Articles of Interest
Uncovering Latent Failures Using Human Factors Approach as a Diagnostic Tool for Quality Improvement in Orthopedic Surgery
Khan, A, Cohen, T, Shappell, S & Boquet, A. (2025). Uncovering Latent Failures Using Human Factors Approach as a Diagnostic Tool for Quality Improvement in Orthopedic Surgery. American Journal of Medical Quality, 40, 255-260. https://doi.org/10.1097/JMQ.0000000000000265
Abstract
Human factors significantly influence medical quality, especially in complex environments like orthopedic surgery, where latent failures can compromise patient safety. A total of 3168 intraoperative events were observed across 40 orthopedic procedures and classified using the Human Factors Analysis and Classification System (HFACS). Three trained coders independently applied HFACS across 4 tiers and 19 causal categories. Interrater reliability was measured through percent agreement and Fleiss' Kappa using unanimous, majority, and reconciled coding conditions. Nearly all observed disruptions (98.97%) were classified as preconditions to unsafe acts, most (68.75%) stemmed from crew resource management failures, distractions from personal electronic devices, poor communication, and sales representative presence. A total of 19.47% of disruptions were due to personal readiness, due to the sales representation supporting role in ensuring technologies. An additional 5.87% were due to physical environment issues like equipment noise.
Conclusions: The HFACS framework demonstrated strong reliability in identifying systemic weaknesses within orthopedic surgical workflows. These findings emphasize the urgent need for structured interventions that reduce distractions, improve team communication, and regulate vendor interactions in the operating room, all essential steps toward advancing safety and enhancing overall patient care quality.
When AI Becomes Overly Agreeable: New Research on the Risks of Sycophantic Chatbots
(2026). When AI Becomes Overly Agreeable: New Research on the Risks of Sycophantic Chatbots. Biomedical Safety & Standards, 56 (10), 185-186. doi: 10.1097/01.BMSAS.0001193568.10898.79.
Excerpt:
Artificial intelligence (AI) chatbots take a helpful, polite, and supportive tone. A new paper published in Science 1 suggests that agreeableness can become problematic when it becomes sycophancy. In “Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence,”1 researchers from Stanford University Department of Computer Science investigated how frequently AI systems tell users what they want to hear, and what effect that has on human judgment and behavior.
It will not surprise users to learn that sycophantic AI appears to be widespread. But even brief interactions with it can reduce accountability, weaken interest in repairing relationships, and increase dependence on the chatbot.
The researchers define sycophancy as excessive agreement, affirmation, or flattery directed toward users. In ordinary conversation, that may seem harmless or even reassuring. But in emotionally charged situations, particularly interpersonal conflict, the chatbots can reinforce a one-sided interpretation of events. The paper1 focuses on a scenario that is already familiar to many users: asking a chatbot for advice about whether they were justified in an argument, breakup, betrayal, or family dispute. According to the study, leading AI models often validate the user's perspective rather than challenge it, even when the behavior described involves deception, illegality, or clear harm.
Potential applications of artificial intelligence with large datasets for predicting food biotoxicity
Xing, Kuoran , Wang, Qiang , Dadol, Glebert Cañete , et al
Potential applications of artificial intelligence with large datasets for predicting food biotoxicity, Food Quality and Safety, vol 26 no 10, February 2026.
Abstract
Food safety is a critical global concern, as toxic substances in food pose serious risks to public health. With the rise of novel
food products such as cell-cultured, fermented, and genetically modified items, there is an urgent need for more efficient
and accurate methods to assess food toxicity. Traditional testing approaches often lack the speed, scalability, and
sensitivity needed to detect emerging toxicants. Omics-based technologies now offer comprehensive insights into
biological responses, enabling the identification of subtle or unknown toxic effects. However, the complexity and scale
of omics data present significant challenges for interpretation. To address this, artificial intelligence (AI) has emerged as
a powerful tool to analyze large datasets and improve toxicity prediction. In this review, we summarize key categories of
food toxicants, introduce omics technologies and publicly available databases, outline general AI modeling workflows,
and highlight recent applications of AI in food safety. Together, AI with large amount of food-related data are shaping
the future of food safety strategies.
Symptom Tracking in Primary Care: Designing and Testing an EHR-Embedded Safety-net System
Talya Salant, MDPhD; James Benneyan, PhD; Anthony Zhong, BS; Jasper Su, BS; Molly Hodsden; Seth J Berkowitz, MD; Russell Phillips, MD; Gordon D Schiff, MD
Joint Commission Journal on Quality and Patient Safety, The,
Abstract
High reliability in healthcare requires individuals and teams that perform at their best and model teamwork and respect, supporting psychological safety. When patterns of disrespectful behaviors undermine a culture of safety and respect, patients may suffer, 1-3 individual and team performance declines, 4-8 and culture is negatively impacted. 9 In response, health systems that pursue high reliability have progressed from ad hoc, reactive approaches to address unprofessional behaviors to research-based strategies that focus on early identification for individuals with emerging patterns of unprofessional behavior. 10,11 A small subset of individuals may be associated with persistent patterns of unprofessional behavior or a single serious event that may violate laws, regulations, or policies. These individuals might struggle with underlying challenges that interfere with their ability to self-regulate and some may need formal assessments to increase their chances of successful remediation.
A recent evaluation of a statewide physician and professional health program 12 offers useful insight into individuals referred for formal evaluation and interventions. Viewed through the lens of professional accountability, Crane et al , reinforces the necessity of a tiered approach to health care professionals’ behavior to ensure the best quality and safety outcomes and promote the right culture for patients and team members.
Global Burden of Adverse Effects of Medical Treatment, 1990-2021: Trends, Inequities, and Projections From the Global Burden of Disease Study 2021
Yang, J. (2026). Global Burden of Adverse Effects of Medical Treatment, 1990-2021: Trends, Inequities, and Projections From the Global Burden of Disease Study 2021. Journal of Patient Safety, 22 (3), 173-181. doi: 10.1097/PTS.0000000000001440.
Abstract
Objectives:
To examine the trends and future projections of adverse effects of medical treatment (AEMT) burden at the global, regional, and national levels based on the latest Global Burden of Disease (GBD) 2021 data.
Methods:
We analyzed the incidence, mortality, and disability-adjusted life years (DALYs) of AEMT based on the GBD 2021 data. Trends were assessed using the estimated annual percentage change (EAPC), and future projections were modeled via the Nordpred age-period-cohort (APC) framework.
Results:
Globally, incident cases of AEMT increased by 68.50% from 1990 to 2021, yet age-standardized incidence rate (ASIR) declined in most countries and territories, and the age-standardized mortality rate (ASMR) decreased. Regionally, high sociodemographic index (SDI) regions experienced a 1.72% annual rise in ASIR, whereas low SDI regions saw a 12.87% rise in mortality. National disparities were pronounced, with the United States and India bearing the highest burdens. Projections suggest declining incidence rates but rising deaths, driven by factors like demographic shifts and health care expansion.
Conclusions:
Despite progress in reducing mortality and disability rates, the absolute AEMT burden remains relatively heavy, with considerable inequities across development settings. The findings reveal a critical paradox that high-resource systems face escalating risks from complex interventions, whereas low-resource settings struggle with preventable fatalities. Our findings inform global safety initiatives by highlighting the paradoxical burdens in high- and low-resource systems, calling for tailored, context-specific policy responses.
Implementing Participation-level Goals to Improve Patient-Centeredness in Pediatric Rehabilitation
Tanner, K. , Boster, J. , Gates, E. , Rospert, A. , Coleman Casto, S. , O’Rourke, S. , Gillespie, J. & Bican, R. (2026). Implementing Participation-level Goals to Improve Patient-Centeredness in Pediatric Rehabilitation. Pediatric Quality and Safety, 11 (2), e874. doi: 10.1097/pq9.0000000000000874.
Abstract
Introduction:
The global goal of pediatric rehabilitation services is to increase the ability of children to participate in meaningful life activities. Services that do not include a participation-centered goal are unlikely to achieve this impact. In this study, a participation component was operationally defined as a reference to a person or place outside the therapy context, regardless of the practice setting or discipline. The aim was to increase the use of participation-level goals among patients seen across all departments in the Division of Clinical Therapies from 50% to 80% by December 31, 2022, and sustain this level for 6 months.
Methods:
We implemented Plan-Do-Study-Act cycles in accordance with the Model for Improvement endorsed by the Institute for Healthcare Improvement. We tailored interventions for 6 participating departments. Strategies included audit and feedback, targeted communication with teams, and the use of champions. Then, we sampled charts across departments. We analyzed data using a statistical process control chart with a baseline mean of 32.5% and a target of 80%.
Results:
The goal of 80% of charts containing a rehabilitation goal with a participation-level component was achieved after 24 months of interventions, reaching a new centerline of 82% and sustaining this level for 6 months.
Conclusions:
We implemented patient goals with a participation component across multiple rehabilitation departments within a large division of a major pediatric hospital, using quality improvement methodology. Departments benefited from general strategies (eg, reminders) and tailored interventions (eg, targeted communication) to achieve and maintain 80% compliance.
.png-638919983642827586.png)
Use this form to request
button.
at the top right corner of the screen.