Pioneering Ideas

More on Rapid Learning Systems for Cancer: Amy Abernethy Featured in Health Affairs

Editor's note: Below is Dr. Amy Abernethy's full Health Affairs Blog post on rapid learning systems for cancer. For more details on Pioneer's work in the field, please reference Tuesday's post here.

Rapid Learning Cancer Care: Getting Serious About Implementation

With respect to rapid learning healthcare, it’s time to get serious about implementation.  National entities, such as the Institute of Medicine and the Robert Wood Johnson Foundation, have helped shape a growing consensus that this new model can help bridge the chasm between research and clinical care, and have convened thought leaders to define it.

As a practicing oncologist and clinical researcher, I see rapid learning healthcare as starting from patient care itself, which is provided individually, patient-by-patient. In the rapid learning healthcare vision, data that are routinely collected in patient care will feed into an ever-growing databank or set of coordinated databases.  The system “learns” by routinely and iteratively: (1) collecting data in a planned, strategic manner; (2) analyzing captured data; (3) generating evidence through retrospective analysis of existing data as well as data from prospective studies; (4) implementing new insights into subsequent clinical care; (5) evaluating outcomes of changes in clinical practice, and; (6) generating new hypotheses for investigation.

Discovery thus becomes a natural outgrowth of patient care. The care of the current patient is informed by all those with like characteristics who came before him/her, and this patient’s care and outcomes are reinvested into the data stream to inform the care of similar individuals in the future. 

Oncology is the natural place to start developing a rapid learning health care system. The flagship demonstration of rapid learning healthcare in action should happen in oncology, given cancer’s severity, threat to life, costliness, strong patient engagement, and population-wide impact.  Continuous investigation, discovery, and evidence implementation are intrinsic to cancer research.  Patient-reported outcomes have been well-studied in oncology, and are widely used in clinical practice as well as research, so reliable patient-reported data can anchor the rapid learning system to patient-centered care.  Cancer is ripe for a new model of integrated clinical/research function, and rapid learning healthcare provides just that.

Implemented in cancer care, a rapid learning system would expand the pace and magnitude of evidence generation for oncologists and our patients.  It would enable increasingly definitive analyses of the comparative effectiveness of current and future treatment options, at both the individual and population level – better matching the right intervention with the right individual at the right time.  It would facilitate and encourage system-wide learning, leveraging the experience of all cancer patients as well as that of clinical trial participants. 

Thus far, substantial progress has been made in developing the cancer-focused tools and infrastructure for a top-down rapid learning strategy in cancer. These include: the National Cancer Institute’s investment in caBIG® and its related toolbox to support an interoperable data infrastructure that accommodates the complexities of cancer research, discovery and patient care; a national track record in registries and data collection to support cancer surveillance and outcomes assessment; and, commitment across the oncology community to research and best practice as evidenced by widely implemented clinical trials programs, clinical practice guidelines, and quality initiatives.  This is overlaid by national investments in health information technology, comparative effectiveness research, and accountable care.

However, what we still lack is a national blueprint for building up the rapid learning healthcare system from the foundational level of daily clinical practice and research.  At Duke, we have begun to implement a rapid learning model in the cancer clinic using linked datasets that coordinate clinical, patient-reported, administrative, financial, clinical research, and basic science information into a pooled resource.  Iterative analyses and reinvestment of lessons learned occur within the context of what clinicians, patients, and researchers define as important questions to answer and hypotheses to test. 

In order for rapid learning healthcare to be truly embedded in clinical practice, we need to understand the nuances of implementation at point-of-care, and the tools needed by the frontline provider and patients such as clinical decision support, data capture solutions, data visualization, real-time analytics, streamlined care models, patient education, and improved communication.  At Duke, our demonstration in the clinic is intended to parallel, and ultimately link up with, national efforts to develop rapid learning healthcare – thereby providing a living laboratory to work through logistical and practical solutions.

 

The Need for a National Rapid Learning System for Cancer: Pioneer Grantee Lynn Etheridge Featured in Health Affairs

Editor's note: Dr. Lynn Etheridge is a Pioneer grantee [more of what we're funding here] working on developing "rapid learning" systems to identify the best practices and promising innovations when it comes to treating a variety of diseases. Yesterday, Etheridge and his colleagues published a special article in the Journal of Clinical Oncology, where they proposed implementing a national rapid-learning cancer strategy. Today, the Health Affairs Blog ran an excellent post by Lynn detailing the need for these RL systmes, which we have included below in-full:

Cancer is among the most complicated group of diseases to research and treat. The progress in the federal government’s “war on cancer” launched in the 1970s has been frustratingly slow.

 

A rapid-learning (RL) cancer system is now possible — using the potential of “in silico” research. Traditional health research has relied on “in vitro” and “in vivo” methods – bench science and experiments. In silico research would add large computerized registries and databases, with many millions of records, Internet-connected research networks, and high speed computers – today’s petaflop computers do a quadrillion operations per second. Such new learning capabilities have critical importance for the cancer system — huge amounts of genetic and other clinical data, on patients and tumors, could be recorded, understood and used in research, treatments, and outcomes analyses. By harnessing all of the quickly-accumulating data on cancer and patient experience, a rapid-learning cancer system could develop knowledge about the optimal treatment for each patient and promptly deliver that information to physicians and patients.

I first proposed a rapid-learning cancer system last year in Health Affairs, and National Cancer Institute (NCI) director John Niederhuber wrote in supportof the proposal. The Institute of Medicine, through its National Cancer Policy Forum, responded with a two-day workshop and written report, released earlier this month. The workshop report discusses the building blocks of a RL cancer system, including registries, grid computing, comparative effectiveness research, treatment guidelines, and decision tools. It describes prototype RL models and patient-driven learning systems, the considerable challenges that lie ahead, and a federal action agenda.

Yesterday, the Journal of Clinical Oncology published a study, based on IOM’s work.  Lead author Amy Abernethy, I, and others proposed that the Department of Health and Human Services, with committed private sector partners, implement a national rapid-learning cancer strategy.

HHS is now well positioned to launch a new cancer learning strategy. It is the major supporter of cancer research and the major payer for cancer care; it regulates cancer drugs, operates a nationwide bio-informatics grid for cancer data-sharing, and funds a national system of cancer registries, as well as national cancer genetics databases. The Obama administration and Congress have added new funds for cancer research, for comparative effectiveness studies, and for electronic health records for all Americans. Francis Collins, the National Institutes of Health director, led the Human Genome Project that provides foundational science for rapid cancer learning; Harold Varmus, nominated to be NCI director, is a Nobel-prize winning cancer genetics researcher.

It is time to make rapid progress on rapid learning.

 

Accelerating Care - 'Rapid Learning' Systems Highlighted in Journal of Clinical Oncology

Today, the Journal of Clinical Oncology (JCO) published a paper proposing a ‘rapid learning’ (RL) system for the nation to accelerate delivery of optimal cancer care and research. Basically, RL systems use large electronic health databases – representing the experience of millions of patients – to extract knowledge from the available data and accelerate research and treatments.   

The JCO article draws on key findings from a workshop held by the National Cancer Policy Forum of the Institute of Medicine on Oct. 5-6, 2009. Its authors (one of whom—Lynn Etheredge—is a Pioneer grantee), envision a cancer-focused rapid learning system that makes practical use of rapidly growing electronic health data repositories, such as electronic medical record systems, disease registries and databases, to hone in on what works best for individual cancer patients.

But the benefits of RL systems have reach well beyond the cancer community. That’s why we’re supporting work by Etheridge to explore how RL systems can produce evidence-based research to identify best practices and promising innovations, including comparative effectiveness analyses. With electronic health records and high-quality databases, more studies could be done, and they could be done more quickly.

 Lynn calls it “in silico” research, meaning that health research to date has relied on traditional “in vitro” and “in vivo” methods – bench science and experiments. “In silico” methods add large computerized databases, with millions of records & high-speed computers, and Internet-connected research networks – as another major tool for research science.  Drawing on many millions of patient records, RL systems will enable constant reassessment of interventions and outcomes that will help clinicians, researchers, policy makers and patients refine knowledge and learn what works best for whom, and when. The blog Health Affairs plans on posting more of Lynn’s thoughts on this tomorrow, and we’ll be sure to post to that here on Pioneering Ideas when that runs.

We know that policy-makers, clinicians and patients are clamoring for this kind of information power – the IOM has already underscored the urgent need to know what works in health care.  We need to accelerate progress in putting this evidence in their hands by creating a rapid-learning health care system that fuels stronger health policy and health care treatment decisions.

OpenNotes Fuels Discussion on Demand for Personalized Health Information

Earlier in the week we announced the launch of OpenNotes, a 12-month study that will evaluate the impact on both patients and physicians of sharing with patients, through online medical record portals, the observations made by physicians in their notes after each patient encounter. Over the last few days, we’ve been noticing some buzz around the announcement, primarily from the health blogger community. It’s great to see such strong support for this study, which ultimately aims to empower patients while delivering some much-needed transparency into the world of health care. Susan Frampton from Planetree summed-up the demand for OpenNotes perfectly in a post from HealthLeaders' John Commins:

Frampton says a growing number of patients—particularly baby boomers—want to be informed about the care their getting." They want to know what is in there, they want to know what the plan is and they want to know what the results are," she says. "People who don't want to see their charts just elect not to look at it if it is presented to them, and that choice needs to be presented."

Today we were thrilled to see Newsweek’s Claudia Kalb echoing those same thoughts, tying OpenNotes into her article on the growing curiosity and interest in personalized genomics:

The idea of knowing your personal genome is still pretty futuristic, but the complaints and symptoms and worries we share with our primary-care docs and the observations they make about us matter right now. And they could have a direct impact on our health habits immediately. Delbanco is passionate about getting physicians and patients to exchange this information and to collaborate: “My view of medicine is that we have a unique body of knowledge that unless you’re a doctor you don’t have. And you have a unique body of knowledge about yourself that I will never have. Our job is to get the two together as close as possible.” Which certainly seems like a simple and enlightened proposition.

As shown in those two quotes above, the concept behind OpenNotes is rooted in the growing movement for increased patient access to highly personalized medical data and information. This demand is controversial among many in health care: as Steve Downs has noted, there are many people with a patient advocacy perspective who think this is so obvious a right and there should be no question about it.  And there are many physicians who think – for very plausible reasons – that this is a terrible idea.  However, as Tom Delbanco mentions, these two approaches to health care need not be mutually exclusive. Combining the medical expertise of doctors with the “self” expertise of the patient can result in better care – and we look forward to seeing if OpenNotes can bring the two together.

 

 

 

OpenNotes Study Kicks Off, May Transform Doctor/Patient Communications

As the push for greater transparency and patient engagement in health care gains momentum, we’re supporting a new study that examines what happens when you add a new layer of openness to a traditionally one-sided element of the doctor-patient relationship – the notes that doctors record during and after patients’ visits. Beginning today, the OpenNotes project will evaluate the impact on both patients and physicians of sharing, through online medical record portals, the observations made by physicians after each patient encounter.

More than 100 primary care physicians and 25,000 patients across three sites – Beth Israel Deaconess Medical Center (BIDMC) in Boston, Geisinger Health System in Pennsylvania and Harborview Medical Center in Seattle – will participate in the 12-month trial. The study is led by primary care physician Tom Delbanco, MD and Jan Walker, RN, MBA, both of whom are on staff at BIDMC and on the faculty of Harvard Medical School.

Doctors have mixed opinions about opening up the notes they record to patients. However, according to RWJF Assistant Vice President Steve Downs, this subtle change could “reposition notes to be for the patient instead of about the patient, which might have a powerful impact on the doctor-patient relationship and, in the long run, lead to better care.”

We’re interested in hearing your thoughts – both from doctors and patients – on how this type of new access could transform the way medical care is managed.

What Inning Are We In?

I was at the Games for Health meeting in Boston last week. This was the fourth year Pioneer has supported the meeting, which has come a long way since its inception.  When I first attended, most of the conversation I heard was an effort by gamers and health practitioners to each understand the other.  From one side, you heard questions that asked, essentially, “What makes a good game?” From the other side, you heard questions that asked, essentially, “Help me understand diseases, therapies, and how health care works. 

 

And from both sides, you heard, “When you say X, what exactly do you mean?”

The conversation this year was significantly different.  Instead of talking to each other, people were talking with each other, trying to figure out how to solve problems. Attendees were frequently working off a common language, though some are more fluent than others.

Given that much of the conversation has moved from discovery to collaboration, it has me wondering what’s needed now to move the field along?  The funding we provided under ourHealth Games Research national program focused on establishing efficacy and exploring game design principles.  Does the field need more of that?  Some of the ideas I heard at the Games for Health conference of what was needed now included research to demonstrate cost-effectiveness and the establishment of a journal devoted to the field of health games research.

Any opinions?