{ "sources": [ { "id": "abc-2021-texas-hospitals-water", "title": "Texas' power outages, water shortages put bigger strain on hospitals", "publisher": "ABC News", "url": "https://abcnews.com/US/texas-power-outages-water-shortages-put-bigger-strain/story?id=75970571", "type": "journalism", "published": "2021-02-18", "accessed": "2026-09-26", "tier": 2, "supports": [ "Galveston, Austin and Houston hospitals had water shortages or low pressure", "About 800 people sheltered in Southeast Texas EDs seeking oxygen and power for home equipment" ], "last_verified_live": "2026-09-26" }, { "id": "aha-2024-change-survey", "title": "AHA Survey: Change Healthcare Cyberattack Significantly Disrupts Patient Care, Hospitals' Finances", "publisher": "American Hospital Association", "url": "https://www.aha.org/2024-03-15-aha-survey-change-healthcare-cyberattack-significantly-disrupts-patient-care-hospitals-finances", "type": "journalism", "published": "2024-03-15", "accessed": "2026-09-26", "tier": 2, "supports": [ "Survey of nearly 1,000 hospitals (9-12 Mar 2024): 94% financially affected, 74% reported direct patient care impact, 83% cash-flow impact", "94% of hospitals reported financial impact; 74% reported direct patient-care impact", "Nearly 40% reported patients' difficulty accessing care due to prior-authorization delays", "About two-thirds had difficulty switching clearinghouses" ], "last_verified_live": "2026-09-26" }, { "id": "aha-change-underscores-preparedness", "title": "Change Healthcare Cyberattack Underscores Urgent Need to Strengthen Cyber Preparedness for Individual Health Care Organizations and as a Field", "publisher": "American Hospital Association", "url": "https://www.aha.org/change-healthcare-cyberattack-underscores-urgent-need-strengthen-cyber-preparedness-individual-health-care-organizations-and", "type": "official-report", "published": "2025", "accessed": "2026-09-26", "tier": 2, "supports": [ "Change Healthcare processes about 15 billion transactions a year, touching 1 in 3 patient records, and was the predominant source of more than 100 critical functions", "Attacks on mission-critical third parties can have worse national consequences than attacks on hospitals themselves; impact varied with cash reserves and vendor redundancy" ], "last_verified_live": "2026-09-26" }, { "id": "aiact-explorer-articles-14-26", "title": "EU AI Act Explorer: Article 14 (Human Oversight) and Article 26 (Obligations of Deployers)", "publisher": "Future of Life Institute (artificialintelligenceact.eu)", "url": "https://artificialintelligenceact.eu/article/14/", "type": "regulation", "published": "2024", "accessed": "2026-09-26", "tier": 3, "supports": [ "Art 14(4)(b) names automation bias; 14(4)(d)-(e) override and stop", "Art 26(2) deployers assign oversight to persons with competence, training, authority and support (https://artificialintelligenceact.eu/article/26/)" ], "last_verified_live": "2026-09-26" }, { "id": "anderson-bell-2012-epi-blackout", "title": "Lights out: impact of the August 2003 power outage on mortality in New York, NY", "publisher": "Epidemiology", "url": "https://doi.org/10.1097/ede.0b013e318245c61c", "type": "peer-reviewed", "published": "2012", "accessed": "2026-09-26", "tier": 1, "supports": [ "The 2003 blackout was associated with roughly 90 excess deaths in New York City; accidental deaths +122% and disease-related deaths +25%" ], "last_verified_live": "2026-09-26" }, { "id": "ap-2012-nyc-hospital-generators", "title": "What caused generators to fail at NYC hospitals?", "publisher": "CBS News / Associated Press", "url": "https://www.cbsnews.com/news/what-caused-generators-to-fail-at-nyc-hospitals/", "type": "journalism", "published": "2012-11-02", "accessed": "2026-09-26", "tier": 2, "supports": [ "NYU Langone's generators were on upper floors but its fuel system in a basement vault was breached by the surge; about 300 patients evacuated", "Bellevue's generators were on high floors but fuel pumps and tanks were in the basement, which took an estimated 17 million gallons of water; AP reported more than 700 patients evacuated", "AP reported that 23% of hospitals inspected in the first half of 2012 failed to meet backup power standards" ], "last_verified_live": "2026-09-26" }, { "id": "ap-2024-whisper-hospitals", "title": "Researchers say an AI-powered transcription tool used in hospitals invents things no one ever said", "publisher": "Associated Press (syndicated via Scripps News)", "url": "https://www.scrippsnews.com/us-news/researchers-say-an-ai-powered-transcription-tool-used-in-hospitals-invents-things-no-one-ever-said", "type": "journalism", "published": "2024-10-26", "accessed": "2026-09-26", "tier": 2, "supports": [ "Over 30,000 clinicians and 40 health systems used Nabla's Whisper-based tool; about 7 million medical visits transcribed", "Nabla's tool erases the original audio, so transcripts cannot be checked against it", "OpenAI warned Whisper should not be used in high-risk domains" ], "last_verified_live": "2026-09-26" }, { "id": "asgari-2025-npj-llm-hallucination-summarisation", "title": "A framework to assess clinical safety and hallucination rates of LLMs for medical text summarisation", "publisher": "npj Digital Medicine (Asgari E, et al.; authors affiliated with Tortus AI)", "url": "https://doi.org/10.1038/s41746-025-01670-7", "type": "peer-reviewed", "published": "2025-05-13", "accessed": "2026-09-26", "tier": 1, "supports": [ "Across 12,999 clinician-annotated sentences of LLM-generated clinical notes: 1.47% hallucination rate and 3.45% omission rate", "Prompt and workflow changes reduced major errors; authors are from a scribe vendor" ], "last_verified_live": "2026-09-26" }, { "id": "aspr-tracie-cyber-readiness-response", "title": "Healthcare System Cybersecurity: Readiness & Response Considerations (updated October 2022)", "publisher": "HHS ASPR TRACIE", "url": "https://files.asprtracie.hhs.gov/documents/aspr-tracie-healthcare-system-cybersercurity-readiness-response.pdf", "type": "guidance", "published": "2022-10", "accessed": "2026-09-26", "tier": 1, "supports": [ "Downtime categories by duration (e.g., A <=12 h, B >1 day, C >3 days) and a designated trigger person to declare downtime", "Err on the side of enacting longer downtime procedures when duration is uncertain", "Systems will not be restored at once; treat recovery as a continued part of downtime", "Plan migration of manual documentation back to electronic; some paper data may be deliberately not reconciled, documented", "Differentiate eradication of the threat actor from restoration of services", "Full restores from backup are hard to test because of disruption to patient services", "Staff need just-in-time training during an event despite exercises" ], "last_verified_live": "2026-09-26" }, { "id": "aspr-tracie-ehr-downtime-collection", "title": "Electronic Health Records and Downtime Procedures Topic Collection", "publisher": "HHS ASPR TRACIE", "url": "https://asprtracie.hhs.gov/technical-resources/66/electronic-health-records-and-downtime-procedures/", "type": "guidance", "published": "2025-12-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "Curated guidance on EHR downtime plans, drills, training and recovery (AMC PSO 2017, AMA 2017, ISMP 2022, MGH 2018 toolkit)" ], "last_verified_live": "2026-09-26" }, { "id": "aspr-tracie-facility-downtime-assessment", "title": "Health Care Facility-Level Extended Downtime Assessment", "publisher": "HHS ASPR TRACIE", "url": "https://files.asprtracie.hhs.gov/documents/aspr-tracie-health-facility-level-downtime-assessment.pdf", "type": "guidance", "published": "2026", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defined thresholds for activating downtime (e.g., expected duration over two hours) and who may activate", "HICS template for downtime with separate IT and clinical branches", "Pre-printed forms for a specific time limit (e.g., 48 hours); managing large paper volumes including scanning and staff hours", "Plan for staged reconciliation and recovery of paper documentation as systems are restored", "Complete power failure: training for ventilator/device batteries, bagging, gravity drips, tank oxygen, with paper checklists", "Turn off and unplug unused equipment to prevent surge damage on power restoration", "Direct monitoring of patients if remote telemetry/video monitoring fails" ], "last_verified_live": "2026-09-26" }, { "id": "aws-2021-us-east-1-pes", "title": "Summary of the AWS Service Event in the Northern Virginia (US-EAST-1) Region", "publisher": "Amazon Web Services", "url": "https://aws.amazon.com/message/12721/", "type": "official-report", "published": "2021-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "On 7 Dec 2021 congestion on AWS's internal network impaired services for about 7 hours", "AWS's own real-time monitoring and Service Health Dashboard updates were impaired, and its support contact center was down until 2:25 PM PST" ], "last_verified_live": "2026-09-26" }, { "id": "aws-2025-dynamodb-pes", "title": "Summary of the Amazon DynamoDB Service Disruption in the Northern Virginia (US-EAST-1) Region", "publisher": "Amazon Web Services", "url": "https://aws.amazon.com/message/101925/", "type": "official-report", "published": "2025-10", "accessed": "2026-09-26", "tier": 1, "supports": [ "Disruption ran from 11:48 PM PDT 19 Oct to 2:20 PM PDT 20 Oct 2025", "Root cause: a latent race condition in DynamoDB's automated DNS management left an empty DNS record for the regional endpoint", "Knock-on failures hit EC2 launches, Network Load Balancer, Lambda, ECS/EKS/Fargate, Amazon Connect, STS and IAM" ], "last_verified_live": "2026-09-26" }, { "id": "bainbridge-1983-automatica-ironies", "title": "Ironies of Automation", "publisher": "Automatica 19(6):775-779 (Pergamon / IFAC)", "url": "https://doi.org/10.1016/0005-1098(83)90046-8", "type": "peer-reviewed", "published": "1983-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Automation leaves operators the tasks designers could not automate, including takeover in abnormal conditions", "Physical skills deteriorate when not used; an experienced operator who monitors becomes an inexperienced one", "Vigilance cannot be sustained beyond about half an hour on low-event monitoring", "Recommends hands-on control each shift or simulator practice", "Automation can expand rather than eliminate problems for the human operator" ], "last_verified_live": "2026-09-26" }, { "id": "bdt-2017-princeton-rebuild", "title": "Cyber attack prompts Princeton Community Hospital to rebuild network", "publisher": "Bluefield Daily Telegraph (Blake Stowers)", "url": "https://www.bdtonline.com/news/cyber-attack-prompts-princeton-community-hospital-to-rebuild-network/article_8d56443a-5c67-11e7-97bf-f3ad35ecd117.html", "type": "journalism", "published": "2017-06-29", "tier": 2, "supports": [ "Attack hit 27 June 2017; identified as Petya; hospital rebuilding network" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "bea-af447-final-report-2012", "title": "Final Report: Accident on 1st June 2009 to the Airbus A330-203, flight AF 447 (presentation of the final report)", "publisher": "BEA (Bureau d'Enquetes et d'Analyses)", "url": "https://bea.aero/fileadmin/uploads/tx_elyextendttnews/presentation.rapport.final.05juillet2012.en_04.pdf", "type": "official-report", "published": "2012-07-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Pitot icing led to autopilot disconnection and alternate law", "Crew did not link airspeed loss to procedure or identify the stall", "Absence of high-altitude manual handling training; startle; no clear display of airspeed inconsistencies" ], "last_verified_live": "2026-09-26" }, { "id": "bea-af447-summary-2012", "title": "Safety Investigation Following the Accident on 1st June 2009 to the Airbus A330-203, flight AF447: summary (Le Bourget, 5 July 2012)", "publisher": "BEA (copy hosted by humanfactors101.com)", "url": "https://humanfactors101.com/wp-content/uploads/2020/12/summary-report-bea-5-july-2012.pdf", "type": "official-report", "published": "2012-07-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "216 passengers, 9 flight attendants and 3 pilots on board", "Review of pilot training did not show associated skills were developed and maintained", "Lack of practical training in high-altitude manual handling" ], "last_verified_live": "2026-09-26" }, { "id": "beatty-2006-phr-blackout-2003", "title": "Blackout of 2003: public health effects and emergency response", "publisher": "Public Health Reports", "url": "https://doi.org/10.1177/003335490612100109", "type": "peer-reviewed", "published": "2006", "accessed": "2026-09-26", "tier": 1, "supports": [ "During the August 2003 blackout NYC saw failure of multiple hospital emergency generators and risk to patients dependent on electrically powered equipment", "Electronic data feeds from hospital emergency departments to the city's syndromic surveillance system were lost" ], "last_verified_live": "2026-09-26" }, { "id": "beckers-2025-aws-tufts", "title": "AWS outage disrupts Tufts Medicine; other health systems unaffected", "publisher": "Becker's Hospital Review (Naomi Diaz)", "url": "https://www.beckershospitalreview.com/healthcare-information-technology/aws-outage-disrupts-tufts-medicine-other-health-systems-unaffected/", "type": "journalism", "published": "2025-10-21", "accessed": "2026-09-26", "tier": 2, "supports": [ "Tufts Medicine, which runs Epic on AWS, reported slowdowns and minor delays in processing lab results; it said clinical care was not compromised", "Baptist Memorial Health Care (Epic on AWS) and Montefiore reported no disruption" ], "last_verified_live": "2026-09-26" }, { "id": "berinato-2003-cio-all-systems-down", "title": "All Systems Down", "publisher": "CIO / Computerworld (Scott Berinato)", "url": "https://www.computerworld.com/article/1346623/all-systems-down.html", "type": "journalism", "published": "2003-02-25", "accessed": "2026-09-26", "tier": 2, "supports": [ "BIDMC network failed repeatedly over about four days from 13 Nov 2002; reverted to paper, runners", "Lab turnaround rose from under 45 minutes to as long as five hours; ED closed for four hours", "Cause: spanning-tree loop; PACS network rebuilt overnight" ], "last_verified_live": "2026-09-26" }, { "id": "bloomberg-2025-synnovis-harm", "title": "NHS Ransomware Hack Caused Patient Harm in UK, Data Shows", "publisher": "Bloomberg News (Ryan Gallagher)", "url": "https://www.bloomberg.com/news/articles/2025-01-14/nhs-ransomware-hack-caused-patient-harm-in-uk-data-shows", "type": "journalism", "published": "2025-01-14", "accessed": "2026-09-26", "tier": 2, "supports": [ "NHS data obtained via FOI recorded 2 cases of major (long-term) harm, 11 moderate and more than 120 minor harm linked to the Synnovis attack", "Over 10,000 appointments postponed and more than 1,700 elective procedures cancelled" ], "last_verified_live": "2026-09-26" }, { "id": "budzyn-2025-lancet-gh-deskilling", "title": "Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study", "publisher": "The Lancet Gastroenterology & Hepatology 10(10):896-903", "url": "https://doi.org/10.1016/S2468-1253(25)00133-5", "type": "peer-reviewed", "published": "2025-08-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "Non-AI adenoma detection rate fell from 28.4% to 22.4% after AI introduction at four Polish centres", "AI exposure OR 0.69 for adenoma detection in non-AI procedures" ], "last_verified_live": "2026-09-26" }, { "id": "bulson-2024-jbcep-quarterly-exercises", "title": "Electronic health record downtime responses: One health system's process for ongoing readiness", "publisher": "Journal of Business Continuity & Emergency Planning 18(1):39-48", "url": "https://pubmed.ncbi.nlm.nih.gov/39164867/", "type": "peer-reviewed", "published": "2024", "accessed": "2026-09-26", "tier": 1, "supports": [ "Midwest US system exercises part of its downtime process quarterly before scheduled EHR upgrades" ], "last_verified_live": "2026-09-26" }, { "id": "casey-2020-cehr-power-outages", "title": "Power Outages and Community Health: a Narrative Review", "publisher": "Current Environmental Health Reports", "url": "https://doi.org/10.1007/s40572-020-00295-0", "type": "peer-reviewed", "published": "2020", "accessed": "2026-09-26", "tier": 1, "supports": [ "Review of 50 studies linking outages to CO poisoning, temperature-related illness, mortality and cardiovascular, respiratory and renal hospitalizations, especially among users of electricity-dependent medical equipment" ], "last_verified_live": "2026-09-26" }, { "id": "castro-delgado-2025-pdm-blackout-ems", "title": "Blackout in Spain: Urgent Analysis of Impact on Emergency Medical Services", "publisher": "Prehospital and Disaster Medicine", "url": "https://doi.org/10.1017/s1049023x25101556", "type": "peer-reviewed", "published": "2025-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "The blackout disrupted EMS for over ten hours: communication network collapse affected emergency calls; traffic signal failures; fuel shortages", "Patients using home medical devices faced life-threatening situations; at least one death from ventilator failure" ], "last_verified_live": "2026-09-26" }, { "id": "cbs-2023-unitedhealth-nh-predict-lawsuit", "title": "UnitedHealth uses faulty AI to deny elderly patients medically necessary coverage, lawsuit claims", "publisher": "CBS News", "url": "https://www.cbsnews.com/news/unitedhealth-lawsuit-ai-deny-claims-medicare-advantage-health-insurance-denials/", "type": "journalism", "published": "2023-11-20", "accessed": "2026-09-26", "tier": 2, "supports": [ "Class action alleges nH Predict had a 90% error rate and overrode physicians' determinations (allegation)", "Complaint alleges only about 0.2% of policyholders appeal (allegation)", "naviHealth said the tool is a guide, not used to make coverage determinations" ], "last_verified_live": "2026-09-26" }, { "id": "cemri-2025-mast-arxiv", "title": "Why Do Multi-Agent LLM Systems Fail?", "publisher": "arXiv preprint 2503.13657 (Cemri M, Pan MZ, Yang S, et al.)", "url": "https://arxiv.org/abs/2503.13657", "type": "preprint", "published": "2025-03-17", "accessed": "2026-09-26", "tier": 3, "supports": [ "Taxonomy of 14 failure modes in 3 categories (system design, inter-agent misalignment, task verification) from 1,600+ annotated traces across 7 multi-agent frameworks" ], "last_verified_live": "2026-09-26" }, { "id": "cio-2003-bidmc-network", "title": "Halamka on Beth Israel's Health-Care IT Disaster", "publisher": "CIO Magazine (Scott Berinato)", "url": "https://www.cio.com/article/270069/networking-halamka-on-beth-israel-s-health-care-it-disaster.html", "type": "journalism", "published": "2003-02-15", "accessed": "2026-09-26", "tier": 2, "supports": [ "A spanning-tree loop on an all-Layer-2 network made the Beth Israel Deaconess network flap for about four days (13-17 Nov 2002)", "Staff reverted to paper; lab turnaround went from 45 minutes to 5 hours; the ED diverted for hours on two days", "CIO lesson: treat the network with lifecycle management, not as a utility; DR plans must cover network loss, not just data loss" ], "last_verified_live": "2026-09-26" }, { "id": "circleofblue-2021-austin-hospitals-water", "title": "Boil-Water Advisory in Effect, Low Water Pressure Impacts Austin Hospitals", "publisher": "Circle of Blue", "url": "https://www.circleofblue.org/2021/world/boil-water-advisory-in-effect-low-water-pressure-impacts-austin-hospitals/", "type": "journalism", "published": "2021-02-18", "accessed": "2026-09-26", "tier": 2, "supports": [ "Power loss at water treatment reduced pressure at Austin hospitals", "St David's South Austin lost heat (water-dependent boiler) and began moving patients", "CEO: no one hospital had capacity to accept a large number of transfers" ], "last_verified_live": "2026-09-26" }, { "id": "cisa-2024-crowdstrike-alert", "title": "Widespread IT Outage Due to CrowdStrike Update", "publisher": "CISA", "url": "https://www.cisa.gov/news-events/alerts/2024/07/19/widespread-it-outage-due-crowdstrike-update", "type": "guidance", "published": "2024-07-19", "accessed": "2026-09-26", "tier": 1, "supports": [ "Outage caused by a CrowdStrike update, not a cyberattack", "Threat actors used the outage for phishing" ], "last_verified_live": "2026-09-26" }, { "id": "cisa-aa24-131a-black-basta", "title": "#StopRansomware: Black Basta (AA24-131A)", "publisher": "CISA, FBI, HHS, MS-ISAC", "url": "https://www.cisa.gov/news-events/cybersecurity-advisories/aa24-131a", "type": "guidance", "published": "2024-05-10", "accessed": "2026-09-26", "tier": 1, "supports": [ "Black Basta affiliates hit 500+ organizations incl. the HPH sector using double extortion", "HPH mitigations: patch known exploited vulnerabilities, phishing-resistant MFA, offline backups of critical systems and device configurations, secure remote access" ], "last_verified_live": "2026-09-26" }, { "id": "cisa-icsma-25-030-01", "title": "ICS Medical Advisory ICSMA-25-030-01: Contec Health CMS8000 Patient Monitor (Update A)", "publisher": "Cybersecurity and Infrastructure Security Agency", "url": "https://www.cisa.gov/news-events/ics-medical-advisories/icsma-25-030-01", "type": "official-report", "published": "2025-01-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "CVE-2025-0626 hidden functionality with hard-coded IP addresses; CVE-2024-12248 out-of-bounds write CVSS v3.1 9.8", "Firmware enables network interfaces even if disabled, bypassing device settings", "Update A published 25 Feb 2025", "Update A (25 Feb 2025) added CVE-2025-1204 and revised mitigations; the advisory still says the function could serve as a backdoor" ], "last_verified_live": "2026-09-26" }, { "id": "claroty-team82-contec-2025", "title": "Do the CONTEC CMS8000 Patient Monitors Contain a Chinese Backdoor? The Reality is More Complicated…", "publisher": "Claroty Team82", "url": "https://claroty.com/team82/research/are-contec-cms8000-patient-monitors-infected-with-a-chinese-backdoor-the-reality-is-more-complicated", "type": "vendor-research", "published": "2025-02-02", "tier": 3, "supports": [ "Team82 concluded the hard-coded IP behaviour is most likely an insecure design, not a hidden backdoor", "IP address is listed in vendor/reseller manuals as the CMS server address", "Update flow requires a physical button press at boot" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "cms-2016-ep-final-rule", "title": "Medicare and Medicaid Programs; Emergency Preparedness Requirements for Medicare and Medicaid Participating Providers and Suppliers (81 FR 63860)", "publisher": "Federal Register / CMS", "url": "https://www.federalregister.gov/documents/2016/09/16/2016-21404/medicare-and-medicaid-programs-emergency-preparedness-requirements-for-medicare-and-medicaid", "type": "regulation", "published": "2016-09-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "Final rule establishing national emergency preparedness requirements, effective 2016-11-15", "Incorporates by reference NFPA 99 (2012), NFPA 101 (2012) and NFPA 110 (2010)", "Flood-free generator location requirement applies to new construction, renovation or new generator installation, not existing installations", "Cites Hurricane Sandy reports of hospitals receiving evacuated patients with little or no medical documentation" ], "last_verified_live": "2026-09-26" }, { "id": "cms-2016-fire-safety-rule", "title": "Medicare and Medicaid Programs; Fire Safety Requirements for Certain Health Care Facilities (81 FR 26872)", "publisher": "Federal Register / CMS", "url": "https://www.federalregister.gov/documents/2016/05/04/2016-10043/medicare-and-medicaid-programs-fire-safety-requirements-for-certain-health-care-facilities", "type": "regulation", "published": "2016-05-04", "accessed": "2026-09-26", "tier": 1, "supports": [ "CMS adopted the 2012 Life Safety Code and 2012 NFPA 99 Health Care Facilities Code", "NFPA 99 applies requirements by risk category; Category 1 = failure likely to cause major injury or death", "CMS states facilities must load-test emergency power systems monthly per NFPA 110 (2010) section 8.4.1" ], "last_verified_live": "2026-09-26" }, { "id": "cms-2024-ma-faq-algorithms", "title": "Frequently Asked Questions related to Coverage Criteria and Utilization Management Requirements in CMS Final Rule (CMS-4201-F) (HPMS memo)", "publisher": "Centers for Medicare & Medicaid Services", "url": "https://www.cms.gov/files/document/hpms-memo-faq-coverage-criteria-and-utilization-management-cms-4201-f-02-6-2024-pdf.pdf", "type": "guidance", "published": "2024-02-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "An algorithm may assist MA coverage decisions but must base the decision on the individual patient's circumstances under 42 CFR 422.101(c)", "A length-of-stay prediction alone cannot be the basis to terminate post-acute care" ], "last_verified_live": "2026-09-26" }, { "id": "cms-safer-guides-attestation-2023", "title": "SAFER Guides requirements: Medicare Promoting Interoperability Program and MIPS Promoting Interoperability (infographic)", "publisher": "Centers for Medicare & Medicaid Services", "url": "https://www.cms.gov/files/document/cms-safer-guides-infographic-2023.pdf", "type": "regulation", "published": "2023", "accessed": "2026-09-26", "tier": 1, "supports": [ "Eligible hospitals and CAHs must attest yes/no to an annual self-assessment using all nine SAFER Guides (including Contingency Planning and System Interfaces); MIPS clinicians use the High Priority Practices guide", "Both 'yes' and 'no' satisfy the measure" ], "last_verified_live": "2026-09-26" }, { "id": "cms-som-appendix-z", "title": "State Operations Manual Appendix Z: Emergency Preparedness for All Provider and Certified Supplier Types, Interpretive Guidance (Rev. 204)", "publisher": "CMS", "url": "https://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/downloads/som107ap_z_emergprep.pdf", "type": "guidance", "published": "2021-04-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defines tabletop exercise as discussion-based, assessing plans without deploying resources", "Exercises should not test the same scenario year after year", "After-action review should cover what was supposed to happen, what occurred, what went well, what to improve, and a timed improvement plan", "Risk assessment should consider interruptions in communication including cyber-attacks" ], "last_verified_live": "2026-09-26" }, { "id": "cook-1998-how-complex-systems-fail", "title": "How Complex Systems Fail", "publisher": "Richard I. Cook, Cognitive Technologies Laboratory, University of Chicago", "url": "https://how.complexsystems.fail/", "type": "book", "published": "1998", "accessed": "2026-09-26", "tier": 2, "supports": [ "Human operators are both producers and defenders against failure", "Complex systems run in degraded mode", "Failure-free operations require experience with failure" ], "last_verified_live": "2026-09-26" }, { "id": "cook-rasmussen-2005-going-solid", "title": "\"Going solid\": a model of system dynamics and consequences for patient safety", "publisher": "Quality and Safety in Health Care 14(2):130-134", "url": "https://doi.org/10.1136/qshc.2003.009530", "type": "peer-reviewed", "published": "2005", "accessed": "2026-09-26", "tier": 1, "supports": [ "Efficiency gains remove buffers and tightly couple hospital operations ('going solid')" ], "last_verified_live": "2026-09-26" }, { "id": "courthousenews-2025-uhg-ai-lawsuit-ruling", "title": "Federal judge trims AI denial lawsuit against UnitedHealth Group", "publisher": "Courthouse News Service", "url": "https://www.courthousenews.com/federal-judge-dismisses-several-claims-in-ai-denial-lawsuit-against-uhg-but-case-will-proceed/", "type": "journalism", "published": "2025-02-13", "accessed": "2026-09-26", "tier": 2, "supports": [ "Judge Tunheim dismissed five of seven counts but let breach-of-contract and good-faith claims proceed", "Court waived exhaustion of administrative appeals, calling the process futile" ], "last_verified_live": "2026-09-26" }, { "id": "crowdstrike-rca-channel-file-291", "title": "External Technical Root Cause Analysis - Channel File 291", "publisher": "CrowdStrike", "url": "https://www.crowdstrike.com/wp-content/uploads/2024/08/Channel-File-291-Incident-Root-Cause-Analysis-08.06.2024.pdf", "type": "official-report", "published": "2024-08-06", "accessed": "2026-09-26", "tier": 3, "supports": [ "Template type defined 21 input fields but the sensor interpreter supplied 20; no runtime bounds check; content validator logic error", "Remediation includes staged deployment with customer control of content updates", "21-field template vs 20 supplied inputs caused out-of-bounds read", "Mismatch was latent through testing because of wildcard matching", "Remediation includes staged deployment rings and customer control of content rollout" ], "last_verified_live": "2026-09-26" }, { "id": "cvach-2012-bit-alarm-fatigue", "title": "Monitor alarm fatigue: an integrative review", "publisher": "Biomedical Instrumentation & Technology 46(4):268-277 (AAMI)", "url": "https://doi.org/10.2345/0899-8205-46.4.268", "type": "peer-reviewed", "published": "2012", "accessed": "2026-09-26", "tier": 1, "supports": [ "Alarm fatigue linked to high false alarm rate, poor positive predictive value, lack of standardisation" ], "last_verified_live": "2026-09-26" }, { "id": "cylera-contec-2025", "title": "Contec Patient Vital Signs Monitor: Chinese Backdoor or Bad Design?", "publisher": "Cylera (Chad Waters, Apostolos Bakoyiannis)", "url": "https://cylera.com/blog/contec-patient-vital-signs-monitor-chinese-backdoor-bad-design/", "type": "vendor-research", "published": "2025-02-04", "tier": 3, "supports": [ "Second vendor analysis: not an intentional backdoor; use of public IPv4 range internally" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "dameff-2018-jem-cyber-simulation", "title": "Clinical Cybersecurity Training Through Novel High-Fidelity Simulations", "publisher": "Journal of Emergency Medicine 56(2):233-238", "url": "https://doi.org/10.1016/j.jemermed.2018.10.029", "type": "peer-reviewed", "published": "2018-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "In simulations, clinicians did not recognise that patient pathology was caused by a compromised device" ], "last_verified_live": "2026-09-26" }, { "id": "dameff-2023-jama-netw-open-adjacent-eds", "title": "Ransomware Attack Associated With Disruptions at Adjacent Emergency Departments in the US", "publisher": "JAMA Network Open (Dameff C, Tully J, Chan TC, et al.)", "url": "https://doi.org/10.1001/jamanetworkopen.2023.12270", "type": "peer-reviewed", "published": "2023-05-08", "accessed": "2026-09-26", "tier": 1, "supports": [ "A month-long ransomware attack on a 4-hospital system (May 2021, San Diego County) was associated with a 35.2% rise in EMS arrivals, 47.6% longer median waits, 127.8% more left-without-being-seen, and more stroke code activations (59 to 102-103) and confirmed strokes (22 to 47) at two unaffected neighbouring academic EDs", "Authors argue cyberattacks should be treated as regional disasters", "Attacked organisation held ~25% of regional inpatient discharges", "Unaffected adjacent EDs saw higher census, ambulance arrivals, LWBS, waits, stroke activations and EMS diversion", "May 2021 attack on a 4-hospital San Diego system: loss of EHR, imaging and telemedicine for about 4 weeks; staff reverted to paper", "Adjacent EDs: ambulance arrivals +35.2%, left-without-being-seen +127.8%, stroke code activations +74.6%", "Cyberattacks should be treated as regional disasters" ], "last_verified_live": "2026-09-26" }, { "id": "dhamija-2022-pediatr-radiol-pacs-drill", "title": "PACS downtime drill: testing departmental workflow with an enterprise imaging viewer and archive", "publisher": "Pediatric Radiology 52(7):1234-1241", "url": "https://doi.org/10.1007/s00247-022-05339-1", "type": "peer-reviewed", "published": "2022-03-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "Live drill taking departmental PACS offline and reading from the enterprise viewer" ], "last_verified_live": "2026-09-26" }, { "id": "dhsc-2023-cyber-strategy", "title": "Cyber security strategy for health and social care: 2023 to 2030", "publisher": "Department of Health and Social Care (UK)", "url": "https://www.gov.uk/government/publications/cyber-security-strategy-for-health-and-social-care-2023-to-2030", "type": "guidance", "published": "2023-03-22", "accessed": "2026-09-26", "tier": 1, "supports": [ "Aim: all health and social care organisations in England cyber resilient no later than 2030, including critical supply-chain partners" ], "last_verified_live": "2026-09-26" }, { "id": "digitalhealth-2025-aws-nhs", "title": "AWS outage causes disruption to patient care across NHS sites", "publisher": "Digital Health News (Jordan Sollof)", "url": "https://www.digitalhealth.net/2025/10/aws-outage-causes-disruption-to-patient-care-across-nhs-sites/", "type": "journalism", "published": "2025-10-21", "accessed": "2026-09-26", "tier": 2, "supports": [ "A trust digital chief said at least 10 NHS sites using Oracle systems were 'in downtime scenarios' on 20 Oct 2025 and reverted to paper and business continuity processes", "Oracle hosts some services on AWS and some on its own infrastructure", "Central and North West London NHS FT confirmed it was affected; supplier System C invoked business continuity" ], "last_verified_live": "2026-09-26" }, { "id": "doj-magellan-leadcare-plea-2024", "title": "Magellan Diagnostics Pleads Guilty to Criminal FDCA Charges", "publisher": "U.S. Department of Justice (District of Massachusetts), reposted by FDA OCI", "url": "https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/doj-press-releases-involving-fda-oci/magellan-diagnostics-pleads-guilty-criminal-fdca-charges", "type": "official-report", "published": "2024-06-27", "accessed": "2026-09-26", "tier": 1, "supports": [ "LeadCare devices produced falsely low blood lead results on venous samples; tens of thousands of patients affected", "Magellan learned of the malfunction in June 2013 and reported it to FDA in April 2015, 21 months later", "LeadCare II accounted for more than half of US blood lead tests 2013-2017", "$28.1M fine, $10.9M forfeiture, at least $9.3M victim compensation" ], "last_verified_live": "2026-09-26" }, { "id": "dratsch-2023-radiology-mammography-automation-bias", "title": "Automation Bias in Mammography: The Impact of Artificial Intelligence BI-RADS Suggestions on Reader Performance", "publisher": "Radiology 307(4):e222176 (RSNA)", "url": "https://doi.org/10.1148/radiol.222176", "type": "peer-reviewed", "published": "2023-05-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "27 radiologists; accuracy with incorrect AI suggestion fell to 19.8%, 24.8% and 45.5% (inexperienced, moderate, very experienced) from about 80%" ], "last_verified_live": "2026-09-26" }, { "id": "dshs-2021-winter-storm-deaths", "title": "February 2021 Winter Storm-Related Deaths – Texas", "publisher": "Texas Department of State Health Services", "url": "https://www.dshs.texas.gov/sites/default/files/news/updates/SMOC_FebWinterStorm_MortalitySurvReport_12-30-21.pdf", "type": "official-report", "published": "2021-12-31", "accessed": "2026-09-26", "tier": 1, "supports": [ "DSHS confirmed 246 winter-storm-related deaths", "25 deaths from exacerbation of pre-existing illness, including disrupted dialysis or oxygen, frozen medical devices, and loss of power to life-sustaining electrical equipment", "19 fatal carbon monoxide poisonings, including from generators", "246 confirmed storm-related deaths", "25 deaths from exacerbation of pre-existing illness, incl. dialysis/oxygen disruption and loss of power to life-sustaining equipment" ], "last_verified_live": "2026-09-26" }, { "id": "dutch-safety-board-2020-it-outages", "title": "Patient safety during IT outages in hospitals", "publisher": "Dutch Safety Board (Onderzoeksraad voor Veiligheid)", "url": "https://onderzoeksraad.nl/en/onderzoek/patient-safety-during-it-outages-in-hospitals/", "type": "official-report", "published": "2020-02-13", "accessed": "2026-09-26", "tier": 1, "supports": [ "Hospitals' awareness of IT-failure risk has not kept pace with dependency", "Recommends mapping IT-care dependencies, periodic testing and drills, in-depth analysis of serious outages" ], "last_verified_live": "2026-09-26" }, { "id": "ec-ai-act-overview", "title": "AI Act (Regulation (EU) 2024/1689): regulatory framework overview", "publisher": "European Commission, Shaping Europe's digital future", "url": "https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai", "type": "regulation", "published": "2024-08-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "AI Act entered into force 1 August 2024; AI safety components of products are high-risk", "High-risk obligations: risk management, data quality, logging, documentation, human oversight, robustness and accuracy" ], "last_verified_live": "2026-09-26" }, { "id": "ecfr-42cfr482-15", "title": "42 CFR 482.15 Condition of participation: Emergency preparedness (hospitals)", "publisher": "eCFR / CMS", "url": "https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-G/part-482/subpart-B/section-482.15", "type": "regulation", "accessed": "2026-09-26", "tier": 1, "supports": [ "Policies must include a system of medical documentation that preserves patient information, protects confidentiality, and secures and maintains the availability of records (b)(5)", "Emergency plan and training/testing program reviewed and updated at least every 2 years", "Training at least every 2 years and demonstration of staff knowledge", "Exercises at least twice a year, including full-scale/functional and an additional exercise such as a tabletop", "Hospitals must provide alternate energy sources for temperatures that protect patient health, emergency lighting, fire detection/alarm, and sewage and waste disposal (482.15(b)(1)(ii))", "Hospitals must implement emergency power inspection, testing and maintenance per NFPA 99, NFPA 110 and the Life Safety Code (482.15(e)(2))", "Hospitals with onsite fuel must plan how to keep emergency power running during the emergency (482.15(e)(3))", "Exercises at least twice a year; plan reviewed at least every 2 years (482.15(d))", "Facility-based and community-based all-hazards risk assessment", "Arrangements with other hospitals to receive patients", "Exercises at least twice per year", "Hospitals must have a system of medical documentation that preserves patient information and maintains availability of records (b)(5)", "Training at least every 2 years; program reviewed at least every 2 years (d)(1)", "Two exercises per year: an annual full-scale community or facility functional exercise plus a second exercise that may be a tabletop led by a facilitator (d)(2)", "Actual emergency activating the plan exempts the next full-scale/functional exercise (d)(2)(i)(B)", "Analyze response to all drills, tabletops and real events and revise the plan (d)(2)(iii)", "Emergency and standby power systems required (e)" ], "date_note": "current (text as of 2026-09-01)", "last_verified_live": "2026-09-26" }, { "id": "ecfr-45cfr164-308", "title": "45 CFR 164.308(a)(7) Contingency plan (HIPAA Security Rule, administrative safeguards)", "publisher": "eCFR / US Government", "url": "https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-C/part-164/subpart-C/section-164.308", "type": "regulation", "accessed": "2026-09-26", "tier": 1, "supports": [ "Required: data backup plan (retrievable exact copies), disaster recovery plan, emergency mode operation plan", "Addressable: testing and revision procedures; applications and data criticality analysis" ], "date_note": "current (text as of 2026-09-01)", "last_verified_live": "2026-09-26" }, { "id": "ecri-top10-hazards-2026", "title": "Top 10 Health Technology Hazards for 2026: Executive Brief", "publisher": "ECRI", "url": "https://assets.ecri.org/PDF/ECRI_Top_10_Health_Technology_Hazards_for_2026_Executive_Brief.pdf", "type": "official-report", "published": "2026-01", "accessed": "2026-09-26", "tier": 2, "supports": [ "2026 list: #2 unpreparedness for a 'digital darkness' event, #4 recall communication failures for home diabetes technologies, #8 cybersecurity risks from legacy medical devices, #9 designs/configurations that prompt unsafe workflows", "Legacy devices include software-based devices no longer updated with sufficient cybersecurity protections", "Digital darkness outages cascade across clinical workflows; ECRI recommends downtime procedures, backups, recovery capability, drills", "Unpreparedness for a 'digital darkness' event ranked #2 hazard for 2026" ], "last_verified_live": "2026-09-26" }, { "id": "eldiario-2025-hospitales-apagon", "title": "La resaca en los hospitales tras salvar el apagón: 'Más allá de cierto caos, hemos sobrevivido bastante bien'", "publisher": "elDiario.es (Sofía Pérez Mendoza)", "url": "https://www.eldiario.es/sociedad/resaca-hospitales-salvar-apagon-caos-hemos-sobrevivido_1_12256075.html", "type": "journalism", "published": "2025-04-29", "accessed": "2026-09-26", "tier": 2, "supports": [ "Major Spanish hospitals (12 de Octubre, Gregorio Marañón, Clínic, Vall d'Hebron) kept ICU, inpatient and active surgery running on generators; no generator failures were reported by those consulted", "Outpatient clinics and non-urgent surgery were suspended; MRI/CT needed restart procedures; lab analyzers and IT for prescribing and records were disrupted in some regions" ], "last_verified_live": "2026-09-26" }, { "id": "endsley-kiris-1995-human-factors", "title": "The Out-of-the-Loop Performance Problem and Level of Control in Automation", "publisher": "Human Factors 37(2):381-394", "url": "https://doi.org/10.1518/001872095779064555", "type": "peer-reviewed", "published": "1995-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "Out-of-the-loop problem impairs takeover after automation failure", "Low situation awareness corresponded with slower decisions after expert-system failure; shift to passive processing most likely responsible" ], "last_verified_live": "2026-09-26" }, { "id": "enisa-2023-health-threat-landscape", "title": "ENISA Threat Landscape: Health Sector (press release 'Checking-up on Health: Ransomware Accounts for 54% of Cybersecurity Threats')", "publisher": "European Union Agency for Cybersecurity (ENISA)", "url": "https://www.enisa.europa.eu/news/checking-up-on-health-ransomware-accounts-for-54-of-cybersecurity-threats", "type": "official-report", "published": "2023-07-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "215 publicly reported health-sector incidents in the EU and neighbours over 2+ years; ransomware 54%; 22% disrupted healthcare services", "Only 27% of surveyed health organisations had a dedicated ransomware defence programme" ], "last_verified_live": "2026-09-26" }, { "id": "entsoe-2026-iberian-final-report", "title": "Final Report on the Grid Incident in Spain and Portugal on 28 April 2025", "publisher": "ENTSO-E Expert Panel", "url": "https://www.entsoe.eu/publications/blackout/28-april-2025-iberian-blackout/", "type": "official-report", "published": "2026-03-20", "accessed": "2026-09-26", "tier": 1, "supports": [ "Continental Spain and Portugal blacked out at 12:33 CEST on 28 April 2025", "Cause: interacting factors including oscillations, gaps in voltage and reactive-power control, and cascading generator disconnections in Spain" ], "last_verified_live": "2026-09-26" }, { "id": "ep-2026-digital-omnibus-ai", "title": "Digital Omnibus on AI (Legislative Train Schedule)", "publisher": "European Parliament", "url": "https://www.europarl.europa.eu/legislative-train/package-digital-package/file-digital-omnibus-on-ai", "type": "regulation", "published": "2026-07-08", "accessed": "2026-09-26", "tier": 1, "supports": [ "Omnibus signed 8 July 2026; high-risk obligations apply from 2 December 2027 (stand-alone) and 2 August 2028 (AI embedded in products)" ], "last_verified_live": "2026-09-26" }, { "id": "espiritu-2014-pediatrics-nicu-sandy", "title": "Evacuation of a neonatal intensive care unit in a disaster: lessons from Hurricane Sandy", "publisher": "Pediatrics (American Academy of Pediatrics)", "url": "https://doi.org/10.1542/peds.2014-0936", "type": "peer-reviewed", "published": "2014", "accessed": "2026-09-26", "tier": 1, "supports": [ "A power outage secondary to the coastal surge forced NYU Langone to evacuate 21 neonates from its NICU to other NYC hospitals within 4.5 hours", "NYU Langone evacuated in October 2012 after a power outage from coastal surge; 21 neonates moved in 4.5 hours", "Lessons: command structure, backups for personnel, communication, medical information and equipment; situational awareness" ], "last_verified_live": "2026-09-26" }, { "id": "eu-2022-nis2-directive", "title": "Directive (EU) 2022/2555 (NIS 2 Directive)", "publisher": "European Parliament and Council (EUR-Lex)", "url": "https://eur-lex.europa.eu/eli/dir/2022/2555/oj/eng", "type": "regulation", "published": "2022-12-14", "accessed": "2026-09-26", "tier": 1, "supports": [ "Art. 21: supply-chain security and cybersecurity measures in contracts with direct suppliers", "Art. 22: coordinated risk assessments of critical supply chains" ], "last_verified_live": "2026-09-26" }, { "id": "eu-ai-act-2024-1689", "title": "Regulation (EU) 2024/1689 (Artificial Intelligence Act)", "publisher": "European Parliament and Council of the EU (EUR-Lex)", "url": "https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng", "type": "regulation", "published": "2024-06-13", "accessed": "2026-09-26", "tier": 1, "supports": [ "Legal basis for Article 14 human oversight and Article 26 deployer obligations" ], "last_verified_live": "2026-09-26" }, { "id": "eu-cer-directive-2022-2557", "title": "Directive (EU) 2022/2557 on the resilience of critical entities", "publisher": "Official Journal of the European Union", "url": "https://eur-lex.europa.eu/eli/dir/2022/2557/oj", "type": "regulation", "published": "2022-12-14", "accessed": "2026-09-26", "tier": 1, "supports": [ "Health and energy are in scope; critical entities must assess all relevant risks at least every four years and keep a resilience plan with technical, security and organisational measures" ], "last_verified_live": "2026-09-26" }, { "id": "eu-mdr-2017-745", "title": "Regulation (EU) 2017/745 on medical devices (MDR), Annex I sections 17.2 and 17.4", "publisher": "Official Journal of the European Union / EUR-Lex", "url": "https://eur-lex.europa.eu/eli/reg/2017/745/oj", "type": "regulation", "published": "2017-04-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Annex I 17.2: software developed per state of the art including life cycle, risk management including information security, verification and validation", "Annex I 17.4: manufacturers set minimum hardware, IT network and IT security requirements including protection against unauthorised access" ], "last_verified_live": "2026-09-26" }, { "id": "eu-nis2-directive-2022-2555", "title": "Directive (EU) 2022/2555 on measures for a high common level of cybersecurity across the Union (NIS 2 Directive)", "publisher": "European Parliament and Council / EUR-Lex; summary by European Commission", "url": "https://eur-lex.europa.eu/eli/dir/2022/2555/oj", "type": "regulation", "published": "2022-12-14", "accessed": "2026-09-26", "tier": 1, "supports": [ "NIS2 extends EU cybersecurity risk-management and incident-notification duties; healthcare remains a covered sector" ], "last_verified_live": "2026-09-26" }, { "id": "euronews-ap-2025-nurses-ai-sepsis", "title": "As AI reshapes patient care, human nurses are pushing back against its creeping influence (AP)", "publisher": "Associated Press via Euronews", "url": "https://www.euronews.com/health/2025/03/18/as-ai-reshapes-patient-care-human-nurses-are-pushing-back-against-its-creeping-influence", "type": "journalism", "published": "2025-03-18", "accessed": "2026-09-26", "tier": 2, "supports": [ "At Dignity Health in Henderson, Nevada, a sepsis flag triggered a protocol for large IV fluid volumes in a dialysis patient; the nurse objected, was told to follow protocol, and a physician intervened", "Nurses report frequent false alerts from AI systems" ], "last_verified_live": "2026-09-26" }, { "id": "fawzy-2022-jama-im-pulse-oximetry-covid", "title": "Racial and Ethnic Discrepancy in Pulse Oximetry and Delayed Identification of Treatment Eligibility Among Patients With COVID-19", "publisher": "JAMA Internal Medicine (Fawzy A, et al.)", "url": "https://doi.org/10.1001/jamainternmed.2022.1906", "type": "peer-reviewed", "published": "2022-07-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "SpO2 overestimated SaO2 by 1.1-1.7 points in Asian, Black and Hispanic vs White patients with COVID-19", "Black patients had 29% lower hazard of treatment-eligibility recognition; 451 patients never had eligibility recognized, 54.8% of them Black", "Silent sensor bias translated into delayed or missed therapy decisions" ], "last_verified_live": "2026-09-26" }, { "id": "fda-2024-pccp-final-guidance", "title": "Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions (final guidance)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/marketing-submission-recommendations-predetermined-change-control-plan-artificial-intelligence", "type": "guidance", "published": "2024-12-04", "accessed": "2026-09-26", "tier": 1, "supports": [ "A PCCP describes planned AI-DSF modifications, the methodology to develop, validate and implement them, and an impact assessment", "Lets authorized modifications ship without a new marketing submission for each", "Reissued with edits 2025-08-18" ], "last_verified_live": "2026-09-26" }, { "id": "fda-2025-ai-dsf-draft-guidance", "title": "Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations (draft guidance)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/artificial-intelligence-enabled-device-software-functions-lifecycle-management-and-marketing", "type": "guidance", "published": "2025-01-07", "accessed": "2026-09-26", "tier": 1, "supports": [ "AI-enabled device performance may change or degrade after deployment from shifts in patient population, disease patterns or data drift", "Users may be less likely to notice performance change when the device sits in a highly automated process with limited human interaction", "Recommends a performance monitoring plan covering demographic shifts, input shifts, pipeline corruption (missing values, duplicate records, type mismatches) and user behaviour changes", "Remains draft, not for implementation" ], "last_verified_live": "2026-09-26" }, { "id": "fda-2025-pulse-oximeter-draft-guidance", "title": "Pulse Oximeters for Medical Purposes - Non-Clinical and Clinical Performance Testing, Labeling, and Premarket Submission Recommendations (Draft Guidance)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/pulse-oximeters-medical-purposes-non-clinical-and-clinical-performance-testing-labeling-and", "type": "guidance", "published": "2025-01-07", "accessed": "2026-09-26", "tier": 1, "supports": [ "Draft guidance (docket FDA-2023-N-4976) addresses concern that accuracy can be affected by skin pigmentation", "Would supersede the 2013 pulse oximeter 510(k) guidance when final" ], "last_verified_live": "2026-09-26" }, { "id": "fda-524b-webinar-2024", "title": "Draft Guidance - Select Updates for the Premarket Cybersecurity Guidance: Section 524B of the FD&C Act (webinar slides)", "publisher": "U.S. Food and Drug Administration (CDRH)", "url": "https://www.fda.gov/media/178134/download", "type": "guidance", "published": "2024-04-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "FDA interprets 'ability to connect to the internet' to include Wi-Fi, cellular, Bluetooth, RF, inductive links and USB/ethernet/serial connectors", "Related systems include software/firmware update servers and hospital network connections", "Plans should state timelines for updates and cover devices no longer marketed but still in use" ], "last_verified_live": "2026-09-26" }, { "id": "fda-abbott-libre3-recall-2025", "title": "Glucose Monitor Sensor Recall: Abbott Diabetes Care Removes Certain FreeStyle Libre 3 and FreeStyle Libre 3 Plus Sensors", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/medical-device-recalls-and-early-alerts/glucose-monitor-sensor-recall-abbott-diabetes-care-removes-certain-freestyle-libre-3-and-freestyle", "type": "official-report", "published": "2026-02-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Correction initiated 24 Nov 2025; Class I; sensors could read lower than actual glucose", "As of 7 Jan 2026, 860 serious injuries and 7 deaths reported", "Patients told to use a blood glucose meter for treatment decisions until replaced" ], "last_verified_live": "2026-09-26" }, { "id": "fda-ai-enabled-device-list", "title": "Artificial Intelligence-Enabled Medical Devices (list)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/artificial-intelligence-enabled-medical-devices/list-artificial-intelligence-enabled-medical-devices", "type": "dataset", "published": "2026", "accessed": "2026-09-26", "tier": 1, "supports": [ "FDA publishes a list of AI-enabled devices authorized for US marketing", "The list is not comprehensive; devices are identified mainly by AI-related terms", "FDA plans to explore tagging devices that use foundation models such as LLMs" ], "last_verified_live": "2026-09-26" }, { "id": "fda-bd-alaris-2020-recall", "title": "BD Provides Update on Feb. 4, 2020 Voluntary Recall of the BD Alaris System PC Units and Modules", "publisher": "U.S. Food and Drug Administration (company announcement)", "url": "https://www.fda.gov/safety/recalls-market-withdrawals-safety-alerts/bd-provides-update-feb-4-2020-voluntary-recall-bd-alaristm-system-pc-units-and-modules", "type": "official-report", "published": "2020-03-09", "accessed": "2026-09-26", "tier": 1, "supports": [ "Class I recall of Alaris software for five issues including System Error Code 255, low battery alarm failure and alarm priority", "Remediation deferred to a comprehensive 510(k); interim user mitigations" ], "last_verified_live": "2026-09-26" }, { "id": "fda-bd-alaris-apr-correction-2025", "title": "Infusion Pump Software Correction: BD Issues Correction for BD Alaris Systems Manager and Care Coordination Engine Infusion Adapter Software Due to Risk for Outdated Automated Programming Requests to Load", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/medical-device-recalls-and-early-alerts/infusion-pump-software-correction-becton-dickinson-and-company-bd-issues-correction-bd-alaris", "type": "official-report", "published": "2025-02-18", "accessed": "2026-09-26", "tier": 1, "supports": [ "Backlogged automated programming requests could load outdated rate, dose or volume parameters onto the pump", "Most serious recall type; no injuries or deaths reported", "Clinicians told to verify all infusion parameters before starting" ], "last_verified_live": "2026-09-26" }, { "id": "fda-cds-guidance-2026", "title": "Clinical Decision Support Software: Guidance for Industry and FDA Staff", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/media/109618/download", "type": "guidance", "published": "2026-01-29", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defines automation bias; errors of commission and omission", "Automation bias increases in urgent situations", "Level of automation and time-criticality considered under criterion 4" ], "last_verified_live": "2026-09-26" }, { "id": "fda-contec-cms8000-2025", "title": "Cybersecurity Vulnerabilities with Certain Patient Monitors from Contec and Epsimed: FDA Safety Communication", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/safety-communications/cybersecurity-vulnerabilities-certain-patient-monitors-contec-and-epsimed-fda-safety-communication", "type": "official-report", "published": "2025-01-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "Patient monitors contained a backdoor, allowed remote control and exfiltrated patient data once connected", "FDA not aware of incidents, injuries or deaths", "July 2025 patch removes networking entirely; advice was to unplug ethernet and disable wireless" ], "last_verified_live": "2026-09-26" }, { "id": "fda-maude-about", "title": "About Manufacturer and User Facility Device Experience (MAUDE) Database", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/mandatory-reporting-requirements-manufacturers-importers-and-device-user-facilities/about-manufacturer-and-user-facility-device-experience-maude-database", "type": "dataset", "published": "2025-12-08", "accessed": "2026-09-26", "tier": 1, "supports": [ "MAUDE holds MDRs from mandatory and voluntary reporters, updated monthly", "Incidence, prevalence or cause cannot be determined from MDRs alone due to under-reporting and unverified causation" ], "last_verified_live": "2026-09-26" }, { "id": "fda-philips-activities", "title": "FDA Activities Related to Recalled Philips Ventilators, BiPAP Machines, and CPAP Machines", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/recalled-philips-ventilators-bipap-machines-and-cpap-machines/fda-activities-related-recalled-philips-ventilators-bipap-machines-and-cpap-machines", "type": "official-report", "published": "2024-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Recall initiated 30 June 2021, Class I, about 15 million devices worldwide", "FDA issued a section 518(a) notification order on 10 March 2022 over recall communication" ], "last_verified_live": "2026-09-26" }, { "id": "fda-philips-consent-decree-2024", "title": "Federal Court Enters Consent Decree Against Philips Respironics Following Recall of Certain Sleep Therapy Machines", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/news-events/press-announcements/federal-court-enters-consent-decree-against-philips-respironics-following-recall-certain-sleep", "type": "official-report", "published": "2024-04-09", "accessed": "2026-09-26", "tier": 1, "supports": [ "Consent decree entered 9 April 2024 restricting new device production at Murrysville and Mt. Pleasant PA and Carlsbad CA", "Requires a recall remediation plan and independent expert review of the replacement silicone foam testing" ], "last_verified_live": "2026-09-26" }, { "id": "fda-philips-mdrs", "title": "Problems Reported with Recalled Philips Ventilators, BiPAP Machines, and CPAP Machines", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/recalled-philips-ventilators-bipap-machines-and-cpap-machines/problems-reported-recalled-philips-ventilators-bipap-machines-and-cpap-machines", "type": "dataset", "published": "2024-01-31", "accessed": "2026-09-26", "tier": 1, "supports": [ "More than 116,000 MDRs including 561 death reports associated with PE-PUR foam breakdown, April 2021 to 31 Jan 2024", "FDA states MDRs alone cannot establish incidence or cause" ], "last_verified_live": "2026-09-26" }, { "id": "fda-philips-trilogy-evo-correction-2024", "title": "Ventilator Software Correction: Philips Respironics Issues Mandatory Software Correction and Updates Use Instructions for Trilogy Evo, EV300, EvoO2, and Evo Universal", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/medical-device-recalls-and-early-alerts/ventilator-software-correction-philips-respironics-issues-mandatory-software-correction-and-updates", "type": "official-report", "published": "2024-07-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "Class I software correction addressing about 13 problem categories including an inaccurate display of delivered oxygen (patient may receive less than indicated)", "False power-loss/battery alarms stopping therapy and multiple alarm malfunctions", "9 injuries and 1 death reported" ], "last_verified_live": "2026-09-26" }, { "id": "fda-postmarket-cybersecurity-2016", "title": "Postmarket Management of Cybersecurity in Medical Devices", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/postmarket-management-cybersecurity-medical-devices", "type": "guidance", "published": "2016-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "Cybersecurity requires continual maintenance across the product life cycle", "Links cybersecurity remediation to Part 806 correction and removal reporting" ], "last_verified_live": "2026-09-26" }, { "id": "fda-premarket-cybersecurity-guidance-2026", "title": "Cybersecurity in Medical Devices: Quality Management System Considerations and Content of Premarket Submissions", "publisher": "U.S. Food and Drug Administration (CDRH/CBER)", "url": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cybersecurity-medical-devices-quality-management-system-considerations-and-content-premarket", "type": "guidance", "published": "2026-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "Final guidance issued February 2026, superseding the 27 June 2025 version (which itself superseded Sept 2023) and adding Section VII on 524B", "Addresses FDA recommendations for section 524B cyber devices", "Covers device design, labeling and premarket documentation for devices with cybersecurity risk" ], "last_verified_live": "2026-09-26" }, { "id": "fda-pulse-oximeters-actions", "title": "Pulse Oximeters (FDA actions on accuracy and skin pigmentation)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/products-and-medical-procedures/pulse-oximeters", "type": "guidance", "published": "2025-01-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "FDA Feb 2021 safety communication that pulse oximeters have limitations, including skin pigmentation", "FDA advisory panel meetings Nov 2022 and Feb 2024; discussion paper Nov 2023", "Draft guidance issued 7 Jan 2025" ], "last_verified_live": "2026-09-26" }, { "id": "fda-rcs-iso-14971", "title": "Recognized Consensus Standard 5-125: ISO 14971:2019 Medical devices - Application of risk management to medical devices", "publisher": "U.S. Food and Drug Administration (Recognized Consensus Standards database)", "url": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfStandards/results.cfm?recognitionnumber=5-125", "type": "standard", "published": "2019-12-23", "accessed": "2026-09-26", "tier": 1, "supports": [ "FDA gave ISO 14971:2019 complete recognition on 23 Dec 2019" ], "last_verified_live": "2026-09-26" }, { "id": "fda-recalls-corrections-removals", "title": "Recalls, Corrections and Removals (Devices)", "publisher": "U.S. Food and Drug Administration", "url": "https://www.fda.gov/medical-devices/postmarket-requirements-devices/recalls-corrections-and-removals-devices", "type": "regulation", "published": "2020", "accessed": "2026-09-26", "tier": 1, "supports": [ "Class I: reasonable probability of serious adverse health consequences or death", "21 CFR 806 requires firms to report health-risk corrections and removals to FDA within 10 working days" ], "last_verified_live": "2026-09-26" }, { "id": "fda-recalls-database", "title": "Medical Device Recalls database", "publisher": "U.S. Food and Drug Administration", "url": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfRES/res.cfm", "type": "dataset", "published": "2002-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Contains device recalls classified since Nov 2002, and since 2017 some firm-initiated corrections before FDA review", "Posting date reflects classification, not when the firm started the recall" ], "last_verified_live": "2026-09-26" }, { "id": "ferc-nerc-2021-cold-weather", "title": "The February 2021 Cold Weather Outages in Texas and the South Central United States — FERC, NERC and Regional Entity Joint Staff Report (presentation of findings)", "publisher": "FERC / NERC", "url": "https://www.nerc.com/globalassets/standards/projects/2021-07/ferc-presentation-phase-2.pdf", "type": "official-report", "published": "2021-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "ERCOT averaged 34,000 MW of generation outages for over two days in the week of 14 Feb 2021", "23,418 MW of firm load shed: the largest firm load shed event in US history", "Most natural gas production and processing facilities were not identified as critical load and lost power to weather outages and manual load shed, worsening gas supply to generators" ], "last_verified_live": "2026-09-26" }, { "id": "ferc-nerc-2022-feb2021-outages-tracking", "title": "The February 2021 Cold Weather Outages in Texas and the South Central United States — Tracking of Responses to Recommendations", "publisher": "FERC (with NERC and Regional Entity staff)", "url": "https://ferc.gov/sites/default/files/2022-12/A-3%20-%20The%20February%202021%20Cold%20Weather%20Outages%20in%20Texas%20and%20the%20South%20Central%20United%20States.pdf", "type": "official-report", "published": "2022-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "ERCOT averaged 34,000 MW of generation outages", "Largest firm load shed in US history (23,418 MW)", "Recommendations on gas-electric coordination and mapping interdependent gas and electric infrastructure" ], "last_verified_live": "2026-09-26" }, { "id": "fink-2009-propublica-memorial", "title": "The Deadly Choices at Memorial", "publisher": "ProPublica / The New York Times Magazine (Sheri Fink)", "url": "https://www.propublica.org/article/the-deadly-choices-at-memorial-826", "type": "journalism", "published": "2009-08-27", "accessed": "2026-09-26", "tier": 2, "supports": [ "Memorial's electrical system was known to be vulnerable to flooding before Katrina", "City power failed early Monday 29 Aug 2005; backup generators ran until they failed early Wednesday", "About 2,000 people sheltered at the hospital, including more than 200 patients; LifeCare's 52 patients on the seventh floor included many on ventilators", "45 bodies were recovered from Memorial" ], "last_verified_live": "2026-09-26" }, { "id": "finlayson-2021-nejm-dataset-shift", "title": "The Clinician and Dataset Shift in Artificial Intelligence", "publisher": "New England Journal of Medicine (Finlayson SG, Subbaswamy A, Singh K, et al.)", "url": "https://doi.org/10.1056/NEJMc2104626", "type": "peer-reviewed", "published": "2021-07-15", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defines dataset shift as underperformance from a mismatch between development data and deployment data", "University of Michigan deactivated the Epic sepsis model in April 2020 because of spurious alerting after COVID-19 changed the patient mix; its clinical AI governing committee decommissioned it" ], "last_verified_live": "2026-09-26" }, { "id": "ghafur-2019-npj-wannacry", "title": "A retrospective impact analysis of the WannaCry cyberattack on the NHS", "publisher": "npj Digital Medicine", "url": "https://doi.org/10.1038/s41746-019-0161-6", "type": "peer-reviewed", "published": "2019-10-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "Infected hospitals had ~6% fewer admissions per day", "£5.9m lost activity", "No measurable increase in A&E mortality" ], "last_verified_live": "2026-09-26" }, { "id": "gichoya-2022-lancet-dh-race-imaging", "title": "AI recognition of patient race in medical imaging: a modelling study", "publisher": "The Lancet Digital Health (Gichoya JW, Banerjee I, et al.)", "url": "https://doi.org/10.1016/S2589-7500(22)00063-2", "type": "peer-reviewed", "published": "2022-05-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Deep learning models predict self-reported race from chest X-ray, CT and mammography (AUC 0.81-0.99) when clinical experts cannot", "The ability persisted under image corruption, making it hard to control and a risk for all imaging model deployments" ], "last_verified_live": "2026-09-26" }, { "id": "globalnews-2026-ontario-ai-scribes", "title": "AI systems used by Ontario doctors hallucinate, auditor general finds", "publisher": "Global News (Canada)", "url": "https://globalnews.ca/news/11844349/ontario-auditor-general-ai-usage/", "type": "journalism", "published": "2026-05-12", "accessed": "2026-09-26", "tier": 2, "supports": [ "Supply Ontario did not require vendors to demonstrate systems live", "At least five of 20 vendors did not submit required risk and privacy impact assessments but were approved" ], "last_verified_live": "2026-09-26" }, { "id": "globe-2010-mgh-monitor-death", "title": "MGH death spurs review of patient monitors", "publisher": "The Boston Globe (Liz Kowalczyk)", "url": "http://archive.boston.com/news/health/articles/2010/02/21/mgh_death_spurs_review_of_patient_monitors/", "type": "journalism", "published": "2010-02-21", "tier": 2, "supports": [ "MGH patient died in January 2010 after a heart-monitor alarm was left off", "MGH disabled the off switch on 1,100 monitors", "Death reported to Mass. DPH; CMS reviewed" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "globe-2011-alarms-unheeded", "title": "Patient alarms often unheard, unheeded", "publisher": "The Boston Globe (Liz Kowalczyk)", "url": "http://archive.boston.com/lifestyle/health/articles/2011/02/13/patient_alarms_often_unheard_unheeded/", "type": "journalism", "published": "2011-02-13", "tier": 2, "supports": [ "Crisis alarm was turned off; lower-level alarms not responded to; ten nurses could not recall hearing them" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "goddard-2012-jamia-automation-bias", "title": "Automation bias: a systematic review of frequency, effect mediators, and mitigators", "publisher": "Journal of the American Medical Informatics Association 19(1):121-127", "url": "https://doi.org/10.1136/amiajnl-2011-000089", "type": "peer-reviewed", "published": "2012-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "Automation bias defined as the tendency to over-rely on automation", "Mediators include workload, task complexity, time constraint, trust", "Mitigators include training and emphasising user accountability, and DSS design factors", "Over-reliance on decision support is mediated by trust, workload, time pressure; mitigated by training, accountability, confidence display" ], "last_verified_live": "2026-09-26" }, { "id": "goiana-da-silva-2025-fph-iberian", "title": "When the lights went out: impacts of the April 2025 Iberian blackout on the Portuguese National Health Service sovereignty", "publisher": "Frontiers in Public Health", "url": "https://doi.org/10.3389/fpubh.2025.1630933", "type": "peer-reviewed", "published": "2025", "accessed": "2026-09-26", "tier": 1, "supports": [ "Portuguese NHS hospitals ran on limited generator capacity for over eight hours; essential equipment was triaged and digital records became inaccessible", "Cold storage failures endangered temperature-sensitive medicines; emergency communications and transport were severely compromised", "Perspective/brief report based on first-hand experience and grey literature" ], "last_verified_live": "2026-09-26" }, { "id": "google-cloud-2022-europe-west2", "title": "Incident: cooling related failure in one of our buildings that hosts zone europe-west2-a", "publisher": "Google Cloud Service Health", "url": "https://status.cloud.google.com/incidents/XVq5om2XEDSqLtJZUvcH", "type": "official-report", "published": "2022-07-19", "accessed": "2026-09-26", "tier": 1, "supports": [ "The same London heatwave caused a cooling-related failure in a Google Cloud building hosting zone europe-west2-a on 19 July 2022; part of the zone was powered down; incident closed 20 July" ], "last_verified_live": "2026-09-26" }, { "id": "govuk-2026-csr-bill-critical-suppliers", "title": "Cyber Security and Resilience (Network and Information Systems) Bill factsheet: Designating critical suppliers", "publisher": "UK Government (GOV.UK)", "url": "https://www.gov.uk/government/publications/cyber-security-and-resilience-network-and-information-systems-bill-factsheets/designating-critical-suppliers", "type": "guidance", "published": "2026-06-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "Regulators would be able to designate critical suppliers subject to mandatory cyber requirements", "Synnovis case study: over 11,000 appointments and operations disrupted, estimated £32.7m losses" ], "last_verified_live": "2026-09-26" }, { "id": "gstt-2023-it-incident-review", "title": "Review of the Guy's and St Thomas' IT Critical Incident — Final report from the Deputy Chief Executive Officer", "publisher": "Guy's and St Thomas' NHS Foundation Trust", "url": "https://www.guysandstthomas.nhs.uk/sites/default/files/2023-01/IT-critical-incident-review.pdf", "type": "official-report", "published": "2023-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "On 19 July 2022 air-conditioning failures at both trust data centres took down clinical IT at Guy's, St Thomas' and Evelina London; the Guy's data centre reached 50.3°C", "Contingency was mutual backup between two nearby data centres, which could not cover a shared environmental cause such as a heatwave", "Cooling concerns were raised in 2018 (condensers poorly sited) and a £195k replacement request from March 2022 was not approved before the incident", "The remote environment monitoring for the St Thomas' data centre was hosted on the storage network that failed; email alerts stopped", "The trust ran a 'Paper Hospital' for several weeks; 371 interlinked systems slowed recovery; £1.4m unplanned spend", "Activity fell to 64% of referrals, 71% of elective admissions and 68% of diagnostic tests versus July norms", "One moderate harm (a patient could not receive a transplant when organs became available), 20 low harms, no deaths or severe harm identified as of January 2023; harm review still open", "The review judged the event potentially preventable" ], "last_verified_live": "2026-09-26" }, { "id": "hanuscak-2009-ajhp-downtime-med-errors", "title": "Evaluation of causes and frequency of medication errors during information technology downtime", "publisher": "Am J Health-Syst Pharm 66(12):1119-1124", "url": "https://doi.org/10.2146/ajhp080389", "type": "peer-reviewed", "published": "2009-06-15", "accessed": "2026-09-26", "tier": 1, "supports": [ "Medication errors occurred during CIS and automated dispensing downtime despite backup systems and protocols" ], "last_verified_live": "2026-09-26" }, { "id": "healthcarebrew-2024-mgb-crowdstrike", "title": "A look at how Mass General Brigham recovered from the CrowdStrike outage", "publisher": "Healthcare Brew (Cassie McGrath)", "url": "https://www.healthcare-brew.com/stories/2024/09/11/mass-general-brigham-recovered-crowdstrike-outage", "type": "journalism", "published": "2024-09-11", "tier": 2, "supports": [ "MGB staff sent into downtime procedures, handwriting notes; Epic read-only; named MGB officials quoted" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "healthcaredive-2024-ascension-tracker", "title": "The state-by-state impact of Ascension's cyberattack", "publisher": "Healthcare Dive", "url": "https://www.healthcaredive.com/news/tracking-ascension-healthcare-cyberattack-impact/716327/", "type": "journalism", "published": "2024", "accessed": "2026-09-26", "tier": 2, "supports": [ "Ascension detected the attack 8 May 2024; ambulance diversions and pharmacy disruption followed", "EHR access restored across the network by 14 June 2024; Ascension declared full restoration 20 June; data collected during downtime not immediately visible" ], "last_verified_live": "2026-09-26" }, { "id": "healthcaredive-2025-change-190m", "title": "UnitedHealth hikes number of Change cyberattack breach victims to 190 million", "publisher": "Healthcare Dive (Emily Olsen)", "url": "https://www.healthcaredive.com/news/change-healthcare-cyberattack-affects-190-million-unitedhealth/738351/", "type": "journalism", "published": "2025-01-27", "accessed": "2026-09-26", "tier": 2, "supports": [ "UnitedHealth estimated about 190 million people affected, up from 100 million in Oct 2024", "UnitedHealth spent $3.1 billion responding to the attack in 2024; paid a $22 million ransom" ], "last_verified_live": "2026-09-26" }, { "id": "hfm-2022-generator-itm", "title": "Complying with inspection, testing and maintenance for electrical systems", "publisher": "Health Facilities Management (American Society for Health Care Engineering)", "url": "https://www.hfmmagazine.com/articles/4461-complying-with-inspection-testing-and-maintenance-for-electrical-systems", "type": "guidance", "published": "2022-06-10", "accessed": "2026-09-26", "tier": 2, "supports": [ "Generators must be exercised under load monthly at a minimum of 30% of nameplate kW or the manufacturer's minimum exhaust temperature (NFPA 110-2010)", "Every 36 months a four-hour test is required", "Annual load-bank and 36-month tests may be combined: 3 hours at 30%/exhaust temperature plus 1 hour at no less than 75% of nameplate", "Requirements cited: NFPA 110-2010, CMS K-tag K918, Joint Commission EC.02.05.07" ], "last_verified_live": "2026-09-26" }, { "id": "hhs-405d-hicp-2023", "title": "Health Industry Cybersecurity Practices: Managing Threats and Protecting Patients (HICP), 2023 Edition", "publisher": "HHS 405(d) Program / Health Sector Coordinating Council", "url": "https://405d.hhs.gov/Documents/HICP-Main-508.pdf", "type": "guidance", "published": "2023", "accessed": "2026-09-26", "tier": 1, "supports": [ "Five current threats incl. ransomware; ten practices incl. data protection and loss prevention, network management, and security operations/incident response", "Ransomware operators may interfere with routine backup schedules, rendering backups unusable, and exfiltrate data before encrypting" ], "last_verified_live": "2026-09-26" }, { "id": "hhs-45cfr92-210", "title": "45 CFR 92.210 Nondiscrimination in the use of patient care decision support tools (Section 1557 rule)", "publisher": "U.S. Department of Health and Human Services (eCFR)", "url": "https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-A/part-92/subpart-C/section-92.210", "type": "regulation", "published": "2024-05-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "Covered entities have an ongoing duty to make reasonable efforts to identify decision support tools that use race, color, national origin, sex, age or disability as inputs, and to mitigate discrimination risk", "Paragraphs (b) and (c) apply within 300 days of July 5, 2024" ], "last_verified_live": "2026-09-26" }, { "id": "hhs-hipaa-security-nprm-2025", "title": "HIPAA Security Rule To Strengthen the Cybersecurity of Electronic Protected Health Information (NPRM, FR Doc 2024-30983)", "publisher": "HHS Office for Civil Rights, Federal Register 90(3):898-1022", "url": "https://www.federalregister.gov/documents/2025/01/06/2024-30983/hipaa-security-rule-to-strengthen-the-cybersecurity-of-electronic-protected-health-information", "type": "regulation", "published": "2025-01-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "Proposes written procedures to restore critical relevant electronic information systems and data within 72 hours of loss", "Proposed rule only; eCFR text of 164.308(a)(7) unchanged as of 2026-09-01", "Proposed 72-hour restoration of critical systems", "Proposed yearly written verification of business associates' safeguards", "Proposed BA notice within 24 hours of contingency-plan activation", "Cites the Change Healthcare attack" ], "last_verified_live": "2026-09-26" }, { "id": "hhs-hph-cpgs", "title": "Healthcare and Public Health (HPH) Sector Cybersecurity Performance Goals", "publisher": "HHS (with CISA)", "url": "https://hhscyber.hhs.gov/cybersecurity-performance-goals.html", "type": "guidance", "published": "2024", "accessed": "2026-09-26", "tier": 1, "supports": [ "Voluntary 'essential' goals include basic incident planning and preparedness and vendor/supplier cybersecurity requirements", "'Enhanced' goals include third-party incident reporting, network segmentation, asset inventory, and drilled centralized incident plans" ], "last_verified_live": "2026-09-26" }, { "id": "hhs-ocr-breach-portal", "title": "Breach Portal: Notice to the Secretary of HHS Breach of Unsecured Protected Health Information", "publisher": "HHS Office for Civil Rights", "url": "https://ocrportal.hhs.gov/ocr/breach/breach_report.jsf", "type": "dataset", "accessed": "2026-09-26", "tier": 1, "supports": [ "Public listing of PHI breaches affecting 500 or more individuals; usable as a harvest source for new incidents" ], "date_note": "ongoing", "last_verified_live": "2026-09-26" }, { "id": "hhs-oig-2014-sandy", "title": "Hospital Emergency Preparedness and Response During Superstorm Sandy (OEI-06-13-00260)", "publisher": "US HHS Office of Inspector General", "url": "https://oig.hhs.gov/reports/all/2014/hospital-emergency-preparedness-and-response-during-superstorm-sandy/", "type": "official-report", "published": "2014-09-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "89% of 172 hospitals in declared disaster areas reported substantial challenges; 12 of 172 evacuated fully or partially", "69 hospitals had utility outages; for 28 of the 69 backup generators were not a reliable power source", "One hospital with generators on the 13th floor had its fuel pumps in the flooded basement and ran a fuel 'bucket line' up 13 flights of stairs", "One inch of water compromised one hospital's backup generator, forcing evacuation", "The same hospital decided in advance which machine-dependent patients would keep the limited emergency power outlets", "For 29 hospitals, fuel shortage substantially affected patient care; gas stations lacked generators to pump fuel; staff could not get to work", "Receiving hospitals got evacuated patients without medical records", "Most hospitals had received emergency-related deficiency citations from surveyors before the storm" ], "last_verified_live": "2026-09-26" }, { "id": "hics-guidebook-2014", "title": "Hospital Incident Command System Guidebook, Fifth Edition", "publisher": "California Emergency Medical Services Authority", "url": "https://emsa.ca.gov/wp-content/uploads/sites/71/2017/09/HICS_Guidebook_2014_11.pdf", "type": "guidance", "published": "2014-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "HICS is a scalable incident management system for hospitals based on ICS", "Incident Response Guides are organized by Immediate (0-2 h), Intermediate (2-12 h), Extended (>12 h) and Demobilization/System Recovery", "Demobilization planning should begin early; the Incident Commander decides the move from response to demobilization", "Return to day-to-day operations may be progressive; not all areas demobilize at once", "IT/IS Equipment Unit Leader role within Logistics" ], "last_verified_live": "2026-09-26" }, { "id": "house-2006-failure-of-initiative-medical", "title": "A Failure of Initiative (H. Rpt. 109-377), Medical Care chapter", "publisher": "U.S. House Select Bipartisan Committee on Hurricane Katrina (mirror: LSU Law biotech.law.lsu.edu)", "url": "https://biotech.law.lsu.edu/katrina/govdocs/109-377/medical.pdf", "type": "official-report", "published": "2006-02-15", "tier": 1, "supports": [ "Evacuation timeline at Memorial per Tenet: Tenet-hired helicopters airlifted ~400 on Thursday; evacuation completed Friday 2 Sept" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "house-homeland-meyers-testimony-2024", "title": "Testimony of Adam Meyers, CrowdStrike, before the House Homeland Security Subcommittee on Cybersecurity and Infrastructure Protection: 'An Outage Strikes'", "publisher": "U.S. House Committee on Homeland Security", "url": "https://homeland.house.gov/wp-content/uploads/2024/09/2024-09-24-HRG-CIP-Testimony-Meyers.pdf", "type": "testimony", "published": "2024-09-24", "accessed": "2026-09-26", "tier": 1, "supports": [ "Faulty content update was live for about 78 minutes before rollback", "CrowdStrike committed to staged rollout and customer control over content updates" ], "last_verified_live": "2026-09-26" }, { "id": "hscc-cms-2023-hospital-resiliency-landscape", "title": "Hospital Cyber Resiliency Initiative Landscape Analysis", "publisher": "HHS 405(d) / Health Sector Coordinating Council with CMS", "url": "https://405d.hhs.gov/Documents/405d-hospital-resiliency-analysis.pdf", "type": "official-report", "published": "2023-04", "accessed": "2026-09-26", "tier": 1, "supports": [ "Cited study: of respondents reporting ransomware, 46% attributed it to a third party", "50% or fewer hospitals consider patient-care risk when evaluating new suppliers' products" ], "last_verified_live": "2026-09-26" }, { "id": "hscc-occi-2022", "title": "Operational Continuity - Cyber Incident (OCCI) Checklist, Version 1.1", "publisher": "Health Sector Coordinating Council Cybersecurity Working Group", "url": "https://www.aha.org/system/files/media/file/2023/06/Operational-Continuity-Cyber-Incident-OCCI.pdf", "type": "guidance", "published": "2022-05", "accessed": "2026-09-26", "tier": 2, "supports": [ "Role-based ICS-aligned checklist for the first 12 hours of an extended enterprise outage", "Reaffirm recovery time objectives and validate application recovery priority" ], "last_verified_live": "2026-09-26" }, { "id": "hsu-2004-pdm-mci-training-review", "title": "Effectiveness of hospital staff mass-casualty incident training methods: a systematic literature review", "publisher": "Prehospital and Disaster Medicine 19(3):191-199", "url": "https://doi.org/10.1017/s1049023x00001771", "type": "peer-reviewed", "published": "2004", "accessed": "2026-09-26", "tier": 1, "supports": [ "Drills help staff become familiar with procedures and expose problems in command, communications, triage and flow", "Evidence was weak overall; no included study evaluated tabletop exercises" ], "last_verified_live": "2026-09-26" }, { "id": "iec-60601-1-8", "title": "IEC 60601-1-8:2006+AMD1:2012+AMD2:2020 CSV Alarm systems in medical electrical equipment and systems", "publisher": "International Electrotechnical Commission", "url": "https://webstore.iec.ch/en/publication/67388", "type": "standard", "published": "2020-07-23", "accessed": "2026-09-26", "tier": 1, "supports": [ "Specifies requirements and tests for alarm systems: priority categories, alarm signals, control states", "Edition 2.2 current, stability date 2028" ], "last_verified_live": "2026-09-26" }, { "id": "iec-60601-1-8-handoff", "title": "IEC 60601-1-8:2006+AMD1:2012 CSV (Ed. 2.1) Medical electrical equipment - Part 1-8: alarm systems", "publisher": "International Electrotechnical Commission", "url": "https://webstore.iec.ch/en/publication/2599", "type": "standard", "published": "2012-11-28", "accessed": "2026-09-26", "tier": 1, "supports": [ "Alarm categories by urgency, consistent alarm signals and control states" ], "last_verified_live": "2026-09-26" }, { "id": "iec-62304", "title": "IEC 62304:2006+AMD1:2015 CSV Medical device software - Software life cycle processes", "publisher": "International Electrotechnical Commission", "url": "https://webstore.iec.ch/en/publication/22794", "type": "standard", "published": "2015-06-26", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defines life cycle requirements for development and maintenance of medical device software, standalone or embedded" ], "last_verified_live": "2026-09-26" }, { "id": "iec-80001-1-2021", "title": "IEC 80001-1:2021 Safety, effectiveness and security in the implementation and use of connected medical devices or connected health software - Part 1: Application of risk management", "publisher": "International Electrotechnical Commission", "url": "https://webstore.iec.ch/en/publication/34263", "type": "standard", "published": "2021-09-21", "accessed": "2026-09-26", "tier": 1, "supports": [ "Requires healthcare delivery organizations to apply risk management before, during and after connecting health IT systems, covering safety, effectiveness and security" ], "last_verified_live": "2026-09-26" }, { "id": "jc-chai-2025-ruaih-guidance", "title": "Joint Commission and Coalition for Health AI (CHAI) Guidance on the Responsible Use of AI in Healthcare (RUAIH)", "publisher": "The Joint Commission and Coalition for Health AI", "url": "https://digitalassets.jointcommission.org/api/public/content/dcfcf4f1a0cc45cdb526b3cb034c68c2", "type": "guidance", "published": "2025-09-17", "accessed": "2026-09-26", "tier": 1, "supports": [ "Seven elements: governance; patient privacy and transparency; data security; ongoing quality monitoring; voluntary blinded reporting of AI safety events; risk and bias assessment; education and training", "Local monitoring is critical because AI tools may not undergo consistent external review in the deploying organization", "Monitoring should include ensuring AI tools rely on up-to-date data and building an AI dashboard", "Treat AI events such as major performance degradation after an update as patient safety events and report via PSOs or FDA pathways" ], "last_verified_live": "2026-09-26" }, { "id": "kashiwagi-2016-ajmq-clear-toolkit", "title": "All CLEAR? Preparing for IT Downtime", "publisher": "American Journal of Medical Quality 32(5):547-551", "url": "https://doi.org/10.1177/1062860616667546", "type": "peer-reviewed", "published": "2016-08-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "Work-area downtime drills with self-report and external audits reveal gaps in performance and plans" ], "last_verified_live": "2026-09-26" }, { "id": "kff-2024-ascension-lapses", "title": "Cyberattack led to harrowing lapses at Ascension hospitals, clinicians say", "publisher": "KFF Health News (Rachana Pradhan) and Michigan Public (Kate Wells); co-published by NPR", "url": "https://kffhealthnews.org/health-industry/hospitals-cyberattacks-ascension-patient-care/", "type": "journalism", "published": "2024-06-20", "accessed": "2026-09-26", "tier": 2, "supports": [ "Clinicians in three states described delayed or lost lab results, medication errors and loss of barcode scanning checks after the 8 May 2024 attack on Ascension's 140 hospitals", "Specific reported events (a death after lab results never arrived; a narcotic given to the wrong patient) were not confirmed or disputed by Ascension", "Expert: most downtime plans cover 'one, two, or three days'", "Clinicians reported medication errors, delayed/lost labs, loss of scanning safety checks", "Hospitals reverted to handwritten notes, faxes, spreadsheets; EHR restored June 14", "Nurse said he had no training for the response; Ascension said staff were trained" ], "last_verified_live": "2026-09-26" }, { "id": "kilbridge-2003-nejm-computer-crash", "title": "Computer crash - lessons from a system failure", "publisher": "New England Journal of Medicine 348(10):881-882", "url": "https://doi.org/10.1056/NEJMp030010", "type": "peer-reviewed", "published": "2003-03-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "NEJM Perspective on the November 2002 Beth Israel Deaconess network failure" ], "last_verified_live": "2026-09-26" }, { "id": "kim-2017-jamia-hit-problems-review", "title": "Problems with health information technology and their effects on care delivery and patient outcomes: a systematic review", "publisher": "Journal of the American Medical Informatics Association 24(2):246-250 (Kim MO, Coiera E, Magrabi F)", "url": "https://doi.org/10.1093/jamia/ocw154", "type": "peer-reviewed", "published": "2017-03", "accessed": "2026-09-26", "tier": 1, "supports": [ "Frequency, scale and severity of health IT problems are not adequately captured in the literature", "IT problems propagate along an information value chain from user interaction to information receipt, decisions, care processes and outcomes" ], "last_verified_live": "2026-09-26" }, { "id": "koenecke-2024-careless-whisper", "title": "Careless Whisper: Speech-to-Text Hallucination Harms", "publisher": "ACM FAccT 2024 (Koenecke A, Choi ASG, Mei KX, Schellmann H, Sloane M); arXiv 2402.08021", "url": "https://arxiv.org/abs/2402.08021", "type": "peer-reviewed", "published": "2024-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "About 1% of Whisper transcriptions contained whole hallucinated phrases or sentences absent from the audio", "38% of hallucinations included explicit harms such as violence, false associations or false authority", "Hallucination rates were higher for speakers with aphasia and longer non-vocal pauses" ], "last_verified_live": "2026-09-26" }, { "id": "konstantinoudis-2026-natcomms-iberian-mortality", "title": "Excess mortality attributable to the 2025 Iberian Peninsula blackout", "publisher": "Nature Communications", "url": "https://doi.org/10.1038/s41467-026-75581-w", "type": "peer-reviewed", "published": "2026-07", "accessed": "2026-09-26", "tier": 1, "supports": [ "Spain saw an estimated +167 excess deaths over the two days after the blackout (95% credible interval +28 to +300; +2.4%), concentrated in women 85+; no comparable effect in Portugal", "Eight fatalities were directly linked to the ~10-hour outage" ], "last_verified_live": "2026-09-26" }, { "id": "larouzee-lecoze-2020-swiss-cheese-critics", "title": "Good and bad reasons: The Swiss cheese model and its critics", "publisher": "Safety Science 126:104660", "url": "https://doi.org/10.1016/j.ssci.2020.104660", "type": "peer-reviewed", "published": "2020", "accessed": "2026-09-26", "tier": 1, "supports": [ "Review of critiques of the Swiss-cheese model; concludes it remains relevant because of its systemic foundations" ], "last_verified_live": "2026-09-26" }, { "id": "larsen-2018-jamia-ehr-downtime-events", "title": "Implications of electronic health record downtime: an analysis of patient safety event reports", "publisher": "Journal of the American Medical Informatics Association 25(2):187 (Larsen E, Fong A, Wernz C, Ratwani RM)", "url": "https://doi.org/10.1093/jamia/ocx057", "type": "peer-reviewed", "published": "2018-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "76 downtime-related safety event reports out of 80,381; 48.7% involved lab orders/results, 14.5% medication", "Downtime procedures were not followed or not in place in over 46% of reports; successfully executed in 27.6%", "Patient identification and communication of clinical information are key risk areas during downtime", "76 downtime-related safety reports; 48.7% lab, 14.5% medication", "46% said downtime procedures not followed or not in place; 27.6% executed successfully", "Risk areas: patient identification and communication", "76 downtime-related reports among 80,381 safety event reports", "48.7% involved lab orders/results; 14.5% medication ordering/administration", "46% of reports indicated downtime procedures were not followed or not in place" ], "last_verified_live": "2026-09-26" }, { "id": "larsen-2019-aci-downtime-laboratory", "title": "Continuing Patient Care during Electronic Health Record Downtime", "publisher": "Applied Clinical Informatics (Larsen E, Hoffman D, Rivera C, Kleiner BM, Wernz C, Ratwani RM)", "url": "https://doi.org/10.1055/s-0039-1692678", "type": "peer-reviewed", "published": "2019-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "During downtime, laboratory results were delayed by an average of 62% versus normal operation at two mid-Atlantic hospitals", "Downtime paper records were incomplete and inconsistent, so the impact of downtime is hard to measure after the fact", "Lab results delayed by average 62% during downtime", "Delays often due to improper procedure execution and incomplete documentation", "Lab result reporting delayed on average 62% during downtime", "Paper downtime records were incomplete, preventing full measurement", "Clinicians kept ordering at normal rates though manual lab capacity is a fraction of automated capacity", "Recommendations: communicate downtime limits, reduce ordering, drills, designated support staff" ], "last_verified_live": "2026-09-26" }, { "id": "leveson-2019-cast-handbook", "title": "CAST Handbook: How to Learn More from Incidents and Accidents", "publisher": "Nancy G. Leveson, MIT Partnership for Systems Approaches to Safety and Security", "url": "https://psas.scripts.mit.edu/home/get_file4.php?name=CAST_Handbook.pdf", "type": "guidance", "published": "2019", "accessed": "2026-09-26", "tier": 1, "supports": [ "CAST looks for why the control structures meant to prevent events failed, rather than for failures and blame", "Root-cause analysis oversimplifies causality" ], "last_verified_live": "2026-09-26" }, { "id": "leveson-2024-cast-healthcare-handbook", "title": "CAST Handbook: A \"Systems Thinking\" Approach to the Investigation of Healthcare Adverse Events (Draft)", "publisher": "Nancy Leveson, MIT", "url": "https://psas.scripts.mit.edu/home/get_file10.php?name=CAST_Handbook_Healthcare.pdf", "type": "guidance", "published": "2024-01-28", "accessed": "2026-09-26", "tier": 1, "supports": [ "Healthcare adverse-event investigation relies too much on blame; CAST offers a systems-theory alternative to standard RCA" ], "last_verified_live": "2026-09-26" }, { "id": "lyell-coiera-2017-jamia-verification-complexity", "title": "Automation bias and verification complexity: a systematic review", "publisher": "Journal of the American Medical Informatics Association 24(2):423-431", "url": "https://doi.org/10.1093/jamia/ocw105", "type": "peer-reviewed", "published": "2017-03-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "Automation bias found in single tasks with high verification complexity, not only multitasking", "Mitigation may focus on cognitive load reduction" ], "last_verified_live": "2026-09-26" }, { "id": "lyon-2023-aci-downtime-uptime", "title": "What Goes Up, Must Come Down: A State-of-the-Art Electronic Health Record Downtime and Uptime Procedure in a Metropolitan Health Setting", "publisher": "Applied Clinical Informatics 14(3):513-520", "url": "https://doi.org/10.1055/s-0043-1768995", "type": "peer-reviewed", "published": "2023-07-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Explicit uptime (recovery) procedure including medication reconciliation and uptime documentation", "Six-monthly drills on randomly selected units testing kit location, backup access and printing", "Doctor-pharmacist pairs reconciled medications; 2-3 hours in high medication-use areas after an 8-hour downtime", "Staff survey: 51% unaware of online resources; 27% felt unsupported" ], "last_verified_live": "2026-09-26" }, { "id": "malek-2025-dmphp-cbrne-training-review", "title": "Effectiveness of CBRNE Event-Response Training in a Hospital Setting: A Scoping Review", "publisher": "Disaster Medicine and Public Health Preparedness 20:e2", "url": "https://doi.org/10.1017/dmp.2025.10274", "type": "peer-reviewed", "published": "2025-12-23", "accessed": "2026-09-26", "tier": 1, "supports": [ "Tabletop exercises were the most common hospital training method; no single superior method identified" ], "last_verified_live": "2026-09-26" }, { "id": "martin-2019-lancet-dh-hit-failures", "title": "The effects and preventability of 2627 patient safety incidents related to health information technology failures", "publisher": "Lancet Digital Health 1(3):e127-e135", "url": "https://doi.org/10.1016/S2589-7500(19)30057-3", "type": "peer-reviewed", "published": "2019-06-27", "accessed": "2026-09-26", "tier": 1, "supports": [ "2627 HIT-failure incidents in England and Wales over 10 years; 4 contributed to death; 75% judged preventable" ], "last_verified_live": "2026-09-26" }, { "id": "mhra-2025-ai-airlock-report", "title": "AI Airlock Sandbox Pilot Programme Report", "publisher": "Medicines and Healthcare products Regulatory Agency (UK)", "url": "https://www.gov.uk/government/publications/ai-airlock-sandbox-pilot-programme-report", "type": "official-report", "published": "2025-10-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "MHRA ran a regulatory sandbox for AI as a medical device from April 2024 to March 2025 with four case studies and recommendations for the GB framework" ], "last_verified_live": "2026-09-26" }, { "id": "mhra-pms-2025", "title": "First major overhaul of medical device regulation comes into force across Great Britain", "publisher": "Medicines and Healthcare products Regulatory Agency (GOV.UK)", "url": "https://www.gov.uk/government/news/first-major-overhaul-of-medical-device-regulation-comes-into-force-across-great-britain", "type": "regulation", "published": "2025-06-16", "accessed": "2026-09-26", "tier": 1, "supports": [ "From 16 June 2025 manufacturers must report serious incidents within 15 days (previously 30)", "Field Safety Notices must be submitted to MHRA before distribution" ], "last_verified_live": "2026-09-26" }, { "id": "microsoft-2024-crowdstrike-8-5m-devices", "title": "Helping our customers through the CrowdStrike outage", "publisher": "Microsoft (David Weston)", "url": "https://blogs.microsoft.com/blog/2024/07/20/helping-our-customers-through-the-crowdstrike-outage/", "type": "official-report", "published": "2024-07-20", "accessed": "2026-09-26", "tier": 3, "supports": [ "About 8.5 million Windows devices affected, under 1% of all Windows machines", "Outsized impact because enterprises running critical services use CrowdStrike" ], "last_verified_live": "2026-09-26" }, { "id": "mitre-atlas", "title": "MITRE ATLAS (Adversarial Threat Landscape for AI Systems), data v5.6.0", "publisher": "MITRE", "url": "https://atlas.mitre.org/", "type": "standard", "published": "2026", "accessed": "2026-09-26", "tier": 2, "supports": [ "Knowledge base of adversary tactics and techniques against AI systems, including agent techniques such as AI Agent Context Poisoning and AI Agent Tool Invocation" ], "last_verified_live": "2026-09-26" }, { "id": "nao-2017-wannacry", "title": "Investigation: WannaCry cyber attack and the NHS", "publisher": "National Audit Office (UK)", "url": "https://www.nao.org.uk/reports/investigation-wannacry-cyber-attack-and-the-nhs/", "type": "official-report", "published": "2017-10-27", "accessed": "2026-09-26", "tier": 1, "supports": [ "At least 81 of 236 trusts affected plus 603 primary care and other organisations (595 GP practices)", "About 19,000 appointments cancelled; five A&E departments diverted patients", "Infected organisations had unpatched or unsupported Windows; attack could have been prevented by basic IT security practice; no mechanism to verify local compliance with alerts", "At least 80 of 236 trusts affected: 34 infected, 46 disrupted", "Precautionary shutdowns forced pen-and-paper working", "603 primary-care and other organisations infected, incl. 595 GP practices", "Five trusts diverted ambulances; 6,912 cancelled appointments counted, ~19,000 estimated", "No patient harm reported; infected organisations had unpatched or unsupported Windows" ], "last_verified_live": "2026-09-26" }, { "id": "nelson-2007-jcc-downtime-procedures", "title": "Downtime procedures for a clinical information system: a critical issue", "publisher": "Journal of Critical Care 22(1):45-50", "url": "https://doi.org/10.1016/j.jcrc.2007.01.004", "type": "peer-reviewed", "published": "2007-03", "accessed": "2026-09-26", "tier": 1, "supports": [ "LDS Hospital did not recover smoothly from planned downtimes in 2000", "Downtime plan specifies which clinical data must be re-entered and includes a regularly scheduled downtime drill like a fire drill" ], "last_verified_live": "2026-09-26" }, { "id": "neprash-2022-jama-hf-ransomware-trends", "title": "Trends in Ransomware Attacks on US Hospitals, Clinics, and Other Health Care Delivery Organizations, 2016-2021", "publisher": "JAMA Health Forum (Neprash HT, McGlave CC, Cross DA, et al.)", "url": "https://doi.org/10.1001/jamahealthforum.2022.4873", "type": "peer-reviewed", "published": "2022-12-29", "accessed": "2026-09-26", "tier": 1, "supports": [ "374 ransomware attacks on US health care delivery organizations 2016-2021; annual count rose from 43 to 91", "Nearly 42 million patients' PHI exposed", "44.4% of attacks disrupted care delivery: EHR/system downtime in 41.7%, cancelled/delayed scheduled care in 10.2%, ambulance diversion in 4.3%" ], "last_verified_live": "2026-09-26" }, { "id": "neprash-2024-j-rural-health-ransomware", "title": "What happens to rural hospitals during a ransomware attack? Evidence from Medicare data", "publisher": "Journal of Rural Health (Neprash HT, McGlave CC, Rydberg K, Henning-Smith C)", "url": "https://doi.org/10.1111/jrh.12834", "type": "peer-reviewed", "published": "2024-03-17", "accessed": "2026-09-26", "tier": 1, "supports": [ "First-week falls in inpatient, outpatient and ED volume at attacked hospitals, comparable in rural and urban settings", "Travel time/distance to the nearest non-attacked hospital was 4-7 times greater for rural attacked hospitals than urban ones" ], "last_verified_live": "2026-09-26" }, { "id": "neprash-2026-aej-policy-hacked-to-pieces", "title": "Hacked to Pieces? The Effects of Ransomware Attacks on Hospitals and Patients", "publisher": "American Economic Journal: Economic Policy 18(1):256-81 (Neprash H, McGlave C, Nikpay S)", "url": "https://www.aeaweb.org/articles?id=10.1257/pol.20240594", "type": "peer-reviewed", "published": "2026-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "Ransomware attacks cut hospital volume 17-24% in the attack week, recovering within about 3 weeks", "In-hospital mortality among patients already admitted when an attack begins rises 34-38% (Medicare claims, event-study design)" ], "last_verified_live": "2026-09-26" }, { "id": "nfpa-110", "title": "NFPA 110, Standard for Emergency and Standby Power Systems (2025 edition)", "publisher": "National Fire Protection Association", "url": "https://www.nfpa.org/product/nfpa-110-standard-for-emergency-and-standby-power-systems/p0110code/nfpa-110-standard-for-emergency-and-standby-power-systems-2025/11025", "type": "standard", "published": "2025", "accessed": "2026-09-26", "tier": 1, "supports": [ "NFPA 110 is the standard for emergency and standby power systems; 2025 is the current edition (CMS enforces the 2010 edition)" ], "last_verified_live": "2026-09-26" }, { "id": "nhs-2018-wannacry-lessons-learned", "title": "Lessons learned review of the WannaCry Ransomware Cyber Attack", "publisher": "Department of Health and Social Care / NHS England (William Smart, CIO for Health and Social Care)", "url": "https://www.england.nhs.uk/wp-content/uploads/2018/02/lessons-learned-review-wannacry-ransomware-cyber-attack-cio-review.pdf", "type": "official-report", "published": "2018-02-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "At least 80 of 236 trusts affected: 34 infected, 46 not infected but disrupted (e.g., took email offline as a precaution); 1% of NHS activity directly affected; no reports of patient harm", "Some diagnostic devices (MRI, blood test analysers) ran unsupported XP; results became unviewable when the viewing workstation was infected or quarantined", "Some local authorities switched off their NHS N3 network link as a precaution; mutual aid shifted A&E demand to neighbours" ], "last_verified_live": "2026-09-26" }, { "id": "nhs-dcb0160", "title": "DCB0160: Clinical Risk Management: its Application in the Deployment and Use of Health IT Systems", "publisher": "NHS England (NHS Standards Directory)", "url": "https://standards.nhs.uk/published-standards/clinical-risk-management-its-application-in-the-deployment-and-use-of-health-it-systems", "type": "standard", "published": "2018-06-07", "accessed": "2026-09-26", "tier": 1, "supports": [ "Care organisations deploying health IT must apply clinical risk management; mandated under section 250 Health and Social Care Act 2012" ], "last_verified_live": "2026-09-26" }, { "id": "nhs-england-synnovis-incident", "title": "Synnovis cyber incident (update of 10 November 2025)", "publisher": "NHS England", "url": "https://www.england.nhs.uk/synnovis-cyber-incident/", "type": "official-report", "published": "2025-11-10", "accessed": "2026-09-26", "tier": 1, "supports": [ "Synnovis, pathology provider co-owned by Guy's and St Thomas', King's College Hospital and SYNLAB, suffered a ransomware attack on 3 June 2024", "Over 11,000 outpatient and elective procedure appointments delayed in South-East London; services fully restored by December 2024", "Stolen data published 20 June 2024; data-scope investigation took over a year because the data were unstructured and fragmented" ], "last_verified_live": "2026-09-26" }, { "id": "nhsbt-2024-o-type-appeal", "title": "O Positive and O Negative donors asked to urgently book appointments to give blood following London hospitals IT incident", "publisher": "NHS Blood and Transplant", "url": "https://www.nhsbt.nhs.uk/news/o-positive-and-o-negative-donors-asked-to-urgently-book-appointments-to-give-blood-following-london-hospitals-it-incident/", "type": "official-report", "published": "2024-06-10", "accessed": "2026-09-26", "tier": 1, "supports": [ "Affected hospitals could not match patients' blood at the usual frequency, raising O-type demand", "Synnovis attack meant affected hospitals could not match blood at usual frequency; O-type blood used", "Operations and appointments postponed or diverted to neighbouring hospitals" ], "last_verified_live": "2026-09-26" }, { "id": "nhse-htm-06-01", "title": "Health Technical Memorandum 06-01: Electrical services supply and distribution", "publisher": "NHS England", "url": "https://www.england.nhs.uk/publication/electrical-services-supply-and-distribution-htm-06-01/", "type": "guidance", "published": "2017-04-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "UK guidance on legal requirements, design, operation and maintenance of healthcare electrical infrastructure; applies to new and existing sites; last updated 11 April 2017" ], "last_verified_live": "2026-09-26" }, { "id": "nist-ai-100-1-rmf", "title": "Artificial Intelligence Risk Management Framework (AI RMF 1.0), NIST AI 100-1", "publisher": "National Institute of Standards and Technology", "url": "https://doi.org/10.6028/NIST.AI.100-1", "type": "standard", "published": "2023-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "Four functions: Govern, Map, Measure, Manage", "MEASURE 2.4: system behaviour monitored in production; MEASURE 2.5: limits of generalizability documented", "MANAGE 2.4: mechanisms and assigned responsibility to supersede, disengage or deactivate AI systems performing inconsistently with intended use", "MANAGE 4.1: post-deployment monitoring including appeal and override, decommissioning, incident response, recovery and change management" ], "last_verified_live": "2026-09-26" }, { "id": "nist-ai-600-1-genai-profile", "title": "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1", "publisher": "National Institute of Standards and Technology", "url": "https://doi.org/10.6028/NIST.AI.600-1", "type": "standard", "published": "2024-07", "accessed": "2026-09-26", "tier": 1, "supports": [ "Defines confabulation as confidently stated but erroneous or false content", "Names automation bias and over-reliance as a GAI risk that can exacerbate confabulation and bias" ], "last_verified_live": "2026-09-26" }, { "id": "nist-sp800-161r1", "title": "NIST SP 800-161 Rev. 1: Cybersecurity Supply Chain Risk Management Practices for Systems and Organizations", "publisher": "NIST", "url": "https://csrc.nist.gov/pubs/sp/800/161/r1/upd1/final", "type": "standard", "published": "2022-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Multilevel guidance for identifying, assessing and mitigating cybersecurity supply-chain risk" ], "last_verified_live": "2026-09-26" }, { "id": "npr-2019-hollywood-hills", "title": "4 Former Staffers Face Charges Over Nursing Home Deaths After Hurricane Irma", "publisher": "NPR", "url": "https://www.npr.org/2019/08/26/754400472/four-former-staffers-face-charges-nursing-home-deaths-after-hurricane-irma", "type": "journalism", "published": "2019-08-26", "accessed": "2026-09-26", "tier": 2, "supports": [ "Irma knocked out a transformer supplying the Hollywood Hills rehabilitation center's air conditioning; power otherwise stayed on", "14 residents died; the county medical examiner ruled 12 deaths homicides; staff faced manslaughter charges" ], "last_verified_live": "2026-09-26" }, { "id": "ntsb-har-19-03-uber-tempe", "title": "Collision Between Vehicle Controlled by Developmental Automated Driving System and Pedestrian, Tempe, Arizona, March 18, 2018 (HAR-19-03)", "publisher": "National Transportation Safety Board", "url": "https://www.ntsb.gov/investigations/Pages/HWY18MH010.aspx", "type": "official-report", "published": "2019", "accessed": "2026-09-26", "tier": 1, "supports": [ "Probable cause: operator visually distracted by personal phone, failed to monitor", "Contributing: lack of adequate mechanisms for addressing operators' automation complacency" ], "last_verified_live": "2026-09-26" }, { "id": "obermeyer-2019-science-racial-bias", "title": "Dissecting racial bias in an algorithm used to manage the health of populations", "publisher": "Science (Obermeyer Z, Powers B, Vogeli C, Mullainathan S)", "url": "https://doi.org/10.1126/science.aax2342", "type": "peer-reviewed", "published": "2019-10-25", "accessed": "2026-09-26", "tier": 1, "supports": [ "A widely used commercial care-management algorithm predicted health-care cost, not illness, so Black patients at a given risk score were sicker than White patients", "Fixing the disparity would raise the share of Black patients receiving extra help from 17.7% to 46.5%", "Convenient proxy labels are a general source of algorithmic bias" ], "last_verified_live": "2026-09-26" }, { "id": "omiye-2023-npj-race-based-llm", "title": "Large language models propagate race-based medicine", "publisher": "npj Digital Medicine (Omiye JA, Lester JC, Spichak S, Rotemberg V, Daneshjou R)", "url": "https://doi.org/10.1038/s41746-023-00939-z", "type": "peer-reviewed", "published": "2023-10-20", "accessed": "2026-09-26", "tier": 1, "supports": [ "All four commercial LLMs tested produced debunked race-based medical content", "Models gave inconsistent answers to the same question asked repeatedly" ], "last_verified_live": "2026-09-26" }, { "id": "onc-2025-hti5-proposed", "title": "HTI-5 Proposed Rule fact sheet: ONC Deregulatory Actions to Unleash Prosperity", "publisher": "ASTP/ONC", "url": "https://www.healthit.gov/topic/laws-regulation-and-policy/hti-5-proposed-rule-fact-sheet", "type": "regulation", "published": "2025-12-22", "accessed": "2026-09-26", "tier": 1, "supports": [ "HTI-5 proposes to reduce the DSI criterion and fully remove the AI 'model card' (source attribute) requirements" ], "last_verified_live": "2026-09-26" }, { "id": "onc-45cfr170-315-b11", "title": "45 CFR 170.315(b)(11) Decision support interventions (ONC Health IT Certification criterion)", "publisher": "ASTP/ONC, U.S. Department of Health and Human Services (eCFR)", "url": "https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-D/part-170/subpart-C/section-170.315", "type": "regulation", "published": "2024-01-09", "accessed": "2026-09-26", "tier": 1, "supports": [ "Certified EHRs must support source attributes for predictive DSIs including intended use, cautioned out-of-scope use, external validation, fairness, validity in local data, and how validity is monitored over time" ], "last_verified_live": "2026-09-26" }, { "id": "onc-safer-contingency-2025", "title": "SAFER Guide: Contingency Planning (2025 edition)", "publisher": "ASTP/ONC, US Department of Health and Human Services", "url": "https://www.healthit.gov/wp-content/uploads/2025/06/SAFER-Guide-2.-Contingency-Planning-Final.pdf", "type": "guidance", "published": "2025", "accessed": "2026-09-26", "tier": 1, "supports": [ "A UPS (battery or flywheel) should maintain the EHR for at least 10 minutes and be tested at least monthly", "The backup generator should maintain critical EHR functions; keep 2 days of fuel on site and weigh the risks of storing more (e.g., 96 hours)", "Keep the UPS and generator in secure locations not likely to flood", "Large 24/7 organizations should have a warm-site backup >50 miles away and >20 miles from the coastline, activatable in <8 hours, tested at least quarterly", "Keep a read-only backup EHR that can print, tested regularly, and connected to a UPS or generator-backed 'red plug' outlet", "Avoid single points of failure; maintain redundant internet paths in different trenches", "Disaster recovery plans reviewed at least annually, including hardware duplication, network redundancy and data replication (1.1)", "Warm-site >50 miles away able to run the whole EHR within 8 hours, tested at least quarterly; redundant internet paths via separate trenches or providers (1.1 guidance)", "Daily off-site encrypted backups, restore-tested optimally monthly; air-gapped backup media for ransomware recovery; read-only backup EHR updated optimally hourly and tested weekly, on UPS or generator power (1.4)", "Unannounced downtime drills at least yearly (2.1); downtime communication independent of EHR infrastructure (2.2); activate warm site ideally before 2 hours of unplanned downtime (2.3)", "Orderly stop/restart of interfaces with buffers sized to prevent data loss and user alerts when an interface is down (2.5)", "Monitor response time (optimally under 2 s) including a synthetic daily test order; root-cause review of unplanned downtimes over 24 hours (3.1-3.3)", "Many organisations have employees who do not know how to function in a paper-based environment", "Practice 2.1: staff trained and tested on downtime; unannounced downtime drills at least yearly; read-only backup EHR training", "Paper forms for key EHR functions; EHR-independent communication; patient ID procedures; review downtimes over 24 hours", "Paper forms for at least 8 hours of care in each area", "Tested read-only backup EHR", "Unannounced downtime drills at least once a year", "13 recommended practices for planned and unplanned EHR unavailability", "Paper forms for at least 8 hours in each care area; process to enter and reconcile paper data after reactivation (1.3)", "Read-only backup EHR updated at least hourly, tested at least weekly, printable, on UPS or red plug (1.4)", "UPS for at least 10 minutes, tested monthly; generator tested monthly; 2 days of fuel on site (1.2)", "Temporary patient record numbers during downtime with reconciliation after (1.5)", "Unannounced downtime drills at least once a year; staff trained and tested on downtime and recovery (2.1)", "Downtime communication independent of EHR infrastructure; mechanism to announce EHR back online (2.2)", "Warm-site activation ideally before 2 hours of unplanned unavailability; policy reviewed every 2 years (2.3)", "Orderly stop/restart of interfaces; empty buffers; in-transit data can be lost silently (2.5)", "Functional downtime defined as hourly mean response time >5 s or >3 SD above mean; target <2 s (3.1, 3.2)", "Root-cause or FMEA review of unexpected downtimes over 24 hours (3.3)" ], "last_verified_live": "2026-09-26" }, { "id": "ontario-ag-2026-ai-report", "title": "Performance Audit: Use of Artificial Intelligence in the Ontario Government (Special Report 2026)", "publisher": "Office of the Auditor General of Ontario", "url": "https://www.auditor.on.ca/en/content/specialreports/specialreports/en26/2026_AI_EN.pdf", "type": "official-report", "published": "2026-05-12", "accessed": "2026-09-26", "tier": 1, "supports": [ "All 20 approved AI Scribe vendors showed at least one inaccuracy in procurement testing", "9 of 20 fabricated information including treatment steps never discussed; 12 of 20 recorded a different drug than prescribed; 17 of 20 missed key mental-health details in at least one test" ], "last_verified_live": "2026-09-26" }, { "id": "owasp-2025-agentic-top10", "title": "OWASP Top 10 for Agentic Applications", "publisher": "OWASP GenAI Security Project", "url": "https://genai.owasp.org/2025/12/09/owasp-top-10-for-agentic-applications-the-benchmark-for-agentic-security-in-the-age-of-autonomous-ai/", "type": "standard", "published": "2025-12-09", "accessed": "2026-09-26", "tier": 2, "supports": [ "ASI06 Memory & Context Poisoning; ASI08 Cascading Failures; ASI09 Human-Agent Trust Exploitation; ASI10 Rogue Agents (cites Replit incident)" ], "last_verified_live": "2026-09-26" }, { "id": "owasp-llm-top10-2025", "title": "OWASP Top 10 for LLM Applications 2025", "publisher": "OWASP GenAI Security Project", "url": "https://genai.owasp.org/llm-top-10/", "type": "standard", "published": "2025", "accessed": "2026-09-26", "tier": 2, "supports": [ "LLM06 Excessive Agency: excessive functionality, permissions and autonomy; mitigate with least privilege and human approval of high-impact actions", "LLM09 Misinformation" ], "last_verified_live": "2026-09-26" }, { "id": "parasuraman-riley-1997-human-factors", "title": "Humans and Automation: Use, Misuse, Disuse, Abuse", "publisher": "Human Factors 39(2):230-253", "url": "https://doi.org/10.1518/001872097778543886", "type": "peer-reviewed", "published": "1997-06", "accessed": "2026-09-26", "tier": 1, "supports": [ "Misuse = over-reliance leading to monitoring failures and decision biases", "Disuse commonly caused by false alarms" ], "last_verified_live": "2026-09-26" }, { "id": "parasuraman-sheridan-wickens-2000-ieee", "title": "A Model for Types and Levels of Human Interaction with Automation", "publisher": "IEEE Transactions on Systems, Man, and Cybernetics Part A 30(3):286-297", "url": "https://doi.org/10.1109/3468.844354", "type": "peer-reviewed", "published": "2000-05", "accessed": "2026-09-26", "tier": 1, "supports": [ "Automation applies to four function classes (information acquisition, analysis, decision selection, action implementation) at levels from manual to fully automatic", "Automation changes human activity and imposes new coordination demands", "Automation applies to four functions (information acquisition, analysis, decision selection, action implementation), each on a continuum from manual to fully automatic", "10-point scale of decision automation based on Sheridan and Verplank 1978", "High-level automation risks reduced situation awareness, complacency and skill degradation; these matter most when automation fails", "Reliability is a key criterion; point estimates of reliability should be treated with skepticism" ], "last_verified_live": "2026-09-26" }, { "id": "perneger-2005-swiss-cheese-holes", "title": "The Swiss cheese model of safety incidents: are there holes in the metaphor?", "publisher": "BMC Health Services Research 5:71", "url": "https://doi.org/10.1186/1472-6963-5-71", "type": "peer-reviewed", "published": "2005", "accessed": "2026-09-26", "tier": 1, "supports": [ "Among 85 quality and safety professionals, interpretation of the Swiss-cheese model varied widely (e.g. only 32.9% identified holes as latent errors)" ], "last_verified_live": "2026-09-26" }, { "id": "perrow-1999-normal-accidents", "title": "Normal Accidents: Living with High-Risk Technologies (Updated Edition)", "publisher": "Princeton University Press", "url": "https://press.princeton.edu/books/paperback/9780691004129/normal-accidents", "type": "book", "published": "1999", "accessed": "2026-09-26", "tier": 1, "supports": [ "Interactive complexity and tight coupling make system accidents 'normal'", "Adding safeguards can add complexity and new accident types" ], "last_verified_live": "2026-09-26" }, { "id": "psi-2024-ma-post-acute-report", "title": "Refusal of Recovery: How Medicare Advantage Insurers Have Denied Patients Access to Post-Acute Care", "publisher": "U.S. Senate Permanent Subcommittee on Investigations (Majority Staff)", "url": "https://www.hsgac.senate.gov/wp-content/uploads/2024.10.17-PSI-Majority-Staff-Report-on-Medicare-Advantage.pdf", "type": "official-report", "published": "2024-10-17", "accessed": "2026-09-26", "tier": 1, "supports": [ "UnitedHealthcare's post-acute prior-authorization denial rate rose from 8.7% (2019) to 22.7% (2022) while its overall denial rate stayed near 7.5%", "The rise coincided with initiatives to automate prior authorization; an internal auto-authorization model test noted 'an increase in adverse determination rate'", "A 2022 naviHealth presentation showed care coordinators completing nH Predict to determine post-acute placement" ], "last_verified_live": "2026-09-26" }, { "id": "reason-2000-bmj-human-error", "title": "Human error: models and management", "publisher": "BMJ 320(7237):768-770", "url": "https://doi.org/10.1136/bmj.320.7237.768", "type": "peer-reviewed", "published": "2000", "accessed": "2026-09-26", "tier": 1, "supports": [ "Swiss-cheese model: latent conditions and active failures align across defensive layers", "Systems approach versus person approach to error" ], "last_verified_live": "2026-09-26" }, { "id": "rossley-2022-jnpd-escape-room", "title": "Implementing a Large-Scale Escape Room: An Intervention to Increase Electronic Health Record Downtime Competence", "publisher": "Journal for Nurses in Professional Development 39(6):316-321", "url": "https://doi.org/10.1097/NND.0000000000000863", "type": "peer-reviewed", "published": "2022", "accessed": "2026-09-26", "tier": 1, "supports": [ "74 nurses in a tabletop escape-room downtime exercise; 90% reported it influenced learning (self-report)" ], "last_verified_live": "2026-09-26" }, { "id": "sae-j3016-2021", "title": "J3016_202104: Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles", "publisher": "SAE International", "url": "https://saemobilus.sae.org/standards/j3016_202104-taxonomy-definitions-terms-related-driving-automation-systems-road-motor-vehicles", "type": "standard", "published": "2021-04-30", "accessed": "2026-09-26", "tier": 1, "supports": [ "Six levels of driving automation, 0-5 (analogue only; not a healthcare standard)" ], "last_verified_live": "2026-09-26" }, { "id": "sciam-2026-ai-nurses-schellmann", "title": "AI enters the exam room, and nurses are left to manage the fallout", "publisher": "Scientific American (Hilke Schellmann)", "url": "https://www.scientificamerican.com/article/ai-is-entering-health-care-and-nurses-are-being-asked-to-trust-it/", "type": "journalism", "published": "2026-02-17", "tier": 2, "supports": [ "Independent firsthand account: St. Rose Dominican, Henderson; AI sepsis alert; dialysis catheter; charge nurse told him to give fluids; physician ordered dopamine instead", "Hospital spokesperson declined to comment on specific case" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "sendelbach-funk-2013-aacn-alarm-fatigue", "title": "Alarm fatigue: a patient safety concern", "publisher": "AACN Advanced Critical Care 24(4):378-386", "url": "https://doi.org/10.1097/NCI.0b013e3182a903f9", "type": "peer-reviewed", "published": "2013", "accessed": "2026-09-26", "tier": 1, "supports": [ "72% to 99% of clinical alarms reported false in research", "Alarm fatigue leads to desensitisation and missed alarms" ], "last_verified_live": "2026-09-26" }, { "id": "sheridan-verplank-1978", "title": "Human and Computer Control of Undersea Teleoperators (MIT Man-Machine Systems Laboratory technical report)", "publisher": "MIT / Office of Naval Research (DTIC AD-A057655)", "url": "https://doi.org/10.21236/ada057655", "type": "official-report", "published": "1978", "accessed": "2026-09-26", "tier": 1, "supports": [ "Origin of the 10-level scale of human-computer control (as reproduced in Parasuraman, Sheridan and Wickens 2000, Table I)" ], "last_verified_live": "2026-09-26" }, { "id": "sittig-2014-ijmi-contingency-survey", "title": "Contingency planning for electronic health record-based care continuity: a survey of recommended practices", "publisher": "International Journal of Medical Informatics 83(11):797-804 (Sittig DF, Gonzalez D, Singh H)", "url": "https://doi.org/10.1016/j.ijmedinf.2014.07.007", "type": "peer-reviewed", "published": "2014", "accessed": "2026-09-26", "tier": 1, "supports": [ "96% of 50 large US integrated health systems reported at least one unplanned EHR downtime in 3 years; 70% had one longer than 8 hours", "Three institutions reported patient injury linked to planned or unplanned downtime", "Very few recommended contingency practices were followed by all organizations" ], "last_verified_live": "2026-09-26" }, { "id": "sjoding-2020-nejm-pulse-oximetry", "title": "Racial Bias in Pulse Oximetry Measurement", "publisher": "New England Journal of Medicine (Sjoding MW, Dickson RP, Iwashyna TJ, Gay SE, Valley TS)", "url": "https://doi.org/10.1056/NEJMc2029240", "type": "peer-reviewed", "published": "2020-12-17", "accessed": "2026-09-26", "tier": 1, "supports": [ "Occult hypoxemia (SaO2 <88% with SpO2 92-96%) in 11.7% of Black vs 3.6% of White patients' paired measurements (Michigan cohort)", "17.0% vs 6.2% in a 178-hospital multicenter cohort", "Black patients had nearly three times the frequency of occult hypoxemia not detected by pulse oximetry" ], "last_verified_live": "2026-09-26" }, { "id": "skarha-2021-jamahf-irma-nursing-homes", "title": "Association of Power Outage With Mortality and Hospitalizations Among Florida Nursing Home Residents After Hurricane Irma", "publisher": "JAMA Health Forum", "url": "https://doi.org/10.1001/jamahealthforum.2021.3900", "type": "peer-reviewed", "published": "2021-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Among 54,095 Florida nursing home residents, power loss after Irma was associated with higher odds of death within 7 days (OR 1.25, 95% CI 1.05–1.48) and 30 days (OR 1.12)" ], "last_verified_live": "2026-09-26" }, { "id": "statnews-2024-crowdstrike-hospitals", "title": "Microsoft global outage forces hospitals to cancel appointments", "publisher": "STAT (Palmer K, Trang B, Ross C)", "url": "https://www.statnews.com/2024/07/19/microsoft-crowdstrike-outage-hospitals-cancellations/", "type": "journalism", "published": "2024-07-19", "accessed": "2026-09-26", "tier": 2, "supports": [ "Mass General Brigham cancelled non-urgent surgeries, procedures and visits on 19 July 2024", "Dana-Farber told patients to stay home; Memorial Sloan Kettering suspended procedures needing anesthesia; Duke reported clinical systems affected" ], "last_verified_live": "2026-09-26" }, { "id": "sujan-2019-bmjhci-human-factors-ai", "title": "Human factors challenges for the safe use of artificial intelligence in patient care", "publisher": "BMJ Health & Care Informatics 26(1)", "url": "https://doi.org/10.1136/bmjhci-2019-100081", "type": "peer-reviewed", "published": "2019-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "Human factors challenges for clinical AI include automation bias, handover between clinicians and AI, and situation awareness" ], "last_verified_live": "2026-09-26" }, { "id": "synnovis-cyber-update", "title": "Synnovis cyber update", "publisher": "Synnovis", "url": "https://synnoviscyberupdate.co.uk/", "type": "official-report", "published": "2025", "accessed": "2026-09-26", "tier": 3, "supports": [ "Attack on 3 June 2024 affected almost all Synnovis IT systems", "Data published 20 June 2024", "Full restoration by late autumn 2024", "Ransomware on 3 June 2024 affected almost all Synnovis IT systems", "Access to all previously available services restored by late autumn 2024; none of the affected infrastructure remains" ], "last_verified_live": "2026-09-26" }, { "id": "texastribune-2021-austin-hospitals", "title": "Austin hospitals lose water pressure and heat amid winter storm", "publisher": "The Texas Tribune", "url": "https://www.texastribune.org/2021/02/17/austin-texas-hospitals-power-outage-water/", "type": "journalism", "published": "2021-02-17", "accessed": "2026-09-26", "tier": 2, "supports": [ "St. David's South Austin Medical Center lost city water pressure; because water feeds its boiler it also lost heat; stable patients were discharged and others transferred", "Officials said no one hospital could accept a large number of transfers because the impact was regional", "Dell Children's asked patients not to shower and limited linen changes" ], "last_verified_live": "2026-09-26" }, { "id": "therecord-2025-synnovis-death", "title": "Ransomware attack contributed to patient's death, says Britain's NHS", "publisher": "The Record by Recorded Future News (Alexander Martin)", "url": "https://therecord.media/ransomware-attack-contributed-patient-death-uk-nhs", "type": "journalism", "published": "2025-06-25", "accessed": "2026-09-26", "tier": 2, "supports": [ "King's College Hospital: a patient safety investigation found a long wait for a blood test result due to the Synnovis attack among contributing factors in a patient's unexpected death", "Blood stocks depleted; hospitals relied on universal-donor blood", "KCH statement: long wait for a blood test result due to the attack was a contributing factor in a patient's death" ], "last_verified_live": "2026-09-26" }, { "id": "therecord-2026-synnovis-ongoing", "title": "Ransomware attack continues to disrupt healthcare in London nearly two years later", "publisher": "The Record (Recorded Future News)", "url": "https://therecord.media/ransomware-nhs-cyberattack-disruption", "type": "journalism", "published": "2026", "accessed": "2026-09-26", "tier": 2, "supports": [ "NHS England reported 10,152 outpatient appointments and 1,710 elective procedures postponed", "SLaM still without fully restored pathology systems; 122 incidents of incorrect, unavailable or delayed results by January 2026" ], "last_verified_live": "2026-09-26" }, { "id": "theregister-2025-qilin-nhs-death", "title": "Qilin ransomware attack on NHS supplier contributed to patient fatality", "publisher": "The Register", "url": "https://www.theregister.com/2025/06/26/qilin_ransomware_nhs_death/", "type": "journalism", "published": "2025-06-26", "accessed": "2026-09-26", "tier": 2, "supports": [ "South East London ICB reported 170 patients harmed, mostly low harm" ], "last_verified_live": "2026-09-26" }, { "id": "theregister-2025-replit-database", "title": "Vibe coding service Replit deleted user's production database, faked data, told fibs galore", "publisher": "The Register", "url": "https://www.theregister.com/2025/07/21/replit_saastr_vibe_coding_incident/", "type": "journalism", "published": "2025-07-21", "accessed": "2026-09-26", "tier": 2, "supports": [ "Replit's coding agent deleted a production database despite an instruction to freeze code changes", "The agent said rollback was impossible; rollback in fact worked" ], "last_verified_live": "2026-09-26" }, { "id": "tjc-em-reference-guide-2022", "title": "Reference Guide: Emergency Management Standards, effective July 1, 2022 (HAP & CAH)", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/7988802d78c54b3d8246f104e41896c5?v=aee3a672", "type": "standard", "published": "2022", "accessed": "2026-09-26", "tier": 1, "supports": [ "EM.13.01.01: written continuity of operations plan, secondary location, succession plan, delegation of authority", "EM.14.01.01: written disaster recovery strategies for assessment and restoration", "EM.16.01.01: two exercises per year", "EM.17.01.01: committee reviews after-action reports/improvement plans; forwarded to senior leadership", "EM.12.02.11: 96-hour sustainability plan; plans for alternative power and utility systems" ], "last_verified_live": "2026-09-26" }, { "id": "tjc-npg-2026-hospital", "title": "National Performance Goals Effective January 2026 for the Hospital Program", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/9ca80055182b4274842a5780a94f2c82", "type": "standard", "published": "2026-01-01", "accessed": "2026-09-26", "tier": 2, "supports": [ "NPG.01.05.01: the hospital improves the safety of clinical alarm systems", "EP 1: identify the most important alarm signals based on risk if unattended or malfunctioning, alarm fatigue, incident history" ], "last_verified_live": "2026-09-26" }, { "id": "tjc-npsg-2025-hospital", "title": "National Patient Safety Goals, Effective January 2025 for the Hospital Program", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/9be383450fc941df806b76c5fbdd9ae6?v=3c600c3a", "type": "standard", "published": "2024-10-30", "tier": 1, "supports": [ "Through 2025 the hospital alarm goal was NPSG.06.01.01" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "tjc-perspectives-2025-07", "title": "Joint Commission Perspectives, July 2025 (Vol. 45, No. 7)", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/integration/ingenta/publications/jcp_45_2025_07.pdf", "type": "guidance", "published": "2025-07-01", "tier": 1, "supports": [ "NPG chapter supersedes/replaces the NPSG chapter for hospitals and CAHs effective 1 January 2026" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "tjc-r3-issue-5-npsg-06-01-01", "title": "R3 Report Issue 5: Alarm system safety (NPSG.06.01.01)", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/aa15f8d34283484d9947f6802aa14122?v=4e9c34ef", "type": "standard", "published": "2013-12-11", "accessed": "2026-09-26", "tier": 1, "supports": [ "NPSG.06.01.01 EPs 1-4 and dates", "Policies must define who may set, change and turn off alarm parameters", "NPSG generally excludes CPOE and other IT alerts" ], "last_verified_live": "2026-09-26" }, { "id": "tjc-sea-50-alarms-2013", "title": "Sentinel Event Alert Issue 50: Medical device alarm safety in hospitals", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/f65e5c9df2b94000a99445e0a7877007", "type": "official-report", "published": "2013-04-08", "accessed": "2026-09-26", "tier": 2, "supports": [ "98 alarm-related sentinel events Jan 2009-Jun 2012, 80 of them deaths", "An estimated 85-99% of alarm signals do not require clinical intervention", "MAUDE had 566 alarm-related deaths reported Jan 2005-Jun 2010, considered an undercount", "98 alarm-related sentinel events Jan 2009-Jun 2012, 80 deaths", "85-99% of alarm signals estimated not to require intervention", "Contributors: alarms inappropriately off (36), not audible (25), improper settings (21); alarm fatigue most common", "FDA MAUDE: 566 alarm-related deaths Jan 2005-Jun 2010", "2010 ICU death after one-hour delayed alarm response", "Recommendations: tailor settings, change electrodes, check audibility, cross-disciplinary team" ], "last_verified_live": "2026-09-26" }, { "id": "tjc-sea-67-cyberattack", "title": "Sentinel Event Alert Issue 67: Preserving patient safety after a cyberattack", "publisher": "The Joint Commission", "url": "https://digitalassets.jointcommission.org/api/public/content/8bd6addf4df943f383a074f2d9d4af64?v=44f03d64", "type": "guidance", "published": "2023-08-15", "accessed": "2026-09-26", "tier": 1, "supports": [ "Be prepared to have life- and safety-critical technology offline for four weeks or longer", "99 cyberattack/ransomware/hacking events reported to TJC over 10 years, 90% hospitals, some with harm", "Case: loss of CDS during EHR downtime contributed to a GI bleed on warfarin with an interacting drug", "Princeton Community Hospital moved to paper within an hour of Petya ransomware and restored computers after 36 hours", "Paper resources should match the format of electronic workflows, not legacy forms", "Drills annually or quarterly depending on staff turnover and prior downtime experience", "MGH iDART three-tier escalation: assessment team, response team, HICS activation", "Restore electronic systems deliberately; assign staff to back-enter data; authenticate all transcribed orders", "Organisations should prepare to deliver safe care through extended cyberattack downtime", "Prioritize services that must stay operational during extended downtime using the hazard vulnerability analysis", "Form a downtime planning committee; designate interdisciplinary response teams; train all staff on downtime operation; keep communication with patients and families; regroup and improve after an attack" ], "last_verified_live": "2026-09-26" }, { "id": "tully-2025-jama-netw-open-crowdstrike", "title": "Patient Care Technology Disruptions Associated With the CrowdStrike Outage", "publisher": "JAMA Network Open (Tully JL, ... Dameff CJ)", "url": "https://doi.org/10.1001/jamanetworkopen.2025.30226", "type": "peer-reviewed", "published": "2025-07-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "759 of 2,232 US hospitals (34.0%) lost response to internet scanning immediately after the update", "239 of 1,098 affected network services (21.8%) were direct patient-facing", "759 of 2,232 US hospitals had detectable service disruptions", "239 of 1,098 disrupted services were patient-facing", "Outages were measured from outside via IP scans and FHIR endpoint scans", "759 of 2232 US hospitals (34.0%) lost response to internet scanning after the 19 July 2024 faulty update", "239 of 1098 affected network services were direct patient-facing", "Internet measurement may be useful for surveillance of health infrastructure" ], "last_verified_live": "2026-09-26" }, { "id": "uhs-2020-10k", "title": "Universal Health Services Form 10-K for fiscal year 2020", "publisher": "Universal Health Services, Inc. / SEC EDGAR", "url": "https://www.sec.gov/Archives/edgar/data/352915/000156459021008851/uhs-10k_20201231.htm", "type": "official-report", "published": "2021", "accessed": "2026-09-26", "tier": 1, "supports": [ "Applications restored on a rolling basis during October 2020; ambulance traffic and elective procedures diverted to competitors; coding and billing delayed into December", "Estimated pre-tax impact about $67 million in 2020" ], "last_verified_live": "2026-09-26" }, { "id": "uhs-2020-8k", "title": "Universal Health Services Form 8-K (information technology security incident)", "publisher": "Universal Health Services, Inc. / SEC EDGAR", "url": "https://www.sec.gov/Archives/edgar/data/352915/000156459020044863/uhs-8k_20200927.htm", "type": "official-report", "published": "2020-09-29", "accessed": "2026-09-26", "tier": 1, "supports": [ "Incident early on 27 Sep 2020; UHS suspended user access to IT applications for US operations; facilities used established back-up processes including offline documentation" ], "last_verified_live": "2026-09-26" }, { "id": "urban-2006-hospitals-katrina", "title": "Hospitals in Hurricane Katrina: Challenges Facing Custodial Institutions in a Disaster", "publisher": "The Urban Institute (Bradford H. Gray, Kathy Hebert)", "url": "https://www.urban.org/sites/default/files/publication/50896/411348-Hospitals-in-Hurricane-Katrina.PDF", "type": "official-report", "published": "2006-07-01", "tier": 2, "supports": [ "45 bodies at Memorial (Tenet said 11 died before the storm)", "Generator failure undisputed", "Generators failed midday Wednesday", "Louisiana AG investigation, 73 subpoenas" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" }, { "id": "usc-21-360n-2-section-524b", "title": "21 U.S.C. 360n-2 (FD&C Act section 524B): Ensuring cybersecurity of devices", "publisher": "Office of the Law Revision Counsel, U.S. House of Representatives", "url": "https://uscode.house.gov/view.xhtml?req=granuleid:USC-prelim-title21-section360n-2&num=0&edition=prelim", "type": "regulation", "published": "2022-12-29", "accessed": "2026-09-26", "tier": 1, "supports": [ "Cyber device sponsors must submit a plan to monitor, identify and address postmarket vulnerabilities and exploits", "Updates and patches on a reasonably justified regular cycle for known vulnerabilities, and as soon as possible out of cycle for critical ones", "Sponsors must provide a software bill of materials including commercial, open-source and off-the-shelf components", "Added by Pub. L. 117-328 (Consolidated Appropriations Act, 2023), enacted 29 Dec 2022" ], "last_verified_live": "2026-09-26" }, { "id": "va-oig-2022-unknown-queue", "title": "The New Electronic Health Record's Unknown Queue Caused Multiple Events of Patient Harm (Report 22-01137-204)", "publisher": "VA Office of Inspector General", "url": "https://www.vaoig.gov/sites/default/files/reports/2022-07/VAOIG-22-01137-204.pdf", "type": "official-report", "published": "2022-07-14", "accessed": "2026-09-26", "tier": 1, "supports": [ "From go-live (Oct 2020) to June 2021 the new EHR at Mann-Grandstaff VAMC sent more than 11,000 orders to an 'unknown queue' instead of the intended service, without alerting the ordering provider", "VHA review of 1,286 event assessments classified 149 adverse events: 2 major, 52 moderate, 95 minor harm", "Mitigation relied on staff manually monitoring the queue; 206 orders still in the queue on 16 May 2022" ], "last_verified_live": "2026-09-26" }, { "id": "vandersijs-2006-jamia-alert-override", "title": "Overriding of drug safety alerts in computerized physician order entry", "publisher": "Journal of the American Medical Informatics Association 13(2):138-147", "url": "https://doi.org/10.1197/jamia.M1809", "type": "peer-reviewed", "published": "2006", "accessed": "2026-09-26", "tier": 1, "supports": [ "Drug safety alerts overridden in 49% to 96% of cases", "Low specificity and workflow disruption are error-producing conditions" ], "last_verified_live": "2026-09-26" }, { "id": "vtdigger-2026-kinney-burt", "title": "A pharmacy chain in Vermont implemented AI for efficiency. It's led to delays, incorrect information and privacy concerns", "publisher": "VTDigger", "url": "https://vtdigger.org/2026/07/29/a-pharmacy-chain-in-vermont-implemented-ai-for-efficiency-its-led-to-delays-incorrect-information-and-privacy-concerns/", "type": "journalism", "published": "2026-07-29", "accessed": "2026-09-26", "tier": 2, "supports": [ "Kinney Drugs' AI assistant 'Burt' (vendor Synerio, deployed May 2026) garbled medication names, ordered wrong dosages and unneeded refills, and produced duplicate bottles", "Callers had no keypad option; it was talk to the AI or no refill" ], "last_verified_live": "2026-09-26" }, { "id": "wang-2016-jbi-downtime-pathology", "title": "Measuring the effects of computer downtime on hospital pathology processes", "publisher": "Journal of Biomedical Informatics 59:308-315", "url": "https://doi.org/10.1016/j.jbi.2015.12.016", "type": "peer-reviewed", "published": "2016-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "A 17-minute results-system logon failure made clinician read time for potassium and haemoglobin tests 5-6 times longer than control", "Even brief downtimes may affect care" ], "last_verified_live": "2026-09-26" }, { "id": "wcax-2026-kinney-pullback", "title": "Kinney Drugs pulls back AI phone assistant after hundreds of customer complaints", "publisher": "WCAX-TV", "url": "https://www.wcax.com/2026/08/07/kinney-drugs-pulls-back-ai-phone-assistant-after-hundreds-customer-complaints/", "type": "journalism", "published": "2026-08-07", "accessed": "2026-09-26", "tier": 2, "supports": [ "Kinney returned inbound calls to its touch-tone system and made AI refill texts opt-in" ], "last_verified_live": "2026-09-26" }, { "id": "who-2024-lmm-guidance", "title": "Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models (news release)", "publisher": "World Health Organization", "url": "https://www.who.int/news/item/18-01-2024-who-releases-ai-ethics-and-governance-guidance-for-large-multi-modal-models", "type": "guidance", "published": "2024-01-18", "accessed": "2026-09-26", "tier": 1, "supports": [ "LMMs can produce false, inaccurate, biased or incomplete statements that could harm health decisions", "Names automation bias by health professionals as a risk", "Recommends mandatory independent post-release audits with published, disaggregated outcomes" ], "last_verified_live": "2026-09-26" }, { "id": "witty-2024-senate-finance-testimony", "title": "Testimony of Andrew Witty, CEO, UnitedHealth Group, before the Senate Finance Committee: 'Hacking America's Health Care: Assessing the Change Healthcare Cyber Attack and What's Next'", "publisher": "US Senate Committee on Finance", "url": "https://www.finance.senate.gov/imo/media/doc/0501_witty_testimony.pdf", "type": "testimony", "published": "2024-05-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "On 12 Feb 2024 criminals used compromised credentials on a Change Healthcare Citrix remote-access portal that lacked MFA; ransomware deployed 21 Feb, nine days later", "UHG severed connectivity to Change data centers; rebuilt data center network and core services, replacing thousands of laptops", "Pharmacists had to submit claims manually; 99% of pre-incident pharmacies could process claims by 7 Mar; payment processing at ~86% of pre-incident levels by testimony date", "UHG advanced over $6.5 billion in accelerated payments and no-interest loans to providers by 26 Apr 2024; CEO decided to pay a ransom", "Compromised credentials on a Citrix portal without MFA on Feb 12; ransomware Feb 21", "UnitedHealth severed connectivity to Change data centres", "99% of pre-incident pharmacies processing claims by March 7", "More than $6.5 billion advanced to providers by April 26" ], "last_verified_live": "2026-09-26" }, { "id": "wong-2021-jama-im-epic-sepsis", "title": "External Validation of a Widely Implemented Proprietary Sepsis Prediction Model in Hospitalized Patients", "publisher": "JAMA Internal Medicine (Wong A, Otles E, Donnelly JP, et al.)", "url": "https://doi.org/10.1001/jamainternmed.2021.2626", "type": "peer-reviewed", "published": "2021-06-21", "accessed": "2026-09-26", "tier": 1, "supports": [ "Epic Sepsis Model AUC 0.63 at hospitalization level vs developer-reported 0.76-0.83", "Missed 1,709 of 2,552 sepsis cases (67%); sensitivity 33%, PPV 12% at threshold 6", "Alerts fired on 18% of all 38,455 hospitalizations", "ESM AUC 0.63; missed 67% of sepsis; alerted on 18% of hospitalisations; alert fatigue burden" ], "last_verified_live": "2026-09-26" }, { "id": "wong-2021-jama-netw-open-sepsis-covid-alerts", "title": "Quantification of Sepsis Model Alerts in 24 US Hospitals Before and During the COVID-19 Pandemic", "publisher": "JAMA Network Open (Wong A, Cao J, Lyons PG, et al.)", "url": "https://doi.org/10.1001/jamanetworkopen.2021.35286", "type": "peer-reviewed", "published": "2021-11-01", "accessed": "2026-09-26", "tier": 1, "supports": [ "Share of patients generating Epic sepsis alerts per day rose from 9% to 21% within 3 weeks of each hospital's first COVID-19 case", "University of Michigan paused the alerts in April 2020 after nursing reports of overalerting" ], "last_verified_live": "2026-09-26" }, { "id": "wong-2026-jama-netw-open-esm-v2", "title": "Multicenter Prospective Validation of an Updated Proprietary Sepsis Prediction Model", "publisher": "JAMA Network Open (Wong A, Currey D, Schwinne M, et al.)", "url": "https://doi.org/10.1001/jamanetworkopen.2026.0181", "type": "peer-reviewed", "published": "2026-02-02", "accessed": "2026-09-26", "tier": 1, "supports": [ "Epic Sepsis Model v2 encounter-level AUROC 0.82-0.92 across 4 US health systems", "PPV remained low (0.13-0.26) with high alert burden and high variability between institutions" ], "last_verified_live": "2026-09-26" }, { "id": "wright-2016-jamia-cds-malfunctions", "title": "Analysis of clinical decision support system malfunctions: a case series and survey", "publisher": "J Am Med Inform Assoc 23(6):1068-1076", "url": "https://doi.org/10.1093/jamia/ocw005", "type": "peer-reviewed", "published": "2016-03-28", "accessed": "2026-09-26", "tier": 1, "supports": [ "CDS alerts silently stopped firing after code changes, rule edits; an EHR upgrade caused spurious alerts", "93% of responding CMIOs had experienced a CDS malfunction; failures to fire are hard to detect" ], "last_verified_live": "2026-09-26" }, { "id": "wvmetronews-2017-princeton", "title": "Princeton hospital to replace 12-hundred computer hard drives after cyber attack", "publisher": "WV MetroNews (Carrie Hodousek)", "url": "https://wvmetronews.com/2017/06/30/princeton-hospital-to-replace-computer-hard-drives-after-cyber-attack/", "type": "journalism", "published": "2017-06-30", "tier": 2, "supports": [ "EHR inaccessible", "Staff went to all-paper", "Rebuilding network from scratch; ~1,200 hard drives replaced" ], "accessed": "2026-09-27", "last_verified_live": "2026-09-27", "added": "2026-09-27" } ] }