David Williams: What is the focus of the company? What are you bringing to the market that has not been available in the past?
Michael Weintraub: We’re aiming to build a census view of health care in America. What I mean by that is to build a large-scale informatics asset that various constituents can tap into to get a perspective on whatever question they might have, whether it’s a disease or a therapeutic area. The focus will vary depending on whether you’re a hospital, a large medical practice, a pharmaceutical manufacturer, biotech, the federal government, etc.
We’re basically building a large-scale factory that is bringing in a vast amount of patient data from the provider setting. By the provider setting I mean hospitals, integrated delivery networks (which include large ambulatory settings and doctors offices), standalone doctors’ practices, and then putting all that information into a common database. It’s protected by world-class security and adheres to the principles of HIPAA. The organizations that treat those patients can certainly see their specific information at a detailed and granular level, but other than that, everything else is done at an aggregate, encrypted, unidentified level.
We’re not studying the individual or the hospital or doctor per se, we’re studying patterns and analyzing the information and the various cohorts to understand meaningful variation; really studying that information to drive value and performance improvement. At my last company PharMetrics, we did this with insurance data. That information provides you with a view but it’s aggregated and it’s not at the granular level that we’re talking about here. So the state of the art today –with the big push by Obama and the federal stimulus dollars and the Health Care Recovery act– is electronic medical records capability. As the level of automation increases our aims becomes possible. We are working with organizations that have electronic medical records and we are bringing that information into our common factory, then applying scientific, statistical and medical techniques to normalize that information so it can be used in the manner I’ve just described.
We have multiple markets. We have provider markets and we have built some exciting applications (which I’ll have some of my colleagues here talk about in a moment) on their behalf that enable the exchange of information. Data comes to us and then we provide/sell those capabilities to our customers. We have other stakeholders in the health care industry who are very interested in understanding how their products are utilized, whether it be a biotech company, a top pharma company or a medical device company. Those organizations (and we’ll comment on that deeper in a minute) are very interested in studying large volumes of patients that might have asthma or type two diabetes and so forth. So having a business intelligence view, a very specific kaleidoscope into a population of interest is really now at a level of clinical specificity that has not been available before, unless one conducts primary market research.
Williams: What is the time lag of the data? It sounds like you’re using the information for aggregated views –which are perhaps not so time sensitive—but also providing information back to providers on an identified basis to improve treatment. Have I got that right?
Weintraub: The state of the art has been less clinical specificity, but also data is often batched or historical. When we implement this capability, we grab several years of retrospective data, but we also have designed our capabilities such that our information is streaming in near real time, in five-minute increments. If you’re a provider, you’re able not only to see your history, but you’re also able to study the population while they’re actually in your care. So it’s got a direct link to your electronic medical records. It’s streaming continuously, so if you’re a hospital looking to study preventable complications that would be an obvious application.
Williams: If you’re a hospital, beyond studying it, could you also be using information for real time clinical decision support or would that be beyond the scope?
Weintraub: That’s a great question. I think there is a fine line here. Absolutely it’s going to be utilized for real time clinical support. What we don’t want to be is the real time patient management system for a variety of reasons. But if you’re a large provider with a significant cardiology unit, you’re going to have an individual from a quality department with a quality nurse assigned to that unit and what has typically been done through silos of information with multiple EMR’s and lots of paperwork and checklists and highlighters can now be done online on a real time basis.
Williams: Tell me about your product offering.
Allen Kamer: We have built two products and have launched one of them. The first product is called “Humedica Minedshare.” We compare performance data and benchmark activities from one hospital or medical group to an aggregate or to others within their system. So for example, a medical group with multiple clinics can identify on a clinic-by-clinic basis how they’re doing in treating certain patient groups and then look at overall how their organization does and then go into a greater level of detail. Additionally, we are building the capability for that particular medical group to compare with all the other medical groups that are participating.
So that’s called Minedshare. It enables clinical analytics where you can benchmark and compare treatment of patients by disease and severity across locations.
Weintraub: If you’re a 25 or 50 or 100-hospital chain or system, you’ll likely want to utilize this for comparisons, inter-hospital as well as intra-hospital, because you have lots of hospitals in your system. You also might want to analyze hospitals to the norms within their region because many hospital chains are structured by region. And then at a third level, as our database grows, you’ll want to analyze performance compared to the benchmark within our broader and ever growing database and based on region, specialty type and so forth.
It’s all about discovering variation; often variation and acting on variation is at the heart of both clinical effectiveness and operational efficiency. The endocrinology department might want to look at average clinical information –like HbA1c scores for the diabetic population by physician– and be aware of the differences demographically or geographically rather than by overall population. Within the practices they can then try to understand attribution and cause so that they can attack it.
That product has been rolled out and is in use by our first customer Christus, a 20-hospital system in Texas. They are an active user of that product.
Kamer: Our second product is called: “Humedica Minedstream.” This a real-time and predictive clinical surveillance tool that has been developed for the hospital to reduce preventable complications and improve performance metrics and really ensure that appropriate compliance is maintained with the performance protocols.
At Humedica we gather, map and normalize the data as patients are admitted into the hospital and tests are given. We use our advanced predictive analytic capabilities and modeling techniques to identify the hospital patients that might be at risk for preventable complication or might require tracking due to the Join Commission’s core measure criteria. The Minedstream product is a dashboard that allows you to track patients. A quality nurse can identify who they are and where they are in the hospital and the physicians whose care they’re under.
It lays out all of the requirements associated with delivering care to those patients and it has a count down mechanism and dashboard tool to ensure that those activities are being adhered to. For example, it a patient comes in with chest pains and then has testing done, we identify that those patients may be heart attack patients. Then there are a number of things that need to happen for a patient who has had a heart attack, such as aspirin needs to be given within 24 hours if their troponin is at a certain level.
There are things that need to be done at discharge so our dashboard first identifies these patients and then tracks whether or not the physician or the nurse has performed those activities that are required for those patients.
Weintraub: Our first product rolled out earlier this fall and the second product is being rolled out this month. Those are all for the inpatient hospital settings. The complementary applications which will leverage the same environment, the same product, the same data factory, the same scientific methods, but for the ambulatory market, will roll out in 2010.
The ambulatory market is very important to us. Much of health care spending is on the ambulatory side. Roughly three of four dollars of prescription drug spending is on the ambulatory side. It’s a heavily growing market due to chronic disease. We have established a long-term, exclusive partnership with the American Medical Group Association (AMGA). They are a very prestigious organization in the D.C./Virginia area and are the gateway organization that is the connection to over 300 large medical groups, integrated networks in this country.
They provide care to one in four Americans. They are in 49 states and touch about 100 million patients a year and several hundred thousand physicians. It’s a very important organization, which includes prestigious organizations you’ve heard named by Obama such as Kaiser, Cleveland Clinic, Intermountain, Mayo, Geisinger, Henry Ford and others. We are their exclusive partner in a clinical informatics capability to address the very issues we’ve been talking about for the past 15 or so minutes.
We are working with roughly ten of their organizations who are early adopters to roll out this capability in the spring of 2010 to the ambulatory market. What’s really important is to be working with hospitals and medical groups so that we can connect patients longitudinally and get an integrated longitudinal view of health care across treatment settings.
Continued to Part 2.