In 2011, the Institute of Medicine's (IOM) Roundtable on Value & Science-Driven Health Care set an ambitious goal: By 2020, 90% of clinical decisions will be supported by accurate, timely, and up-to-date clinical information and will reflect the best available evidence.1 As articulated in a national workshop organized by the IOM Roundtable, a key element for reaching that goal is the designing of a health care system that embeds real-time learning for continuous improvement in the quality, safety, and effectiveness of care, while generating new knowledge and evidence about what works best.1
In the IOM workshop, Don Berwick, former director of the Centers of the US Medicare and Medicaid Services (CMS), spoke about the powerful benefit to the health care system of deepening systems knowledge and taking action based on that knowledge, a theme advocated previously by health care leaders such as neurosurgeon Atul Gawande.2 Berwick argued that one of the major challenges to systems thinking is to recognize the value of dynamic learning and local adaptation as scientific learning progresses. By “dynamic learning,” Berwick was referring to learning that is achieved through the application of what he calls “Plan-Do-Study-Act” cycles, a form of inquiry that capitalizes on processes and knowledge growth in a nonlinear fashion. It is an approach to continuous quality improvement in medicine that Berwick popularized through his Institute for Healthcare Improvement.3
Berwick advocated that dynamic systems learning will require a reexamination of the traditional hierarchy of scientific evidence as the basis for evaluating clinical practices. He contended that placing randomized controlled trials (RCTs) at the top of the scientific hierarchy—as typically done when considering against other forms of traditional inquiry—does not acknowledge the fact that most learning in complex systems occurs in local and individual settings using nonlinear techniques such as the Plan-Do-Study-Act cycles. Berwick argued that the chasm between formal clinical trials and local improvement strategies is enormous, the cost to systems knowledge growth in health care is very high, and scientific journals need to open their review process and pages to systems science and the knowledge it produces.
Numerous speakers at the workshop promoted a far more pragmatic approach to systems science, such as the approach used in the engineering field, which harvests the knowledge that accumulates through innovation in local settings. Throughout the workshop, participants discussed the many ways in which the US health care system is substantially underperforming and the significant opportunities for systems to learn and develop into systems that yield the highest value for their patients.
There are encouraging signs that our colleagues in medicine are moving in the direction of systems science and are taking significant steps toward deepening systems knowledge in medicine and surgery through efforts such as the High Value Healthcare Collaborative (HVHC), a consortium of 17 health care delivery systems and The Dartmouth Institute for Health Policy and Clinical Practice.4 The mission of the HVHC is to improve health care quality and outcomes over costs, across time, for its service population, while serving as a model for larger-scale health care reform. The HVHC follows a systems science approach advocated by Gawande (among others), which includes:
Measure: Define, test, and disseminate advanced measures and tools to support clinicians, health care systems, and payers in their efforts to deliver high-value care.
Innovate: Identify, test, and rapidly disseminate best-practice care models and payment models that are safe, improve care, have better outcomes, and reduce costs.
Replicate: By establishing a collaborative “learning network,” encourage membership by other organizations, help implement best practices in new member organizations (and learn from them), and distribute findings publicly so that they can be more broadly considered for implementation.
One of the HVHC's first initiatives was a study of variation in total knee replacement (TKR) delivery across 5 institutions located across the United States.5 Among the major findings: the health care system with the lowest in-hospital complication rate had brought together patients with a multispecialty team prior to the surgery, including members from anesthesiology and internal medicine, to co-manage patients who are medically complex. In addition, the fastest operating times (and shortest patient stays) were at a hospital where patients with TKR were served by a team of anesthesia doctors, scrub techs, and nurses specializing in arthroplasty. And the health care system that involved patients prior to surgery in their discharge planning process (and managed patient expectations about disposition after hospitalization) had shorter hospitalizations. Given the HVHC focus on the acute care and surgical processes, it is not surprising that they did not focus on what happened after hospital discharge following a TKR. Interestingly, they did report, but did not investigate, the effects of the enormous variation in hospital discharge post-TKR: discharge home with “self-care” ranged from 88.6% in one institution to 5% in another, whereas the use of inpatient rehabilitation hospital stays ranged from 1.1% to 13.1% across these 5 institutions.
How can physical therapy and the rehabilitation field at large move toward the goal where 90% of our clinical decisions are supported by accurate, timely, and up-to-date clinical information and reflect the best available evidence? I believe that relying on RCTs will not get us where we need to be by 2020. In my opinion, the urgent need is for the creation of a High Value Rehabilitation Care Collaborative (HVRC), which has as its core mission taking a pragmatic approach to rehabilitation systems science that will accumulate and disseminate new systems knowledge through innovation in local post–acute care and rehabilitation settings. As with the HVHC, creating the HVRC would require a small group of innovative institutions (ie, positive deviants) to join together around a common rehabilitation condition or shared set of clinical questions. For example, an HVRC could focus on a question regarding the variation in post–acute care services and settings following TKR or on a broader question, such as the role of post–acute care and rehabilitation services in altering the risk of hospital readmission within 30 days of discharge.
The challenges to forming an HVRC are considerable and beyond the scope of one editorial to discuss. A fundamental requirement is that HVRC members would need to agree on the use of a common set of advanced process and outcome measures so that critical data elements can be aggregated and analyzed to address each question of interest. Fortunately, the creation of entities such as the Physical Therapy Outcomes Registry6 may make it more feasible today to implement a common set of measures across institutions to allow an HVRC to take root and begin to apply systems science in rehabilitation.
In future editorials, I will discuss the challenge of how we disseminate practice innovations once they are achieved, so that the results of real-time learning regarding the quality, safety, and effectiveness of rehabilitation care can be adopted in a widespread manner across the US health care system.
- © 2016 American Physical Therapy Association