The Safety and Quality of Health Care¶
Chapter 7 | Part 1: The Profession of Medicine
KEY CLINICAL POINTS¶
- Nearly one in four hospitalized patients experiences harm related to medical care, with adverse drug events (ADEs) being the most common type of adverse event
- Safety theory distinguishes between 'slips' (automatic behavior errors) and 'mistakes' (errors in conscious decision-making), requiring different prevention strategies
- Systems theory emphasizes that most accidents result from a series of small failures ('Swiss cheese model') rather than individual provider negligence
- Quality of care in the US averages only 55% of recommended care across preventive, acute, and chronic conditions
- Effective improvement strategies include computerized physician order entry (CPOE), checklists, team-based care, and pay-for-performance incentives
1. DEFINITION & OVERVIEW¶
Safety and quality represent two central dimensions of health care that are closely related but do not completely overlap. The Institute of Medicine has established that safety is the foundational component of quality—health care systems must first guarantee safe care delivery. While safety improvements are more tangible to the public, greater net clinical benefit may ultimately derive from quality improvements. Both dimensions are increasingly subject to measurement, accountability, and performance-based payment systems.
1.1 Relationship Between Safety and Quality¶
Safety is considered the first part of quality. A health care system must first guarantee safe care before addressing broader quality measures. Public demands for measurement and accountability are driving improvements in both areas, with payment increasingly tied to performance metrics.
1.2 Key Definitions¶
Adverse Event: An injury caused by medical management rather than by the patient's underlying disease, resulting in death, disability at discharge, or prolonged hospital stay by at least 2 days. Preventable Adverse Event: An adverse event that could have been avoided through reasonable intervention or system design. Ameliorable Adverse Event: An adverse event whose severity could have been reduced through appropriate action.
2. SAFETY IN HEALTH CARE¶
Safety theory clearly demonstrates that individuals make errors regularly as part of normal cognitive functioning. The challenge is designing systems that minimize error occurrence and capture errors before they cause patient harm.
Factors Contributing to Healthcare Errors¶
| Factor | Mechanism | Mitigation Strategies |
|---|---|---|
| Fatigue | Cognitive impairment from sleep deprivation; 33% increased error rate on 24-hour shifts | Work hour limitations; adequate rest periods; recognition of impairment |
| Stress | High-pressure situations impair decision-making | Protocols for high-stress scenarios; simulation training; team support |
| Interruptions | Incomplete actions; forgotten tasks | Minimize non-urgent interruptions; urgency stratification tools |
| Complexity | Data overload; missed critical findings | Clinical decision support; alert systems; data visualization |
| Transitions | Information loss between providers/settings | Structured handoff tools; standardized communication (SBAR) |
2.1 Safety Theory and Systems Theory¶
Two fundamental types of errors occur in healthcare: SLIPS are low-level, semi-automatic behavior errors that occur during routine activities (e.g., forgetting to write an order while completing another action). MISTAKES are higher-level errors occurring in new or non-stereotypic situations requiring conscious decisions (e.g., dosing an unfamiliar medication). Prevention strategies differ for each type. Systems theory (Swiss Cheese Model) suggests that most accidents result from a series of small failures that align in a specific instance, allowing an accident to occur. Most providers are trying to deliver safe care; accidents typically result from system defects rather than individual negligence. Systems should be designed to make errors less likely AND to identify errors that inevitably occur.
2.2 Factors Increasing Error Likelihood¶
FATIGUE: House officers on 24-hour shifts are approximately one-third more likely to make errors compared to those who slept 8 hours the previous night. The 80-hour workweek addresses total hours but not extended-duty shifts, the primary cause of fatigue-related errors. STRESS AND WORKLOAD: High-pressure situations (e.g., cardiac arrests) increase error rates. Protocols and situational awareness can mitigate risk. INTERRUPTIONS: Common in healthcare and frequently lead to incomplete actions. Strategies include minimizing interruptions and using tools to define urgency. COMPLEXITY: Providers face streams of data requiring identification of clinically significant abnormalities. Decision support tools can help emphasize important findings. TRANSITIONS: Care transitions between providers and settings represent vulnerability points, especially with the 80-hour workweek. Structured information exchange tools can improve safety.
3. EPIDEMIOLOGY OF ADVERSE EVENTS¶
The frequency of adverse events in healthcare has been extensively studied, particularly in inpatient settings. The data reveal concerning rates of preventable harm across multiple care environments.
Adverse Event Categories in Hospitalized Patients¶
| Category | Percentage of All Adverse Events | Key Considerations |
|---|---|---|
| Adverse Drug Events | 19% (37% of nonoperative) | Most common; extensively studied; many prevention strategies available |
| Category | Percentage of All Adverse Events | Key Considerations |
|---|---|---|
| Wound Infections | 14% | Preventable with surgical bundles and protocols |
| Technical Complications | 13% | Often procedure-related |
| Diagnostic Mishaps | 15% of nonoperative | Cognitive errors; system failures |
| Therapeutic Mishaps | 14% of nonoperative | Treatment selection/execution errors |
| Falls | 5% of nonoperative | Environmental and patient factors |
3.1 Inpatient Setting¶
Harvard Medical Practice Study (New York): Adverse event rate of 3.7%, with 58% considered preventable. Colorado and Utah replication: Similar rates confirmed. Current estimates: The frequency of harm related to medical care now approaches ONE IN FOUR admissions. Reasons for apparent increase include improved detection techniques using 'triggers' (e.g., unexpected ICU transfer), electronic records enabling easier searching, and continued growth in care complexity.
3.2 Types of Adverse Events (Harvard Medical Practice Study)¶
ADVERSE DRUG EVENTS (ADEs): 19% of all adverse events—the most common category. Wound Infections: 14%. Technical Complications: 13%. Surgical Procedures: Associated with almost half of all adverse events. NONOPERATIVE EVENTS BREAKDOWN: ADEs 37%, Diagnostic mishaps 15%, Therapeutic mishaps 14%, Procedure-related mishaps 13%, Falls 5%.
3.3 Adverse Drug Events - Detailed Epidemiology¶
Inpatient ADEs: 6-10% of patients admitted to US hospitals experience an ADE. Studies using multiple detection methods (chart review, computerized ADE monitors) find higher rates than single-method studies, suggesting true underlying rates are higher. Outpatient ADEs: Rate of 21 ADEs per 100 patients per year. Severity generally lower than inpatient setting. Approximately one-third are preventable. Post-Discharge Period: Adverse event rate of 19% after medical service hospitalization. One-third preventable, one-third ameliorable. ADEs are the single leading error category in the post-discharge period.
4. PREVENTION STRATEGIES¶
Multiple evidence-based strategies have been developed to prevent adverse events, with particular focus on nosocomial infections and adverse drug events.
Technology-Based Medication Safety Interventions¶
| Technology | Function | Evidence of Effectiveness |
|---|---|---|
| CPOE with Decision Support | Order entry with alerts, default doses, interaction checking | 55% reduction in serious medication errors |
| Bar Code Medication Administration | Verifies right patient, medication, time | Promising early results; studies ongoing |
| Electronic Medication Administration Record | Documents medication administration | Improves tracking; supports reconciliation |
| Smart Pumps | Dose limit warnings for IV medications | Additional safety layer; prevents overdoses |
4.1 Checklists and Bundles¶
Implementation of checklists has dramatically reduced rates of: Catheter-associated bloodstream infections and Ventilator-associated pneumonia (VAP). The checklist concept is based on identifying several specific actions that reduce complication frequency; consistent application for every patient produces substantial improvements. Example: VAP bundle includes 5 measures including head of bed elevation. These practices have been successfully disseminated across wide regions (e.g., Michigan state initiative).
4.2 Computerized Physician Order Entry (CPOE)¶
CPOE with clinical decision support reduces serious medication error rates by 55%. As of 2019, over 95% of US hospitals have implemented CPOE, though decision support remains limited. CPOE CAPABILITIES: Default dose suggestions; Completeness verification (dose, route, frequency); Allergy checking; Drug-drug interaction alerts; Drug-laboratory alerts. DECISION SUPPORT FOR RENAL DOSING: Without support - appropriate dose given only one-third of the time; With support - appropriate dosing increases to approximately two-thirds; Patients with renal insufficiency discharged half a day earlier.
4.3 Bar Coding and Electronic Medication Administration Records¶
Bar coding linked with electronic medication administration records ensures the right patient gets the right medication at the right time. Electronic records simplify determination of what medications a patient has received. Early results are promising for impact on medication safety.
4.4 Smart Pumps¶
Programmable infusion pumps set according to medication and dose. Provide warnings if excessively high doses are about to be administered. Serve as an additional safety layer for IV medication delivery.
5. NATIONAL SAFETY RECOMMENDATIONS¶
Multiple organizations have developed recommendations for improving healthcare safety that all hospitals are expected to implement.
Error-Prone Abbreviations to Avoid¶
| Abbreviation | Intended Meaning | Potential Misinterpretation | Preferred Alternative |
|---|---|---|---|
| U | Units | 0 (zero), 4, or cc | Write 'units' |
| IU | International Units | IV or 10 | Write 'international units' |
| QD/QOD | Daily/Every other day | Each mistaken for other or QID | Write 'daily' or 'every other day' |
| Trailing zero (1.0) | 1 | 10 | Never use trailing zero |
| Lack of leading zero (.5) | 0.5 | 5 | Always use leading zero |
5.1 Key Organizations¶
National Quality Forum: Released recommendations on practices most likely to improve care safety. The Joint Commission: Accreditation body with safety standards. Agency for Healthcare Research and Quality: Research and implementation support. Centers for Medicare and Medicaid Services: Payment policy and quality reporting.
5.2 National Quality Forum Safe Practices¶
READBACK: All verbal orders must be recorded and immediately read back to the physician to verify accuracy. STANDARD ABBREVIATIONS: Consistent use of approved abbreviations and dose designations. Some designations are particularly error-prone (e.g., '7U' may be read as '70').
6. MEASUREMENT OF SAFETY¶
Measuring healthcare safety is challenging and expensive due to the relative rarity of adverse events.
Safety Measurement Methods Comparison¶
| Method | Sensitivity | Advantages | Limitations |
|---|---|---|---|
| Spontaneous Reporting | ~5% (1 in 20) | Simple; immediate; promotes culture | Severe underreporting; reporting bias |
| Chart Review | Moderate-High | Detailed information; context | Labor-intensive; expensive; retrospective |
| Trigger Tools | Moderate | Automated; consistent; efficient | Requires electronic systems; false positives |
| Claims Data | Low-Moderate | Large populations; available | Better for surgery; coding inaccuracies |
| Electronic Record Mining | Under evaluation | Automated; comprehensive | Not yet standardized; requires EHR |
6.1 Current Methods and Limitations¶
SPONTANEOUS REPORTING: Most hospitals rely on this approach. Sensitivity is very low—only approximately 1 in 20 ADEs reported. Important for individual event learning but inadequate for epidemiology. ELECTRONIC RECORD SEARCHING: Promising research techniques involve searching for signals suggesting adverse events. Not yet in wide use but expected to become routine. CLAIMS DATA: Used to identify adverse event frequency. Works better for surgical than medical care. Requires additional validation. TRIGGER TOOLS: Automated identification of events suggesting complications (e.g., naloxone administration suggests opioid overdose).
6.2 Provider Responsibilities¶
All providers have responsibility to report safety problems as identified. All hospitals have spontaneous reporting systems. Reported events serve as lessons for subsequent improvement. Organizational culture supporting reporting is essential.
7. QUALITY THEORY AND MEASUREMENT¶
Quality assessment tools have improved substantially, enabling more rigorous measurement of healthcare performance across multiple dimensions.
Donabedian Quality Framework with Examples¶
| Dimension | Definition | Examples | Measurement Challenges |
|---|---|---|---|
| Structure | Organizational characteristics and resources | ICU staffing ratios; EHR availability; cath lab presence | Doesn't guarantee quality care delivery |
| Process | Actions taken in delivering care | Aspirin for MI; diabetes screening; hand hygiene | May not correlate with outcomes |
| Dimension | Definition | Examples | Measurement Challenges |
|---|---|---|---|
| Outcome | Results of care | Mortality rates; readmission rates; functional status | Requires risk adjustment; attribution challenges |
7.1 Donabedian Framework¶
STRUCTURE: Characteristics applicable to a setting. Examples: Does the hospital have a catheterization laboratory? Does the clinic use an electronic health record? PROCESS: How care is delivered. Examples: Was a Pap smear performed at recommended intervals? Was aspirin given for suspected myocardial infarction? OUTCOME: What happens to patients. Examples: Mortality rate in myocardial infarction; functional status; patient satisfaction. KEY PRINCIPLE: Good structure and process do not always result in good outcomes. A patient may receive all recommended care and still have a poor outcome due to disease severity.
7.2 Continuous Quality Improvement Theory¶
Organizations should evaluate care continuously and make incremental improvements. Overall quality improves more by raising performance of all providers rather than punishing poor performers. System changes are especially effective because they affect many providers simultaneously.
7.3 Plan-Do-Check-Act (PDCA) Cycle¶
A tool for rapid cycle improvement of processes. PLAN: Identify potential improvement strategies. DO: Try out strategies in small 'tests of change'. CHECK: Measure whether strategies have made a difference. ACT: Adopt, adapt, or abandon strategies based on results. Example application: Reducing time from pneumonia diagnosis to antibiotic administration. Statistical tools such as control charts are often used to determine progress.
8. CURRENT STATE OF QUALITY IN THE US¶
Comprehensive studies reveal substantial gaps between recommended and delivered care across all domains of healthcare.
Quality of Care in the United States (RAND Study)¶
| Care Type | Percentage of Recommended Care Delivered |
|---|---|
| Preventive Care | 54% |
| Acute Care | 54% |
| Chronic Conditions | 56% |
| Overall | 55% |
8.1 RAND Corporation Study Findings¶
Patients in the United States received only 55% of recommended care overall. Breakdown by care type: Preventive care: 54%, Acute care: 54%, Chronic conditions: 56%. CONCLUSION: The chances of receiving high-quality care in the United States are little better than winning a coin flip.
8.2 Dartmouth Atlas Findings¶
Substantial regional variation in utilization and quality exists. No positive correlation between utilization and quality at the regional level (higher spending does not equal better quality). Providers with larger volumes for specific conditions (especially surgical) have better outcomes.
9. STRATEGIES FOR IMPROVING QUALITY¶
Multiple strategies exist for quality improvement at both individual and system levels, with varying degrees of evidence and effectiveness.
Strategies for Quality Improvement¶
| Strategy | Level | Effectiveness | Key Considerations |
|---|---|---|---|
| Rationing | Individual/System | Limited areas | Often resisted; compliance issues |
| Education | Individual | Short-term effective | Effect decays over time |
| Feedback | Individual | Moderate | Best if individualized and timely |
| Incentives | Individual/System | Promising | Requires adequate magnitude |
| Penalties | Individual | Produces resentment | Rarely used in healthcare |
| Reminders | System | Effective for preventive care | Requires information systems |
| Bundles | System | Highly effective | Requires high fidelity implementation |
| Team Changes | System | Most effective for chronic disease | Requires practice redesign |
| Case Managers | System | Effective for complex patients | Resource-intensive |
Chronic Care Model Components¶
| Component | Description | Examples |
|---|---|---|
| Self-Management Support | Help patients manage their own health | Goal setting; action plans; patient education |
| Delivery System Design | Reorganize care for proactive management | Planned visits; team care; case management |
| Decision Support | Integrate evidence into daily practice | Guidelines; specialist involvement; provider education |
| Clinical Information Systems | Organize data for efficient care | Registries; reminders; feedback on performance |
| Health System Organization | Create culture supporting improvement | Leadership; incentives; quality improvement strategy |
| Community Resources | Mobilize resources to meet patient needs | Community programs; patient organizations |
9.1 Individual-Level Strategies¶
RATIONING: Effective in specific areas (e.g., formulary compliance). Generally resisted by providers and patients. EDUCATION: Effective in the short run; necessary for knowledge transfer. Effect decays fairly rapidly with time. FEEDBACK: Most effective if individualized and in close temporal proximity to original events. Group feedback less effective than individual feedback. INCENTIVES: Believed to be key to quality improvement. Effectiveness depends on incentive magnitude and design. Pay-for-performance shows promise but requires adequate incentive levels. PENALTIES: Produce provider resentment. Rarely used in healthcare.
9.2 System-Level Strategies¶
REMINDERS: Electronic or paper reminders about needed actions improve performance in preventive services and chronic disease management. BUNDLES: Groups of quality measures implemented together with high fidelity. Example: Ventilator-associated pneumonia bundle with 5 measures. SCAMPs (Standardized Clinical Assessment and Management Plans): Care guidelines developed by clinicians identifying key workflow steps and decisions. Improve process outcomes through standardization.
9.3 Chronic Care Model¶
Developed by Wagner and colleagues for improving chronic disease care. Requires combination of strategies: Self-management support, Delivery system design changes, Decision support, Clinical information systems. Team care is essential—multiple providers, not just physicians. Evidence from diabetes care shows team changes and case managers are most effective strategies for improving HbA1c.
10. NATIONAL QUALITY MEASUREMENT INFRASTRUCTURE¶
Quality measurement has become standardized across US hospitals through collaborative efforts of multiple organizations.
Hospital Quality Initiative - Core Measure Sets¶
| Condition | Example Measures | Reporting Status |
|---|---|---|
| Acute Myocardial Infarction | Aspirin at arrival; beta-blocker prescribed; PCI within 90 minutes | Mandatory for most hospitals |
| Heart Failure | Discharge instructions; LVEF assessment; ACEI/ARB for LVSD | Mandatory for most hospitals |
| Pneumonia | Blood cultures before antibiotics; appropriate antibiotic selection | Mandatory for most hospitals |
| Surgical Care Improvement | Prophylactic antibiotics timing; VTE prophylaxis | Mandatory for most hospitals |
10.1 Hospital Quality Initiative¶
Collaborative involving: Hospital Quality Alliance, The Joint Commission, National Quality Forum, Agency for Healthcare Research and Quality, Centers for Medicare and Medicaid Services. Conditions measured: Myocardial infarction, Congestive heart failure, Pneumonia, Surgical infection prevention. 20 measures included in all. Data publicly released on Hospital Compare website.
10.2 Electronic Clinical Quality Measures (ECQMs)¶
Major recent change: Medicare moving from claims-based to EHR-extracted measures. Historical measurement: Primarily through claims data. Current direction: Data extracted from electronic health records. Challenge: New metrics require validation.
10.3 Trends in Performance¶
Joint Commission analyses show: Performance on measures has improved over time. Lower performers have improved more than higher performers. Substantial regional variation in quality persists. Important differences among hospitals remain.
11. PUBLIC REPORTING AND PAY-FOR-PERFORMANCE¶
Transparency through public reporting and financial incentives through pay-for-performance represent key strategies for driving quality improvement.
Pay-for-Performance Programs Comparison¶
| Program/Country | Incentive Level | Observed Effects | Concerns |
|---|---|---|---|
| US Hospital P4P | 1-2% of payment | Limited evidence; studies ongoing | May be insufficient incentive |
| UK GP Quality Framework | 40% of salary | Substantial reported improvements | Unclear if real improvement vs better reporting |
| Medicare Value-Based Purchasing | Variable | Evolving evidence | Risk adjustment challenges |
11.1 Public Reporting¶
Availability: Commercial websites have quality data for most US regions. National hospital data available (Hospital Compare). IMPACT: Patients have not made much use of data historically. Patients rely on provider reputation for choices. Data have had important effects on provider and organization behavior. ONGOING CHALLENGES: Many metrics can be 'gamed'. EHRs nearly universal but most metrics from inaccurate claims data. More EHR-derived metrics needed.
11.2 Pay-for-Performance (P4P)¶
THEORY: Providers paid more for higher-quality care will invest in improvement strategies. KEY ISSUES: Effectiveness evidence limited; studies ongoing. Incentive levels: US typically 1-2% of payment. UK General Practitioners: 40% of salary at risk based on performance—associated with substantial improvements. Potential for perverse consequences. CONCERNS: Sickest patients may be transferred to avoid poor outcomes. Incentivized measures may receive disproportionate attention. Gaming of metrics may occur.
12. SPECIAL CONSIDERATIONS¶
Several special situations require particular attention in safety and quality improvement efforts.
12.1 Transitions of Care¶
The period immediately after hospital discharge is very risky: Adverse event rate of 19% in one study. One-third preventable, one-third ameliorable. ADEs are the leading error category. Structured handoff processes and follow-up systems are essential.
12.2 Outpatient vs Inpatient Settings¶
Much more known about inpatient safety epidemiology. Limited data available for outpatient settings. Outpatient ADEs: 21 per 100 patients per year. Generally lower severity than inpatient. More research needed on outpatient safety.
12.3 Culture of Safety¶
Organizational culture supporting safety reporting is essential. Leadership commitment required. Non-punitive environment encourages disclosure. Team training and simulation support safe practices.
13. KEY POINTS & CLINICAL PEARLS¶
Healthcare safety and quality improvement requires systematic approaches combining technology, team-based care, standardization, and appropriate incentives.
Summary: Evidence-Based Safety Interventions¶
| Intervention | Target | Evidence Level | Implementation Status |
|---|---|---|---|
| CPOE with CDS | Medication errors | Strong (55% reduction) | >95% US hospitals |
| Checklists for CLABSI | Central line infections | Strong (dramatic reduction) | Widely implemented |
| Checklists for VAP | Ventilator pneumonia | Strong | Widely implemented |
| Bar coding | Medication errors | Promising | Increasing adoption |
| Structured handoffs | Transition errors | Strong | Variable implementation |
| Smart pumps | IV medication errors | Moderate | Common in ICUs |
| Work hour limits | Fatigue-related errors | Moderate | Mandated (ACGME) |
13.1 Safety Key Points¶
- Nearly one in four hospitalized patients experiences harm—safety improvement is urgent. 2. Most errors result from system failures, not individual negligence—design better systems. 3. Slips and mistakes require different prevention strategies. 4. CPOE with decision support reduces serious medication errors by 55%. 5. Checklists have dramatically reduced catheter infections and VAP in ICUs. 6. Spontaneous reporting captures only ~5% of adverse events—better measurement needed. 7. Transitions of care represent particularly vulnerable periods.
13.2 Quality Key Points¶
- US patients receive only 55% of recommended care—quality improvement opportunity is substantial. 2. Structure, process, and outcome measures each have value and limitations. 3. The Chronic Care Model emphasizes team-based care and patient engagement. 4. Continuous quality improvement with PDCA cycles enables rapid process improvement. 5. Pay-for-performance incentives need to be sufficient (>1-2%) to drive change. 6. Public reporting influences provider behavior even if patients don't use the data. 7. Electronic clinical quality measures from EHRs are replacing claims-based measures.
13.3 Clinical Pearls¶
- Report all safety events—each is a learning opportunity for the system.
- Use checklists consistently—they work by ensuring all steps are completed.
- The 80-hour workweek addresses total hours but not extended shifts, the primary fatigue risk.
- Structured handoffs (e.g., I-PASS) reduce errors during care transitions.
- For chronic diseases, patient self-management support is as important as clinical interventions.
- Volume-outcome relationships exist for surgical procedures—consider referral for complex cases.
- Good process doesn't guarantee good outcomes—communicate uncertainty to patients.
- Quality metrics can be 'gamed'—focus on genuine improvement, not just measurement.