Skip to content
All lessons
Cross-cutting · Patient Safety

Systems-Based Practice & Care Transitions

A board-focused walkthrough of systems-based practice and care transitions: how to classify errors (active vs latent, near-miss, sentinel, never event), match the analysis tool (RCA retrospectively vs FMEA prospectively vs PDSA), and pick the safest next step at handoffs — medication reconciliation, structured I-PASS/SBAR sign-out, forcing functions, and honest error disclosure.

13 min readHigh yield

Systems-based practice treats most medical errors as failures of systems and processes, not individual incompetence — the systems approach (redesign the process) replaces the person approach (blame-and-shame). Reason's Swiss cheese model frames each safeguard (pharmacy check, barcode scan, nurse verification) as a slice with holes; harm reaches the patient only when the holes across successive layers momentarily align, letting an error slip through every defense. Care transitions — admission, transfer, shift-change handoffs, and discharge — are the highest-risk moments because responsibility and information are passed between teams. Per the Joint Commission, communication failures are among the leading root causes of sentinel events. Boards test three moves: correctly classify the error, pick the right analysis tool (retrospective vs prospective), and choose the safest next step at a transition.

Reason's Swiss cheese model: stacked defensive layers each with holes, where an accident trajectory passes through aligned holes to reach a patient
Reason's Swiss cheese model — harm results only when weaknesses (holes) in successive layers of defense line up. · Wikimedia Commons — Davidmack — CC BY-SA 3.0, via Wikimedia Commons
Error taxonomy (know these labels cold)
  • Active error ("sharp end"): occurs at the point of patient contact; effect is immediate (e.g., nurse gives the wrong drug)
  • Latent error ("blunt end"): hidden system/design flaw awaiting a trigger (look-alike vials, understaffing, unsafe EHR defaults)
  • Near miss: an error that did not reach the patient (caught in time or by chance) — still report it; a free warning of a latent flaw
  • Adverse event: harm caused by medical care rather than the underlying disease; may be preventable or not
  • Sentinel event: a safety event reaching the patient with death, permanent harm, or severe temporary harm (or the risk thereof) → triggers mandatory root cause analysis
  • Never event: egregious, largely preventable error (wrong-site surgery, retained foreign body, ABO-incompatible transfusion, wrong-patient procedure)
  • Slip vs lapse: skill-based execution errors — correct plan, wrong action; a slip is an attention failure in the action, a lapse is a memory failure
  • Mistake: knowledge/rule-based planning error — a flawed plan carried out correctly

Which improvement / analysis tool?

MethodTimingCore idea
Root cause analysis (RCA)Retrospective — after a sentinel/adverse eventNon-punitive team asks the "5 Whys" to expose latent system causes
FMEA (Failure Mode & Effects Analysis)Prospective — before launching a processAnticipate how a new process could fail; prioritize high-risk steps
PDSA cycleContinuous, iterativePlan–Do–Study–Act small tests of change, then scale up
LeanOngoingEliminate waste / non–value-added steps
Six Sigma (DMAIC)OngoingReduce process variation and defects
Circular Plan-Do-Check-Act cycle showing iterative continuous quality improvement
The PDSA/PDCA cycle: small, iterative tests of change used for continuous quality improvement. · Wikimedia Commons — Karn-b - Karn Bulsuk (http://www.bulsuk.com). Originally published at http://www.bulsuk.com/2009/02/taking-first-step-with-pdca.html — CC BY 4.0, via Wikimedia Commons
Vignette — sentinel event

Vignette: A patient receives 10× the intended insulin dose after a nurse pulls a look-alike vial from a shared bin, then becomes profoundly hypoglycemic. Two weeks later the hospital convenes a multidisciplinary, non-punitive team that maps the event timeline and repeatedly asks why each step failed.

Process = Root cause analysis (RCA). It is retrospective, aims to uncover latent system errors (look-alike packaging, unsafe storage), and is explicitly not about blaming the individual nurse.

Contrast: analyzing a newly proposed insulin-dispensing workflow to predict failures before rollout would be FMEA (prospective).

Best durable fix = a forcing function: stock only standardized/prefilled concentrations so the wrong dose physically cannot be drawn — stronger than re-educating staff.

Safe care transitions
  • Medication reconciliation at every transition (admission, transfer, discharge): compare the full med list to catch omissions, duplications, dose errors, and interactions — the single most tested transition safeguard
  • Handoffs are the highest-risk communication event — use a standardized, structured tool, not free recall
  • Structured tools: I-PASS for shift/sign-out handoffs, SBAR for urgent verbal escalation; both use closed-loop / read-back so the receiver repeats critical data (verbal orders, critical labs)
  • Discharge bundle: reconciled medication list, a scheduled follow-up appointment, pending/outstanding labs flagged, patient education via teach-back, and a summary sent to the PCP
  • Timely post-discharge follow-up reduces readmissions and closes the loop on pending results
  • Verbal / telephone orders and critical values → write down and read back to confirm
SBAR & I-PASS

Two board-classic handoff frameworks:

SBAR — rapid verbal escalation (e.g., calling about a deteriorating patient):

  • Situation — what is happening now
  • Background — relevant history/context
  • Assessment — what you think is going on
  • Recommendation — what you need or are asking for

I-PASS — structured shift handoff:

  • Illness severity (stable / "watcher" / unstable)
  • Patient summary
  • Action list (to-dos)
  • Situation awareness & contingency planning ("if X, then Y")
  • Synthesis by receiver (read-back to confirm)
Vignette — disclosing an error

Vignette: A resident realizes a heparin infusion ran at 10× the ordered rate overnight; the patient now has painless hematuria and a supratherapeutic aPTT but is hemodynamically stable. After the team stops the infusion and stabilizes the patient, a family member asks what happened.

Next best step: promptly and honestly disclose the error to the patient/family, express empathy and apology, and explain what will be done — regardless of the degree of harm. Then file a non-punitive incident (safety) report so the latent system flaw can be analyzed.

Do NOT: conceal or minimize the error, wait for the patient to ask, or scapegoat an individual. Disclosure is led by the responsible attending/team. Honest disclosure is ethically required and does not hinge on whether the harm turns out to be permanent.

Error prevention hierarchy + Just Culture
  • Hierarchy of effectiveness (strongest → weakest): forcing functions / constraints > automation & standardization > protocols & checklists > double checks / reminders > education & warnings alone. Relying on memory and vigilance is the weakest control
  • Forcing function: design makes the error impossible (incompatible connectors so an enteral line can't join IV tubing; "can't remove the key unless in park")
  • CPOE (computerized order entry) cuts transcription, legibility, and dosing errors; barcode medication administration reinforces the 5 rights (patient, drug, dose, route, time)
  • Checklists & bundles: WHO Surgical Safety Checklist, central-line insertion bundle; the surgical time-out (Universal Protocol) verifies correct patient, site, procedure to prevent wrong-site surgery
  • Just Culture: separates human error (console/support) from at-risk behavior (coach) and reckless behavior (discipline) — accountability with a blame-free reporting climate
  • Non-punitive voluntary reporting maximizes near-miss capture

Practice Patient Safety now

Board-style questions, spaced-repetition flashcards, and a Socratic AI tutor — free to start.