical concerns highlighted by patients, carers and staff in this review. This includes concerns around informed consent, patient confidentiality, data protection and potential iatrogenic harms. They should also include procedures for informing patients about the use of surveillance technologies, routinely monitoring and discussing their usage, and investigating and addressing misuse of technology and data. Consideration should also be given to measures to protect patients’ dignity and privacy and prevent (re)traumatisation, for example, pixelating or blurring footage of sensitive areas of patients’ bodies when using video monitoring technologies. Wider systemic challenges, including issues such as staffing shortages, power imbalances and reliance on restrictive approaches to risk management, also need to be acknowledged and actively addressed.

It is essential that all stakeholders, particularly patients, are meaningfully involved in all stages of future research, implementation, evaluation and decision-making regarding surveillance technology use in inpatient mental health settings.

Implications for research

The literature base identified in this review is largely characterised by uncontrolled and poor-quality studies presenting inconsistent results. Over a quarter of papers identified in this review had declared conflicts of interest, and an additional potential undisclosed conflict of interest was also identified.

Future research on surveillance in inpatient wards should be funded and conducted independently to ensure the rigour and validity of the methods and findings. Conflicts of interest should also be declared in any published reports or articles. Research led by those with lived experience of mental health inpatient care generally, and surveillance technologies specifically, would be particularly valuable in evaluating potential harms missed by academic or clinical researchers. Care should also be taken to ensure that the perspectives of those who are unwell, or may need support to express their views, are captured in any future research on technologies in these settings [95]. Further synthesis of data on surveillance from other locations where people with mental health problems present may be helpful, for example in crisis services or mental health presentations in emergency departments.

Future primary research in this area should more purposefully aim to (i) investigate the harms caused by surveillance, including a full exploration of the psychological impact and an exploration of changes in care protocols due to the technology, (ii) explore and establish best practice and ethical guidelines for the use of surveillance in inpatient units (and across all mental health services and settings) which fully consider the experiences of patients who have negative views and adverse responses to surveillance, and (iii) include those with lived experience in study design, analysis, interpretation and dissemination.

Conclusions

This systematic review has found insufficient evidence to suggest that surveillance technologies in inpatient mental health settings are achieving their intended outcomes, such as improving safety, enhancing care quality and reducing costs. Qualitative findings indicate that patient, staff and carer experiences of surveillance technologies are mixed and complex. Whilst some people perceived potential benefits, such as reductions in incidents, less intrusive monitoring and provision of evidence to aid investigations, others raised concerns about their potential harms, including ethical concerns about their impact on patients’ human rights, privacy, dignity, and recovery. The reviewed studies were generally of low methodological quality, lacked lived experience involvement, and a considerable proportion disclosed conflicts of interest. Further independent, co-produced research is needed to more comprehensively evaluate the impact of surveillance technologies in inpatient settings, including their full range of potential benefits and harms.

Lived experience commentary, by Georgia Johnson and Rachel Rowan Olive

We are unsurprised by the poor quality and inconsistent results of the evidence. In our experience surveillance technology—like most restrictive practice—is rapidly rolled out in response to institutional anxiety following serious incidents. Surveillance technology’s illusion of control alleviates that anxiety, promising potential benefits well beyond the evidence base. Surveillance’s damage, however, is more concrete. Most researchers did not look for iatrogenic harm, thus compounding said harm by invalidating our fears and experiences.

But we know these harms intimately, because we have experienced them. These digital technologies strip away our most basic dignity, and are, by an extension, an affront to our very humanity. It is when professionals stop treating us like humans, and see only a cluster of symptoms, that restrictive practice becomes its most abusive self. Other people’s fear is not a justification for abusing us in this way.

The UK’s psychiatric system is not one where meaningful consent for surveillance can be implemented; however, blithely manufacturers and evaluators state that consent is always obtained. When Oxevision was piloted on Georgia’s ward, she was not given the opportunity to consent: she only discovered the system existed after a nurse said, “Oh you’re in the bathroom, I couldn’t see you on the camera.” Staff didn’t know whether patients were allowed to refuse it. The distress caused was so great that the response team had to be called. After turning the cameras off, they were turned back on during another shift. When Georgia objected, staff said that no such cameras existed and that she was experiencing psychosis. Would she like a cup of tea instead? FOI data from StopOxevision [96] shows this is not an isolated event, with patient leaflets and posters frequently omitting any mention of functionalities such as camera surveillance.

Finally, we highlight the contrast in attitudes to staff surveilling patients versus patients filming staff. On being illegally detained during a mental health crisis, Rachel began recording those detaining her, knowing we are frequently disbelieved when making complaints. Outraged staff wearing body-worn cameras promptly insisted, “we are not here to be filmed”.

This is a common response to patients documenting poor experiences; it puts paid to any illusion that institutional surveillance could lessen the violent disbelief we face. Staff control when and how cameras are used. Surveillance within this system only cements power imbalances and causes lasting trauma.