Everywhere we turn these days it seems “Big Data” is being touted as a solution for physicians and physician groups who want to participate in Accountable Care Organizations, (ACOs) and/or accountable care-like contracts with payers.
We disagree, and think the accumulated experience about what works and what doesn’t work for care management suggests that a “Small Data” approach might be good enough for many medical groups, while being more immediately implementable and a lot less costly. We’re not convinced, in other words, that the problem for ACOs is a scarcity of data or second rate analytics. Rather, the problem is that we are not taking advantage of, and using more intelligently, the data and analytics already in place, or nearly in place.
For those of you who are interested in the concept of Big Data, Steve Lohr recently wrote a good overview in his column in the New York Times, in which he said:
“Big Data is a shorthand label that typically means applying the tools of artificial intelligence, like machine learning, to vast new troves of data beyond that captured in standard databases. The new data sources include Web-browsing data trails, social network communications, sensor data and surveillance data.”
Applied to health care and ACOs, the proponents of Big Data suggest that some version of IBM’s now-famous Watson, teamed up with arrays of sensors and a very large clinical data repository containing virtually every known fact about all of the patients seen by the medical group, is a needed investment. Of course, many of these data are not currently available in structured, that is computable, format. So one of the costly requirements that Big Data may impose on us results from the need to convert large amounts of unstructured or poorly structured data to structured data. But when that is accomplished, so advocates tell us, Big Data is not only good for quality care, but is “absolutely essential” for attaining the cost efficiency needed by doctors and nurses to have a positive and money-making experience with accountable care shared-savings, gain-share, or risk contracts.
The promotional literature for Big Data is peppered with jargon and catch phrases — “close to the point of care,” “synthesizing large amounts of information,” “transformational analytics,” and so on — that promise to “de-fragment” the current health care environment and offer predictive insights that the doctors, nurses, and patients do not now possess.
This all may be true. But why wait for Big Data to be put in place, when what we’ll call “Small Data” is already available and can offer information and analytical insights sufficient to get a good start on care management programs capable of improving quality and reducing some unnecessary costs?
What we’re recommending is that physician groups take stock of their existing sources of data and information, look for ways to inexpensively extend the value of their current EHR investments for analytical resources, and learn from the literature and their peers about what relatively low tech, high touch, data-driven interventions have been successful in care management. Before you invest in Big Data systems, see what Small Data systems can accomplish, and then plan accordingly to fill the gaps.
Let’s take as our sample scenario a Patient-Centered Medical Home, PCMH, with 5-10 physicians, or a multi-specialty physician group with several PCMH practices as the core of its primary care capability, and look at what they can do to collect and analyze data to manage care more appropriately. A focus on this type and size of group makes sense because they are the primary care core upon which any Accountable Care Organization will depend for care management.
Focus on Your High Risk Patients — You Know Who They Are
One of the things that we know about care management and coordination is that the opportunities for quality improvements and costs savings are concentrated in the small number of patients who are sickest and at highest risk for exacerbations that lead to hospitalizations, extra tests, specialty referrals, ER visits, and so on. Most PCMH practices know who these individuals and families are, simply because they take care of them on a regular basis and understand their medical, social, and economic situations better than anyone else. Particularly when participants in care teams in the outpatient clinic setting are in regular contact with those in the inpatient, extended care, and home care settings, awareness of the presence of high risk patients undergoing crises and transitions of care is more a matter of a commitment to making lists, communicating with peers, and then to coordination of care, than it is a function of data analytics.
In well managed PCMH and multi-specialty group practices, the providers know their patients and have established long-term relationships with many of the most frail or high risk individuals within the practice. Making all the team members more aware of who these people are and when they are undergoing transitions of care is a prerequisite for finding ways to better coordinate the care.
Establish Registries for Specific Conditions and High Risk Patients
One of the most commonly mentioned failings of the modern crop of EHRs is their inability to provide population-level information to help practices guide their care management efforts, for example by identifying the high risk individuals within a population of patients with diabetes. While we acknowledge that this is true in a general sense, it is also the case that the remedy is not necessarily a rip-and-replace strategy in which expensive EHRs from vendors who have established Big Data solutions are installed anew. Rather, we suggest that most PCMHs and multi-specialty group practices look for less expensive and more practical ways to extend the usefulness of their EHRs to feed data to registries, that is, to databases that aggregate specific kinds of information on specific populations of patients, either by diagnosis/condition, or by risk factors that are likely to impact their likelihood of benefit from care management services.
Most multi-specialty group practices, and a growing number of PCMHs, now have EHRs that have been certified by ONC for Stage 1 Meaningful Use, which means these systems have been certified as capable of exporting patient specific data in structured formats, either the Continuity of Care Record, CCR, or the Continuity of Care Document, CCD CDA, for a small but relatively important amount of clinical data. (In Stage 2 Meaningful Use starting in 2014, both of these will likely be replaced with the consolidated CDA.) These data include problem lists, current medications and dosages, allergies, recent laboratory test results, allergies, and surgical procedures. Provided they have been entered into the EHR in a reliable and consistent manner by physicians, nurses, and other staff members in the group practice, it is becoming easier and less costly to extract these data from the EHR’s database and aggregate them in a registry or small clinical data repository for analysis and for asking questions that can direct quality improvement efforts aimed at particular groups or classes of patients. Simple data mining and other small data examples from the field are easily found and include:
- Identifying patients who carry diagnoses of both chronic of pulmonary disease, COPD, and heart failure, for specialized nurse case management. (Keystone Beacon Community, Central Pennsylvania.)
- Reporting physician-specific patient benchmarks for patients with diabetes, including percentages of patients meeting targets for HgA1C, LDL, and blood pressure. (Holston Medical Group, Tennessee.)
- Finding patients with frequent ER visits, and assigning them care managers. (Colorado Beacon Community.)
- Recognizing preventive care gaps in care for diabetic patients at the time of presentation to the clinic for both routine and emergency/unscheduled visits, and assigning a care manager to assure the gaps are filled. (We-Care onsite clinics.)
Tracking for non-reimbursable outpatient post-operative visit codes, which, if not supplied by the surgeon may indicate important post-op visits were skipped, and following up on those patients by case managers. (University of Rochester.)
Partner with Payers for Data and Analytics
Health plans are increasingly recognizing the need to work more closely with physicians. The logic here is straightforward — even in a reformed health system, it is still the physician’s pen (or iPad) that will control over 70% of health care costs.
Health plan business models have been turned inside out by health reform legislation. They are becoming increasingly aware that it makes sense to enable and support physicians, particularly as payment models change to align interests under some form of accountable care arrangement. Payers are recognizing that gathering and analyzing both claims data and clinical data is critical to improving care and lowering costs — and that they need to share this information with providers who are in the position to use the information to improve patient care.
Here are some examples:
- Anthem Blue Cross Blue Shield took three years of claims data and downloaded it to the respective databases of 5 physician groups.
- Aetna will implement its broad technology stack and care management services to support Banner Health doctors and their patients. The agreement includes:
·Health information exchange technology
·Point-of-care clinical decision support services and a desktop-based workflow tool to track, monitor, coordinate and report on patient health outcomes; and
·Smartphone and online appointment setting and pre-registration services for patients
- WellPoint and Aetna recently announced plans to significantly increase their fee schedules for primary care practices that abide by principles of the PCMH, such as by offering increased provider access and availability and coordinating patient care.
A few caveats here. First, health plans are in varying stages of gaining enlightenment and providing support to physicians — your mileage may vary.
Second, health plans are engaging mostly with larger physician groups — so far. In researching this essay, we found dozens of examples of health plans partnering with physicians groups under the labels of ACO, PCMH, or similar. Today’s reality is that most of these partnerships are with larger physicians groups — it’s simply more efficient for a health plan to deal with one group of 100 physicians than 10 groups of 10 physicians.
However, we anticipate that health plan engagement with physicians quickly will spread to include smaller groups. Why? Because that’s where the physicians are: data from the American Medical Association(2008) showed that 59% of physicians are practicing in groups of 9 or fewer physicians.
Focus on Transitions of Care
One of the things that has been learned through numerous care management and care coordination pilots and demonstration programs is that success in saving money is related to “high touch” approaches in which more time is spent in face-to-face time with patients, and when these interventions occur at moments that are “just right” for the patients, usually prior to or during transitions. Making this happen in any PCMH or PCMH-neighborhood does not require the analytics of Watson, but it may require the intuition, clinical judgement, and good common sense of providers to act when they are needed, along with a determination to keep lists of high risk patients in front of care givers on a regular basis through group meetings, e-mail communications, and other social networking at the practice level. We are particularly hopeful that Direct message exchange, a product of the Direct Project, will make it easier for care givers to communicate via secure e-mail across organizational and vendor technology boundaries as EHRs become Direct-enabled and Direct-compliant over the next couple of years..
As a concrete step, research has shown that using care coordinators (typically nurses) is one of the most effective ways to coordinate care for high risk patients and to assure smooth transitions of care. Again, your health plan might be willing to help — many are experimenting by increasing payments for primary care or by adding separate care coordination payments. Some are even directly placing care coordinators in physician offices at the health plans expense.
While Big Data holds distant promises, don’t wait. Use the Small Data you have today and your knowledge of your patients to jump start effective care management programs.
David C. Kibbe, MD, MBA, is a Family Physician and Senior Advisor to the American Academy of Family Physicians who consults on healthcare professional and consumer technologies. Vince Kuraitis JD, MBA, is a health care consultant and primary author of the e-CareManagement blog, where this post first appeared.
I enjoyed this post. I agree, too often, valuable data from small-scale studies are overlooked. A good synthesis of findings from “small data” can be a less expensive way to meet research needs. Also, it can increase confidence in results and focus resources for “big data” analyses where they can do the most good.
I like what you guys are up too. This kind of clever
work and coverage! Keep up the excellent works guys I’ve included you guys to my personal blogroll.
Good article, realistic aspirations, and an attainable goal.
Big Data ain’t gonna provide guidance for individualizing the care of Big Momma when she comes in with her pulmonary embolus. Small data probably won’t either.
Big Momma needs Big CT Scanner, Big Bed, Big Gowns, Big Doses, and Big Believers that MU and Big Data is a Big Expense with Big Nothing for most patients, big or small
Curly, what will provide guidance? There was an interesting editorial in NEJM discussing whether IABP was useful in cardiogenic shock – it sounds to me like there’s a huge need for better big data analysis for Big Momma when she comes in with a PE. Are you saying there’s nothing more to learn? Or are you saying that we have enough data already? I respectfully suggest we would be foolish to say we don’t need to explore the possibility that there are new ways of learning from the data we already have but aren’t utilizing for learning better ways to solve difficult problems.
I agree that there isn’t enough done with small data, and that’s a good place to start to gain expertise in dealing with bigger data as it becomes available. I hope you’re not saying claims data is adequate or even desirable. I think as patients look at their claims data they will find enough errors to make the accuracy of such data questionable. I’m not criticizing – the important data is what’s used for clinical purposes, not what’s used for billing. As I recently saw a billing clerk quoted as saying, “It all goes in the same pot.” And likely it’s primarily unintentional. At least that’s my guess. But I don’t think it’s accurate enough to base any conclusions on it. With genomics coming, I think new capabilities are going to be developed to take advantage of this new data that won’t be possible without big data technology.
Jordan Shlain is a concierge internist in San Francisco who founded HealthLoop to help doctors keep track of the sickest patients and to figure out which ones need to be checked in on with a phone call and when. It’s a great idea that started simply with a spreadsheet that Shlain kept for himself.
My own doctor does a great job with a pen and paper. He’s in NJ and I now live in California. I’m not a high risk patient, but he likes to call all of his patients within two months after seeing them to make sure everything’s going ok. During my appointment I watched him write in his notebook to make sure to call me on PST time, and sure enough the other day he did!
Thank you for making a clear statement of the obvious: big data is only “better” when it leads to material benefits to patients and payors. In the absence of “big impact,” big data is – today – a resource intensive distraction for most of the healthcare establishment.
To your point, “small” data when appropriately provided, particularly in the context of predictable life/health transitions, has transformative power today. It’s encouraging to hear the viewpoint espoused!
Great article. It really does pain me to see some provider organizations buying massive IT analytics projects and consulting services for many healthcare reform initiatives and requirements particularly at the stages of development they find themselves. Many organization are starting to capture a lot of new data through their EMR systems trying to get up to speed with meaningful use but “useful” analysis and integration into the clinical workflow are sorely lacking (sometimes just being warehoused with only cursory analysis being undertaken). When all you really need to accomplish the first stage of this whole shebang for most organizations would be something like OpenEMR, R, and Some Open Source GUIs to work with R to automate a lot reporting to different departments, that would seem like something almost any organization on any budget could accomplish. Once many organizations can do this type of rapid cycle iterative feedback process then yes get some “big data” in there for real time point of care analytics. Perhaps it is better to start with the basics and work up to “big data” systems though for most.
We have also been big proponents of the power of small data. We find that it is generally more accurate and high value because it is collected very close to the source. You can read more here: http://captricity.com/the-small-data-revolution/