Patient Safety & Quality Improvement
Swiss cheese, RCA vs FMEA, PDSA, never events, and SBAR
How errors happen: a systems view
Active vs latent. Active errors occur at the "sharp end": frontline staff whose actions have immediate effects, such as giving the wrong drug. Latent errors sit at the "blunt end": system or organizational flaws such as poor staffing, look-alike packaging, or bad equipment design. They lie dormant until they enable an active error. Understaffing that sets the stage for a later medication error is a latent error.
Reason's Swiss cheese model. Every layer of defense (prescriber, pharmacist, nurse, technology) has holes. An adverse event occurs only when the holes in all layers momentarily align and a hazard passes straight through. Prevention means adding redundant, independent barriers so that no single hole can cause harm.
| Type | What fails | Example |
|---|---|---|
| Slip | Execution of a correct plan (skill-based, attentional) | A nurse who knows the right drug grabs a look-alike vial |
| Lapse | Execution through a memory failure | Forgetting to give a scheduled dose |
| Mistake | The plan itself (knowledge- or rule-based) | A wrong diagnosis leads to the wrong treatment |
| Violation | A deliberate deviation from a rule | Skipping the surgical time-out to save time |
Just culture. Most errors come from faulty systems, not bad individuals. A just culture pairs a nonpunitive, systems approach with accountability instead of hunting for a "bad apple." Human error is consoled and the system redesigned; at-risk behavior is coached; only reckless behavior is sanctioned. This encourages voluntary, blame-free reporting, which surfaces near misses so the system is fixed before harm occurs. The answer to a tenfold insulin overdose from look-alike vials is to ask why the system allowed two vials to look identical, not to discipline the nurse.
Event taxonomy: near misses to never events
| Event | Definition | Response |
|---|---|---|
| Near miss | Error caught before it reaches the patient (e.g., a pharmacist intercepts an overdose) | Report it; a free lesson in system weakness |
| Adverse event | Harm caused by medical care; preventable if it stems from error | Disclose, report, analyze |
| Sentinel event (Joint Commission) | Patient safety event, not primarily from the natural course of illness, causing death, permanent harm, or severe temporary harm | Immediate root cause analysis |
| Never event | Serious, largely preventable error that should never occur | RCA; CMS does not reimburse the hospital for these conditions |
Classic never events are wrong-site or wrong-patient surgery, a retained foreign body, an ABO-incompatible transfusion, and air embolism. Medicare (CMS) will not reimburse hospitals for these hospital-acquired conditions. Wrong-site surgery is prevented by the Universal Protocol: pre-procedure verification, marking the operative site, and a time-out immediately before the incision.
Analyzing events and testing change
| Feature | Root cause analysis (RCA) | Failure mode and effects analysis (FMEA) |
|---|---|---|
| Timing | Retrospective and reactive: after an adverse or sentinel event | Prospective and proactive: before harm occurs |
| Question | Why did it happen? | What could fail? |
| Method | "5 Whys" and cause-and-effect (fishbone) diagrams trace surface errors to underlying system causes, not individuals | Map each process step, anticipate how it could fail and with what effect, then redesign the highest-risk steps |
| Example | Team traces why a patient died of a chemotherapy overdose | New surgical center designs out wrong-site surgery before opening |
Model for Improvement. It asks three questions: What are we trying to accomplish? How will we know a change is an improvement? What change can we make? It then tests answers with rapid Plan-Do-Study-Act (PDSA) cycles. Small, iterative tests limit risk and speed learning, so they beat large untested rollouts.
- PlanDesign the change and predict the result.
- DoTest it on a small scale (e.g., one unit).
- StudyAnalyze the data against the prediction.
- ActAdopt, adapt, or abandon, then run the next cycle.
A hospital that pilots a discharge checklist on one unit, reviews data weekly, and adjusts before spreading it is running PDSA cycles.
Measuring quality
The IOM's six aims (STEEEP). Quality care is Safe, Timely, Effective, Efficient, Equitable, and Patient-centered.
Donabedian's framework measures quality along three axes:
- Structure, the resources: staffing ratios, equipment, an electronic record
- Process, what is done: the percentage of patients with diabetes who get an annual foot exam, hand-hygiene compliance
- Outcome, the results: mortality, infection rates, HbA1c control
QI projects also track balancing measures, unintended consequences elsewhere in the system, such as readmissions rising after a push to shorten length of stay.
Safety tools: communication, reconciliation, checklists, disclosure
Standardized handoffs. Communication failures are a leading contributor to sentinel events, and handoffs and care transitions are especially high-risk. SBAR (Situation, Background, Assessment, Recommendation) gives a concise shared script that reduces omissions; I-PASS is a similar structured handoff tool. Read back verbal orders and critical results.
Medication reconciliation at every transition (admission, transfer, discharge) catches omissions, duplications, and dosing errors.
Checklists and bundles. Checklists standardize high-risk care. The WHO Surgical Safety Checklist lowered surgical mortality, and the central-line insertion bundle sharply cut central line–associated bloodstream infections (CLABSI; Pronovost).
Strong fixes change the system: forcing functions such as incompatible connectors, standardization, removing look-alike vials, barcode scanning, and computerized order entry. Weak fixes rely on memory: retraining, warning memos, "be more careful."
Disclose and report. Disclose harmful errors to the patient honestly and apologize. Disclosure is an ethical duty, not an admission of legal liability. Separately, report errors and near misses through the institutional safety system so the system itself can improve.