Management Atlas · Narrative Review
Culture, reliability, and executive accountability, a narrative review
Executive synthesis
A strong safety culture is not a favorable survey score or a communication campaign. It is a governed system in which hazards can be reported, interpreted, acted upon, and verified without allowing accountability or learning to disappear.
What this review establishes
Safety culture is often reduced to staff perceptions collected at intervals. Those perceptions are important because they reveal whether people experience teamwork, communication openness, handoffs, response to error, staffing, and leadership support as reliable. Validated instruments such as the Hospital Survey on Patient Safety Culture 2.0 provide a common structure for measurement, and recent validation work strengthens their use across settings (Bagnasco et al., 2025). Yet a survey is a signal, not the culture itself. Culture is reproduced through daily decisions about scheduling, escalation, production pressure, investigations, resource allocation, and whose knowledge is treated as credible.
Leadership research repeatedly associates supportive, authentic, servant, safety-focused, and transformational behaviors with safety culture, voice, empowerment, and reporting (Demeke et al., 2025; Dirik & Intepeler, 2024; Ünal et al., 2025; Ystaas et al., 2023). These findings should be interpreted cautiously because many studies are cross-sectional, but their convergence is operationally meaningful. Leaders shape the perceived consequences of reporting and dissent. When executives ask for concerns and then fail to respond, tolerate disrespect, or punish the messenger, the organization teaches silence more effectively than any policy can teach safety.
Just culture is sometimes misunderstood as a promise that adverse events will not produce consequences. Its purpose is more disciplined: differentiate human error, at-risk behavior, reckless behavior, system contribution, and organizational responsibility so that the response fits the behavior and context. Recent work on transparency and risk reporting emphasizes that people report more reliably when the process is credible and fair (Davis et al., 2026). Toxic leadership and emotional exhaustion have also been associated with safety-incident reporting behavior, reinforcing the need to examine management conditions rather than treating underreporting as a staff attitude (Poku et al., 2026).
Executives should require a consistent review method with trained reviewers, conflict-of-interest safeguards, and documented reasoning. The analysis should examine work design, supervision, staffing, equipment, information, handoffs, incentives, and latent conditions before assigning individual accountability. Fairness also requires timeliness and communication. Staff need to know that a report was received, that the risk was evaluated, and that a decision followed. Confidentiality must be protected, but silence about the organizational response can look indistinguishable from inaction. Just culture becomes credible when comparable cases receive comparable analysis and leaders are accountable for system conditions within their control.
Incident reports capture only part of the safety picture. Reporting depends on recognition, time, usability, psychological safety, feedback, and the belief that the information will matter. Educational level and experience have been examined in relation to reporting and safety perception, while leadership, burnout, job satisfaction, and culture have been linked with documentation quality and safety voice (Gambashidze et al., 2025; Herak et al., 2026; Qian et al., 2025). Executives should therefore avoid interpreting a fall in reports as automatic improvement or a rise as automatic deterioration.
A mature surveillance system integrates voluntary reports, trigger tools, complaints, claims, infection data, pharmacy signals, staffing and workload, safety rounds, and qualitative observations. System-wide analysis of narrative incident data shows the potential for structured content analysis to uncover patterns that conventional categories may miss (Engstrom et al., 2025). The aim is not to automate judgment. It is to improve signal detection and direct expert attention. Each signal should enter a defined triage path based on severity, recurrence, detectability, population impact, equity, and potential for catastrophic harm. The resulting actions should be traceable to an owner and verification date.
Full narrative review
High-reliability language can become ceremonial unless it changes management routines. Qualitative research on high-reliability transformation describes the importance of leadership commitment, frontline engagement, structured communication, and sustained attention to systems (Leonard et al., 2024). The comparison between surgical safety and NASA also emphasizes disciplined learning, crew coordination, and leadership behavior, though cross-industry transfer must account for different technologies, regulation, and operating contexts (Al-Gholmy et al., 2026).
For healthcare executives, preoccupation with failure should appear in the operating calendar. Daily huddles identify current risk, tiered escalation moves unresolved risk to people with authority, leader rounds test whether controls work at the frontline, and weekly reviews examine recurring patterns rather than isolated incidents. Deference to expertise means decisions move toward the person with relevant knowledge, not automatically toward rank. Sensitivity to operations requires leaders to understand demand, staffing, workarounds, and handoffs as they exist, not as the procedure describes them. Resilience includes recovery capacity, redundancy for critical functions, and rehearsed response to degradation.
Human-factors analysis treats performance as an interaction among people, tasks, tools, environment, information, and organizational conditions. Studies of retained foreign objects, checklists, medication documentation, and missed care show why leaders should examine the design of the work rather than relying on attention or vigilance alone (Cromie et al., 2025; Gambashidze et al., 2025; Labrague & Cayaban, 2025; Palm & Braut, 2026). A checklist can fail when timing is wrong, roles are unclear, interruptions are high, or hierarchy prevents challenge. A technically correct alert can become noise when volume exceeds human capacity.
Controls should be layered. Standardization can reduce unnecessary variation; forcing functions can prevent certain errors; independent checks can protect high-risk transitions; visual management can reveal abnormal conditions; and recovery plans can limit harm when prevention fails. Each control also creates workload and may introduce new failure modes. Executives should require usability testing and balancing measures before enterprise deployment. Frontline adaptation is valuable evidence about system fit, but normalized workarounds may also indicate that the designed process is unworkable. Leaders need a safe way to surface both.
Leadership is frequently recommended as a safety intervention, but broad training without behavioral transfer produces limited assurance. Authentic leadership training has shown potential to improve empowerment and safety outcomes in a quasi-experimental setting, while systematic and integrative reviews describe associations between leadership and patient or workforce outcomes (Barcellos Morcelli et al., 2023; Dirik & Intepeler, 2024; Ystaas et al., 2023). The evidence does not identify one universally superior style. It does support developing behaviors that improve clarity, fairness, listening, follow-through, and escalation.
Development should be tied to operating expectations. Managers can be observed conducting safety huddles, responding to a concern, distinguishing urgency from blame, reviewing workload risk, and closing a corrective action. Coaching should use real cases with psychological safeguards. Executive rounds should evaluate whether staff can describe how to escalate, whether prior concerns received feedback, and whether leaders remove obstacles. Performance management should include safety leadership measures that do not reward suppression of reports. A leader with fewer reports is not necessarily safer; a leader who converts credible signals into verified controls provides stronger assurance.
Interventional studies and long-term program reports show that safety culture can change, but context, measurement, and sustainability matter (Salarian et al., 2025; So et al., 2025). A corrective action is not complete because education occurred, a policy was revised, or equipment was purchased. Those are implementation outputs. Leaders need evidence that the control is present in practice, is used as intended, influences the targeted process, and produces an acceptable balance of benefit and burden. The observation period must be long enough to detect drift and infrequent harm.
Action tracking should distinguish containment, corrective action, and prevention. Containment reduces immediate exposure. Corrective action addresses the identified mechanism. Prevention examines where similar mechanisms could occur elsewhere. Each phase should have an owner, completion evidence, and effectiveness review. Aggregate measures can hide site or population differences, so review should stratify where appropriate. The board should see a small set of leading and lagging measures with known definitions: unresolved high-risk hazards, time to interim containment, overdue effectiveness reviews, recurrence, severe harm, safety-climate signals, and workforce concerns.
Boards require enough information to oversee safety without substituting for operational management. National and multicenter studies demonstrate substantial variation in safety culture across hospitals and groups, which means an enterprise average can conceal local risk (Binh et al., 2026; Qu et al., 2026). Policy and payment environments may also influence organizational safety conditions, as suggested by research on Medicaid expansion and hospital safety culture (Forbes & Arrieta, 2025). Board reporting should therefore combine trend, variation, narrative explanation, and action assurance.
A useful board view connects four questions: What risks are emerging? Which controls are intended to manage them? What evidence shows those controls operate? Where does residual risk remain? The board should receive information about high-risk open hazards, overdue corrective actions, recurring event patterns, disparities, workforce safety signals, and management response. It should periodically review a closed case to understand the quality of analysis and verification. The aim is not more pages. It is a traceable line from signal to executive decision, control, effect, and remaining exposure.
Safety systems can fail when leaders treat staffing, civility, workload, burnout, and missed care as separate workforce matters. A systematic review and meta-analysis found that incivility in hospitals is common and is associated with safety culture and outcomes, while another systematic review connected safety culture with missed nursing care (Freedman et al., 2025; Labrague & Cayaban, 2025). Research on toxic leadership and emotional exhaustion further suggests that management conditions can influence incident-reporting behavior (Poku et al., 2026). These relationships should not be interpreted as simple causal estimates, but they support an executive view in which workforce capacity and relational conduct are part of the control environment.
Leaders need signals that reveal when demand is exceeding the system’s ability to perform safely. Relevant indicators may include staffing variance, overtime, missed breaks, use of contingency staff, turnover, vacancy, work compression, missed care, safety voice, incivility, and unresolved equipment or supply problems. The value lies in examining patterns together and close to the work. A staffing number within a nominal range does not prove that skill mix, workload distribution, supervision, or recovery capacity is adequate. Workforce dashboards and safety dashboards should share definitions, review forums, and escalation thresholds where the underlying risk is connected.
Civility should also be managed as an operational expectation. Disrespect can narrow communication, suppress challenge, and weaken coordination at the moments when a team most needs accurate information. Standards should apply across profession, rank, employment status, and facility. Reports should enter a fair process with safeguards against retaliation. Leaders should distinguish conflict about the work from demeaning conduct, and they should address patterns rather than relying on individual resilience. A sustainable safety strategy protects the human capacity needed to notice, communicate, and respond to risk.
Patients and families observe transitions, inconsistencies, communication failures, and recovery problems that internal systems may not capture. Complaints, grievances, bedside concerns, patient-reported outcomes, and family narratives should therefore be treated as safety signals when they describe potential harm or control failure. The system needs a defined route from patient relations to safety triage so that serious concerns do not remain inside a service-recovery workflow. Staff should also know when and how to escalate a patient or family concern that reveals a broader hazard.
Equity is part of the same design. Enterprise averages can conceal differences in exposure, recognition, response, or recovery across language, disability, race and ethnicity, age, geography, insurance, and other relevant characteristics. A control may exist but be less accessible to a patient who needs interpretation, uses an assistive device, receives fragmented care, or has less power to challenge a decision. Executives should identify where stratification is feasible and valid, where qualitative review is needed, and how privacy will be protected. Differences should trigger disciplined inquiry rather than unsupported causal conclusions.
Patient and family participation in improvement should be meaningful and supported. Representatives need a clear role, accessible materials, preparation, and feedback about how their contribution influenced the decision. Their presence should not substitute for systematic collection of experience across affected groups. Leaders should compare patient signals with event reports, claims, and operational data to identify converging patterns. When accounts conflict, the discrepancy is itself useful evidence about communication, documentation, or perspective.
Recovery after harm is also a safety capability. Organizations should have a reliable process for clinical response, communication, disclosure consistent with applicable requirements, patient and family support, staff support, investigation, and follow-up. Board assurance should include whether patients and families can access the process, whether commitments are completed, and whether lessons affect the wider system. Respectful recovery does not erase harm, but weak recovery can compound it and suppress future reporting.
Annual or biennial culture surveys provide valuable benchmarking and domain-level insight, but the interval is too long for operational control. Leaders need a layered measurement system that combines periodic validated surveys with short pulse measures, direct observation, rounding, qualitative review, workforce data, reporting behavior, and control-performance measures. The validated survey should remain stable enough to support comparison; pulse questions should be limited to decisions the organization is prepared to make. Frequent measurement without visible response can create fatigue and reduce trust.
The measurement plan should distinguish climate, capability, activity, reliability, and outcome. Climate concerns perceptions of communication, teamwork, staffing, and response to error. Capability concerns whether leaders and teams have the skills and structures to manage risk. Activity measures count rounds, huddles, reports, reviews, or training, but activity does not prove effect. Reliability measures show whether intended controls operate under real conditions. Outcomes include harm, recovery, missed care, and other consequences. A balanced executive view uses each type for the question it can answer.
Local variation deserves attention. A high enterprise score can coexist with a unit where staff do not speak, actions remain open, or leadership turnover has disrupted routines. Results should be examined with denominators, response patterns, and contextual information. Small groups may require aggregation or qualitative follow-up to protect confidentiality. Leaders should avoid public ranking that encourages score management. The objective is to find material risk and support improvement, not to produce a competition among units with different case mix and staffing conditions.
Assurance also requires independent challenge. Quality, risk, compliance, internal audit, or another qualified function can test whether management evidence supports closure and whether reported improvements are reproducible. Independence should not create a second safety system; it should test the integrity of the operating system. The board can periodically commission a focused deep dive where signals conflict, improvement stalls, or consequences are high. The result should clarify residual exposure and management decisions rather than merely restating policy.
Finally, leaders should communicate what was learned. A concise internal safety review can describe the signal, the system mechanism, the action, the verification, and the lesson without exposing protected information. This makes improvement visible, supports transfer, and demonstrates that reporting leads to organizational response. Sustainable culture develops when people repeatedly observe that credible signals become fair analysis, responsible action, and verified learning.
Executive operating model
Eight connected controls move evidence from an emerging signal to accountable executive action.
Every control requires a named owner, a decision point, and a verification measure.
| Control point | Executive question | Evidence-informed action | Assurance measure |
|---|---|---|---|
| Signal access | Can staff, patients, and partners report risk without unnecessary friction? | Provide multiple channels, confidentiality safeguards, and clear escalation. | Access testing, reporting latency, and workforce feedback. |
| Triage | Which signals require immediate containment or executive attention? | Use severity, recurrence, detectability, equity, and catastrophic potential. | Documented triage decision and containment time. |
| Fair review | Did the analysis examine behavior, context, and system contribution? | Apply a consistent just-culture and human-factors method. | Review quality audit and consistency check. |
| Action design | Does the action target the identified failure mechanism? | Prefer system controls and specify burden and new hazards. | Mechanism-to-control map and implementation evidence. |
| Effectiveness | Did the control reduce risk in practice? | Observe use, measure process and outcomes, and monitor balancing effects. | Effectiveness review with defined follow-up period. |
| Learning spread | Where else could the same mechanism create harm? | Search related workflows and share transferable lessons. | Enterprise prevention review and tracked spread. |
| Leadership practice | Do leaders make speaking up consequential and fair? | Coach response, escalation, rounding, and closure behaviors. | Response reliability and frontline validation. |
| Board assurance | What residual risk remains after management action? | Connect hazards, controls, reliability, outcomes, and disparities. | Concise assurance view with unresolved exposure. |
Evidence boundaries
The recent safety-culture literature contains many cross-sectional surveys and self-reported measures. Associations among leadership, culture, voice, reporting, burnout, and outcomes do not by themselves demonstrate causation. Single-site and country-specific studies may reflect staffing models, professional hierarchies, legal environments, and reporting norms that differ from a U.S. health system. Intervention studies provide stronger temporal information but may lack randomization, long follow-up, or control for concurrent initiatives.
Safety culture is also measured with different instruments, versions, domains, and outcome definitions. Comparisons should account for those differences rather than treating every score as equivalent. Report counts are especially vulnerable to misinterpretation because reporting behavior and event occurrence change simultaneously. Executives should use the literature to select mechanisms and controls, then test them with local process, outcome, balancing, equity, and qualitative measures. High-risk clinical controls may require a focused systematic review beyond this executive narrative synthesis.
90-day executive agenda
Define one enterprise method for triage, just-culture review, human-factors analysis, action design, and effectiveness verification.
Create a unified hazard view that combines incident reports, complaints, claims, trigger data, staffing, rounds, and other surveillance signals.
Set response-time standards for acknowledgement, interim containment, executive escalation, corrective-action approval, and effectiveness review.
Audit a sample of event reviews for fairness, system analysis, consistency, and explanatory quality across facilities and professional groups.
Redesign the executive safety dashboard to separate reporting volume, harm, unresolved exposure, control reliability, and action closure.
Observe managers conducting huddles, escalation, and feedback; coach the behaviors that determine whether staff voice remains credible.
Reopen overdue or administratively closed high-risk actions that lack implementation or effectiveness evidence.
Provide the board with one traceable case from signal through residual-risk decision and use it to test the quality of assurance.
Peer-reviewed evidence
25 peer-reviewed sources support this narrative review.
A narrative review of hospital command centers, including governance, decision rights, patient flow, technology, performance, and executive assurance.
Read analysis
A hospital executive narrative review of administrative burden, workflow redesign, responsible automation, workforce outcomes, and policy simplification.
Read analysis
Learn how hospitals can align demand, functional capacity, decision rights, throughput, discharge, and safety through an enterprise patient-flow operating system.
Read analysis