Reinventing Electronic Support for Physicians: A Look at Kaiser Permanente's Patient Panel Support Tool and Care Delivery: Part 1
EXECUTIVE SUMMARY:
A Web-based tool that extracts information from the electronic health record helps physicians improve care and manage their entire panel of patients. Two Kaiser Permanente studies examine the effectiveness of the tool in a large, diverse patient population.
In October, the Oakland, Calif.-based Kaiser Permanente integrated health system released the first large studies of its proprietary Panel Support Tool. That tool, first rolled out in 2006, uses Web-based software that extracts information from Kaiser's HealthConnect electronic medical record (EMR) to help physicians improve and manage their patients' care. Kaiser officials say the two studies are the first to examine the effectiveness of the Panel Support Tool in a large, diverse patient population.
Kaiser Permanente leaders developed the Panel Support Tool in-house in order to help their primary care physicians manage the care of individual patients, groups of patients, or their entire panels. It works by comparing the care a patient is receiving to the care recommended by national guidelines. A physician can, for example, query the Panel Support Tool in advance of a patient visit, to find out if that patient needs a screening test or a vaccination. Physicians can also ask the Panel Support Tool to display a list of all of their patients who are overdue for a mammogram or colon cancer screening; or a list of their diabetic patients whose blood sugars are too high, or those who need a foot exam or an eye exam.
The retrospective, longitudinal study, “Effect of a Patient Panel-Support Tool on Care Delivery,” which was published in the October issue of The American Journal of Managed Care, followed 204 primary care teams who are using the Panel Support Tool to manage care of 48,344 patients with diabetes and/or heart disease. After three years, for patients with diabetes, the percentage of care recommendations met every month increased from 67.9 percent to 72.6 percent; for heart disease patients, the percentage rose from 63.5 percent to 70.6 percent.
The second study, “Improving Population Care with an Integrated Electronic Panel Support Tool,” which was published online in Population Health Management, involved 207 primary-care teams that were using the Panel Support Tool to manage the care of 263,509 adult patients, some of whom were relatively healthy and others who had chronic diseases. The study looked at 13 different care recommendations and found that, after 20 months, the Panel Support Tool improved performance in delivering care according to recommendations, from 72.9 percent to an average of 80 percent. The researchers also found that during the first year of tool use, performance in delivering care according to recommendations improved by a statically significant degree every four months.
The Panel Support Tool monitors recommendation pertaining to medical management and screening for co-morbidities in six chronic conditions: asthma, diabetes, coronary artery disease, heart failure, hypertension, and chronic kidney disease. The tool measures preventative care measures, such as administering adult immunizations and screening for breast, cervical, and colorectal cancer, hyperlipidemia, and osteoporosis. For each care recommendation, the Panel Support Tool indicates what action needs to be taken, if any.
Recently, HCI Managing Editor John DeGaspari interviewed Robert Unitan, M.D., one of the authors of the first study, and Yvonne Zhou, Ph.D., lead author of the second study, about the events that led to the development of the Panel Support Tool and what it will mean for Kaiser Permanente's primary care physicians.
In Part 1 of the interview, Drs. Unitan and Zhou speak about the formation and early development of the Panel Support Tool, and the improvements in care that Kaiser Permanente's physicians have seen with its use.
Healthcare Informatics: Explain the genesis of the Panel Support Tool. How did it develop?
Robert Unitan, M.D.: I am physician leader at Kaiser Permanente, the director of operations for medical specialties. And we are always looking for ways to take better care of our membership.
We in the Northwest Region were the first in Kaiser Permanente to implement [Verona, Wis.-based Epic Systems Corp.'s] EMR [electronic medical record]; we piloted that in 1994, and went region-wide with it here in 1996. Kaiser went down a different road nationally, but ultimately, when we got a new CEO, Mr. George Halvorson, he very quickly got on track for the rest of the company, essentially, to implement Epic. It was his expectation that, just having EMR, with the decision support that you have, in an EMR, would be sufficient to get our quality performance up to the 90th percentile, as measured by the HEDIS [Healthcare Data and Information Set] measures, from NCQA [National Committee for Quality Assurance].
[Halvorson] in a very big national meeting, essentially announced that it was his expectation that we would all be at the 90th percentile by the end of the decade. This was back in 2004, and 2003 when it was made. And I knew, coming from the [Northwest Kaiser experience], where we already had Epic ambulatory EMR for eight years, that it was going to take more than that.
WHEN I WENT OUT AND TRAINED CLINICIANS AND WE WENT BUILDING BY BUILDING TO TRAIN THE CLINICIANS AND STAFF IN HOW TO USE THIS TOOL AND EXPLAIN WHY THEY SHOULD USE THIS TOOL, THE FACT THAT IT WAS PHYSICIAN-LED WAS VERY IMPORTANT IN OUR ORGANIZATION.-ROBERT UNITAN, M.D.
And I happened to meet a couple of physician colleagues [from Kaiser Permanente's Hawaii region], who had this concept that they called Total Panel Ownership. It was about how a primary care doctor should not only be responsible for the patients who are on their schedule each day, the 20 to 24 patients that they see, but instead they are responsible for the health of the entire panel [of] between 1,500 and 2,000 Kaiser members who call that doctor their primary care physician. One of the things that these docs from Hawaii said would be absolutely essential was to have basically a spreadsheet that would allow them to find the gaps in care for all of the patients in the panel, Not just the people who would be coming in to see them.
I took it upon myself to partner with [physicians from Kaiser Permanente's Hawaii region] and with Kaiser Permanente Information Technology to get this built. We built this pretty rapidly in the second half of 2005; here in the Northwest it was completely rolled out across our region starting in the spring of 2006 through the end of 2006. We basically brought nearly identical copies of the Panel Support Tool up in both regions. We built it in Hawaii first, but it was a happy coincidence that same people who run the data warehouse for the Northwest region also run the data warehouse for the Hawaii region. One of the things I wanted to do was to demonstrate to our program leadership in Oakland that physicians from different regions could actually collaborate on this sort of thing, and agree on the rules that would drive a care engine such as this.
HCI: How important was the fact that physicians took the lead in developing this tool?
Unitan: It was very important in terms of when we rolled it out it had a lot more credibility, in part because we had developed it ourselves, and we had developed it in a way that we do business. HMOs are different than fee-for-service medicine; we've got some different business drivers and different incentives around the health of our population. But when I went out and trained clinicians and we went building by building to train the clinicians and staff in how to use this tool and explain why they should use this tool, the fact that it was physician-led was very important in our organization.
HCI: Does anything like the Panel Support Tool exist in the commercial market?
Unitan: Not that we know of. There are other health systems that we believe have built a similar tool, although they have not publically reported their results, and we don't think these are in wide use. And we have not seen an electronic medical record that has this functionality.
HCI: Besides what has been derived from HEDIS, what other criteria did you use for recommendations to be monitored in the Panel Support Tool?
Unitan: For example, there is an initiative that we call ‘ALL,’ which stands for Aspirin, Lisinopril, and Lovastatin; it's basically that all patients who have diabetes or cardiovascular disease should be on a statin and a blood pressure medicine; an ACE [angiotensin converting enzyme] inhibitor, which is what Lisinopril is. There is not a HEDIS measure around that. Furthermore, the cardiovascular patients should be taking a [baby] aspirin a day. That is not something with a HEDIS measure, but it is something that within Kaiser, the evidence suggests that it will prevent myocardial infarctions. So that was an internal Kaiser measure, not HEDIS; but it is certainly a population care measure for those distinct care populations.
Yvonne Zhou, Ph.D.: The population health management article included many measures of preventive care, for example, pap-smear screening. Also, it includes [13] chronic condition care gaps. We picked up 13 mainly to have the consistent data measures. HEDIS measures sometimes change from year to year. We used the measures that were consistently defined during the study period, so that we could really evaluate before and after in terms of performance increments.
HEDIS MEASURES SOMETIMES CHANGE FROM YEAR TO YEAR. WE USED THE MEASURES THAT WERE CONSISTENTLY DEFINED DURING THE STUDY PERIOD, SO THAT WE COULD REALLY EVALUATE BEFORE AND AFTER IN TERMS OF PERFORMANCE INCREMENTS.-YVONNE ZHOU, PH.D.
HCI: What quantifiable performance improvements have you seen with the Panel Support Tool?
Unitan: Depending on the measure, we were already doing pretty well [before the Panel Support Tool roll out] because we had an electronic medical record. We had an integrated delivery system, where we are able to track a patient's labs and pharmacy results, pharmacy dispenses very closely.
So we were already doing very well, although we weren't at the 75th-and certainly not at the 90th-percentile in many of the HEDIS measures that were built into the Panel Support Tool. At this point, after four years of use, we are on track to finish the year with over 65 percent of all of the HEDIS measures above the 90th percentile in the Northwest. That represents thousands and thousands of mammograms, pap smears, [and] fecal occult blood tests. It represents thousands of pediatric immunizations, and adult immunizations. So it is very clear that it has had a remarkable impact in terms of those sorts of measures.
In terms of improving health outcomes, often, you have to invest up front. The cancer that you are preventing is not a cancer that might not have occurred for several years, so actually seeing hard improvement in clinical outcomes is going to take more time; although we do have some evidence here in the Northwest that we are diagnosing colon cancer much earlier. The percentage of patients that we are diagnosing, who have stage 1 colon cancer, is much greater than it was before we started down this road. And much of that improvement is directly attributable to the Panel Support Tool; when you catch those cancers earlier, the patients get treatments sooner and have an enhanced chance of survival.
HCI: What is it about Panel Support Tool that has allowed you to catch those at risk of cancer earlier?
Unitan: It's one of those things that you have to think of how it works for a patient. A patient might have a sore throat or a sore back or they twisted their ankle. Even if they have diabetes or cardiovascular disease, or even if they don't have chronic disease, they just are a general patient coming in with a complaint. Those complaints are completely unrelated to the sorts of gaps in care that we are tracking with the tool.
So in the past a person would come in with their own agenda and the clinician would take on the presenting problem. Only a very compulsive clinician would take the time to comb through the entire chart, to see, for each of these 75 gaps in care, whether the patient was up to date on things or not.
The Panel Support Tool makes it instantly apparent, in just a matter of a few seconds, what needs to be done. And furthermore, we engaged our staff, the assistants who support the physicians. They are the ones who previously would take the patients from the waiting room, ask them what their chief complaint was, and would take their blood pressure and the vital signs. And that was it.
And what we have been able to do, with the combination of the Panel Support Tool and the electronic medical record, is that, for many of the gaps in care, the medical assistant can go and stage that work for the clinician. So the clinician still comes in and deals with the sore throat or sore back or twisted ankle. But also, the medical assistant has already placed the lab orders, or the order for the mammogram, or the order for the stool test. All the physician has to do is sign it and tell the patient, ‘Oh, by the way, we noticed you are due for these other things. So in addition to picking up the medication for your sore back or sore throat, we are going to have you stop at the lab and get this, or stop at the pharmacy and get this needed med to take care of your lipids.'
The Panel Support Tool has made it very readily apparent and it was much more readily accepted than the sorts of alerts and reminders that can be built in to the standard electronic medical record. Clinicians tend to ignore those. They are like banner ads on the Internet; they are not why you went to the Web page. You opened that chart to take care of the patient's presenting problem.
Alert fatigue is a very real phenomenon in the world of electronic medical records, and clinicians will blow right past it, saying they don't have time. The nice thing about the Panel Support Tool is that you are intentionally going to the tool to look for these other things that the patient might be missing. By making it the medical assistant's job, to do that and make sure that every time we have an opportunity to close those gaps in care, we take advantage of it, we make it as easy for the physician as possible. That's very different than having a big ‘banner ad’ pop up in your face when you go into the chart.
HCI: Is it easier for the physician because the Panel Support Tool dovetails with physician workflow?
Unitan: Yes. They have only got 20 minutes with the patient for the most part and there's a lot they have to take of [with] the chief complaint that brought the patient in.
And we are only talking about the in-reach workflow, in which a patient comes into the office and you discover other things they need. There is another, just as important, outreach workflow, for all of the patients who aren't coming into the clinic. Those patients have care gaps as well; and the tool allows you to go through to send letters and to call patients, and to tell them, that we noticed that you are missing this. That is equally important in terms of the improvement in our performance metrics that we have seen in the last four years.
One of the keys to the [overall workflow scheme] was making sure that we have everybody working to their appropriate scope of practice. If there is work that a medical assistant can do that frees the physician up, that medical assistant is performing a vital function. And it has been remarkable, in terms of the job satisfaction the medical assistants have had [because] they are participating that much more actively in the care of the patients. It's a team-based approach to care, that has been very well received, at a time when primary care physicians are really struggling. It's overwhelming to be a primary care doctor these days. So anything that can help leverage the staff to be able to take work off of their plate, and make things easier for them to do the right thing, we want to make sure that we take advantage of that.
Next month, the authors will comment on the tool's integration with HealthConnect, Kaiser Permanente's EMR system, improvements with patient care and outreach, and future refinements. Healthcare Informatics 2011 February;28(2):40-44