Building Bridges Between Researchers and Managers: Can it be Done? David R. Nerenz, Ph.D. Director, Center for Health Services Research Henry Ford Health System Detroit, MI

Henry Ford Health System Large, vertically integrated health care system –Hospitals –Medical Group –HMO –Nursing Homes –Ancillary Services In-house health services research center since 1980

Researchers and Managers – Different Worlds Health Services Research –Years to complete and publish typical project –Interested in universal, cause-effect relationships –p <.05 is key –Faith in universality of phenomena under study –“Management by data”, or “evidence- based management” Health Care Administration –Answers needed in days –Interested in practical, applicable solutions to problems –p <.30 is fine, if implementable and likely to yield big difference –“Yes, but can it work HERE?” –Management by vision, principles, relationships, and incentives

“All {Quality Improvement/System Change/Relevance of Research Findings} is Local” Our experience is shaped by, and dependent on, local circumstances that don’t exist in precisely the same way anywhere else. Relationships among individuals and organizations are unique – lessons learned may or may not apply elsewhere.

Researchers and Managers – Different Worlds Researcher’s approach to question of what works better – put two alternatives in an RCT and run a study (or do an observational study with propensity score analysis...) Manager’s approach to question of what works better – implement alternative that can be implemented, then work on “tweaking” to improve gradually over time and make it work better.

Significance or Effect Size? “Change is painful. The magnitude of difference {for a new approach} has to be around 25%. I won’t tear the organization apart for a 10% difference – certainly not 5%.” –William Conway, M.D., Senior Vice President, Chief Medical Officer, Henry Ford Hospital

Types of Studies that Typically Don’t Influence Managers Studies of organizational factors RCTs of “big things” – quality improvement, EMRs, incentive systems Any study whose results are expressed as beta coefficients or odds ratios

Types of Studies that Influence Managers - I Identify, describe, or quantify an important problem –Leape – medical errors –Wennberg – small area variation –McGlynn – quality of care in primary care settings

Types of Studies that Influence Managers - II Studies that develop a tool or a metric or a classification system that addresses a management problem –DRGs (hospital payment) –HEDIS (quality of care measures) –ACGs (severity/risk adjustment)

Types of Studies that Influence Managers - III Demonstration or QI projects that show that something MIGHT work –Hospitalists (Simmer et al) –Group visits –Cancer care coordination –Patient safety initiatives Some of these projects may not be research at all – will not produce generalizable knowledge as their primary aim

Examples of Researcher-Manager Collaboration Center for Health Management Research –University of Washington, University of California, HRET, 10 health care organizations, 13 other universities AHRQ’s ACTION (Accelerating Change and Transformation in Organizations and Networks) program Veterans Administration HSR Centers of Excellence

Building the Bridge to the Other Side – A Different Research Paradigm Clinician researchers – those who actually do patient care should be those who design, test, and refine health care delivery innovations. Research in clinic and inpatient unit “laboratories” – example – Mayo Clinic’s SPARC unit. Research done in organizations, not about organizations. Research on truly new things to determine whether they can work, rather than on big, already-implemented things to determine whether they do work (or did work!). Focus on managers’ problems rather than policy-makers’ problems – e.g., reducing medical errors, reducing no-shows, enhancing interpreter services, reducing inefficiency and duplication…. (industrial engineering) Small, bite-size problems rather than large, mega-problems (not, “Do EMRs enhance quality of care?”, but 100 specific questions on how to create or enhance an effect of a specific EMR system on quality of care) Explicit study attention paid to local context effects and interactions rather than use of randomization designs and regression models to eliminate them.

What We Can Do