Public Health Sciences

Patient Safety — Never Events and Root Cause Analysis

~6 min read4 sections
⭐ High-yield🎯 Drill Public Health Sciences
Contents (4)

Modern patient safety rests on a systems view: most harm results from predictable failures in system design rather than individual carelessness, so the correct response is almost never to discipline the person at the sharp end.

  • The Swiss cheese model: layered defences each have holes; harm occurs when holes align. Fixing the system means adding or thickening layers, not blaming the individual whose hole was last.
  • Sentinel events are unexpected occurrences involving death or serious harm, requiring immediate investigation. Never events are serious, largely preventable events that should not occur — wrong-site or wrong-patient surgery, retained foreign body, severe medication errors, and others.
  • Root cause analysis is a retrospective, structured, non-punitive review asking why the event happened at each level, aiming at contributing system factors. Failure mode and effects analysis is its prospective counterpart, examining a process for how it could fail before it does.
  • Quality improvement uses the Plan–Do–Study–Act cycle with measurable aims.
  • Error types: active errors at the point of care, latent errors built into the system; slips and lapses (skill-based) versus mistakes (knowledge- or rule-based).
  • A just culture distinguishes human error from at-risk behaviour and from recklessness — supporting the first, coaching the second, and holding the third accountable.

(Seed article — remaining sections to be written and reviewed.)

Event vocabulary (know the ladder)

  • Adverse event: harm caused by medical care rather than the underlying disease. It may or may not be preventable — an anaphylactic reaction to a first-ever penicillin exposure is an adverse event but not an error.
  • Medical error: a failure of a planned action to be completed as intended, or use of a wrong plan. A near miss (close call) is an error that is intercepted or otherwise does not reach the patient. An error that reaches the patient but causes no injury is a no-harm event. Both must be reported, because they map the holes before they align.
  • Sentinel event (The Joint Commission term): a safety event reaching a patient with death, permanent harm, or severe temporary harm requiring intervention to sustain life. It "signals the need for immediate investigation and response."
  • Never event: the lay term for the National Quality Forum Serious Reportable Events list — surgical, product, patient-protection, care-management, environmental, radiologic, and criminal categories. CMS overlaps this via its hospital-acquired condition policy, withholding additional payment for conditions such as retained foreign body, air embolism, ABO-incompatible transfusion, catheter-associated UTI, and stage III/IV pressure ulcers acquired in hospital.

Analytic tools

  • Root cause analysis: retrospective and interdisciplinary; uses the five whys and the fishbone (Ishikawa) diagram sorted by people, equipment, environment, policy, and process. (The Joint Commission now calls the review it requires after a sentinel event a comprehensive systematic analysis, of which RCA — and the RCA² approach — is the prototypical form.) Output is an action plan with owners and metrics, not a name.
  • Failure mode and effects analysis: prospective; each failure mode gets a risk priority number = severity × probability of occurrence × probability of non-detection, and the highest scores are redesigned first.
  • Donabedian framework: quality is measured as structure (ICU nurse ratio), process (percent receiving DVT prophylaxis), and outcome (VTE rate). Quality improvement adds a balancing measure to catch harm from the fix itself.
  • Hierarchy of effectiveness: forcing functions and constraints (removing concentrated potassium from floor stock) beat automation and standardization, which beat checklists and reminders, which beat education and "be more careful" — the weakest and most common wrong answer.

The stem: A patient scheduled for a left inguinal hernia repair undergoes a right-sided repair. The site was never marked, the surgeon arrived late, and the time-out was performed while the attending was still scrubbing. A junior resident who noticed the discrepancy on the consent did not speak up. The question asks for the most appropriate next step.

Work it in order

  • Stabilize and treat the patient first. Immediate clinical needs precede any administrative action.
  • Disclose the error to the patient, promptly, factually, with an apology, by the responsible attending. Disclosure is expected in US practice and is not contingent on legal review; "consult risk management before telling the patient" and "tell the family only if they ask" are distractors.
  • File a report in the institutional safety event reporting system, which is non-punitive and separate from the medical record.
  • Convene a root cause analysis within the timeframe required by The Joint Commission sentinel event policy. This is the usual correct answer when the stem describes a completed serious harm event and asks what the hospital should do.

What the RCA should surface: the active error is operating on the wrong side; the latent errors are absent site marking, a time-out performed without the full team present, production pressure from a delayed start, and a hierarchy that silenced the resident. Under just culture, proceeding without a proper time-out is at-risk behaviour normalized by repetition — coached and re-engineered, not punished.

Corrective action and testing: hard-stop enforcement of the Universal Protocol (site marking with the patient awake, time-out with every team member stopped and participating), plus AHRQ TeamSTEPPS tools such as the two-challenge rule and CUS language so juniors can escalate. Test the change on one operating room through PDSA cycles, tracking time-out completeness as a process measure and wrong-site events as the outcome measure.

  • Disclose promptly and personally — and then document the disclosure conversation in the chart (the incident report stays separate). When a stem describes harm from an error, the single best next step after stabilizing the patient is honest disclosure by the responsible physician. Concealing an error, or telling a colleague to disclose it, is always wrong.
  • Report the near miss. Errors intercepted before reaching the patient still go into the non-punitive reporting system — they are the cheapest data on where the Swiss cheese holes are.
  • Retrospective = RCA, prospective = FMEA. The classic one-line discriminator. "After a wrong-drug death" → RCA; "before rolling out a new insulin pump" → FMEA.
  • Two patient identifiers, never the room number. A Joint Commission National Patient Safety Goal, and a favourite distractor when the stem describes a transfusion or specimen mislabelling event.
  • Wrong-site surgery is prevented by the Universal Protocol: pre-procedure verification, site marking, and a time-out. A skipped or rushed time-out is the at-risk behaviour at the sharp end; the latent failure is the system that allowed the case to proceed without a hard stop.
  • Systems fixes beat education. Forcing functions (incompatible connectors for enteral versus IV tubing, removal of concentrated electrolytes from ward stock, computerized order entry with dose limits) outrank "re-educate the nursing staff" or "add a policy reminder," which are the perennially attractive wrong answers.
  • Just culture is not a no-accountability culture. Human error → console and redesign; at-risk behaviour → coach; reckless disregard → disciplinary action. Only the third justifies punishment.
  • Know who owns which list. The Joint Commission defines sentinel events and requires the comprehensive systematic analysis; the National Quality Forum maintains the Serious Reportable Events ("never events") list; CMS withholds incremental payment for selected hospital-acquired conditions. Do not attribute the never-event list to CMS or the sentinel event policy to AHRQ.

Related topics

← Back to library