BMJ Quality & Safety最新文献

Objective: Overconfidence is an important source of medical error. This review analyses experimental studies of confidence in medical diagnosis to identify factors affecting clinicians' confidence in their diagnoses and how confidence impacts patient care.

Method: A scoping review of medical and psychological literature was conducted. Articles were categorised according to methodology and clinical specialty. Findings were analysed thematically. Our review methodology adheres to the JBI's Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist.

Data sources: We searched SCOPUS, MEDLINE, PsycINFO and Global Health. We then performed citation tracking within these papers' references to identify additional articles.

Eligibility criteria: Papers were included if they reported quantitative results from an empirical study in which participants reported their confidence or certainty during a diagnostic decision. Studies comprised several medical subdisciplines.

Results: 77 articles met the inclusion criteria. Across these articles, confidence was not found to be well-calibrated to true diagnostic accuracy regardless of clinician experience. We organised articles under two main themes: the determinants of confidence and the uses of confidence during the patient's care pathway. Confidence is found to be affected by several factors, including case complexity, early diagnostic differentials and the healthcare environment. Factors that affect confidence, but not accuracy, demonstrate how the two can become decoupled, resulting in overconfidence/underconfidence. Confidence is found to affect patient testing, medication administration and referral rates, among other clinical actions.

Conclusions: Improving the calibration of confidence should be a priority for medical education and clinical practice (eg, via decision aids). We propose a theoretical model of factors that affect diagnostic confidence/certainty. Such a model can inform future work on how appropriate diagnostic confidence can be prompted and communicated among clinicians.

Statistical process control (SPC) charts provide a natural approach to analysing time series data for healthcare quality improvement (QI) initiatives. A problem arising in practice is that having established baseline control limits, there is no accepted objective and transparent approach to deciding when to establish new control limits for a given chart. We present the Stable Shift Algorithm, a new algorithm to aid analysts by identifying when control limits should be re-established, partitioning a control chart of time series data into distinct time periods. The algorithm aims to achieve this while (1) using only the theory of SPC, familiar to many QI practitioners, (2) avoiding re-establishing limits prematurely and (3) remaining flexible to choice of basic parameters of typical control chart use in QI. This is achieved through the commonly used shift rule of control charts, applied to establish whether shifts warrant new control limits or not. We conducted a simulation study to evaluate the effectiveness of the algorithm in achieving its aims, and a case study demonstrating application of the algorithm to 557 time series of accident and emergency care measures for providers in England and Scotland. Simulation results show that the algorithm avoids premature re-establishment of limits more often than simply re-establishing at every shift rule break. Application of the algorithm to the accident and emergency measures demonstrates this is not achieved at the cost of excessive additional rule breaks that might indicate control limits do not represent the underlying process. The Stable Shift Algorithm offers a potentially highly valuable tool for QI practitioners and researchers undertaking SPC analyses, providing an automated, consistent and rigorous approach facilitating large-scale analyses.

Background: Valproate exposure during pregnancy causes major congenital malformations in 11% and neurodevelopmental disorders in 30-40% of exposed children. Despite regulatory interventions, including the European 2018 Pregnancy Prevention Programme, pregnancy exposures continued. We evaluated the impact of a Medicines and Healthcare products Regulatory Agency National Patient Safety Alert (NPSA), which mandated two-specialist approval for valproate prescribing in patients under 55 years.

Methods: We conducted a population-based interrupted time series analysis using England National Health Service (NHS) prescribing data from January 2022 to April 2025. In total, we analysed 4 879 978 valproate prescriptions-both new and repeat-across ~8000 general practices. The primary outcome was monthly patients prescribed valproate. Secondary analyses examined regional variation, possible prescribing displacement to hospitals and an equivalent time series design with lamotrigine and levetiracetam as control medications to distinguish valproate-specific effects from broader trends.

Results: Following the NPSA alert, there was a significant decrease in people prescribed valproate in primary care in England. Compared with the preintervention trend (prescribing had been increasing by +12.6 people per month), postintervention prescribing declined by -72.6 people per month, yielding a total change in trend of -85.2 people per month (95% CI -122.9 to -47.5, p<0.001). There was no immediate step change. All seven NHS regions demonstrated significant declines. Postintervention decreases were also seen for hospital prescribing. Lamotrigine and levetiracetam showed no intervention effect.

Conclusions: The 2023 NPSA alert achieved valproate prescribing reductions across England through mandatory procedural requirements. The two-specialist approval proved effective where previous educational interventions failed. However, given this restricts access to a highly effective antiepileptic drug, particularly for generalised epilepsies, prospective studies evaluating seizure control outcomes following valproate withdrawal are needed to assess the impact of these regulatory restrictions.

Background: Artificial intelligence large language models (LLMs) are increasingly used to inform clinical decisions but sometimes exhibit human-like cognitive biases when facing nuanced medical choices.

Methods: We tested whether new chain-of-thought reasoning LLMs might mitigate cognitive biases observed in physicians. We presented medical scenarios (n=10) to models released by DeepSeek, OpenAI and Google. Each scenario was presented in two versions that differed according to a specific bias (eg, surgery framed in survival vs mortality statistics). Responses were categorised and the extent of bias was measured by the absolute discrepancy between responses to different versions of the same scenario. The extent of intransigence (also termed dogma or inflexibility) was measured by Shannon entropy. The extent of deviance in each scenario was measured by comparing the average model response to the average practicing physician response (n=2507).

Results: DeepSeek-R1 mitigated 6 out of 10 cognitive biases observed in practicing physicians by generating intransigent all-or-none responses. The four biases that persisted were post hoc fallacy (34% vs 0%, p<0.001), decoy effects (44% vs 5%, p<0.001), Occam's razor fallacy (100% vs 0%, p<0.001) and hindsight bias (56% vs 0%, p<0.001). In every scenario, the average model response deviated substantially from the average response of practicing physicians (p<0.001 for all). Similar patterns of persistent specific biases, intransigent responses and substantial deviance from practicing physicians were also apparent in OpenAI and Google.

Conclusion: Some biases persist in chain-of-thought reasoning LLMs, and models tend to produce intransigent recommendations. These findings highlight the role of clinicians to think broadly, respect diversity and remain vigilant when interpreting chain-of-thought reasoning artificial intelligence LLMs in nuanced medical decisions for patients.

Introduction and aims: Women residing in lower socioeconomic status (SES) areas have lower breast cancer survival but it is not clear how differences in the quality of care received contribute to these disparities. We compared adherence to clinical practice guidelines (CPG) for the diagnosis and treatment of breast cancer and subsequent breast cancer survival between women residing in lower versus higher SES areas.

Methods: We conducted a multicentre population-based study of all new cases of invasive breast cancer in women diagnosed 2010-2014 in six Spanish provinces with population-based cancer registries (n=3206). Clinical data were extracted in the framework of the European Cancer High Resolution studies and vital status follow-up covered a minimum of 5 years. SES of the patient's residence was measured with the 2011 Spanish Deprivation Index. Adherence to CPG was measured with 16 indicators based on European and Spanish guidelines. Relative survival was modelled using flexible parametric models.

Results: There were no differences in the type of treatment received but women living in the lowest SES areas were less likely to undergo a sentinel lymph node biopsy, reconstruction after mastectomy, surgery within 30 days after pathological diagnosis and adjuvant treatment within 6 weeks after surgery. After accounting for demographic and clinical factors, women residing in lower SES areas had higher risk of death, HR=1.57 (95% CI 1.04, 2.36). Further accounting for adherence to CPG in the model, in particular having undergone a sentinel lymph node biopsy, eliminated the significant effect of SES.

Conclusions: Despite the overall coverage of the Spanish health system, women living in more deprived areas were less likely to receive care in line with CPG and had shorter survival.

Background: Self-management support (SMS) is a recommended component of cancer survivorship care that improves health-related quality of life and reduces healthcare utilisation. However, widespread implementation has been difficult to achieve, with a gap in the literature on system-wide implementation efforts. This study examines contextual factors perceived to influence SMS adoption and sustainment in cancer centres in the Republic of Ireland.

Method: Semistructured interviews were conducted with 47 key informants from 20 cancer organisations across community and hospital settings. Participants were asked to report the level of adoption and sustainment of SMS at their organisation. This information was used to categorise organisations as low, medium or high implementers. We conducted cross-case analysis following the principles of Framework Analysis. Using the Consolidated Framework for Implementation Research as a menu of constructs, we examined factors influencing adoption and sustainment and variation in levels of implementation.

Results: National policy, external accreditation, external financing opportunities and the presence of champions in organisations are influential early in the implementation process driving adoption. Healthcare provider-led programmes and evidence of SMS improving patient outcomes and aligning with an organisation's priorities are necessary to secure buy-in, particularly among senior leadership. An organisational culture of entrepreneurship enables adoption and sustainment, with resources and a culture supporting staff well-being enabling sustainment.

Conclusion: While national policy is a driver, additional factors related to programme attributes and local contextual features such as the presence of champions, organisational readiness and culture influence implementation. The results may be used for future evaluations of SMS implementation in cancer survivorship care and to inform the development of tailored implementation strategies.

Introduction: QI education is essential for resident physicians with established requirements from the Accreditation Council for Graduate Medical Education outlining the necessary components. Literature supports the inclusion of both didactic and experiential learning, however, most studies review knowledge and attitude based assessments of residency QI programs. In 2012, our pediatric residency program identified a gap in resident engagement in QI, which led to the formalization of a QI education program grounded in the Institute for Healthcare Improvement (IHI) Model for improvement with objective measures of QI projects.

Methods: Over 13 years, our program implemented interative interventions across three phases to enchance QI training. Initial core interventions involved the structure of didactics to teach core principles of QI and with time focusing on more individualized mentorship. Our efforts were guided through a residency QI committee. Our aim was to (1) increase the percentage of resident QI projects with SMART aims, and (2) increase the use of QI graphs (run charts and SPC charts) of resident QI projects. We tracked graduating resident exit survey satisfaction with the QI program as a balancing measure.

Results: From 2012 through 2025, over 300 resident completed 390 QI projects that were reviewed. The percentage of of projects with SMART aims and QI graphs increased over time. The percentage of projects applying for American Board of Pediatrics Part IV applications and presenting at conferences also increased. Resident satisfaction remained with with an average "Poor" rating of only 2.6%.

Conclusions: Using the model for improvement to assess objective measures in a residency QI program is both feasible and effective. This 13-year intiative demonstrates how systematic, iterative improvement can improve the rigor of resident QI projects. Future goals including leveraging informatics to further support residents with their projects and track downstream patient outcomes.