A23/B23: Patient Harm in US Hospitals: How Much? 23rd Annual National Forum on Quality Improvement in Health Care December 6, 2011 Objectives Summarize the findings of three recent studies measuring adverse events with the IHI Global Trigger Tool Describe the methods used in these studies Summarize the conclusions from each of the three studies
Faculty & Disclosures Fran Griffin, RRT, MPA Employed by BD David Classen, MD, MS Employed by CSC Lee Adler, DO Lead Physician in the OIG Studies on National Incidence of Harm in the Medicare Population Paul Sharek, MD, MPH Nothing to disclose 2011 The Institute for Healthcare Improvement Session Outline Brief description of IHI Global Trigger Tool Papers: Health Affairs / IHI Study NEJM / IHI- NC study DHSS OIG study Summary and implications Q&A Fran Griffin Fran Griffin Paul Sharek Lee Adler David Classen ALL
IHI Global Trigger Tool & Health Affairs Paper Fran Griffin IHI Global Trigger Tool Retrospective review of closed patient records Check for triggers or clues to harm Examples: transfusions, Benadryl, Narcan Count all unintended consequences of medical care Focus on events of commission not omission Faster than reading records Uses sampling for measure over time
Process Random selection of records Review using trigger tool process by 2 independent mid-level reviewers (clinical, nonphysician) Consensus reviewed by physician Determine harm from patient s viewpoint without regard for preventability Assign level of harm to each individual event Categories of Harm (from NCC MERP Index) E - F - G - H - I - Temporary harm, intervention required Temporary harm, initial or prolonged hospitalization Permanent patient harm Life sustaining intervention required Contributing to death
Health Affairs: April 2011 Classen DC, Resar R, Griffin F, et al. Global Trigger Tool shows that adverse events in hospitals may be ten times greater than previously measured. Health Affairs. 2011 Apr;30(4):581-589 Study Goal & Design GOAL Evaluate the overall performance sensitivity and specificity, as well as the positive and negative predictive value, of three detection methods: IHI Global Trigger Tool Voluntary Reporting AHRQ Patient Safety Indicators DESIGN Reviews of data from three tertiary care centers with well-established (criteria based) patient safety programs
Review Methodology Inter-rater reliability established prior to study Random samples obtained by hospitals with specific directions adult (age on admission > 18) inpatients (length of stay > 24 hours) admitted during period October 2004 Each record reviewed by 1 primary reviewer Two physicians reviewed findings of each and came to consensus Local review team at one hospital (A) for detailed analysis Population & Total Amount of Harm 3 tertiary care hospitals in US 795 records from Oct 2003 reviewed 393 adverse events total 33% of admissions 49 / 100 admissions 91 adverse events / 1000 patient days Classen DC, Resar R, Griffin F, et al. Global Trigger Tool shows that adverse events in hospitals may be ten times greater than previously measured. Health Affairs. 2011 Apr;30(4):581-589.
Types of Harm Classen DC, Resar R, Griffin F, et al. Global Trigger Tool shows that adverse events in hospitals may be ten times greater than previously measured. Health Affairs. 2011 Apr;30(4):581-589. Methods of Detection Classen DC, Resar R, Griffin F, et al. Global Trigger Tool shows that adverse events in hospitals may be ten times greater than previously measured. Health Affairs. 2011 Apr;30(4):581-589.
Analysis 393 Adverse Events GTT found 354 (90%) AHRQ found 35 (9%) Voluntary reports found 4 (1%) Sensitivity & Specificity GTT both high AHRQ PSI s: very low sensitivity & high specificity Voluntary reporting: very insensitive Three Key Points from this Study 10 years post the IOM s report To Err is Human, rates of adverse events in hospital patients are still high. Our study (GTT) detected far more adverse events in hospitalized patients than have been found in prior studies. Voluntary reporting & AHRQ PSI s fail to detect most adverse events.
Establishing a National Harm Study Methodology: The North Carolina Patient Safety Study Paul Sharek, MD, MPH Associate Professor of Pediatrics, Stanford University Medical Director, Center for Quality and Clinical Effectiveness and Chief Clinical Patient Safety Officer, Lucile Packard Children s Hospital Study Goals and Design No information available to determine if harm is changing over time Pilot study in 10 randomly selected North Carolina adult hospitals to inform a future national harm study Questions to be answered Operator characteristics of IHI Global Trigger Tool? Who performs better: Internal vs External reviewers? How Internal/External compare to a gold standard? Is there change in harm rates over time in North Carolina (2002-2007)? 18
Population Hospital sampling: 10 hospitals (North Carolina) Stratify: Rural vs Urban, Teaching vs Not, Large vs Small Sampling: 10 records per quarter, randomly selected, over 24 quarters (2002-2007)= total 240/hospital Eligibility: Inpatient for 24+ hours Age 18+ on admission Not admitted for Psychiatric diagnosis Rehabilitation Hospice Discharge date: 1/1/2002 thru 12/31/2007 Methodology-Internal/External Review IHI white paper method: 2 mid-level reviewers, consensus developed Cases with events only presented to physician reviewer for final determination Our method*: 1 primary reviewer (non MD, usually RN) Review charts When identify potential event, summarize in event form Present to 2 MD reviewers independently (verbally) MD reviewer completes event form MD reviewers (both) come to consensus on event *Consistent with Resar, Classen, James, Griffin 2009 GTT study
Harm Determination Critical Recommendations from SAG NCCMERP harm scale Include Level E harm Exclude harm from errors of omission Include harm noted at the time admission (to get a sense of the frequency of this) Assess preventability of event Likert scale (with scores ranging from 1 for definitely not preventable to 4 for definitely preventable ) Score of 3 or 4 = preventable Physician reviewers with final determination Harm Determination Note differences from IHI method as to what events were counted as harm omission vs. commission and whether preventability was assessed and how
Results Results
Analysis In a study of 10 North Carolina hospitals, we found that harms remain common, with little evidence of widespread improvement. Three Key Points from this Study 1. No evidence of change in total or preventable harm rates over time (2002-2007) Base model: internal or external reviewers Multivariate model: internal or external reviewers No evidence of any specific hospital with change over time 2. IHI GTT is a useful instrument for internal reviewers to use within hospitals to track harms over time. 3. Sensitivity and specificity of the IHI GTT are very good, particularly compared to other methods presently used to identify harm
Three Key Points from this Study Let s each note 3 (and only 3) key points from each study Adverse Events in Hospitalized Medicare Patients Reported to Congress by the OIG
Study Goal Objectives o Capture incidence of patient harm o Assess preventability of adverse events o Estimate costs to Medicare Congressional Mandates o Deficit Reduction Act of 2005 hospitalacquired conditions o Tax Relief and Health Care Act of 2006 Sample for National Incidence Study October 2008 999,645 780 Sample month Medicare beneficiaries discharged from acute care hospitals Sample Medicare beneficiaries 661 Hospitals represented
Sample Medicare Patient Statistics 59% 84% 5% 60% Female 65 years and older Stay lasting longer than 14 days (ALOS = 5.2) Entered from Emergency Department IHI vs. OIG - Global Trigger Tool (GTT) IHI METHOD Clinical reviewer Physician serves as authenticator/arbitrator Excludes omission Includes preadmission events Identifies adverse events OIG METHOD Nurse reviewer Physicians reviews every flagged record & discuss Includes omission Excludes preadmission events Screening for adverse events
Incidence Rates of all beneficiaries 13.5% 0.6% 1.0% 13.5% Adverse Events (NQF, HAC, F I Level) NQF Serious Reportable Events Medicare Hospital-Acquired Conditions Temporary Harm Events (E Level) Percent of Adverse Events by Category CLINICAL CATEGORY % of EVENTS Medication 31% Patient Care 28% Surgical/Procedural 26% Infections 15%
Preventability Results all events PREVENTABILITY % of EVENTS Preventable 44% Not preventable 51% Unable to determine 5% % Preventable by Clinical Category CLINICAL CATEGORY % of EVENTS Infection 60% Medication 50% Patient care 51% Surgery and other procedures 17%
Cost of Events to Medicare $324M 3.5% $4.4B 2/3 rds Cost to Medicare for sample month % of total Medicare inpatient expenditures Cost to Medicare extrapolation for full year Result of additional hospital stays Medicare Inpatient Financial Impact 84% $5,601 $13,745 $0 Of events did not raise Medicare payment Average cost of event during initial hospital stay Average cost of event-related additional stay Cost reductions from Medicare HAC policy
OIG Study Takeaways Events are common Current methods to identify and measure events is lacking It is possible to assess preventability Reducing events is key to reducing costs OIG Recommendations OIG Call to Action Broaden the definition of adverse events Enhance efforts to identify events Provide incentives for hospitals to improve care DHHS Agency Responses AHRQ funding research CMS National Goal set by Don Berwick to reduce patient harm by 40%, CMS Center of Innovation, VBP, ACO Pilots
Summary & Implications David Classen Results Across Studies IHI GTT OIG NC Harm % Harm 33% 28% 18% Per 100 admissions Sample differences 49 36 25 795 patients Ages 18+ October 2004 3 Tertiary care hospitals high case mix index 780 patients Medicare only October 2008 Multiple hospitals & types (random sample of beneficiaries) POA excluded 2341 patients Ages 18+ Jan 02 Dec 07 10 hospitals, various types
Harm Distribution 60% 50% 40% 30% IHI OIG NC 20% 10% 0% E F G H I Adverse Events: What We Are Learning What We Know Adverse Events in the Hospital Increase risk of death and risk of more adverse events Increase LOS, Costs Adverse Events in the Hospital Increase Risk of Readmission with 30 days Increase Use of Outpatient services over 6 months Increase mortality over 18 months What We Will Know Predictors of Adverse Events Predict, Prevent, Mitigate Real time indicators
Automation of Triggers Many triggers accessible from databases in US hospitals: speeds up review time Pharmacy billing data (administered) Laboratory information systems Coding Systems Electronic Medical Records (EMR) option Move towards automated reviews and real-time intervention Several Organizations Beginning this Journey Early experience suggest this is a significant technical and organizational challenge E Triggers at LDS Hospital Adverse Drug Event Report 02/26/2000 02/26/2000.0600 Loperamide 02/26/2000.0350 RR < 8 b/m Doe, Jane F 67yo F W635 MR#20022 DOC: 211 Doe, Scott R. Admit: 02/12/2000.1012 Admit Dx: Pneumonia Allergies: Sulfas, Peanuts Current Drugs: 02/26/2000.0600 Loperamide 2mg oral as directed 02/21/2000.0400 Ceftriaxone 1gm IV every 24 hours 02/25/2000.2315 Morphine PCA 30mg/30cc IV if needed 02/15/2000.1113 Albuterol 5mg/ml inhalation as directed 02/12/2000.1830 Famotidine 20mg oral twice a day Discontinued Drugs: 02/20/2000.0900 Azithromycin 250mg IV every 24 hours
Sepsis Patient Identification in Real Time Where Can We Get Started? Secure financial and risk management leadership in building the business case unlike anything we've seen in patient safety and quality to date, this business case is substantial Develop incrementally This is a difficult journey, and requires a significant investment and expertise Build a plan through assessment and prioritization
Questions