Assessment of machine learning algorithms in national data to classify the risk of self-harm among young adults in hospital: A retrospective study.

BACKGROUND: Self-harm is one of the most common presentations at accident and emergency departments in the UK and is a strong predictor of suicide risk. The UK Government has prioritised identifying risk factors and developing preventative strategies for self-harm. Machine learning offers a potential method to identify complex patterns with predictive value for the risk of self-harm. METHODS: National data in the UK Mental Health Services Data Set were isolated for patients aged 18-30 years who started a mental health hospital admission between Aug 1, 2020 and Aug 1, 2021, and had been discharged by Jan 1, 2022. Data were obtained on age group, gender, ethnicity, employment status, marital status, accommodation status and source of admission to hospital and used to construct seven machine learning models that were used individually and as an ensemble to predict hospital stays that would be associated with a risk of self-harm. OUTCOMES: The training dataset included 23 808 items (including 1081 episodes of self-harm) and the testing dataset 5951 items (including 270 episodes of self-harm). The best performing algorithms were the random forest model (AUC-ROC 0.70, 95%CI:0.66-0.74) and the ensemble model (AUC-ROC 0.77 95%CI:0.75-0.79). INTERPRETATION: Machine learning algorithms could predict hospital stays with a high risk of self-harm based on readily available data that are routinely collected by health providers and recorded in the Mental Health Services Data Set. The findings should be validated externally with other real-world, prospective data. FUNDING: This study was supported by the Midlands and Lancashire Commissioning Support Unit.

How safe is primary care? A systematic review.

IMPORTANCE: Improving patient safety is at the forefront of policy and practice. While considerable progress has been made in understanding the frequency, causes and consequences of error in hospitals, less is known about the safety of primary care. OBJECTIVE: We investigated how often patient safety incidents occur in primary care and how often these were associated with patient harm. EVIDENCE REVIEW: We searched 18 databases and contacted international experts to identify published and unpublished studies available between 1 January 1980 and 31 July 2014. Patient safety incidents of any type were eligible. Eligible studies were critically appraised using validated instruments and data were descriptively and narratively synthesised. FINDINGS: Nine systematic reviews and 100 primary studies were included. Studies reported between <1 and 24 patient safety incidents per 100 consultations. The median from population-based record review studies was 2-3 incidents for every 100 consultations/records reviewed. It was estimated that around 4% of these incidents may be associated with severe harm, defined as significantly impacting on a patient's well-being, including long-term physical or psychological issues or death (range <1% to 44% of incidents). Incidents relating to diagnosis and prescribing were most likely to result in severe harm. CONCLUSIONS AND RELEVANCE: Millions of people throughout the world use primary care services on any given day. This review suggests that safety incidents are relatively common, but most do not result in serious harm that reaches the patient. Diagnostic and prescribing incidents are the most likely to result in avoidable harm. SYSTEMATIC REVIEW REGISTRATION: This systematic review is registered with the International Prospective Register of Systematic Reviews (PROSPERO CRD42012002304).

Assessing the relationship between chronic pain and cardiovascular disease: A systematic review and meta-analysis.

BACKGROUND AND AIMS: Chronic pain is a potentially disabling condition affecting one in three people through impaired physical function and quality of life. While the psychosocial impact of chronic pain is already well established, little is known about the potential biological consequences. Chronic pain may be associated with an increased prevalence of cardiovascular disease, an effect that has been demonstrated across a spectrum of chronic pain conditions including low back pain, pelvic pain, neuropathic pain and fibromyalgia. The aim of this study was to review and summarize the evidence for a link between chronic pain and cardiovascular disease. We sought to clarify the nature of the relationship by examining the basis for a dose-response gradient (whereby increasing pain severity would result in greater cardiovascular disease), and by evaluating the extent to which potentially confounding variables may contribute to this association. METHODS: Major electronic databases MEDLINE, EMBASE, Psychinfo, Cochrane, ProQuest and Web of Science were searched for articles reporting strengths of association between chronic pain (pain in one or more body regions, present for three months or longer) and cardiovascular outcomes (cardiovascular mortality, cardiac disease, and cerebrovascular disease). Meta-analysis was used to pool data analysing the association between chronic pain and the three principal cardiovascular outcomes. The impact of pain severity, and the role of potentially confounding variables were explored narratively. RESULTS: The searches generated 11,141 studies, of which 25 matched our inclusion criteria and were included in the review. Meta-analysis (of unadjusted study outcomes) demonstrated statistically significant associations between chronic pain and mortality from cardiovascular diseases: pooled odds ratio 1.20, (95% confidence intervals 1.05-1.36); chronic pain and cardiac disease: pooled odds ratio 1.73 (95% confidence intervals 1.42-2.04); and chronic pain and cerebrovascular disease: pooled odds ratio 1.81 (95% confidence intervals 1.51-2.10). The systematic review also found evidence supporting a dose-response relationship, with greater pain intensity and distribution producing a stronger association with cardiovascular outcomes. All of the included studies were based on observational data with considerable variation in chronic pain taxonomy, methodology and study populations. The studies took an inconsistent and incomplete approach in their adjustment for potentially confounding variables, making it impossible to pool data after adjustments for confounding variables, so it cannot be concluded that these associations are causal. CONCLUSIONS: Our review supports a possible dose-response type of association between chronic pain and cardiovascular disease, supported by a range of observational studies originating from different countries. Such research has so far failed to satisfactorily rule out that the association is due to confounding variables. What is now needed are further population based longitudinal studies that are designed to allow more robust exploration of a cause and effect relationship. IMPLICATIONS: Given the high prevalence of chronic pain in developed and developing countries our results highlight a significant, but underpublicized, public health concern. Greater acknowledgement of the potentially harmful biological consequences of chronic pain may help to support regional, national and global initiatives aimed at reducing the burden of chronic pain.

Harms from discharge to primary care: mixed methods analysis of incident reports.

BACKGROUND: Discharge from hospital presents significant risks to patient safety, with up to one in five patients experiencing adverse events within 3 weeks of leaving hospital. AIM: To describe the frequency and types of patient safety incidents associated with discharge from secondary to primary care, and commonly described contributory factors to identify recommendations for practice. DESIGN AND SETTING: A mixed methods analysis of 598 patient safety incident reports in England and Wales related to 'Discharge' from the National Reporting and Learning System. METHOD: Detailed data coding (with 20% double-coding), data summaries generated using descriptive statistical analysis, and thematic analysis of special-case sample of reports. Incident type, contributory factors, type, and level of harm were described, informing recommendations for future practice. RESULTS: A total of 598 eligible reports were analysed. The four main themes were: errors in discharge communication (n = 151; 54% causing harm); errors in referrals to community care (n = 136; 73% causing harm); errors in medication (n = 97; 87% causing harm); and lack of provision of care adjuncts such as dressings (n = 62; 94% causing harm). Common contributory factors were staff factors (not following referral protocols); and organisational factors (lack of clear guidelines or inefficient processes). Improvement opportunities include developing and testing electronic discharge methods with agreed minimum information requirements and unified referrals systems to community care providers; and promoting a safety culture with 'safe discharge' checklists, discharge coordinators, and family involvement. CONCLUSION: Significant harm was evident due to deficits in the discharge process. Interventions in this area need to be evaluated and learning shared widely.

A cross-sectional mixed methods study protocol to generate learning from patient safety incidents reported from general practice.

INTRODUCTION: Incident reports contain descriptions of errors and harms that occurred during clinical care delivery. Few observational studies have characterised incidents from general practice, and none of these have been from the England and Wales National Reporting and Learning System. This study aims to describe incidents reported from a general practice care setting. METHODS AND ANALYSIS: A general practice patient safety incident classification will be developed to characterise patient safety incidents. A weighted-random sample of 12,500 incidents describing no harm, low harm and moderate harm of patients, and all incidents describing severe harm and death of patients will be classified. Insights from exploratory descriptive statistics and thematic analysis will be combined to identify priority areas for future interventions. ETHICS AND DISSEMINATION: The need for ethical approval was waivered by the Aneurin Bevan University Health Board research risk review committee given the anonymised nature of data (ABHB R&D Ref number: SA/410/13). The authors will submit the results of the study to relevant journals and undertake national and international oral presentations to researchers, clinicians and policymakers.

Safety incidents in the primary care office setting.

BACKGROUND: In the United Kingdom, 26% of child deaths have identifiable failures in care. Although children account for 40% of family physicians' workload, little is known about the safety of care in the community setting. Using data from a national patient safety incident reporting system, this study aimed to characterize the pediatric safety incidents occurring in family practice. METHODS: We undertook a retrospective, cross-sectional, mixed methods study of pediatric reports submitted to the UK National Reporting and Learning System from family practice. Analysis involved detailed data coding using multiaxial frameworks, descriptive statistical analysis, and thematic analysis of a special-case sample of reports. Using frequency distributions and cross-tabulations, the relationships between incident types and contributory factors were explored. RESULTS: Of 1788 reports identified, 763 (42.7%) described harm to children. Three crosscutting priority areas were identified: medication management, assessment and referral, and treatment. The 4 incident types associated with the most harmful outcomes are errors associated with diagnosis and assessment, delivery of treatment and procedures, referrals, and medication provision. Poor referral and treatment decisions in severely unwell or vulnerable children, along with delayed diagnosis and insufficient assessment of such children, featured prominently in incidents resulting in severe harm or death. CONCLUSION: This is the first analysis of nationally collected, family practice-related pediatric safety incident reports. Recommendations to mitigate harm in these priority areas include mandatory pediatric training for all family physicians; use of electronic tools to support diagnosis, management, and referral decision-making; and use of technological adjuncts such as barcode scanning to reduce medication errors.

Errors in the management of cardiac arrests: an observational study of patient safety incidents in England.

OBJECTIVE: The aim of this qualitative study is to better understand the types of error occurring during the management of cardiac arrests that led to a death. METHODS: All patient safety incidents involving management of cardiac arrests and resulting in death which were reported to a national patient safety database over a 17-month period were analysed. Structured data from each report were extracted and these together with the free text, were subjected to content analysis which was inductive, with the coding scheme emerged from continuous reading and re-reading of incidents. RESULTS: There were 30 patient safety incidents involving management of cardiac arrests and resulting in death. The reviewers identified a main shortfall in the management of each cardiac arrest and this resulted in 12 different factors being documented. These were grouped into four themes that highlighted systemic weaknesses: miscommunication involving crash number (4/30, 13%), shortfalls in staff attending the arrest (4/30, 13%), equipment deficits (11/30, 36%), and poor application of knowledge and skills (11/30, 37%). CONCLUSION: The factors identified represent serious shortfalls in the quality of response to cardiac arrests resulting in death in hospital. No firm conclusion can be drawn about how many deaths in the study population would have been averted if the emergency had been managed to a high standard. The effective management of cardiac arrests should be considered as one of the markers of safe care within a healthcare organisation.

What is the risk of death or severe harm due to bone cement implantation syndrome among patients undergoing hip hemiarthroplasty for fractured neck of femur? A patient safety surveillance study.

OBJECTIVE: To estimate the risk of death or severe harm due to bone cement implantation syndrome (BCIS) among patients undergoing hip hemiarthroplasty for fractured neck of femur. SETTING: Hospitals providing secondary and tertiary care throughout the National Health Service (NHS) in England and Wales. PARTICIPANTS: Cases reported to the National Reporting and Learning System (NRLS) in which the reporter clearly describes severe acute patient deterioration associated with cement use in hip hemiarthroplasty for fractured neck of femur (assessed independently by two reviewers). OUTCOME MEASURES: Primary-number of reported deaths, cardiac arrests and periarrests per year. Secondary-timing of deterioration and outcome in relation to cement insertion. RESULTS: Between 2005 and 2012, the NRLS received 62 reports that clearly describe death or severe harm associated with the use of cement in hip hemiarthroplasty for fractured neck of femur. There was one such incident for every 2900 hemiarthroplasties for fractured neck of femur during the period. Of the 62 reports, 41 patients died, 14 were resuscitated from cardiac arrest and 7 from periarrest. Most reports (55/62, 89%) describe acute deterioration occurring during or within a few minutes of cement insertion. The vast majority of deaths (33/41, 80%) occurred on the operating table. CONCLUSIONS: These reports provide narrative evidence from England and Wales that cement use in hip hemiarthroplasty for fractured neck of femur is associated with instances of perioperative death or severe harm consistent with BCIS. In 2009, the National Patient Safety Agency publicised this issue and encouraged the use of mitigation measures. Three-quarters of the deaths in this study have occurred since that alert, suggesting incomplete implementation or effectiveness of those mitigation measures. There is a need for stronger evidence that weighs the risks and benefits of cement in hip hemiarthroplasty for fractured neck of femur.

Patient-safety-related hospital deaths in England: thematic analysis of incidents reported to a national database, 2010-2012.

BACKGROUND: Hospital mortality is increasingly being regarded as a key indicator of patient safety, yet methodologies for assessing mortality are frequently contested and seldom point directly to areas of risk and solutions. The aim of our study was to classify reports of deaths due to unsafe care into broad areas of systemic failure capable of being addressed by stronger policies, procedures, and practices. The deaths were reported to a patient safety incident reporting system after mandatory reporting of such incidents was introduced. METHODS AND FINDINGS: The UK National Health Service database was searched for incidents resulting in a reported death of an adult over the period of the study. The study population comprised 2,010 incidents involving patients aged 16 y and over in acute hospital settings. Each incident report was reviewed by two of the authors, and, by scrutinising the structured information together with the free text, a main reason for the harm was identified and recorded as one of 18 incident types. These incident types were then aggregated into six areas of apparent systemic failure: mismanagement of deterioration (35%), failure of prevention (26%), deficient checking and oversight (11%), dysfunctional patient flow (10%), equipment-related errors (6%), and other (12%). The most common incident types were failure to act on or recognise deterioration (23%), inpatient falls (10%), healthcare-associated infections (10%), unexpected per-operative death (6%), and poor or inadequate handover (5%). Analysis of these 2,010 fatal incidents reveals patterns of issues that point to actionable areas for improvement. CONCLUSIONS: Our approach demonstrates the potential utility of patient safety incident reports in identifying areas of service failure and highlights opportunities for corrective action to save lives.

Identifying systems failures in the pathway to a catastrophic event: an analysis of national incident report data relating to vinca alkaloids.

BACKGROUND: Catastrophic errors in healthcare are rare, yet the consequences are so serious that where possible, special procedures are put in place to prevent them. As systems become safer, it becomes progressively more difficult to detect the remaining vulnerabilities. Using inadvertent intrathecal administration of vinca alkaloids as an example, we investigated whether analysis of incident report data describing low-harm events could bridge this gap. METHODS: We studied nine million patient safety incidents reported from England and Wales between November 2003 and May 2013. We searched for reports relating to administration of vinca alkaloids in patients also receiving intrathecal medication, and classified the failures identified against steps in the relevant national protocol. RESULTS: Of 38 reports that met our inclusion criteria, none resulted in actual harm. The stage of the medication process most commonly involved was 'supply, transport and storage' (15 cases). Seven cases related to dispensing, six to documentation, and four each to prescribing and administration. Defences most commonly breached related to separation of intravenous vinca alkaloids and intrathecal medication in timing (n=16) and location (n=8); potential for confusion due to inadequate separation of these drugs therefore remains. Problems involved in six cases did not align with the procedural defences in place, some of which represented major hazards. CONCLUSIONS: We identified areas of concern even within the context of a highly controlled standardised national process. If incident reporting systems include and encourage reports of no-harm incidents in addition to actual patient harm, they can facilitate monitoring the resilience of healthcare processes. Patient safety incidents that produce the most serious harm are often rare, and it is difficult to know whether patients are adequately protected. Our approach provides a potential solution.

Comparison of intensive care unit medication errors reported to the United States' MedMarx and the United Kingdom's National Reporting and Learning System: a cross-sectional study.

The objective was to compare the characteristics of medication errors reported to 2 national error reporting systems by conducting a cross-sectional analysis of errors reported from adult intensive care units to the UK National Reporting and Learning System and the US MedMarx system. Outcome measures were error types, severity of patient harm, stage of medication process, and involved medications. The authors analyzed 2837 UK error reports and 56 368 US reports. Differences were observed between UK and US errors for wrong dose (44% vs 29%), omitted dose (8.6% vs 27%), and stage of medication process (prescribing: 14% vs 49%; administration: 71% vs 42%). Moderate/severe harm or death was reported in 4.9% of UK versus 3.4% of US errors. Gentamicin was cited in 7.4% of the UK versus 0.7% of the US reports (odds ratio = 9.25). There were differences in the types of errors reported and the medications most often involved. These differences warrant further examination.

Identification of poor performance in a national medical workforce over 11 years: an observational study.

BACKGROUND: Poorly performing doctors are a source of harm but do not commonly feature in discussions of patient safety. Few countries have national mechanisms to deal with these doctors; most opt for suspension and/or exclusion from clinical practice. This study reports on the 11-year experience of dealing with concerns about doctors' performance in the UK National Health Service (NHS). The aim of this study was to describe the frequency with which doctors were referred due to performance-related concerns, examine demographic and specialty differences, and identify the nature of the concerns prompting referral. METHODS: This observational study uses data collected by the National Clinical Assessment Service for each referral (n=6179 doctors) over an 11-year period (April 2001-March 2012) in England to examine the rate at which concerns about doctors' performance occur, understand differences in rates between practitioner groups, and changes over time. FINDINGS: The annual referral rate was five per 1000 doctors (95% CI 4.6 to 5.4). Doctors whose first medical qualification was gained outside the UK were more than twice as likely to be referred as UK-qualified doctors; male doctors were more than twice as likely to be referred as women doctors; and doctors in the late stages of their career were nearly six times as likely to be referred as early career doctors. DISCUSSION: The UK holds a consistently collected national dataset on performance concerns about doctors. This allows risk groups to be identified so that preventive action and early intervention can be targeted most effectively to reduce harm to patients. A feature of past handling of poor clinical performance has been late presentation and a lack of thematic study of causation.

Epidemiology of lifetime fracture prevalence in England: a population study of adults aged 55 years and over.

BACKGROUND: fractures remain a substantial public health problem but epidemiological studies using survey data are sparse. This study explores the association between lifetime fracture prevalence and socio-demographic factors, health behaviours and health conditions. METHODS: fracture prevalence was calculated using a combined dataset of annual, nationally representative health surveys in England (2002-07) containing 24,725 adults aged 55 years and over. Odds of reporting any fracture was estimated separately for each gender using logistic regression. RESULTS: fracture prevalence was higher in men than women (49 and 40%, respectively). In men, factors having a significant independent association with fracture included being a former regular smoker [odds ratios, OR: 1.18 (1.06-1.31)], having a limiting long-standing illness [OR: 1.47 (1.31-1.66)] and consuming >8 units of alcohol on the heaviest drinking day in the past week [OR: 1.65 (1.37-1.98)]. In women, significant factors included being separated/divorced [OR: 1.30 (1.10-1.55)], having a 12-item General Health Questionnaire (GHQ-12) score of 4+ [OR: 1.59 (1.27-2.00)], consuming >6 units of alcohol in the past week [OR: 2.07 (1.28-3.35)] and being obese [OR: 1.25 (1.03-1.51)]. CONCLUSION: a range of socio-demographic, health behaviour and health conditions, known to increase the risk of chronic disease and premature death, are also associated with fracture occurrence, probably involving the aetiological pathways of poor bone health and fall-related trauma.

Promoting patient safety in India: situational analysis and the way forward.

Unsafe healthcare is a well-recognized issue internationally and is attracting attention in India as well. Drawing upon the various efforts that have been made to address this issue in India and abroad, we explore how we can accelerate developments and build a culture of patient safety in the Indian health sector. Using five international case studies, we describe experiences of promoting patient safety in various ways to inform future developments in India. We offer a roadmap for 2020, which contains suggestions on how India could build a culture of patient safety.

Reducing the burden of surgical harm: a systematic review of the interventions used to reduce adverse events in surgery.

OBJECTIVE: To perform a systematic review of interventions used to reduce adverse events in surgery. BACKGROUND: Many interventions, which aim to improve patient safety in surgery, have been introduced to hospitals. Little is known about which methods provide a measurable decrease in morbidity and mortality. METHODS: MEDLINE, EMBASE, and Cochrane databases were searched from inception to Week 19, 2012, for systematic reviews, randomized controlled trials (RCTs), and cross-sectional and cohort studies, which reported an intervention aimed toward reducing the incidence of adverse events in surgical patients. The quality of observational studies was measured using the Newcastle-Ottawa Scale. RCTs were assessed using the Cochrane Collaboration's tool for assessing risk of bias. RESULTS: Ninety-one studies met inclusion criteria, 26 relating to structural interventions, 66 described modifying process factors. Only 17 (of 42 medium to high quality studies) reported an intervention that produced a significant decrease in morbidity and mortality. Structural interventions were: improving nurse to patient ratios (P = 0.008) and Intensive Care Unit (ITU) physician involvement in postoperative care (P < 0.05). Subspecialization in surgery reduced technical complications (P < 0.01). Effective process interventions were submission of outcome data to national audit (P < 0.05), use of safety checklists (P < 0.05), and adherence to a care pathway (P < 0.05). Certain safety technology significantly reduced harm (P = 0.02), and team training had a positive effect on patient outcome (P = 0.001). CONCLUSIONS: Only a small cohort of medium- to high-quality interventions effectively reduce surgical harm and are feasible to implement. It is important that future research remains focused on demonstrating a measurable reduction in adverse events from patient safety initiatives.