Background: The U.S. Institute of Medicine estimated that over 100,000 Americans die each year because of medical errors, and many more sustain illnesses or injuries. Adherence to evidence-based best practices and reduction in unwanted variation in care would significantly improve patient safety.

Objectives: 1. To review and assess the literature on effective practices to prevent events that should never happen, such as objects left in the body after surgery, medication errors, and central line and other hospital-acquired infections. 2. To review and assess the literature on effective implementation of safety improvement projects. 3. To suggest a research agenda to support safety improvement.

Methods: We searched The Cochrane Library, major data bases, and evidence generated to support guidelines for systematic reviews on patient safety improvement. We then reviewed and critically appraised the literature we found. We also reviewed the evidence and behavior change techniques used to support effective safety improvements at Kaiser Permanente Medical Centers in Northern California.

Results: Most literature in this area consists of observational studies, including cohorts and case series that would be classified as low quality in the GRADE system. Some studies are 'action research,’ applying checklists and protocols and tracking results. Nevertheless, use of protocols and guidelines has resulted in ongoing, measurable improvements in several areas.

Conclusion: The lower quality studies, checklists and organizational interventions reduced adverse event rates substantially. Because they might deny one cohort protection from adverse events, randomized controlled trials (RCTs) may not be ethical in safety sensitive areas. Research should use other designs to prove or disprove the effectiveness of processes to protect patient safety, and focus on methods to ensure consistent application of proven procedures.