Introduction

‘Hybrid’ interventions in which some intervention components are fixed across sites and others are flexible (locally created) are thought to allow for adaptation to the local context while maintaining fidelity. However, there is little evidence regarding the challenges and facilitators of implementing hybrid interventions. This paper reports on a process evaluation of a patient safety hybrid intervention called Your Care Needs You (YCNY). YCNY was tested in the Partners at Care Transitions (PACT) randomised controlled trial and aimed to enhance older patients and their families’ involvement in their care in order to achieve safer transitions from hospital to home.

Methods

The process evaluation took place across eight intervention wards taking part in the PACT trial. 23 interviews and 37 informal conversations were conducted with National Health Service (NHS) staff. Patients (n=19) were interviewed twice, once in hospital and once after discharge. Interviews with staff and patients concerned the delivery and experiences of YCNY. Ethnographic observations (n=81 hours) of relevant activities (eg, multidisciplinary team meetings, handovers, etc) were undertaken.

Results

The main finding relates to how staff understood and engaged with YCNY, which then had a major influence on its implementation. While staff broadly valued the aims of YCNY, staff from seven out of the eight wards taking part in the process evaluation enacted YCNY in a mostly task-based manner. YCNY implementation often became a hurried activity which concentrated on delivering fixed intervention components rather than a catalyst for culture change around patient involvement. Factors such as understaffing, constraints on staff time and the COVID-19 pandemic contributed towards a ‘taskification’ of intervention delivery, which meant staff often did not have capacity to creatively devise flexible intervention components. However, one ward with a sense of distributed ownership of YCNY had considerable success implementing flexible components.

Discussion

Hybrid interventions may allow aspects of an intervention to be adapted to the local context. However, the current constrained and pressured environment of the NHS left staff with little ability to creatively engage with devising flexible intervention components, despite recognising the need for and being motivated to deliver the intervention.

Background

Search engines often serve as a primary resource for patients to obtain drug information. However, the search engine market is rapidly changing due to the introduction of artificial intelligence (AI)-powered chatbots. The consequences for medication safety when patients interact with chatbots remain largely unexplored.

Objective

To explore the quality and potential safety concerns of answers provided by an AI-powered chatbot integrated within a search engine.

Methodology

Bing copilot was queried on 10 frequently asked patient questions regarding the 50 most prescribed drugs in the US outpatient market. Patient questions covered drug indications, mechanisms of action, instructions for use, adverse drug reactions and contraindications. Readability of chatbot answers was assessed using the Flesch Reading Ease Score. Completeness and accuracy were evaluated based on corresponding patient drug information in the pharmaceutical encyclopaedia drugs.com. On a preselected subset of inaccurate chatbot answers, healthcare professionals evaluated likelihood and extent of possible harm if patients follow the chatbot’s given recommendations.

Results

Of 500 generated chatbot answers, overall readability implied that responses were difficult to read according to the Flesch Reading Ease Score. Overall median completeness and accuracy of chatbot answers were 100.0% (IQR 50.0–100.0%) and 100.0% (IQR 88.1–100.0%), respectively. Of the subset of 20 chatbot answers, experts found 66% (95% CI 50% to 85%) to be potentially harmful. 42% (95% CI 25% to 60%) of these 20 chatbot answers were found to potentially cause moderate to mild harm, and 22% (95% CI 10% to 40%) to cause severe harm or even death if patients follow the chatbot’s advice.

Conclusions

AI-powered chatbots are capable of providing overall complete and accurate patient drug information. Yet, experts deemed a considerable number of answers incorrect or potentially harmful. Furthermore, complexity of chatbot answers may limit patient understanding. Hence, healthcare professionals should be cautious in recommending AI-powered search engines until more precise and reliable alternatives are available.

Background

The raison d’etre of healthcare profession regulators across the globe is to protect patients and the public from the risk of harm. In cases of serious misconduct, remediation is deemed to be an important factor when considering the risk of harm from a practitioner under investigation. Yet, we know very little about how regulators account for remediation in their decision-making, and whether it is consistent with the aim of risk reduction. This paper explores the role of remediation in decision-making in cases of serious misconduct before UK healthcare regulators.

Methods

We conducted interviews with 21 participants from across eight of the nine UK healthcare profession regulators, covering a range of roles in the decision-making process in misconduct cases. Interviews were conducted remotely by video call and digitally transcribed. Data were analysed using the framework analysis method. The initial framework was developed from existing literature and guidance documents from the regulators, and was subsequently refined through the various rounds of coding.

Results

Remediation influenced decision-making in three ways: (1) Some types of misconduct were deemed more inherently remediable than others. In cases involving dishonesty or sexual misconduct, remediation was less likely to serve as a mitigating factor. (2) Decision-makers often view remediation as a proxy indicator of practitioner insight. (3) Whether a practitioner had demonstrated their commitment to change through undergoing remediation was more likely to feed into decision-making at the point where current impairment was under consideration.

Conclusions

Remediation plays a key role in decision-makers’ judgements in cases of misconduct, particularly when these cases relate to clinical misconduct. In such cases, remediation informs judgements on the levels of practitioner insight and the risk of such misconduct being repeated. Our results suggest a need to develop remediation interventions that are explicitly geared towards the regulatory function of developing practitioner insight. Regulators should also consider the structure of their fitness to practise processes and whether there are appropriate opportunities for judgements on remediation to feed into decisions and to facilitate balanced and proportionate outcomes.

Introduction

Quality improvement (QI) efforts are critical to promoting health equity and mitigating disparities in healthcare outcomes. Equity-focused QI (EF-QI) interventions address the unique needs of equity-deserving groups and the root causes of disparities. This scoping review aims to identify themes from EF-QI interventions that improve the health of equity-deserving groups, to serve as a resource for researchers embarking on QI.

Methods

In adherence with Preferred Reporting Items for Systematic Reviews and Meta-analysis guidelines, several healthcare and medical databases were systematically searched from inception to December 2022. Primary studies that report results from EF-QI interventions in healthcare were included. Reviewers conducted screening and data extraction using Covidence. Inductive thematic analysis using NVivo identified key barriers to inform future EF-QI interventions.

Results

Of 5,330 titles and abstracts screened, 36 articles were eligible for inclusion. They reported on EF-QI interventions across eight medical disciplines: primary care, obstetrics, psychiatry, paediatrics, oncology, cardiology, neurology and respirology. The most common focus was racialised communities (15/36; 42%). Barriers to EF-QI interventions included those at the provider level (training and supervision, time constraints) and institution level (funding and partnerships, infrastructure). The last theme critical to EF-QI interventions is sustainability. Only six (17%) interventions actively involved patient partners.

Discussion

EF-QI interventions can be an effective tool for promoting health equity, but face numerous barriers to success. It is unclear whether the demonstrated barriers are intrinsic to the equity focus of the projects or can be generalised to all QI work. Researchers embarking on EF-QI work should engage patients, in addition to hospital and clinic leadership in the design process to secure funding and institutional support, improving sustainability. To the best of our knowledge, no review has synthesised the results of EF-QI interventions in healthcare. Further studies of EF-QI champions are required to better understand the barriers and how to overcome them.

Generative artificial intelligence (AI) technologies have the potential to revolutionise healthcare delivery but require classification and monitoring of patient safety risks. To address this need, we developed and evaluated a preliminary classification system for categorising generative AI patient safety errors. Our classification system is organised around two AI system stages (input and output) with specific error types by stage. We applied our classification system to two generative AI applications to assess its effectiveness in categorising safety issues: patient-facing conversational large language models (LLMs) and an ambient digital scribe (ADS) system for clinical documentation. In the LLM analysis, we identified 45 errors across 27 patient medical queries, with omission being the most common (42% of errors). Of the identified errors, 50% were categorised as low clinical significance, 25% as moderate clinical significance and 25% as high clinical significance. Similarly, in the ADS simulation, we identified 66 errors across 11 patient visits, with omission being the most common (83% of errors). Of the identified errors, 55% were categorised as low clinical significance and 45% were categorised as moderate clinical significance. These findings demonstrate the classification system’s utility in categorising output errors from two different AI healthcare applications, providing a starting point for developing a robust process to better understand AI-enabled errors.

The healthcare delivered in high-income countries is riddled with defects in value. One in 10 patients experiences harm when receiving medical care, while nearly 13% of health expenditures are spent managing that harm.1 Half of patients with chronic disease are not on recommended therapy and suffer avoidable hospitalisations and ED visits, all while healthcare costs continue to increase as a percentage of GDP.2 3

Policymakers, health plans and health systems have responded to these challenges by working to improve value. While these efforts continue to mature, physicians are running up against the efficiency-thoroughness trade-off: to complete an increasing number of tasks in service of hitting quality metrics across their entire attributed population, they must decrease the time spent caring for each individual patient or increase the total amount of time they spend working. This paradox, however, is itself a product of how our...

Healthcare systems are currently buckling under the pressure of trying to manage the increasing demand for services. Nowhere is this pressure more acute than in intensive care units (ICUs). Technological developments, an ageing population, increased comorbidities and societal expectations about healthcare delivery and services have all driven demand for critical care resources to exceed capacity.1 ICUs amalgamate all medical and surgical specialties and support services to provide the best care for the most vulnerable and sickest hospital patients; they have been referred to as the ‘heart of the hospital’.2 Because of their pivotal role in providing complex care to different patient cohorts, ICUs require a flexible, nimble and adaptable workforce because when demand for ICU increases, the need for staff surges to meet this demand.3 Responding to resource challenges, increasing bed demands and the need for skilled and experienced staff requires significant adaptability...

In this issue of BMJ Quality & Safety, Hampton and colleagues report a process evaluation of an intervention trial intended to encourage older patients’ involvement in their hospital care.1

The logic of the intervention, Your Care Needs You (YCNY), was that more patient involvement in aspects of care in hospital will carry over to home after discharge, preventing avoidable repeat admissions. YCNY was described as a ‘hybrid’ intervention. Ward-level staff were obliged to deliver ‘fixed’ components—a booklet, and advice sheet and a video. But they were also invited to design and deliver ‘flexible’ components, that is, any other components that the ward team thought would also encourage patients to take part in the selected aspects of their care (some examples were offered by the investigators). One of their eight wards went all in, embracing the challenge of designing flexible components. But the others chose differently, keeping with...

The last 5 years have seen a rapid growth in research applying artificial intelligence or machine learning to improve the quality and safety of healthcare. This coincides with the release of web interfaces (such as ChatGPT from OpenAI and Copilot from Microsoft) that have enabled the general public (including health professionals and researchers) to easily access the latest generation of large language models (LLMs).

LLMs have fundamentally changed how machine learning is used across domains. Unlike previous generation systems that required careful data curation for specific tasks before training, modern LLMs work well with just a few examples or a simple problem description. This progress is mainly due to training on large volumes of web data that allows them to develop an ‘understanding’ of both language and general knowledge which they can then apply to a wide range of tasks.1

To fully comprehend the capabilities and associated...

Performance problems among healthcare professionals can have significant implications for patient safety. Estimates suggest approximately 6–12% of physicians experience performance issues,1 while about one in three healthcare professionals report encountering a poorly performing colleague within the past year.2 Performance problems can arise from individual-level causes including physical illness, substance use disorders, cognitive impairment, mood or personality disorders, and failure to acquire or maintain the knowledge and skills necessary to safely carry out their responsibilities.3 Furthermore, broader systemic issues, including excessive workloads, inadequate resources, lack of institutional support and poor workplace culture, can contribute to or exacerbate performance problems.4 The performance of healthcare professionals is generally evaluated against a set of standards or core competencies of a particular profession that commonly require health professionals to maintain the knowledge, procedural proficiencies, communication skills and professionalism to effectively care for patients. Deficiencies in any...

Background

Healthcare systems are operating under substantial pressures. Clinicians and managers are constantly having to make adaptations, which are typically improvised, highly variable and not coordinated across teams. This study aimed to identify and describe the types of everyday pressures in intensive care and the adaptive strategies staff use to respond, with the longer-term aim of developing practical and coordinated strategies for managing under pressure.

Methods

We conducted qualitative semi-structured interviews with 20 senior multidisciplinary healthcare professionals from intensive care units (ICUs) in 4 major hospitals in the UK. The interviews explored the everyday pressures faced by intensive care staff and the strategies they use to adapt. A thematic template analysis approach was used to analyse the data based on our previously empirically developed taxonomy of pressures and strategies.

Results

The principal source of pressure described was a shortage of staff with the necessary skills and experience to care for the increased numbers and complexity of patients which, in turn, increased staff workload and reduced patient flow. Strategies were categorised into anticipatory (in advance of anticipated pressures) and on the day. The dynamic and unpredictable demands on ICUs meant that strategies were mostly deployed on the day, most commonly by flexing staff, prioritisation of patients and tasks and increasing modes of communication and support.

Conclusions

ICU staff use a wide variety of adaptive strategies at times of pressure to minimise risk and maintain a reasonable standard of care for patients. These findings provide the foundation for a portfolio of strategies, which can be flexibly employed when under pressure. There is considerable potential for training clinical leaders and teams in the effective use of adaptive strategies.

Maintaining the safety and continuity of medicines at care transitions is a long-standing healthcare challenge and a global priority.1 Medication errors at hospital discharge are common and harmful: a systematic review reported a median rate of medication error and unintentional medication discrepancies of approximately 50% of adult and elderly patients, and adverse drug events affected a fifth of all discharged adult and older patients.2 Older people are particularly at risk, and more susceptible to the impact of errors, yet approximately a third to two-thirds of medication-related harm experienced by older people after discharge is considered preventable.3 Problems can often arise from care fragmentation when two or more healthcare organisations operating distinct work systems are involved in the care transition.4 Between those systems the tasks, tools and technologies are misaligned, and the resulting processes cause error and confuse patients.5

In...

Sutton et al report in this edition of BMJ Quality and Safety the findings from a qualitative process evaluation of a pilot patient-activated escalation system in four surgical wards in three English hospitals.1 The study is part of a £2.5 million National Institute for Health and Care Research grant2 testing an escalation system with the name ‘Early 3S’ (See it Early, Speak up Early and Save lives Early). In the participating wards, the system was publicised to patients through leaflets and posters. In two sites, the responder to telephone calls for help from patients and family was a nurse on the actual ward team; in one of these sites the programme was discontinued after 2 months and in the second site most calls were not picked up by the responder. In the third site, a critical care outreach team received seven calls in 8 months but identified no...

Objective

To develop and evaluate measures of patient work system factors in medication management that may be modifiable for improvement during the care transition from hospital to home among older adults.

Design, settings and participants

Measures were developed and evaluated in a multisite prospective observational study of older adults (≥65 years) discharged home from medical units of two US hospitals from August 2018 to July 2019.

Main measures

Patient work system factors for managing medications were assessed during hospital stays using six capacity indicators, four task indicators and three medication management practice indicators. Main outcomes were assessed at participants’ homes approximately a week after discharge for (1) Medication discrepancies between the medications taken at home and those listed in the medical record, and (2) Patient experiences with new medication regimens.

Results

274 of the 376 recruited participants completed home assessment (72.8%). Among capacity indicators, most older adults (80.6%) managed medications during transition without a caregiver, 41.2% expressed low self-efficacy in managing medications and 18.3% were not able to complete basic medication administration tasks. Among task indicators, more than half (57.7%) had more than 10 discharge medications and most (94.7%) had medication regimen changes. Having more than 10 discharge medications, more than two medication regimen changes and low self-efficacy in medication management increased the risk of feeling overwhelmed (OR 2.63, 95% CI 1.08 to 6.38, OR 3.16, 95% CI 1.29 to 7.74 and OR 2.56, 95% CI 1.25 to 5.26, respectively). Low transportation independence, not having a home caregiver, low medication administration skills and more than 10 discharge medications increased the risk of medication discrepancies (incidence rate ratio 1.39, 95% CI 1.01 to 1.91, incidence rate ratio 1.73, 95% CI 1.13 to 2.66, incidence rate ratio 1.99, 95% CI 1.37 to 2.89 and incidence rate ratio 1.91, 95% CI 1.24 to 2.93, respectively).

Conclusions

Patient work system factors could be assessed before discharge with indicators for increased risk of poor patient experience and medication discrepancies during older adults’ care transition from hospital to home.

Background

The management of acute deterioration following surgery remains highly variable. Patients and families can play an important role in identifying early signs of deterioration but effective contribution to escalation of care can be practically difficult to achieve. This paper reports the enablers and barriers to the implementation of patient-led escalation systems found during a process evaluation of a quality improvement programme Rescue for Emergency Surgery Patients Observed to uNdergo acute Deterioration (RESPOND).

Methods

The research used ethnographic methods, including over 100 hours of observations on surgical units in three English hospitals in order to understand the everyday context of care. Observations focused on the coordination of activities such as handovers and how rescue featured as part of this. We also conducted 27 interviews with a range of clinical and managerial staff and patients. We employed a thematic analysis approach, combined with a theoretically focused implementation coding framework, based on Normalisation Process Theory.

Results

We found that organisational infrastructural support in the form of a leadership support and clinical care outreach teams with capacity were enablers in implementing the patient-led escalation system. Barriers to implementation included making changes to professional practice without discussing the value and legitimacy of operationalising patient concerns, and ensuring equity of use. We found that organisational work is needed to overcome patient fears about disrupting social and cultural norms.

Conclusions

This paper reveals the need for infrastructural support to facilitate the implementation of a patient-led escalation system, and leadership support to normalise the everyday process of involving patients and families in escalation. This type of system may not achieve its goals without properly understanding and addressing the concerns of both nurses and patients.

Introduction

Deprescribing (medication dose reduction or cessation) is an integral component of appropriate prescribing. The extent to which deprescribing recommendations are included in clinical practice guidelines is unclear. This scoping review aimed to identify guidelines that contain deprescribing recommendations, qualitatively explore the content and format of deprescribing recommendations and estimate the proportion of guidelines that contain deprescribing recommendations.

Methods

Bibliographic databases and Google were searched for guidelines published in English from January 2012 to November 2022. Guideline registries were searched from January 2017 to February 2023. Two reviewers independently screened records from databases and Google for guidelines containing one or more deprescribing recommendations. A 10% sample of the guideline registries was screened to identify eligible guidelines and estimate the proportion of guidelines containing a deprescribing recommendation. Guideline and recommendation characteristics were extracted and language features of deprescribing recommendations including content, form, complexity and readability were examined using a conventional content analysis and the SHeLL Health Literacy Editor tool.

Results

80 guidelines containing 316 deprescribing recommendations were included. Deprescribing recommendations had substantial variability in their format and terminology. Most guidelines contained recommendations regarding for who (75%, n=60), what (99%, n=89) and when or why (91%, n=73) to deprescribe, however, fewer guidelines (58%, n=46) contained detailed guidance on how to deprescribe. Approximately 29% of guidelines identified from the registries sample (n=14/49) contained one or more deprescribing recommendations.

Conclusions

Deprescribing recommendations are increasingly being incorporated into guidelines, however, many guidelines do not contain clear and actionable recommendations on how to deprescribe which may limit effective implementation in clinical practice. A co-designed template or best practice guide, containing information on aspects of deprescribing recommendations that are essential or preferred by end-users should be developed and employed.

Trial registration number

osf.io/fbex4.

Background

Severe mental illness (SMI) incorporates schizophrenia, bipolar disorder, non-organic psychosis, personality disorder or any other severe and enduring mental health illness. Medication, particularly antipsychotics and mood stabilisers are the main treatment options. Medication optimisation is a hallmark of medication safety, characterised by the use of collaborative, person-centred approaches. There is very little published research describing medication optimisation with people living with SMI.

Objective

Published literature and two stakeholder groups were employed to answer: What works for whom and in what circumstances to optimise medication use with people living with SMI in the community?

Methods

A five-stage realist review was co-conducted with a lived experience group of individuals living with SMI and a practitioner group caring for individuals with SMI. An initial programme theory was developed. A formal literature search was conducted across eight bibliographic databases, and literature were screened for relevance to programme theory refinement. In total 60 papers contributed to the review. 42 papers were from the original database search with 18 papers identified from additional database searches and citation searches conducted based on stakeholder recommendations.

Results

Our programme theory represents a continuum from a service user’s initial diagnosis of SMI to therapeutic alliance development with practitioners, followed by mutual exchange of information, shared decision-making and medication optimisation. Accompanying the programme theory are 11 context-mechanism-outcome configurations that propose evidence-informed contextual factors and mechanisms that either facilitate or impede medication optimisation. Two mid-range theories highlighted in this review are supported decision-making and trust formation.

Conclusions

Supported decision-making and trust are foundational to overcoming stigma and establishing ‘safety’ and comfort between service users and practitioners. Avenues for future research include the influence of stigma and equity across cultural and ethnic groups with individuals with SMI; and use of trained supports, such as peer support workers.

PROSPERO registration number

CRD42021280980.

Identifying high and poorly performing organisations is common practice in healthcare. Often this is done within a frequentist inferential framework where statistical techniques are used that acknowledge that observed performance is an imperfect measure of underlying quality. Various methods are employed for this purpose, but the influence of chance on the degree of misclassification is often underappreciated. Using simulations, we show that the distribution of underlying performance of organisations flagged as the worst performers, using current best practices, was highly dependent on the reliability of the performance measure. When reliability was low, flagged organisations were likely to have an underlying performance that was near the population average. Reliability needs to reach at least 0.7 for 50% of flagged organisations to be correctly flagged and 0.9 to nearly eliminate incorrectly flagging organisations close to the overall mean. We conclude that despite their widespread use, techniques for identifying the best and worst performing organisations do not necessarily identify truly good and bad performers and even with the best techniques, reliable data are required.

Healthcare regulators are having trouble keeping up. There is always a lag between regulators getting on top of things and fast-paced changes in health systems. Care is continuously becoming more complex.1 Rapid technological shifts (eg, new-generation drugs, artificial intelligence (AI) and advances in genomics) are accelerating. This confers new opportunities for better care, but it also implies new risks which need to be regulated differently. Yet the current paradigm is largely predicated on regulators mainly inspecting and investigating harmful events in retrospect, responding after they occur.2 This is despite developments and innovations in proactive inspection methods, and more collaborative approaches.

Another problem is that regulators often argue for system-based approaches to adverse events, but then often act by sanctioning individuals—in part because this is what they are empowered to do. The regulatory logic is: assemble objective evidence and assess this against compliance to the standard...