Patient Safety & Systems-Based Practice

Healthcare quality measures: process, outcome, structure

Clinical Overview and When to Suspect Quality Gaps

— Structure: the fixed attributes of the system delivering care (staffing ratios, EHR availability, ICU beds, board certification, accreditation status)

— Process: what clinicians actually do to or for patients (aspirin within 24h of MI, A1c checked q3 months, DVT prophylaxis ordered)

— Outcome: what happens to the patient (mortality, readmission, HbA1c <7%, patient-reported function, HCAHPS satisfaction)

— Variation in care exists between clinicians/units for the same condition

— Outcomes lag national benchmarks (CMS Hospital Compare, NCQA HEDIS)

— Sentinel events, near-misses, or readmissions cluster

— Disparities exist across race, language, insurance, or geography

Board pearl: When the stem asks "which measure best reflects quality of care delivered?" the answer is usually a process measure — they are most actionable, most attributable to clinicians, and least confounded by patient factors. When it asks "what matters most to the patient?" the answer shifts to outcome measures. Structure measures are correct when the stem emphasizes infrastructure, certification, or staffing — they are necessary but rarely sufficient for quality.

Donabedian framework is the foundational model for healthcare quality, dividing measures into three interrelated domains: structure, process, and outcome

Step 3 expects you to classify a measure when given a clinical scenario and to know which type best drives improvement in a given context

Suspect a quality gap when:

Quality measures feed into value-based payment (MIPS, MACRA, HRRP, HAC Reduction Program, Star Ratings) — bad metrics translate directly to financial penalties and public reporting

Balancing measures (a 4th, often-missed category) track unintended consequences — e.g., aggressive sepsis bundles increasing C. difficile or fluid overload

Presentation Patterns and Key History — How Measures Show Up on Step 3

— Percentage of patients receiving a guideline-recommended action (statin at discharge post-MI, mammogram q2y in women 50–74, pneumococcal vaccine in COPD)

— Time-based interventions (door-to-balloon <90 min, antibiotics within 1h of severe sepsis, blood cultures before antibiotics)

— Documentation completeness (medication reconciliation at admission/discharge, smoking cessation counseling documented)

— 30-day mortality after AMI, CHF, pneumonia, CABG

— 30-day all-cause readmission (HRRP-tracked conditions: AMI, HF, PNA, COPD, CABG, elective hip/knee)

— Hospital-acquired conditions: CLABSI, CAUTI, SSI, CDI, pressure ulcers stage III/IV, falls with injury

— Patient-reported outcomes (PROMs): functional status, pain, HRQoL

— HCAHPS patient experience scores

— Nurse-to-patient ratios, intensivist staffing of ICU, 24/7 in-house surgeon

— EHR with CPOE and clinical decision support

— Magnet nursing designation, Joint Commission accreditation, trauma center level

— Board certification rates, residency program affiliation

Key distinction: Process measures answer "Did we do the right thing?" Outcome measures answer "Did the patient do well?" Structure measures answer "Do we have what we need to do the right thing?" A hospital can have great structure and process but poor outcomes if case mix is sicker — which is why risk adjustment is essential for outcome comparisons.

Stems will not say "this is a process measure" — they describe a scenario and you classify it

Process measure stems typically describe:

Outcome measure stems describe:

Structure measure stems describe:

Key history clues: words like "the hospital tracks," "CMS reports," "the QI committee is reviewing" should prompt you to ask: is this an action (process), a result (outcome), or a resource (structure)?

Physical Exam Findings — Auditing the System, Not the Patient

— Numerator: patients who received the action / had the outcome

— Denominator: eligible population (with appropriate exclusions — e.g., contraindications, hospice, patient refusal)

— Risk adjustment: required for outcome measures to compare across institutions fairly (case-mix index, comorbidity scores)

— Attribution: which clinician or system "owns" the measure? Often contentious for outcomes

Board pearl: Special cause variation demands investigation of a specific event or person (e.g., a new nurse, a broken machine). Common cause variation demands system redesign (the process itself is unreliable). Confusing the two leads to blame culture and missed opportunities — a classic Step 3 patient-safety trap.

In QI, the "exam" is the chart audit, dashboard review, and run chart — Step 3 expects familiarity with these tools

Run chart: data plotted over time; look for shifts (≥6 consecutive points on one side of median), trends (≥5 ascending/descending), or astronomical points

Control chart (Shewhart chart): adds upper/lower control limits (±3 SD); points outside limits = special cause variation (investigate); points within = common cause variation (system-level redesign needed)

Pareto chart: bar graph ranking causes by frequency — applies the 80/20 rule to prioritize the vital few drivers

Fishbone (Ishikawa) diagram: categorizes root causes (People, Process, Equipment, Environment, Materials, Management)

Driver diagram: links aim → primary drivers → secondary drivers → interventions

Five Whys: iterative questioning to reach root cause of a defect or sentinel event

PDSA cycle (Plan-Do-Study-Act): rapid small-scale tests of change before full implementation

When evaluating a specific measure:

Reliability (consistent across raters/time) and validity (measures what it claims) must both be established before a metric is adopted

Diagnostic Workup — Choosing the Right Measure Type

— Evidence strongly links the process to outcomes (e.g., aspirin in AMI → mortality)

— You need rapid feedback (process events occur often; outcomes are rare)

— You want to hold clinicians accountable for actions within their control

— Sample sizes are small (process gives more data points than rare outcomes)

— Multiple processes contribute and you want the "bottom line"

— Patient-centeredness is the goal (PROMs, HCAHPS)

— Stakes are high enough to justify risk adjustment costs

— Caveats: confounded by case mix, social determinants, and chance — require risk adjustment and adequate sample size

— You're assessing readiness or capacity (new program launch, accreditation)

— Process/outcome data not yet available

— Caveat: structure alone does not guarantee quality

— AMI: Process = door-to-balloon <90 min, aspirin/statin/beta-blocker at discharge; Outcome = 30-day mortality, readmission; Structure = 24/7 cath lab capability

— Diabetes (HEDIS): Process = annual A1c, retinal exam, nephropathy screen; Outcome = A1c <8%, BP <140/90; Structure = certified diabetes educator on staff

— Sepsis: Process = SEP-1 bundle compliance; Outcome = sepsis mortality; Structure = rapid response team

Step 3 management: When designing a QI project, always pair a process measure with an outcome measure and a balancing measure. Process tells you if you're doing the work; outcome tells you if it matters; balancing tells you what you broke while improving. Missing any leg = incomplete project and a likely wrong answer choice.

First step: define the QI aim using SMART criteria (Specific, Measurable, Achievable, Relevant, Time-bound)

Then select measures across all three Donabedian domains plus a balancing measure

Process measures — best when:

Outcome measures — best when:

Structure measures — best when:

Examples by condition:

Advanced Concepts — Measurement Science and Risk Adjustment

— Statistical models (often logistic regression or hierarchical models) account for age, comorbidities, severity at presentation, sometimes socioeconomic factors

— CMS uses hierarchical condition categories (HCCs) and condition-specific models

— Without it, hospitals caring for sicker/poorer patients appear "worse" — a major equity concern

— O/E >1 = worse than expected; O/E <1 = better than expected

— Used in mortality, readmission, and complication reporting

— HCAHPS domains: communication with doctors/nurses, responsiveness, pain management (removed), discharge info, cleanliness, quietness, overall rating, would recommend

— CMS: Hospital Compare, HRRP, HAC Reduction, VBP

— NCQA: HEDIS (ambulatory/health plan focus)

— Joint Commission: ORYX core measures, accreditation

— AHRQ: PSIs (Patient Safety Indicators), IQIs

— Leapfrog: hospital safety grades

Board pearl: A hospital with high crude mortality but O/E <1 is actually performing better than expected for its case mix. Step 3 stems use this to test whether you understand that raw outcomes mislead without risk adjustment — a common trap in stems comparing safety-net to suburban hospitals.

Risk adjustment is mandatory for fair outcome comparison:

O/E ratio (observed/expected): a risk-adjusted outcome metric

Sample size and statistical power: rare outcomes (e.g., perioperative death) require large denominators or long observation windows

Composite measures: bundle several related processes (e.g., perfect care for AMI = aspirin + beta-blocker + ACEI + statin + smoking counseling) — penalize all-or-nothing failure but may obscure individual gaps

Patient-reported outcome measures (PROMs) and experience measures (PREMs): increasingly weighted in value-based care

National measure stewards:

Never events (NQF Serious Reportable Events): wrong-site surgery, retained foreign object, ABO-incompatible transfusion, infant abduction — non-reimbursable by CMS and publicly reportable

Risk Stratification — Prioritizing QI Targets

— Impact: How many patients affected? How severe?

— Feasibility: Resources, time, political will, data availability

— Alignment: Does it match organizational strategy, CMS priorities, payer contracts?

— Evidence: Is the intervention proven (e.g., bundle compliance) vs. exploratory?

— HRRP conditions (AMI, HF, PNA, COPD, CABG, hip/knee) — direct Medicare penalty

— HAC Reduction Program: bottom-quartile hospitals lose 1% of Medicare payments

— Core measures with public reporting

— Health equity gaps (CMS now stratifies measures by dual-eligibility, race)

— Early/diagnostic phase: structure + baseline outcome to define the problem

— Intervention phase: process measures for rapid feedback

— Sustainment phase: outcome measures with continuous monitoring

— Throughout: balancing measures

— Preoccupation with failure

— Reluctance to simplify interpretations

— Sensitivity to operations

— Commitment to resilience

— Deference to expertise (not hierarchy)

Step 3 management: When a QI committee asks "where do we start?" the answer hierarchy is: (1) any never event or sentinel event — immediate RCA; (2) publicly reported/penalized metrics with current poor performance; (3) high-volume, high-variation processes; (4) frontline-identified pain points. Starting with what's easy but low-impact is the wrong answer; starting with what's flashy but unmeasurable is also wrong.

Not all gaps deserve equal attention. Prioritize using:

High-priority targets typically include:

Triple Aim (IHI): improve population health + improve patient experience + reduce per-capita cost

Quadruple Aim: adds clinician well-being (recognizing burnout as a quality threat)

Quintuple Aim: adds health equity

Choosing the right measure type for the stage of improvement:

High-reliability organization (HRO) principles:

Pharmacotherapy of Improvement — Core QI Methodologies

— Plan: define question, prediction, data plan

— Do: small-scale test (one nurse, one shift, one patient)

— Study: compare data to prediction

— Act: adopt, adapt, or abandon → next cycle

— Key: start small, ramp up; multiple cycles > one big rollout

— Eliminate waste (muda): overproduction, waiting, transport, overprocessing, inventory, motion, defects, underutilized talent

— Value stream mapping; 5S (Sort, Set, Shine, Standardize, Sustain)

— Gemba walks: leaders observe work where it happens

— Statistical, defect-reduction focus (target: 3.4 defects per million)

— Best for high-volume, well-defined processes

— What are we trying to accomplish? (aim)

— How will we know a change is an improvement? (measures)

— What changes can we make that will result in improvement? (ideas)

— Examples: central line bundle, ventilator bundle, sepsis bundle

— Bundle compliance is a process measure; reduced CLABSI/VAP is the outcome

Board pearl: PDSA cycles must be small and fast — testing a new discharge checklist on one patient on one unit tomorrow is the right Step 3 answer, not "convene a committee and pilot for 6 months." Rapid-cycle testing with iterative refinement beats big-bang implementation nearly every time on board questions.

PDSA (Plan-Do-Study-Act) — the workhorse of iterative improvement:

Lean (Toyota Production System adapted):

Six Sigma (DMAIC): Define-Measure-Analyze-Improve-Control

Lean Six Sigma: hybrid combining waste reduction with statistical rigor

Model for Improvement (IHI): three questions + PDSA

Change management: Kotter's 8 steps, ADKAR; physician engagement is the rate-limiting step

Bundles (IHI): small sets of 3–5 evidence-based interventions performed together, all-or-nothing

Procedures of Quality — RCA, FMEA, and Event Analysis

— Retrospective, triggered by a sentinel event, never event, or serious safety event

— Multidisciplinary team, blameless, focuses on system failures (Swiss cheese model — Reason)

— Tools: Five Whys, fishbone, timeline reconstruction

— Output: action plan with assigned owners, deadlines, and strong actions (forcing functions, automation) preferred over weak actions (education, policy)

— Joint Commission requires RCA within 45 days of sentinel events

— Prospective — done before implementing a new process to anticipate failures

— Each step assessed for: severity, probability, detectability → Risk Priority Number (RPN)

— High RPN = redesign before go-live

— Required by Joint Commission at least annually on a high-risk process

— Strongest: architectural/equipment changes, forcing functions, automation, simplification

— Intermediate: checklists, redundancies, enhanced communication

— Weakest: training, policies, double-checks by humans

CCS pearl: After a wrong-site surgery, the correct Step 3 action sequence is: (1) disclose to patient/family transparently and apologize; (2) stabilize and treat the patient; (3) report to risk management and regulatory bodies; (4) convene RCA (blameless, system-focused); (5) implement strong corrective actions (e.g., hard-stop time-out in EHR). Disciplining the individual surgeon is rarely the right first answer.

Root Cause Analysis (RCA):

Failure Mode and Effects Analysis (FMEA):

Action hierarchy (strong → weak):

Sentinel event (Joint Commission): unexpected occurrence involving death or serious physical/psychological injury, or risk thereof — requires RCA

Never event (NQF): subset of clearly preventable, serious events

Adverse event: harm caused by medical care (not underlying disease)

Near miss / close call: error reached patient but no harm (or didn't reach patient) — reporting these is highest-yield for system learning

Just Culture: distinguishes human error (console), at-risk behavior (coach), reckless behavior (discipline) — replaces punitive culture

Special Populations — Ambulatory and Health Plan Measures (HEDIS)

— Diabetes care: A1c testing, A1c control <8%, retinal exam, nephropathy screening, BP control, statin therapy

— Cardiovascular: statin therapy in patients with CVD or diabetes age 40–75, BP control <140/90

— Cancer screening: breast (mammogram 50–74 q2y per USPSTF; HEDIS 50–74), cervical (Pap or HPV per ages), colorectal (45–75)

— Immunizations: childhood combo, adolescent, influenza, pneumococcal, Tdap

— Behavioral health: antidepressant medication management, follow-up after hospitalization for mental illness (7-day and 30-day), SUD treatment initiation/engagement

— Medication management: statin adherence, RAS antagonist adherence, oral diabetes med adherence (PDC ≥80%)

— Care coordination: transitions of care, medication reconciliation post-discharge

— Categories: Quality, Cost, Improvement Activities, Promoting Interoperability

— Composite score → bonus or penalty on Medicare Part B payments

Board pearl: When the stem describes a primary care office tracking "percentage of diabetic patients with A1c checked in the last 6 months" — that is a HEDIS process measure, and the correct intervention is usually registry-based outreach (population health), not chart-by-chart review. Panel management + automated reminders is the Step 3 answer to "how do you improve this metric?"

HEDIS (Healthcare Effectiveness Data and Information Set) — NCQA's measure set used by >90% of US health plans

Used for plan accreditation, Medicare Advantage Star Ratings, and pay-for-performance

Predominantly process measures because ambulatory outcomes are slow and confounded

High-yield HEDIS measures Step 3 expects you to recognize:

Star Ratings (CMS Medicare Advantage): 1–5 stars; plans ≥4 stars get bonus payments

MIPS (Merit-based Incentive Payment System): for clinicians under MACRA

Special Populations — Pediatrics, Maternity, and Equity Measures

— Well-child visits in first 30 months, adolescent well-care

— Childhood immunization status (Combo 10)

— Developmental screening at 12/24/36 months

— BMI percentile documentation, counseling for nutrition/activity

— ADHD medication follow-up (initiation phase visit within 30 days)

— Asthma medication ratio (controller:total) — surrogate for appropriate controller use

— Early elective delivery <39 weeks (process — should be low/zero; Joint Commission PC-01)

— Cesarean delivery rate in low-risk first births (NTSV C-section) — outcome/process hybrid

— Exclusive breastfeeding at discharge

— Severe maternal morbidity (SMM) — outcome

— Postpartum depression screening

— PHQ-9 screening and follow-up

— Collaborative care model uptake (structure)

— Stratification of all measures by race/ethnicity, language, disability, dual-eligibility, rurality

— Screening for social determinants of health (SDOH): food insecurity, housing, transportation, utilities, interpersonal safety — now a CMS measure

— Health Equity Index in Medicare Advantage Star Ratings (2027)

— Discharge summary to PCP within 48h

— Medication reconciliation at discharge

— Follow-up appointment scheduled before discharge (especially for HF: within 7 days)

— Teach-back method for patient education

Step 3 management: For a postpartum patient or recently discharged HF patient, the most impactful process measures are medication reconciliation + early follow-up (within 7 days) + warm handoff to outpatient team. These three together drive the outcome measure of 30-day readmission — a frequent CCS and MCQ pivot point.

Pediatric quality measures (CMS Child Core Set):

Maternity care:

Behavioral health integration:

Health equity measures (CMS focus area):

Transitions of care (high-yield Step 3):

Complications of Measurement — Unintended Consequences

— Clinicians optimize what's measured at the expense of what isn't

— e.g., aggressive sepsis bundle compliance → fluid overload in HF/ESRD patients (a balancing measure issue)

— Upcoding comorbidities to improve risk adjustment

— Diagnostic relabeling (e.g., "encephalopathy" vs. "altered mental status")

— Discharge against medical advice or transfer to avoid mortality attribution

— Cherry-picking healthier patients ("lemon dropping" vs. "cream skimming")

— Safety-net hospitals serving sicker, poorer patients penalized by HRRP

— Led to peer-grouping in HRRP (2019) — hospitals compared within dual-eligible quintile

— Track balancing measures alongside primary measures

— Audit data integrity periodically

— Listen to frontline staff and patients for unintended effects

Key distinction: A measure with high reliability and validity at the population level can still cause individual harm if applied rigidly (e.g., perioperative beta-blocker for everyone → POISE trial showed increased stroke/death). Step 3 expects you to balance guideline-driven measures with individualized clinical judgment, especially in the elderly and complex patients.

Measures change behavior — sometimes in unwanted ways. Recognize these distortions:

Teaching to the test / measure fixation:

Gaming:

Tunnel vision: ignoring non-measured but important care domains (compassion, dignity, complex undiagnosed symptoms)

Numerator/denominator manipulation: excluding patients liberally to improve ratios

Goodhart's Law: "When a measure becomes a target, it ceases to be a good measure"

Disparities amplified by poor risk adjustment:

Provider burnout from documentation burden and metric fatigue → Quadruple Aim recognition

Patient harm from overuse: aggressive A1c targets in frail elderly → hypoglycemia; aggressive BP targets → falls, AKI

Measure obsolescence: e.g., pain as the "5th vital sign" contributed to opioid crisis

Detection:

When to Escalate — Sentinel Events, RCA Triggers, and Regulatory Reporting

— Sentinel event (Joint Commission): wrong-site/wrong-patient/wrong-procedure surgery, unanticipated death of full-term infant, suicide of patient receiving care, severe maternal morbidity, retained surgical item, fall with death/permanent harm, fire, etc. → RCA within 45 days

— Never events (NQF SREs): subset, always reportable

— Hospital-acquired conditions (HACs): CLABSI, CAUTI, SSI, CDI, falls, pressure injuries → not reimbursable, publicly reported

— Reportable diseases: state-mandated lists (TB, STIs, certain foodborne, COVID, etc.) → public health

— Mandatory reporting: child abuse, elder abuse, dependent adult abuse, certain wounds (gunshot/stab in most states), impaired drivers (varies), impaired colleagues to state medical board

— Risk management / patient safety officer: any adverse event with potential litigation or regulatory exposure

— Ethics committee: value-laden conflicts (capacity, surrogate disputes, futility)

— Compliance: HIPAA breaches, billing concerns

— Quality department: pattern of near-misses, measure performance drops

— CMO / leadership: systemic safety threats

— Internal incident reporting (e.g., RL Solutions, Datix) — should be non-punitive, voluntary, easy

— Patient Safety Organizations (PSOs) — federally protected, confidential aggregate analysis

— FDA MedWatch — devices and drug adverse events

— ISMP — medication errors

CCS pearl: When a near-miss occurs (e.g., pharmacist catches a 10x insulin overdose order before administration), the correct CCS actions are: (1) document and file an incident report even though no harm occurred, (2) thank the pharmacist (reinforce reporting culture), (3) review whether system fix is needed (e.g., max-dose alert in CPOE). Doing nothing because "no harm" was done is wrong — near-misses are the richest source of system learning.

Mandatory escalation triggers:

Who to involve:

Reporting systems:

Key Differentials — Distinguishing Measure Types in Stems

— Nurse-to-patient ratio of 1:4 on medical floor

— Hospital is a Level I trauma center

— ICU staffed by board-certified intensivists 24/7

— EHR with CPOE and clinical decision support

— Hospital is Joint Commission accredited / Magnet-designated

— Pharmacy has 24/7 in-house pharmacist

— Stroke center certification

— Hospitalist program in place

— % MI patients receiving aspirin within 24h

— Door-to-balloon time <90 min

— % surgical patients receiving prophylactic antibiotics within 1h of incision

— VTE prophylaxis ordered within 24h of admission

— Hand hygiene compliance rate

— Time to antibiotics for severe sepsis <1h

— % diabetic patients with A1c checked in past year

— Medication reconciliation completed at discharge

— Smoking cessation counseling documented

— 30-day all-cause mortality after AMI

— 30-day readmission after HF discharge

— CLABSI rate per 1000 line-days

— Surgical site infection rate

— In-hospital fall rate per 1000 patient-days

— Pressure ulcer incidence

— Patient satisfaction (HCAHPS) scores

— Functional status improvement after hip replacement (PROM)

— Stroke disability at 90 days (mRS)

— Hypoglycemia rate while tightening A1c targets

— Hypotension rate while pushing rapid sepsis fluid bundles

— Opioid overdose rate after implementing pain-as-vital-sign

Board pearl: If the measure describes an action by clinicians/staff → process. If it describes what happened to the patient (health state) → outcome. If it describes the system/resources/setting itself → structure. Patient satisfaction is outcome (it's a patient-reported result), not process — a common tripping point.

Classify these examples (a frequent Step 3 task):

Structure measures:

Process measures:

Outcome measures:

Balancing measures:

Key Differentials — Quality vs. Safety vs. Value vs. Equity

— Safe: avoid harm from care

— Timely: reduce waits and delays

— Effective: evidence-based care to those who benefit, withholding from those who won't

— Efficient: avoid waste

— Equitable: care doesn't vary by personal characteristics

— Patient-centered: respectful and responsive to patient preferences

— Adverse events, medical errors, HACs, never events

— Tools: RCA, FMEA, just culture, HRO principles

— Improving value means raising quality, lowering cost, or both

— Value-based purchasing (VBP), bundled payments, ACOs, capitation are payment models aligning incentives with value

— Choosing Wisely: campaigns to reduce low-value care (unnecessary imaging, antibiotics, labs)

— Care quality and outcomes shouldn't vary by race, ethnicity, language, gender, sexual orientation, disability, geography, or SES

— Disparities: differences linked to social/economic/environmental disadvantage

— Requires data stratification, SDOH screening, and targeted interventions (community health workers, language access, transportation support)

— Health outcomes of a group, including distribution within the group

— Triple → Quadruple → Quintuple Aim progression

Key distinction: Reducing unnecessary CT scans for low-back pain improves value (lower cost, same/better quality) and may improve safety (less radiation). Increasing CT availability improves access. Doing CT only when guideline-indicated improves effectiveness (an IOM aim). Step 3 stems often layer these — pick the domain that matches the primary intent stated in the question.

These overlap but Step 3 expects you to distinguish:

Quality (IOM 6 aims — STEEEP):

Patient safety: subset of quality focused on preventing harm from care itself

Value = Quality / Cost

Health equity:

Population health:

Public health vs. clinical quality: public health acts on populations (vaccination campaigns, water fluoridation); clinical quality acts on individual encounters and panels

Long-Term Plan — Sustaining Improvement and Spread

— Embedding into workflow: order sets, default options, EHR templates, automated reminders (forcing functions > reminders > education)

— Ongoing measurement: dashboards visible to frontline; control charts to detect drift

— Accountability structure: named owner, regular review cadence (monthly/quarterly)

— Engaged leadership: executive sponsor, visible commitment

— Frontline ownership: those doing the work co-design the change

— Linkage to incentives: alignment with MIPS/VBP/contracts where possible

— Continuous PDSA: small adjustments as context changes

— Move from one unit/clinic to system-wide

— Adapt — don't simply transplant — to local context

— Use change agents and early adopters (Rogers' diffusion of innovations)

— Document the change package: aim, measures, key interventions, lessons learned

— Mix of structure (occasional review), process (continuous), outcome (continuous with risk adjustment), balancing (continuous)

— Retire measures that "top out" (>95% compliance with low variation) and add new gaps

— Re-baseline annually

Step 3 management: When a successful pilot QI project is asked to be sustained, the strongest action is to embed the change into the EHR as a default order or hard stop (e.g., default DVT prophylaxis order in admission set). Education and reminders alone fade within months — this is a high-yield distractor trap.

Improvement that isn't sustained reverts within 6–12 months. Sustainability requires:

Spread (scaling up):

Long-term measure portfolio:

Learning health system: continuously generates and applies evidence; data → knowledge → practice change → data

Continuous quality improvement (CQI) / Total Quality Management (TQM): organization-wide commitment, not a project

Public reporting (Hospital Compare, Care Compare) creates external pressure for sustainment

Patient and family advisory councils (PFACs): integrate patient voice into governance

Follow-Up and Monitoring — Dashboards and Continuous Surveillance

— Process measures: real-time or daily/weekly dashboards (high event frequency)

— Outcome measures: monthly/quarterly (lower frequency, need risk adjustment)

— Structure measures: annual or upon change (staffing, accreditation)

— Balancing measures: same cadence as primary measure

— Run charts: simplest, plot data over time; look for non-random patterns

— Statistical Process Control (SPC) / control charts: distinguish signal from noise

— Funnel plots: compare units/hospitals adjusting for sample size

— Benchmarking: internal trend + external comparison (state, national, peer)

— Dashboards: aggregate multiple measures; ideally tier (frontline, unit, executive)

— Most effective when: baseline performance low, feedback frequent, from a respected source, includes specific targets and action plan, written + verbal

— Individual clinician-level feedback (with peer comparison) outperforms aggregate

— "Nudges" via EHR (default options, opt-out structures) augment feedback

— HCAHPS surveys post-discharge (random sample, 48h–6 weeks post-discharge, telephone/mail/electronic)

— CG-CAHPS for ambulatory clinician/group experience

— Real-time feedback (kiosks, post-visit texts) supplements but doesn't replace standardized surveys

— Mini-Z, Maslach Burnout Inventory

— Burnout → patient safety risk; track as system health indicator

Board pearl: When a stem asks how to convince a skeptical physician that they need to change practice, the best answer is individualized audit and feedback with peer comparison, ideally delivered by a respected peer or department chair, with specific patient cases. Generic education lectures and policy memos rank far lower in evidence for behavior change.

Measurement cadence by type:

Tools for ongoing monitoring:

Audit and feedback (Cochrane-supported intervention):

Patient and family follow-up:

Clinician well-being monitoring:

Ethical, Legal, and Patient Safety Considerations

— Ethical and (increasingly) legal duty to disclose adverse events causing harm

— Many states have apology laws protecting expressions of sympathy from use as admission of liability

— Communication and Resolution Programs (CRPs) (e.g., Michigan Model): proactive disclosure, investigation, apology, fair compensation — reduce litigation and improve safety culture

— Disclose: what happened, what's being done for the patient, what's being done to prevent recurrence, an apology

— Peer review protected in most states (cannot be used in malpractice discovery)

— Patient Safety Organizations (PSOs) provide federal protection for shared data

— RCAs typically protected — but disclosure of facts to patient is still required

— Distinguish human error (console + system fix), at-risk behavior (coach), reckless behavior (disciplinary action)

— Punishing human error suppresses reporting and harms safety

— Reckless behavior (e.g., diverting opioids, refusing time-outs) requires accountability

— Risk adjustment manipulation, denominator gaming

— Financial incentives (P4P) can distort behavior — design carefully

— Most QI doesn't require IRB review (vs. research) — but the line is sometimes blurry

— When intervention deviates from standard care and is generalizable knowledge generation → likely research → IRB + consent

— Quality improvement activities still must respect patient autonomy

— Discharge med rec errors, missed test follow-ups, delayed PCP appointments

— "Test results pending at discharge" — explicit handoff is a safety requirement

Step 3 management: A resident discovers a CT result with a suspicious lung nodule reported after the patient was discharged. The correct action sequence: (1) contact the patient directly to inform them, (2) arrange follow-up (pulm referral, repeat imaging per Fleischner), (3) notify the PCP, (4) document the communication, (5) report to the QI/safety system so a process to prevent missed post-discharge results is reviewed. Waiting for the attending or assuming the PCP will see it is wrong — closing the loop is the clinician's duty.

Disclosure of medical errors:

Confidentiality of QI/peer review data:

Just Culture:

Conflicts of interest in measurement:

Equity and informed consent in QI:

Transitions of care = highest-risk safety zone:

High-Yield Associations and Rapid-Fire Facts

Board pearl: When a Step 3 stem describes a verbal telephone order or critical lab result, the required safety practice is read-back/repeat-back (closed-loop communication) — this is an NPSG and the highest-yield answer for "what would have prevented this error?" Standardized handoffs (SBAR, I-PASS) are the answer for transition-of-care error stems.

Donabedian = structure / process / outcome (1966)

IOM "To Err Is Human" (1999): 44,000–98,000 deaths/yr from medical errors → launched modern patient safety movement

IOM "Crossing the Quality Chasm" (2001): 6 aims (STEEEP)

Triple Aim (Berwick/IHI 2008): population health, experience, cost → Quadruple (clinician well-being) → Quintuple (equity)

HRRP (Hospital Readmissions Reduction Program, 2012): penalizes up to 3% Medicare payment for excess 30-day readmissions; conditions: AMI, HF, PNA, COPD, CABG, elective hip/knee

HAC Reduction Program: bottom-quartile hospitals lose 1% Medicare payment

Value-Based Purchasing (VBP): ±2% redistribution based on quality

MACRA (2015) → MIPS + APMs: replaced SGR; clinician-level value payment

NQF (National Quality Forum): endorses measures; stewards Serious Reportable Events list

AHRQ PSIs: Patient Safety Indicators derived from admin data

Leapfrog Hospital Safety Grade: A–F public rating

NCQA HEDIS: ambulatory measure set; >90% plans report

Joint Commission National Patient Safety Goals (NPSGs): annual list — identify patients with 2 identifiers, improve staff communication (read-back), medication safety, hand hygiene, prevent wrong-site surgery, suicide risk assessment

Universal Protocol: pre-procedure verification, site marking, time-out

Read-back/closed-loop communication: critical results, verbal orders

SBAR: standardized handoff (Situation, Background, Assessment, Recommendation)

CUS words (TeamSTEPPS): "I'm Concerned, I'm Uncomfortable, this is a Safety issue"

Two-challenge rule: assertively voice concern twice before escalating

Board Question Stem Patterns

— Stem: "The hospital tracks the percentage of CHF patients receiving an ACE inhibitor at discharge. This is best classified as..."

— Answer: Process measure

— Variants substitute: door-to-balloon (process), 30-day mortality (outcome), 24/7 cath lab (structure), HCAHPS (outcome)

— Stem: A team is reducing CLABSI. They've implemented a central line bundle. What measure is most actionable for daily feedback?

— Answer: Bundle compliance (process) — outcome (CLABSI rate) is too rare for daily action

— Variant: "Which measure best reflects patient-centered care?" → outcome/PROM

— Stem describes 8 consecutive points above the median after an intervention

— Answer: Special cause variation / signal of improvement (a "shift")

— Retrospective after an event = RCA

— Prospective before new process = FMEA

— Stem: medication error after RCA. Best next step?

— Answer: Forcing function / EHR hard stop / automated dose check > policy change > education

— Nurse made a single dosing error in an otherwise reliable system → console, system review, no discipline

— Nurse repeatedly bypassed barcode scanning → at-risk behavior → coach

— Provider diverted opioids → reckless → disciplinary

— Lab pending at discharge, PCP not notified → answer is closed-loop communication / direct PCP handoff + patient notification

— HF readmission at day 5 → likely missed early follow-up; answer is 7-day post-discharge appointment + med rec

— HRRP conditions, MIPS, Star Ratings stems test recognition

Key distinction: "Most actionable for QI" → process measure. "Most meaningful to patient" → outcome / PROM. "Most needed before launching a new program" → structure (resources) and FMEA (anticipate failures). Use the stem's verb to pick the type.

Pattern 1: Classify the measure

Pattern 2: Choose the right measure for the project

Pattern 3: Interpret a control chart or run chart

Pattern 4: RCA vs. FMEA

Pattern 5: Strongest corrective action

Pattern 6: Disclosure / Just Culture

Pattern 7: Transition of care

Pattern 8: Public reporting / value-based payment

One-Line Recap

Quality measurement in healthcare follows Donabedian's triad — structure (the resources and system in place), process (what clinicians do), and outcome (what happens to the patient) — and Step 3 expects you to classify any given measure, choose the right type for the QI question at hand, and pair every primary measure with a balancing measure to detect unintended harm.

Board pearl: If you can answer "Did we do the right thing? (process) — Did it work? (outcome) — Did we have what we needed? (structure) — Did we break anything else? (balancing)" for any clinical scenario, you've mastered the Step 3 quality framework.

Structure = resources/capacity (ICU intensivist coverage, EHR, certification); necessary but not sufficient

Process = clinician action (aspirin in AMI, A1c testing, bundle compliance); most actionable, fastest feedback, best for QI projects when evidence links action to outcome

Outcome = patient result (mortality, readmission, HCAHPS, PROMs); most patient-centered, requires risk adjustment and adequate sample size to be fair

Balancing measures detect unintended consequences (hypoglycemia while chasing A1c, fluid overload while chasing sepsis bundles) — always include one

PDSA drives iterative change; RCA investigates events retrospectively (system, not blame); FMEA anticipates failures prospectively

Strong corrective actions (forcing functions, EHR hard stops, automation) >> weak ones (education, policy memos)

Just Culture distinguishes human error (console), at-risk behavior (coach), reckless behavior (discipline)

Closed-loop communication, SBAR handoffs, time-outs, read-backs are NPSG-rooted safety practices and high-yield CCS answers

HRRP, HAC, VBP, MIPS, HEDIS, Star Ratings are the payment/reporting programs that translate measure performance into dollars

Transitions of care (med rec, 7-day follow-up, discharge summary, closing the loop on pending results) are the highest-risk and highest-yield Step 3 zone