PRELIMINARY—DO NOT CITE

Meeting Summary:

Patient Safety at the Clinical Interface

Holiday Inn O'Hare International, Rosemont, Illinois
Thursday, November 30, 2000

8:00 a.m. - 4:30 p.m.

Organized by:
the Agency for Healthcare Research and Quality
For:
the Quality Interagency Coordination Task Force

Background

In response to the Institute of Medicine’s (IOM) report, To Err is Human: Building a Safer Health System, the Quality Interagency Coordination Task Force (QuIC) endorsed the IOM’s recommendation for making a national priority of enhancing the existing knowledge base in the area of patient safety. In particular, the QuIC called for a National Summit on Medical Errors and Patient Safety Research, which was held in Washington, D.C., on September 11, 2000, to begin the process of setting a national research agenda.

During the National Summit, which had representation from diverse interest groups (e.g., consumers, state and local policymakers, purchasers, health care professionals, and researchers), the QuIC and other patient safety funding organizations heard from the users of patient safety research about their pressing needs and the specific research questions related to those needs. To ensure that the users’ input led to a coordinated private–public action plan by funders of patient safety research, representatives from 14 public- and private-sector organizations that fund patient safety research were asked review the input from the users. After the review, the representatives met to develop an integrated, coordinated, user-driven, patient-focused, system-based patient safety research agenda. Given the extent of both the users’ comments as well as of the patient safety field, the current Preliminary Research Agenda is rather broad in scope. Future discussions and meetings, such as “Patient Safety at the Clinical Interface,” have been scheduled to provide more detail for the current research framework.

“Patient Safety at the Clinical Interface” was designed to obtain feedback concerning the Preliminary Research Agenda produced from the National Summit, to articulate in greater depth some of the key issues that clinicians want to have researched so that they can improve patient safety, to explore the role of the professional societies in reducing errors, and to discuss means by which to address various implementation issues.

To reach those stated objectives, the basic format for the meeting involved two different sets of breakout sessions – with a focus on research topics as well as implementation issues – and a working lunch focused on the patient/consumer research agenda. In each of the breakout sessions, a member from one of the 14 funders of patient safety research at the National Summit served as a moderator while two other individuals with differing viewpoints served to catalyze the discussion. NOTE: This summary attempts to highlight the key issues brought up at the sessions. Many of the same issues were brought up at several sessions, and no attempt was made to remove redundancy caused by this. Nor is there an attempt to provide an overall summary of the issues raised in the various sessions. Further, no effort was made to validate the comments presented during the discussion sessions.

Morning Sessions

I. What do clinicians and managers most want to know about the epidemiology of errors to help focus their efforts to improve safety?

Moderator: Teresa Horan

Discussion panel: Lucian Leape, Tammy Lundstrom

Scribe: Judy Goldfarb

In this brainstorming session, the panel and participants were asked to focus on identifying the patterns of health care errors (i.e., the distributions and circumstances that lead to errors) and, based on this, formulate research questions that would be most valuable to clinicians. Specifically, the group considered what are the most useful ways of grouping errors: Should they be categorized by specialty, by disease classification, by root cause or contributing factors, or by some other approach? Are these groupings interrelated and should they be studied together? Identification of these patterns and related issues would lead to the formulation of the most useful questions about errors for initial research funding.

Other questions that were addressed related to the epidemiology of errors include: What are the causes and types of medical errors and close calls among different clinical specialties and settings of care (e.g., acute, ambulatory, long term care, pre-hospital care)? What are the causes and types of medical errors in different patient populations such as by age group and racial and ethnic minority or disability status? What costs are attributed to errors by type and causation within different population groups, care settings, and clinical specialties?

This breakout session met twice, and the following consolidates the major themes and unique suggestions offered by the panel representatives and the session participants, and are considered applicable across all settings of health care.

Dr. Lucian Leape (Harvard Medical School) addressed the following key points:

  • It is important to make known those changes and recommendations for safety that are established as successful (i.e., “best practices”).
  • Comparison of error rates should be avoided, as reporting is usually voluntary and thus unreliable; measures of error are woefully inadequate to the task. Though often sought by administrators and purchasers, comparison to a benchmark, as typically used, is probably a bad idea — error measurements are so imprecise as to make comparisons meaningless. Furthermore, such a comparison is often used to determine if an organization is above average (or above some benchmark) when the objective should be to continually improve safety. The goal, according to Leape, is not to be better than the “other guy”, but to be better than you were. Similarly, it is not workable to use error rates to choose a provider or hospital because the measures are inadequate. Consumers do not want to “shop” for safety or quality, Leape says. They want, and health care organizations should deliver, high quality from ALL providers. Patients do not wish to be forced to compare and choose the “best” practitioner, but to assured that their practitioner is safe.
  • The word “error” should be abandoned for “adverse event,” because the word error generally implies blame and is typically applied to individuals. The objective of reporting and investigations of accidental injuries, however, is to identify the underlying systems failures rather than assign blame to an individual. In addition, it is often not known at the time of discovery of an adverse event what “errors” have occurred. However, we are interested in “near miss” errors or “close calls,” which provide valuable information and which health care personnel can report with less concern for punishment or retribution. Audience participants agreed with this terminology.
  • The meaning of the epidemiology of adverse events must be broad. It should include not only the identification and classification of specific adverse events that are to be addressed on a priority basis, but also the causes and contributing factors of adverse events, their classification, and their pre-disposing conditions.
  • In determining what to study, it is important not to restudy what is already known. Rather, we need to learn what is known from the basic science of error prevention and then seek to relate this information to health care. That is, how does health care differ and how do we apply human factors concepts to redesign of our systems? Three areas illustrate how this might be done: 1) How does team work differ in different health care settings, what are the barriers to developing teams, and how should we train staff to promote team work? 2)How can skills and competence be assessed reliably? How do we identify marginal performance? Can we develop standardized criteria and measures of performance? 3) What needs to be changed to implement humane working hours and work loads? How can continuity be maintained with shortened shifts for house officers? What are appropriate staffing ratios for nurses?

In this breakout session, Dr. Tammy Lundstrom initiated a discussion of what is known about the epidemiology of errors and how best to collect data . Key points in her presentation focused on the Center for Disease Control’s successful National Nosocomial Infections Surveillance System (NNIS) as a model for the study of errors. The attributes of this model are described more fully in the section “How to Study,” below.

At the completion of the panelists’ presentations, the audience participants offered the following discussion points:

  • What to study? Four research areas were identified for study:
  1. Near Misses. Data on these “close calls” or near misses can be obtained more easily than that on actual, known adverse events since near misses are thought to be reported more willingly. This is believed to be so because reporting them can be accomplished in a blame-free manner, without harm or liability. Key elements in reporting near misses must include: a) a clear definition of the term, i.e., what constitutes a “near miss;” b) accurate and available data sources (e.g., Are incident reports of value?); c) non-punitive methods of reporting; d) reason(s) why the event did not occur, or, in other words, what “hero event” or success prevented the adverse event; e) the “consequence potential” if the event were to occur, in terms of the harm itself plus the cost of the harm—establishing this potential is necessary to prove that prevention of the adverse event is cost-effective at the individual and systems levels, and it is a challenge to institutional and national leadership to involve the actual providers of the care in this determination; f) what interventions for prevention have been identified and why are they or are they not used routinely; g) extensive involvement of the clinicians in studying near misses. (NOTE: Identification of a model to study near misses would be valuable. This model might include a “potential severity index” [the study of the consequences if the adverse event is not caught] multiplied by the “probability of recurrence” [ risk], which would provide a continuous causal and predictive analysis of particular classes of near misses.)
  2. Teams. Teamwork provides a “best environment” for safety. It combines the expertise of many individuals and enhances the ability to change behavior due to their additional commitment. Key elements in studying the effects of teams include: a) a study of how they work; b) a demonstration of the effectiveness of proven approaches—i.e., it is important to study the value of teams and how to observe a team functioning [e.g., consider the operating room and how observations during a procedure positively affect the team’s functioning]; c) learning from other industries outside of health care for team successes [e.g., consider industrial engineering and the study of human factors]. NOTE: The study of the team approach and how best to study teams in action can be undertaken at the institutional level as well as at the national level.
  3. Skills and Competencies of Health care Workers. Key questions for study include: a) How should these be assessed objectively among many disciplines? b) How to build these into the culture of safety? c) How to build upon what is already known? d) How to define competence? e) What are different degrees of competence and how are they measured and dealt with? f) How do different degrees of competence affect patient safety?
  4. Working Conditions. A prime area of research is the study of the health care delivery setting and how it influences behavior. Key questions for study include: a) How is the impact of working conditions on provider skills and competence best studied? b) Can we learn from the application of human factors engineering to the health care environment? c) How do we apply all that is learned from studies? d) Consider all the same questions outlined under the previous point (“Skills and Competencies”).
  • How to study? What model of patient safety research should be used?
  1. Identify successful models. A successful model should contain those attributes identified in successful surveillance models such as the National Nosocomial Infections Surveillance System (NNIS) of the Center for Disease Control. These attributes include: maintaining confidentiality and anonymity; being voluntary; providing a standard definition of numerator and denominator for error use; providing training for collection and use of data; assuring validation and risk-adjusted rates; having comparability, including by population variations; providing feedback of aggregated data, including publication of rates; serving as credible and available sources of information; providing staffing data; being evidence-based and able to serve as guidelines; having been proven through cost/benefit analyses to be effective sources of surveillance data. NOTE: Of great importance, a useful model should be active up to date in the provision of data.
  2. Consider standards and guidelines for appropriateness across all settings. It is necessary to identify sources of continuity-of-care information that can be used across various care delivery sites. Open systems create a complexity in tracking information. Key questions in the study of information sources include: a) What are the best, most consistent sources of information about errors and near misses in health care, if any exist, or how best can they be created? b) How can one get to the information? c) What is the culture for providing information about errors and near misses and how can or should that culture be changed? d) Do hotlines for reporting near misses and errors work? If so, how can this method be utilized more effectively? e) What are the best methods for disseminating information? (NOTE: A conceptual framework for tracking information that is suitable in various settings could set an institutional and national standard or guideline for gathering the necessary information.)
  • What are the challenges and barriers to performing research on patient safety?

1. What are these challenges and barriers? It is important to identify what we need to study but also, what prevents us from getting the information and then using successfully what is learned. Key questions to address this issue include: a) What attitudinal changes must occur to foster change-agent behaviors in personnel? For example, do personnel activate policies that give them permission to intervene when they observe an error in the system, such as stopping a potentially harmful surgical operation just as one would stop an assembly line with a potential quality problem in a manufacturing setting? b) Is there technology/software available that successfully guides the collection of necessary information? c) What prevents implementation of what is already a known, successful intervention?

2. How should the research be performed? Should a cohort of patients be studied across settings or should the events themselves be studied, both looking backward for the initial breakdown point and forward, to examine the consequence or outcome of the event? For example, if a wrong medication is given, one could study why the medication was given, as well as the outcome of this event.

3. How to identify the top 10 adverse events and close calls? How do we get to common data that has usefulness across all health care settings and among various populations? Would the matching of academic data with existing insurers’ experience data be helpful—i.e., what lawsuits occur for what common reasons [i.e., what is the low hanging fruit?], and would premium relief research provide information about incentives for behavioral change?

4. What are the different levels of interventions? It is important to consider interventions that work at the individual level and the systems level, and identify the interrelationships that exist between them, if any. For example, one could study team approaches in dealing with system interventions [providing a product such as a central line catheter] and with individual performance [assess the competence of the clinician in placement of the catheter, using a follow-up x-ray].

  • How do we work collaboratively to perform and benefit from patient safety research?

1. Learn the lessons from various health care disciplines. It is important to utilize what is already known in error research and safety design. A review of research and interventions across providers would provide important information. Key questions include: a) How do we get this information categorized and available for access? b) How is knowledge gained in one setting transferred across the various health care settings? c) How is resistance to change overcome and attitudinal shifts accomplished? d) What proven technologies can be tapped for expediting immediate and evidence-based successes? e) What personality types are more likely to accept change? (NOTE: It would be helpful to examine the epidemiology studies conducted in the hospital and nonhospital settings, such as in the ambulatory care setting, home health arena, and, if possible, physician offices. Further, looking at successful change agents, such as pharmacy sales representatives who change physician ordering practices may prove to be a valuable exploration.)

2. Work with partners from other disciplines. It is important to examine industrial engineering and knowledge about the study of human factors when considering how to affect patient safety through application of lessons learned. The consumer, or patient, should be an active participant in the study of patient safety research.