International healthcare experts' consensus on the key requirements of a potential international patient safety learning system: a modified online Delphi study

Through a two-round modified Delphi study involving 21 international healthcare experts, this research establishes a consensus-based framework defining the purposes, key functions, and feasibility of a global Patient Safety Learning System, while generating novel criteria for identifying incidents of international concern to guide future implementation by organizations like the WHO.

The Gist

Imagine the world of healthcare as a massive, global neighborhood. Right now, every country has its own "safety logbook" where doctors and nurses write down when things go wrong (like a medication error or a broken machine). But here's the problem: these logbooks are all written in different languages, use different formats, and no one is reading the whole neighborhood's logs.

If a hospital in Brazil discovers a dangerous flaw in a specific type of IV pump, that information might stay stuck in Brazil. Meanwhile, a hospital in Japan might keep using that same dangerous pump, not knowing the risk exists.

This paper is about a group of 21 global safety experts (doctors, researchers, and leaders) coming together to design a Global Safety Super-App. They used a method called the "Delphi study," which is like a structured, anonymous group chat where experts vote on ideas until they all agree on the best path forward.

Here is the breakdown of what they decided, using simple analogies:

1. The Goal: Why do we need this?

Think of the current system as a bunch of isolated islands. If a storm hits one island, the others don't know until it's too late.
The experts agreed that a Global Patient Safety Learning System (PSLS) should act like a Global Weather Radar.

• It doesn't just record the storm (the accident); it predicts where the next storm might hit.
• It shares the forecast: If a dangerous "weather pattern" (like a contaminated batch of medicine) is found in one country, the radar instantly alerts every other country so they can prepare.
• It helps us learn: Instead of just saying "oops," the system helps countries figure out how to fix the problem so it doesn't happen again.

2. What should go into the Global App?

The experts had to decide: "What kind of bad news is important enough to share with the whole world?"
They created a "Red Flag List."

• High Priority (Share Immediately): If a machine breaks, a drug is contaminated, or a mistake could kill many people across borders, that goes straight to the top of the list. It's like a "Tornado Warning."
• Medium Priority: Things that are new to a specific country or involve common errors (like mixing up IV tubes).
• The "No-Go" Zone: They realized some things are too local or too vague to share globally. For example, a specific cultural misunderstanding in one small village might not need a global alert.

3. How would it actually work? (The Framework)

Imagine a central hub (like a massive airport control tower) that connects all the local safety logbooks.

1. Input: A hospital in France reports a bad reaction to a new vaccine.
2. Analysis: The Global System aggregates this with similar reports from Germany and Canada.
3. Action: The system realizes, "Hey, this isn't just a local glitch; it's a global manufacturing defect!"
4. Output: It sends an alert to the vaccine manufacturer and a "Safety Tip" to every hospital in the world, saying, "Stop using Batch #123 immediately."

4. The Hurdles: Why don't we have this yet?

Even though everyone agrees it's a great idea, building it is like trying to build a bridge between countries that speak different languages and have different laws.

• The "Dictionary" Problem: One country calls a "medication error" something different than another country. The experts said we need a Universal Safety Dictionary (a standard way to name things) before we can start.
• The "Money" Problem: Building and maintaining this global system costs a lot of cash. Who pays? The experts agreed this is the biggest barrier.
• The "Trust" Problem: Doctors are afraid that if they report a mistake, they will get in trouble. The system needs to be a "safe space" where people can admit errors without fear of being fired, or they won't share the data.

5. The Bottom Line

This paper is the blueprint for a global safety net.

• What they achieved: They got 21 top experts to agree on what the system should do, what data to share, and how to handle the challenges.
• What's next: Now, the World Health Organization (WHO) and other groups need to take this blueprint and actually start building the system. But first, they need to solve the money and language issues.

In short: The world is ready to stop reinventing the wheel every time a medical mistake happens. This study is the agreement to finally build a global "Safety Wiki" where we can all learn from each other's mistakes to keep patients safer, no matter where they live.

Technical Summary

1. Problem Statement

Despite the global recognition that patient safety learning can transcend national boundaries, no International Patient Safety Learning System (PSLS) currently exists. While many countries have established national PSLSs, they vary significantly in design, governance, and data classification, leading to a lack of systematic cross-border sharing and learning. Existing international initiatives (e.g., pharmacovigilance networks) address only specific aspects of safety and lack the capacity for routine, broad cross-national learning. There is a critical gap in expert consensus regarding the purpose, key requirements, and feasibility of a unified international system capable of aggregating data, identifying emerging risks, and generating transferable safety interventions.

2. Methodology

The study employed a two-round modified online Delphi technique conducted between June and November 2020.

• Study Design: A modified Delphi approach was used, where a prior systematic literature review and semi-structured key-informant interviews (with experts from healthcare and safety-critical industries like aviation) replaced the traditional first qualitative round.
• Panel Composition: 73 experts were invited; 21 completed Round 1 (29% response rate), and 15 completed Round 2 (71% retention). The panel represented all six continents and included medical doctors, researchers, academics, healthcare leaders, a patient representative, and a medical coder, with a combined 314 years of experience in patient safety.
• Data Collection:
  • Round 1: Panellists rated 62 statements on a 9-point Likert scale (1=strongly disagree, 9=strongly agree).
  • Round 2: Panellists reviewed de-identified group statistics (median, IQR, agreement %) and re-rated non-consensus statements plus 41 new statements generated from qualitative feedback.
• Consensus Criteria:
  • A priori: ≥70% agreement (ratings 7–9) with an Interquartile Range (IQR) ≤2.00.
  • Post-hoc adjustment: To identify the strongest areas of agreement, the threshold was raised to ≥80% agreement for the final analysis.
• Domains: The study focused on four thematic areas: (1) Purposes of an international PSLS, (2) Key functions and features, (3) Patient safety incidents relevant to international sharing, and (4) Enablers and challenges.

3. Key Contributions

• First International Consensus: This is the first study to establish expert consensus on the purposes, functions, and feasibility of a global PSLS.
• Novel Taxonomies: The study generated consensus-based lists of specific patient safety incidents deemed relevant for international sharing and defined criteria for what constitutes a risk of "international concern."
• Operational Framework: It produced a proposed framework for the design and operationalization of an international PSLS, conceptualizing how national systems feed data into a central hub for analysis and feedback.
• Distinction from Reporting: The study clarifies the distinction between a purely incident reporting system and a broader learning system that utilizes diverse data sources (audits, case studies, investigations) beyond voluntary reports.

4. Key Results

Across 103 evaluated statements, the panel reached consensus (≥80% agreement) on 85 statements (83%).

• Purposes (15/19 consensus):
  • Highest agreement (100%) was on identifying R&D priorities, bridging evidence-practice gaps across similar contexts, and sharing affordable design-based interventions.
  • The system was endorsed for both reactive (learning from incidents) and proactive (surveillance for emerging risks) functions.
• Key Functions & Features (17/22 consensus):
  • Top priorities included structured investigation processes for identified risks (100%), informing manufacturers of equipment/medication incidents (95%), and generating transferable safety recommendations (95%).
  • Lower consensus (76%) existed on operational data-sharing mechanics, such as triggering alarms or sharing exemplars for improvement.
• Incidents for International Sharing (11/16 consensus):
  • 100% consensus on sharing incidents involving products/devices as major factors, risks of severe harm/death upon recurrence, and drug/equipment safety incidents.
  • 95% consensus on manufacturing/supply chain issues (e.g., contaminated fluids) and faulty medical devices.
  • Non-consensus items included specific error types like ten-times medication errors and nasogastric tube positioning errors.
• Criteria for International Concern (8/8 consensus - 100%):
  • The panel unanimously agreed that a risk constitutes international concern if it threatens large numbers of individuals across multiple countries, has significant morbidity/mortality impact, or requires action from major stakeholders (e.g., pharma companies).
• Enablers & Challenges:
  • Top Enablers (100%): Standardized international taxonomy and using patient experience to define harm.
  • Top Barriers (100%): Funding difficulties, establishment/maintenance costs, lack of data governance strategy, and the "value proposition" problem across jurisdictions.
  • Cultural Challenge: Significant concern regarding countries being at different maturity levels regarding safety culture (87% agreement).

5. Significance and Implications

• Foundation for Implementation: The consensus-derived framework provides a concrete starting point for organizations like the World Health Organization (WHO) to pilot an international PSLS, aligning with the Global Patient Safety Action Plan 2021–2030.
• Structural Prerequisites: The study highlights that implementation cannot proceed without addressing structural prerequisites, specifically:
  • Adoption of a standardized international taxonomy (e.g., WHO ICPS).
  • Robust data governance frameworks.
  • Mechanisms for equitable participation, particularly for low- and middle-income countries.
• Future Research: The findings necessitate a shift from expert consensus to frontline validation. The next critical step involves engaging end-users (clinicians, nurses) to assess the acceptability and usability of the proposed system, as the current panel was dominated by policymakers and academics.
• Systemic Learning: By aggregating data across borders, such a system could identify rare but catastrophic events (e.g., specific medication errors like vincristine administration) that individual national systems might miss due to low volume or fragmented data.

Conclusion: The study confirms broad international expert agreement on the necessity and design of a global patient safety learning system. While significant barriers regarding funding, governance, and cultural maturity remain, the consensus provides a validated roadmap for moving from concept to feasibility research and potential pilot implementation.