Leadership Accountability
Research Paper Below
Special Report | Executive Research Edition
When Leadership Breaks the Compact
What healthcare workers should demand when mission promises fail: an evidence synthesis, governance framework, and testable accountability protocol.
The Hypothesis Question
Is it realistic for healthcare workers to stop accepting empty promise mission language as evidence of leadership performance and demand an outcome-based leadership model linked to Leadership Accountability and a Compact that governs patient care, staff engagement, proven performance models, safety, voice, compensation, structural wellbeing, executive incentives, and trust repair?
What the evidence does and does not support
The record-high claim is not supported
NSI reported 2025 hospital turnover of 18.5% and RN turnover of 17.6%, both below the 2021 pandemic-era peaks of 25.9% and 27.1%. Federal quit data likewise show no record for the combined health care and social assistance sector.
The comforting interpretation is also wrong
Hospital and RN turnover increased again in 2025, first-year turnover reached 29.5%, frontline support roles exceeded 30%, and the average hospital’s estimated annual RN turnover loss was $5.19 million.
Governance failure
A promise has no accountability value unless it has an owner with authority, a funded intervention, a common metric, a deadline, worker decision rights, an escalation rule, and a consequence for repeated nonperformance.
Five practices that should no longer count as adequate leadership responses
Substituting resilience training for the correction of unsafe workload, avoidable administrative burden, or schedule instability.
Surveying employees without publishing a response covenant that names decisions, owners, dates, and declined requests.
Claiming safe staffing while withholding unit, shift, acuity, missed-care, and variance evidence.
Protecting executive incentive compensation from workforce stability, safety, trust, and first-year retention results.
Using confidentiality, branding, or aggregate reporting to conceal recurring nonperformance at the unit or leader level.
Executive action agenda
| Timing | Required action | Executive evidence of completion |
|---|---|---|
| 0 to 30 days | Publish the baseline, definitions, owners, and red-line thresholds; install anti-retaliation protections. | Signed data dictionary, unit dashboard, owner register, and protected reporting pathway. |
| 31 to 60 days | Seat elected frontline representatives with voting rights; fund the first operating corrections. | Charter, voting record, budget, and documented response to workforce recommendations. |
| 61 to 90 days | Launch unit pilots; place a material share of executive incentives at risk; publish the commitments register. | Pilot protocol, incentive weights, milestones, status, and escalation rules. |
| Quarterly | Review outcomes and implementation fidelity at the board level; intervene after repeated misses. | Minutes, variance review, remedies, consequences, and worker verification. |
| 12 months | Commission independent evaluation and trust reassessment; renew, redesign, or replace failed leadership controls. | Audit opinion, effect estimates, equity review, and disposition decision. |
Decision standard
The test is not whether leadership launches another initiative. The test is whether controllable working conditions improve, whether workers can verify the change, whether turnover and safety outcomes move without harm shifting to another group, and whether leaders experience consequences when avoidable failure persists.
Methods and Evidence Discipline
This special report is a structured evidence synthesis, not a registered systematic review. It prioritizes national workforce surveillance, transparent industry datasets, peer-reviewed systematic reviews and meta-analyses, large adjusted observational studies, and federal systems-level guidance, and it is explicit about where causal claims stop.
Research question and hypotheses
H1: Outcome
Units implementing an audited, funded, and consequence-linked Workforce Accountability Compact will experience a greater reduction in risk-adjusted 12-month voluntary turnover than comparable units under advisory-only programs.
H2: Mechanism
The effect of the Compact on turnover will be mediated by improved staffing adequacy, schedule control, psychological safety, perceived organizational support, and confidence in leadership.
H3: Governance
Fidelity and outcome improvement will be stronger when workforce measures are tied to board oversight, executive incentive compensation, and mandatory escalation than when disclosed without enforceable consequences.
Evidence hierarchy and intended use
| Evidence stream | Contribution | Primary strength | Primary limitation | Decision weight |
|---|---|---|---|---|
| BLS JOLTS | Current macro quit context | Federal recurring series | Health care combined with social assistance | Contextual |
| NSI 2026 | Hospital, RN, tenure, role, and cost estimates | 527 hospitals; large denominators | Industry survey, not a probability sample | High for burden |
| Press Ganey 2026 | Engagement, trust, and turnover relationship | More than 2.6 million workers and physicians | Proprietary observational analysis | Moderate |
| Systematic reviews | Support, staffing, and intervention synthesis | Cross-study convergence | Heterogeneity; many cross-sectional studies | Moderate |
| Adjusted observational studies | Effect size and mechanism signals | Covariate-adjusted estimates | Residual confounding; often intention outcomes | Moderate |
| NIOSH guidance | Systems intervention logic | Federal, operational, and participatory | Guidance, not a causal turnover trial | High for design |
| Organizational theory | Psychological and governance mechanisms | Mature explanatory constructs | No specific proof of this Compact | Mechanistic |
Four inferential rules
1. Association is not causality
Cross-sectional association was never described as proof of causality.
2. Triangulate proprietary data
Industry and proprietary reports were checked against federal or peer-reviewed evidence where possible.
3. Heterogeneity is substantive
Differences in turnover definitions, settings, and follow-up were treated as substantive rather than cosmetic.
4. Evidence plus mechanism
A recommendation was considered stronger when both empirical evidence and a plausible governance mechanism supported it.
Correction to the opening premise
The defensible statement is not that healthcare turnover is at an all-time high. It is that turnover remains strategically unacceptable, worsened again in 2025 for hospitals and RNs, is exceptionally high among early-career and frontline roles, and reflects organizational conditions that leaders are obligated to measure and improve.
The Turnover Reality: Severe, Uneven, and Governable
Figure 1. Hospital and staff RN turnover, 2021 to 2025
Hospital turnover declined from 25.9% in 2021 to 18.3% in 2024, and staff RN turnover declined from 27.1% to 16.4%. That trajectory reversed in 2025: hospital turnover rose to 18.5% and RN turnover rose to 17.6%. The increase does not recreate the 2021 peak, but it interrupts improvement and signals that post-pandemic normalization cannot be assumed to continue without management action.
Source: NSI Nursing Solutions (2026). Interactive rebuild of the report figure.
Figure 2. Turnover by selected hospital role, 2025
First-year hospital employee turnover was 29.5%. A system that treats a nearly one-in-three first-year exit rate as a recruiting problem is misclassifying a design and credibility problem. The roles with the highest turnover are central to care flow, infection prevention, surveillance, and workload distribution, even though they often hold the least organizational power.
Source: NSI Nursing Solutions (2026). Selected occupations shown; values reflect reported annual turnover.
The aggregate hides enormous variation
NSI reported hospital turnover ranging from 7.6% to 33.1% and RN turnover from 5.6% to 40.0%. Such dispersion is incompatible with an explanation based only on a single national labor market. Wide organizational variation indicates substantial room for operating practice and governance to influence outcomes.
Source: NSI Nursing Solutions (2026). Reported organizational ranges with the national rate marked.
Trust is an operating variable
Press Ganey’s 2026 analysis reported that 35% of employees and 37% of providers lacked confidence in senior leadership. Disengaged caregivers were reported as 2.6 times more likely to leave. Because the analysis is proprietary and observational, it should not be treated as an independent causal estimate, but its scale and direction align with peer-reviewed evidence on organizational support, practice environment, and psychological safety.
Trust changes whether employees disclose hazards, interpret leadership messages as credible, invest discretionary effort, remain through a difficult period, or leave before another promise is broken.
Source: Press Ganey (2026). Observed turnover in disengaged versus high-engagement environments.
The financial burden is material and incompletely governed
The governance defect is not simply that turnover costs money. It is that multimillion-dollar exposures are often not measured or assigned a target. On NSI’s estimate, the 1.2-point increase from 2024 to 2025 corresponds to roughly $354,000 for an average hospital before local adjustment.
RN turnover cost model
Estimate the annual financial exposure of RN turnover for your organization and the value of a targeted reduction. The engine applies the NSI 2026 average cost per bedside RN separation, which you can adjust to your local cost model.
Model: separations = headcount × rate; cost = separations × cost per separation. NSI values each RN turnover percentage point at approximately $295,000 for the average hospital. A valid internal model should also measure vacancy days, agency and overtime premiums, recruitment, preceptor time, productivity ramp, and lost capacity.
Govern upstream conditions instead of waiting for exits
| Lagging outcome | Leading condition to govern | Why it matters |
|---|---|---|
| Voluntary turnover | Leadership confidence and organizational support | Detects withdrawal before separation is recorded. |
| First-year exit | Expectation accuracy, preceptor load, and schedule reality | Tests whether recruitment promises hold up under operational contact. |
| Vacancy and agency use | Staffing-plan attainment and capacity escalation | Shows whether demand is covered safely or deferred to premium labor. |
| Safety event | Psychological safety, violence exposure, and missed-care signals | Makes risk visible before patient or worker harm. |
Board question
Why does the organization review operating margin every month but tolerate annual, aggregated, or definitionally unstable reporting of first-year turnover, unsafe staffing shifts, and leadership trust?
From Mission Promise to Workforce Withdrawal
Figure 3. Theory-informed pathway linking broken promises to withdrawal
Mission promise
Explicit or implied reciprocal obligations between worker and organization.
Perceived breach
Cognitive recognition that an obligation was unmet.
Felt violation
Betrayal, anger, or moral injury when the breach is serious, repeated, and poorly repaired.
Appraisal
Magnitude, attribution, fairness, support, and safety shape the response.
Withdrawal
Silence, reduced discretionary effort, departure intention, and exit.
Source: Author synthesis informed by Morrison and Robinson (1997), Colquitt et al. (2001), Rhoades and Eisenberger (2002), and Edmondson (1999).
Four mechanisms that connect leadership behavior to withdrawal
Psychological contract breach
Leaders intensify violation when they deny observable conditions, redefine metrics after failure, ask workers to absorb the consequences, suppress voice, or continue to reward executives. They reduce it by acknowledging the breach, disclosing constraints, making restitution, sharing decision authority, and verifying correction.
Organizational justice
Distributive, procedural, interpersonal, and informational justice explain why the same difficult condition produces different reactions. Meta-analytic evidence relates these dimensions to work attitudes and behavior, making justice a practical design requirement rather than a public-relations concept.
Perceived organizational support
A 2024 meta-analysis of eight cross-sectional studies involving 5,754 nurses estimated a pooled correlation of −0.32 between organizational support and turnover intention: moderate, consistent, and not proof of causality, but decisive evidence that support is not peripheral.
Psychological safety
Psychological safety affects whether staff question an order, report deterioration, disclose near misses, or challenge a staffing decision. An anti-retaliation policy is necessary but not sufficient: workers need independent reporting, observable protection, transparent case closure, and discipline when leaders suppress concerns.
Betrayal is not a metaphor
In a 2024 study of 1,066 healthcare and hospital workers, reported betrayal was associated with markedly higher odds of mental distress and post-traumatic stress symptoms. The study was cross-sectional, single-system, and drawn from an extraordinary period, so it should not be generalized as a prevalence estimate. It does show that perceived institutional betrayal can be clinically and operationally consequential (Park et al., 2024).
A 2024 meta-analysis of 85 studies and 288,581 nurses found nurse burnout associated with worse safety and quality outcomes, including medication errors, falls, infections, and adverse events. When leaders frame exhaustion as an individual resilience deficit while preserving the work design that produces it, they transfer organizational risk back onto workers and patients (Li et al., 2024).
What Predicts Departure, and What Appears to Help
Figure 4. Adjusted associations with planned departure among Michigan nurses, 2023
Friese and colleagues analyzed Michigan nurse survey data. Emotional exhaustion and workplace abuse or violence were associated with higher adjusted odds of planned departure, while a favorable practice environment and an excellent clinical safety rating were associated with substantially lower odds. Response rates were low and the outcome was intention rather than actual exit, so these are strong warning signals rather than causal turnover multipliers.
Source: Friese et al. (2024). Points are adjusted odds ratios; bars are 95% confidence intervals; the dashed line marks no association (OR = 1.0). Logarithmic scale.
Departure intention is rising
Michigan nurses planning to leave their position (Friese et al., 2024). Workload was the most frequently cited reason.
Physician burnout: better is not acceptable
Physicians reporting at least one burnout symptom (Shanafelt et al., 2025). Recovery from an extreme peak does not establish an acceptable operating state.
Staffing, retention, and the intervention gap
Unsafe staffing tracks actual turnover
Bae’s systematic review found inadequate or unsafe nurse staffing associated with actual turnover across the included literature, while evidence on scheduling was less developed (Bae, 2024).
130 studies, only 9 interventions
Buckley and colleagues synthesized 130 studies but found only nine intervention studies. The field knows far more about correlates than about implemented, rigorously evaluated organizational remedies (Buckley et al., 2025).
Change the work
Aust and colleagues found 68% of included organizational interventions improved at least one primary outcome, with evidence strongest for job and task modification. NIOSH reaches the same operational conclusion (Aust et al., 2024; NIOSH, 2024).
Non-negotiable interpretation
Resilience resources may be useful as an optional support. They are not an adequate primary intervention when the principal exposure is unsafe workload, unstable scheduling, violence, administrative burden, or absent decision-making power.
Evidence strength and the boundary of inference
| Finding | Best supporting evidence | Confidence | Governance implication |
|---|---|---|---|
| Turnover remains severe and reversed direction in 2025 | Large 2026 hospital industry survey; federal quit context | Moderate | Treat trend reversal and first-year concentration as board risks. |
| Exhaustion, violence, practice environment, and safety track departure risk | Adjusted observational nurse study | Moderate | Govern leading indicators at the unit and shift level. |
| Organizational support relates to lower turnover intention | Meta-analysis of eight cross-sectional studies | Low to moderate | Measure support and test it as a mediator, not as proof. |
| Unsafe staffing relates to actual turnover | Systematic review | Moderate | Require acuity, competency, variance, and capacity controls. |
| Work-environment interventions can reduce burnout | Systematic review of organizational interventions | Moderate | Fund operational redesign before individual coping programs. |
| Betrayal can accompany significant distress | Cross-sectional single-system study | Emerging | Install explicit breach acknowledgment and trust-repair processes. |
| The proposed Compact will reduce turnover | Theory-informed intervention, not yet trialed | Unknown | Implement with a rigorous comparative evaluation and stop rules. |
Leadership Accountability Without Causal Overreach
Accountability is not the same as blame. Blame asks who deserves condemnation after an outcome. Accountability asks who had authority, resources, and a duty to prevent, detect, correct, and learn. A chief executive does not control every resignation. The executive team and board do control the system.
A disciplined attribution framework
Direct control
Staffing model, schedule rules, span of control, reporting pathway, incentive design, pay architecture, onboarding, security resources.
Duty: set standards, resource, monitor, correct, impose consequences.
Failure is directly governable.
Material influence
Team climate, workload, local labor supply response, commute mitigation, career access, supervisor quality, technology burden.
Duty: use available levers, test alternatives, disclose constraints.
Outcome is shared, but inaction is not neutral.
Limited control
Retirement, family relocation, spouse employment, illness, licensure pipeline, macroeconomic shock.
Duty: forecast, accommodate where feasible, avoid false attribution.
Departure may be unavoidable; system response remains governable.
Unknown or mixed
Exit reason recorded as personal, better opportunity, or culture.
Duty: confidential qualitative review and multiple indicators.
Do not use a vague code to close the inquiry.
The minimum anatomy of accountability
A workforce commitment becomes governable only when eight elements are present:
Governance equation
Accountability = owner + authority + funding + metric + deadline + worker decision rights + consequence + verified repair.
Four tests for a credible accountability claim
| Board test | Passing evidence |
|---|---|
| Controllability | The owner can identify which levers were available, which were used, and which constraints were independently verified. |
| Specificity | The promise, metric, denominator, affected group, deadline, and expected change are explicit. |
| Reciprocity | Workers receive protection, decision rights, and repair proportionate to the risk and sacrifice expected of them. |
| Consequence | Repeated avoidable failures change incentive pay, authority, performance status, or leadership assignment. |
What workers should no longer concede
That aggregate reporting is transparent when it conceals high-risk units, shifts, occupations, or groups. That a listening session is shared governance when management retains every vote. That an employee assistance program is a staffing intervention, that a bonus is compensation justice, or that a new dashboard is accountability when missing targets produce no action.
The Workforce Accountability Compact: Ten Demands
The Compact is a proposed governance intervention, deliberately more demanding than an engagement plan. Each requirement specifies a minimum standard, observable proof, and a consequence. Select any demand to review its full specification.
1. Audited workforce transparency
Minimum standard: monthly unit-level turnover, first-year exits, vacancies, overtime, agency use, injuries, violence, burnout, trust, and pay-compression indicators with common definitions.
Proof: board-reviewed dashboard, data dictionary, denominators, trend, targets, and named owner.
Consequence: corrective plan within 30 days for missing data or adverse outliers; audit committee escalation after two missed cycles.
2. Safe staffing and workload controls
Minimum standard: acuity- and competency-based staffing plan, shift escalation rules, relief coverage, and authority to limit capacity when safe care cannot be supported.
Proof: percent of shifts meeting plan, workload variance, missed breaks, missed care, overtime, closure or diversion events.
Consequence: mandatory operating review and capacity action when red-line thresholds are breached.
3. Meaningful worker decision rights
Minimum standard: elected frontline representation with voting authority on staffing, workflow, scheduling, safety, technology, and wellbeing priorities.
Proof: charter, voting record, documented management response, and implementation log.
Consequence: board review of rejected recommendations and written rationale within 30 days.
4. Psychological safety and anti-retaliation
Minimum standard: independent reporting, just culture review, confidentiality protections, and an explicit prohibition on retaliation for raising safety or staffing concerns.
Proof: case closure time, substantiation patterns, retaliation allegations, culture measures, and remedy completion.
Consequence: independent investigation and leadership discipline for retaliation or suppression.
5. Schedule control and recovery
Minimum standard: predictable schedules, limits on mandatory overtime, protected breaks, usable paid leave, and participatory scheduling where operations permit.
Proof: schedule-change notice, mandatory overtime, missed breaks, denied leave, fatigue events, and schedule-control scores.
Consequence: unit redesign and supervisor correction when thresholds are repeatedly missed.
6. Violence prevention and physical safety
Minimum standard: staffed security plan, hazard assessment, rapid response, post-event support, prosecution policy where appropriate, and environmental controls.
Proof: assault rate, injury severity, response time, lost workdays, reporting completeness, and corrective actions.
Consequence: executive safety review after sentinel workforce events and capital remediation when hazards persist.
7. Fair compensation and career mobility
Minimum standard: transparent market review, compression analysis, differentials, internal mobility, paid development, clinical ladders, and equitable access to advancement.
Proof: market position, compression gaps, promotion rates, internal fill rates, tuition use, and exit reasons.
Consequence: funded adjustment plan with deadlines when material inequities are documented.
8. Structural wellbeing intervention
Minimum standard: operational redesign of workload, administrative burden, staffing, workflow, and team conditions before relying on individual resilience programs.
Proof: funded intervention portfolio, baseline, process measures, outcome measures, and stopped low-value work.
Consequence: reallocate wellbeing spending away from ineffective programs and redesign the work.
9. Executive and board consequences
Minimum standard: workforce stability, safety, trust, and first-year retention are included in executive evaluation and incentive compensation.
Proof: published scorecard weights, thresholds, board minutes, and action after misses.
Consequence: reduced incentive pay, narrowed authority, a formal improvement plan, or leadership replacement after repeated avoidable failure.
10. Trust repair after breach
Minimum standard: specific acknowledgment of what failed, disclosure of decisions and constraints, restitution where possible, co-designed correction, and scheduled follow-up.
Proof: written breach review, commitments register, completion status, worker verification, and independent reassessment.
Consequence: escalation to the board and an external review if leaders deny, minimize, or repeat the breach.
Demand architecture
The ten demands work as a system. Transparency without staffing authority exposes harm but may not reduce it. Worker voice without anti-retaliation invites risk. Compensation without schedule control may buy short-term tolerance at the expense of worsening fatigue. Wellbeing funding without task redesign individualizes a structural problem. Executive incentives without reliable denominators create gaming. Trust repair without consequences becomes another cycle of apologies.
From Compact to Board Operating System
Figure 6. Five-stage workforce accountability maturity model
Organizations often confuse measurement with accountability. Measurement is only the third of five maturity stages.
| Stage | Diagnostic question | Required move |
|---|---|---|
| 1. Denial | Can leadership state the rate, the denominator, the concentration, and the named owner? | Establish the factual baseline and stop normalizing the variance. |
| 2. Promise | Does every commitment have funding, a deadline, and an escalation rule? | Convert statements into a commitments register. |
| 3. Measurement | Can workers alter decisions, and do misses change management action? | Transfer defined decision rights and prescribe responses. |
| 4. Shared governance | Are authority, resources, and worker power sufficient to correct the exposure? | Close implementation gaps and audit fidelity. |
| 5. Consequence | Are outcomes, repair, and leadership consequences independently verifiable? | Maintain external review and recalibrate incentives. |
Figure 7. Illustrative workforce accountability heatmap
Heatmaps should direct attention, not substitute for the underlying data. Select any cell to see the prescribed red-line response for that measure. Status values are fictional and serve only to display logic.
| Measure | Unit A | Unit B | Unit C | Unit D |
|---|---|---|---|---|
| Voluntary turnover | On target | Watch | Red line | On target |
| First-year exits | Watch | Red line | Watch | On target |
| Staffing plan attainment | On target | Watch | Red line | Watch |
| Workplace violence | On target | On target | Watch | Red line |
| Schedule control | Watch | On target | Watch | On target |
| Leadership confidence | Red line | Watch | On target | Watch |
| Commitment closure | On target | On target | Red line | On target |
Select a heatmap cell to display the prescribed red-line response for that measure.
Minimum board scorecard specification
| Measure | Minimum definition | Stratification | Red-line response |
|---|---|---|---|
| Voluntary turnover | Voluntary separations / average headcount; rolling 12-month and quarterly flow | Unit, role, shift, status, tenure, demographic equity | Corrective review after an adverse special-cause signal or two target misses. |
| First-year exits | Separations within 365 days, with hires and risk-set definition disclosed | Hiring cohort, recruiter, unit, supervisor, role | Root-cause review and onboarding or job-design correction within 30 days. |
| Staffing plan attainment | Shifts meeting acuity- and competency-adjusted plan / eligible shifts | Unit, shift, weekday, service line | Capacity, diversion, or staffing escalation when the safety threshold is crossed. |
| Workplace violence | Reported assaults, injury severity, lost days, and response time | Location, shift, role, source of violence | Sentinel workforce event review and capital or security remedy. |
| Schedule control | Notice, changes, mandatory overtime, missed breaks, and denied leave | Unit, shift, employment status, caregiver status | Supervisor and scheduling redesign after repeated breaches. |
| Leadership confidence | Validated survey item with response rate and nonresponse analysis | Unit, role, tenure, and key equity groups | Listening plus published response covenant; independent review if decline persists. |
| Commitment closure | Verified commitments completed by deadline / commitments due | Owner and demand category | Board escalation after two missed cycles; consequence per charter. |
Incentive design and data integrity
Balanced incentive design
Pure outcome incentives can encourage risk selection, coding changes, or suppression. Pure process incentives can reward activity without improvement. A balanced design combines outcomes, leading indicators, implementation measures, and guardrails such as safety events, retaliation allegations, and equity gaps. The compensation committee should approve definitions before the performance year and prohibit retroactive denominator changes.
Worker verification
The data dictionary must define inclusion of transfers, retirements, involuntary separations, per diem workers, contractors, and leave. Worker representatives should be able to challenge definitions and append a dissent statement. An independent audit should focus on missingness, denominator drift, suppressed small cells, repeated reclassification, and whether recorded completion matches frontline experience.
Quarterly board challenge
Show the three largest adverse workforce variances, the workers who experience them, the corrective actions already funded, the commitments past due, and the executive consequences that will follow another missed cycle.
Implementation Sequence
Implementation must move fast enough to establish credibility and slowly enough to preserve measurement quality.
Figure 8. Compact implementation roadmap
First 30 days: acknowledge and disclose
Issue a specific breach statement. Publish the baseline dashboard with definitions, denominators, missingness, targets, and owner names. Freeze retaliation risk with independent reporting and rapid protection review. Identify immediate hazards requiring capacity limits, security action, relief coverage, or schedule correction.
Days 31 to 90: transfer power and fund correction
Seat the workforce council through a transparent election with protected time and technical support. Approve staffing, schedule, safety, and workflow red lines with automatic escalation. Redirect wellbeing resources toward work redesign. Publish executive incentive weights, commitments, and deadlines before claiming the Compact is operational.
Quarterly: verify outcomes and fidelity together
A turnover reduction cannot validate the program if it resulted from a hiring freeze, changed eligibility, or involuntary exits. A missed target cannot invalidate a well-implemented control during an external labor shock. The board needs both the result and the causal context.
12 months: independent evaluation
Estimate effects, test equity, identify unintended consequences, and recommend renewal, redesign, scale, or discontinuation.
The communication covenant
Every workforce survey or listening exercise should begin with a response covenant. Leadership should state in advance when results will be published, which decisions workers can influence, how declined recommendations will be explained, how confidentiality will be protected, and when progress will be revisited. The covenant prevents listening from becoming extraction: workers should not repeatedly disclose harm while leaders retain the information and the discretion to do nothing.
Implementation failure modes and controls
| Implementation risk | Early signal | Required control |
|---|---|---|
| Symbolic adoption | Compact signed, but no budget, red lines, or incentive exposure | Do not declare launch until minimum fidelity criteria are met. |
| Data gaming | Definitions, exclusions, or denominators change after a miss | Freeze definitions prospectively and require audit approval for change. |
| Voice without protection | Participation rises while retaliation allegations or transfers increase | Independent investigation, interim protection, and board escalation. |
| Burden shifting | One group improves while overtime, violence, or vacancies rise elsewhere | Equity and spillover analysis across occupations, shifts, and units. |
| Initiative overload | Milestones accumulate while frontline capacity deteriorates | Stop low-value work and fund implementation time. |
A Research-Grade Test of the Compact
The intervention is not the presence of a written Compact. A unit is exposed only when the required elements are operational, and fidelity should be scored independently. This prevents a null result caused by partial implementation from being misinterpreted as evidence that accountability does not work.
Figure 9. Illustrative stepped-wedge implementation
The preferred pragmatic design is a stepped-wedge rollout across comparable units or facilities. Every participating unit eventually receives the Compact, but the transition is staggered, allowing contemporaneous comparison between usual governance and the intervention. Where staggered rollout is infeasible, use difference-in-differences with matched comparison units and at least 12 months of baseline data.
Source: author illustration. Transition timing should be randomized where operationally and ethically feasible. Follow-up should extend 18 to 24 months because turnover, hiring cohorts, and trust repair evolve more slowly than implementation activity.
Core evaluation protocol
| Element | Specification |
|---|---|
| Design | Stepped-wedge implementation or difference-in-differences comparison across matched units or hospitals, with at least 12 months of baseline and 18 to 24 months of follow-up. |
| Primary outcome | Risk-adjusted 12-month voluntary turnover by unit, occupation, shift, employment status, and tenure; time to exit when person-level de-identified data are available. |
| Secondary workforce outcomes | First-year turnover, vacancy, internal transfer, overtime, agency hours, schedule control, emotional exhaustion, psychological safety, organizational support, and confidence in leadership. |
| Patient and operating outcomes | Safety events, missed-care indicators, patient experience, closed capacity, time-to-fill, labor cost, orientation cost, and turnover cost. |
| Mechanism test | Estimate whether changes in staffing adequacy, organizational support, safety, voice, and trust mediate changes in turnover. |
| Adjustment | Occupation, wage changes, patient acuity, service-line mix, regional unemployment and labor supply, retirement eligibility, seasonality, and baseline trend. |
| Governance contrast | Compare advisory-only implementation with implementation tied to board oversight, executive incentives, deadlines, and documented consequences. |
| Decision rule | Pre-specify minimum meaningful change, confidence intervals, equity checks, and stop or redesign criteria before launch. |
Primary model
g(E[Yit]) = α + β(Compactit) + γt + ui + θXit
β estimates the intervention effect, γt captures common period shocks, ui captures stable unit differences, and Xit contains pre-specified time-varying confounders. Report absolute risk differences as well as relative effects, examine pre-trends, and do not declare success from statistical significance alone.
The four dispositions
Scale
Meaningful improvement in the primary outcome, acceptable fidelity, no serious harm to guardrails, and equitable benefit.
Continue with modification
Partial improvement or fixable fidelity gaps identified by the review.
Pause and redesign
A null outcome with high fidelity argues against the current intervention theory; a null outcome with low fidelity argues against the implementation.
Stop and investigate
Harm, retaliation, or data suppression requires immediate investigation regardless of the turnover outcome.
Guardrails
A retention improvement that depends on involuntary conversion, suppressed reporting, unsustainable overtime, or disproportionate burden on a lower-power group is not successful. Guardrails include missed care, adverse events, workplace injuries, violence severity, sick leave, closed capacity, and patient experience.
Compact Readiness Self-Assessment
Rate your organization on each of the ten demands. The scoring mirrors the maturity logic of the report: a demand that is merely promised is not measured, and a demand that is measured is not enforced. Results map to the five-stage maturity model.
1. Audited workforce transparency
2. Safe staffing and workload controls
3. Meaningful worker decision rights
4. Psychological safety and anti-retaliation Critical
5. Schedule control and recovery
6. Violence prevention and physical safety
7. Fair compensation and career mobility
8. Structural wellbeing intervention
9. Executive and board consequences Critical
10. Trust repair after breach
Awaiting responses
Complete all ten ratings and select Score readiness.
Interpretation follows the report’s maturity model. A rating of 0 on anti-retaliation (Demand 4) or executive consequences (Demand 9) caps the result at Stage 3, because measurement without protection or consequence does not constitute accountability. This self-check is educational and does not replace a formal fidelity audit.
Suggested worker resolution
We request adoption of the Workforce Accountability Compact within 90 days, beginning with audited unit-level disclosure, protected worker decision rights, safe staffing and workload red lines, executive incentive exposure, and a board-approved consequence framework. We further request an independent evaluation of outcomes, equity, implementation fidelity, and trust repair at 12 months.
References and Limitations
Six limitations that narrow, but do not erase, the pattern
1
Rapid evidence synthesis, not a registered systematic review; relevant studies may be omitted.
2
NSI is a large industry survey rather than a probability sample of all US healthcare organizations.
3
BLS combines health care with social assistance, providing macro context rather than a precise hospital measure.
4
Press Ganey analyses are proprietary and observational.
5
Much of the nursing literature is cross-sectional and uses intention to leave rather than verified turnover.
6
The Compact and maturity model are author-developed constructs not yet validated as a package.
Reference library
- Aust, B., Leduc, C., Cresswell-Smith, J., et al. (2024). The effects of different types of organizational workplace mental health interventions on mental health and wellbeing in healthcare workers: A systematic review. International Archives of Occupational and Environmental Health.DOI
- Bae, S. H. (2024). Assessing the impacts of nurse staffing and work schedules on nurse turnover: A systematic review. International Nursing Review, 71(1), 168-179. DOI
- Buckley, L., McGillis Hall, L., Price, S., Visekruna, S., & McTavish, C. (2025). Nurse retention in peri- and post-COVID-19 work environments: A scoping review. BMJ Open, 15(3), e096333. DOI
- Bureau of Labor Statistics. (2026). Job Openings and Labor Turnover Survey, Table 4: Quits levels and rates by industry, May 2026. Link
- Colquitt, J. A., Conlon, D. E., Wesson, M. J., Porter, C. O. L. H., & Ng, K. Y. (2001). Justice at the millennium: A meta-analytic review of 25 years of organizational justice research. Journal of Applied Psychology, 86(3), 425-445. DOI
- Edmondson, A. C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383. DOI
- Friese, C. R., Medvec, B. R., Marriott, D. J., et al. (2024). Changes in registered nurse employment plans and workplace assessments. JAMA Network Open, 7(7), e2421680. DOI
- Galanis, P., Moisoglou, I., Papathanasiou, I. V., et al. (2024). Association between organizational support and turnover intention in nurses: A systematic review and meta-analysis. Healthcare, 12(3), 291. DOI
- Goens, B., & Giannotti, N. (2024). Transformational leadership and nursing retention: An integrative review. Nursing Research and Practice, 2024, 3179141. DOI
- Li, L. Z., Yang, P., Singer, S. J., Pfeffer, J., Mathur, M. B., & Shanafelt, T. D. (2024). Nurse burnout and patient safety, satisfaction, and quality of care: A systematic review and meta-analysis. JAMA Network Open, 7(11), e2443059. DOI
- Morrison, E. W., & Robinson, S. L. (1997). When employees feel betrayed: A model of how psychological contract violation develops. Academy of Management Review, 22(1), 226-256. DOI
- National Institute for Occupational Safety and Health. (2024). Impact Wellbeing Guide: Taking action to improve healthcare worker wellbeing. Link
- Nembhard, I. M., & Edmondson, A. C. (2006). Making it safe: The effects of leader inclusiveness and professional status on psychological safety and improvement efforts in health care teams. Journal of Organizational Behavior, 27(7), 941-966. DOI
- NSI Nursing Solutions. (2026). 2026 NSI National Health Care Retention & RN Staffing Report. Link
- Park, S., Thrul, J., Cooney, E. E., et al. (2024). Betrayal-based moral injury and mental health problems among healthcare and hospital workers serving COVID-19 patients. Journal of Trauma & Dissociation, 25(2), 202-217. DOI
- Press Ganey. (2026). The state of healthcare employee experience 2026. Link
- Rhoades, L., & Eisenberger, R. (2002). Perceived organizational support: A review of the literature. Journal of Applied Psychology, 87(4), 698-714. DOI
- Shanafelt, T. D., West, C. P., Sinsky, C., et al. (2025). Changes in burnout and satisfaction with work-life integration in physicians and the general US working population between 2011 and 2023. Mayo Clinic Proceedings, 100(7), 1142-1158. DOI
Data and figure note: all charts and infographics in the source report are original. Figures 1 and 2 use values transcribed from the 2026 NSI report. Figure 4 reproduces reported adjusted odds ratios and 95% confidence intervals from Friese et al. (2024). Figures 3 and 5 through 9 are author-developed conceptual or illustrative graphics. The illustrative heatmap does not represent a real organization.