Business Intelligence Needed In Healthcare to Provide Relevant Information
Laura Madsen, in her book “Healthcare Business Intelligence: A Guide to Empowering Successful Data Reporting and Analytics” tells a story that in her job, her senior managers ask her to:
“create the omnibus platform for coordination, population health, and care management,
that integrates heterogeneous data from large and small affiliated provider organizations,
supplying retrospective and prospective analytics,
to improve quality, safety and efficiency.”
Here’s a question – Thinking about Ms. Madsen’s challenge, how might you white board the project?
And the answer – Business intelligence is important key to all of her requirements.
If you want to create better care coordination, you must have information on where and when to coordinate. If you want better population health and care management you need to know which population you should work on and what work will really make a difference. If you want great quality, safety and efficiency, you need the same type of data on costs, care points, and how the institution delivers evidenced based medicine.
Three Questions to Guide Healthcare Business Intelligence Discussions
There are several questions we must answer to help make sense of the healthcare business intelligence market.
- In all the cases above you could apply both analytics and business intelligence to help solve the problem. How these fit together and why they need to be coordinated?
- Big data versus small data within business intelligence tools; how should either be integrated into decision support?
- Data warehousing; how does this play into the equation?
Data warehousing and access
Business intelligence solutions start out with the need for data. In healthcare, most solutions have been based on claims data as this was the most available until recently. However, now that we have relatively large quantities of clinical data stored in EMRs or HIEs, we have access to more useful data if we can simply mine it for what we need. There is an ongoing debate about “how much data is enough data”, and “is it okay to impute data when you don’t have a complete set”? Most users of healthcare business intelligence have at least created some type of clinical data warehouse, or are using claims data, or both. Utilizing both types of data would be ideal, but claims data will at times take as long as 120-180 days to access, unfavorable timing for certain disease types or population health management programs.
Type of data needed varies
Which data, and how much data you have, brings up two questions that need to be answered before we can move into how to use the data. The first is: Do we have the type of data we need and do we have all of it in order to make the best decision support tools we can from business intelligence? In most cases, EMRs themselves have a fairly robust data set from a single location or health system. They do not, however, always have the data they need from outside their system when patients use outside labs, physicians who are not on the hospitals network or EMR, and other types of providers of care who may not use an EMR at all. As a vendor of integration services, we are often surprised by how many organizations buy healthcare business intelligence solutions to provide decision support, yet don’t factor in the source and quantity of data to make the program work.
You can have too little data, too much data, and either the right kind or the wrong kind of data, depending on what you are trying to accomplish. This next statement may be surprising, but healthcare can be trend driven. And no, not lab-coat cut and color kind of trend, but rather trends driven by technological advances. While it is important to stay abreast of innovation, riding the trend train is not always the correct strategy for an organization. For example, big data is a popular trend that can have a major impact on the cost and quality of healthcare in time. There are however, plenty of situations where a couple of different types of small data, such as a specific biomarker within lab data or from a blood pressure cuff or both, can be all you need to manage a specific population once diagnosed. A combination of both big data and small data are very important to the success of a strong healthcare business intelligence decision support program.
Analytics and business intelligence fit together for coordinated approach
Assuming we have the data needed, let’s examine what organizations are looking to accomplish with business intelligence tools. In most applications, organizations want to answer questions whereby data is comparing two things.
Broad question: How well is my organization is adhering to best practices?
Comparison set #1: In this case, best practices are encapsulated inside evidenced based medicine with regard to delivering care.
Comparison set #2: The other best practice is the cost of delivering care compared to local, state and national benchmarks.
So what might this look like as a specific question?
How does my hospital compare on delivery of care to diabetics and how much does it costs me to do that, versus my closest competitors, and those who do it best in the entire nation?
Once an organization knows the answers to those questions, they can then strategize on ways to make improvements. As in this example, not all business intelligence tools actually prescribe solutions. Many will identify the problem, allowing drill-down into individual patient situations. Now the challenge becomes clarity; the answers sought with business intelligence aren’t always clear – in this case, how to deal with the patients.
Can business intelligence in healthcare help close the gap?
Healthcare currently has a dilemma; how to provide true care for patients, yet maintain growth as a business industry. Clearly, healthcare must identify where in the process problems are happening yielding high costs and care gaps. Tools to do this type of identification are often referred to as “gap” tools because they find patients or care situations where there is a gap between the ideal care delivery, based on evidenced based medicine, and the care that is actually being delivered. Once identification has been made, the next step is to further define patients who need more or different care versus the ones that already received all the care they need. Prediction is that a new set of tools is coming that will further enhance healthcare’s ability to apply care at higher quality with lower associated costs. We’ll dive into this later in the discussion.
Benefits for Healthcare Using Business Intelligence
With business intelligence tools, healthcare can accomplish much. Synthesized down, here are three things we haven’t been able to do previously:
- Understand patients and how the disease they have is affecting them relative to normal care delivery processes. For example, physicians know that if a patient takes their medications on time, every time, and get more exercise, they will improve their outcomes. This outcome has foundation in evidenced based medicine. So now research can be undertaken to understand the cost and quality impact on patients who follow this perscription. And, most importantly, a physician can make the determination if the patient in-front of them, or on a list within their office, is going to get the same type of result when that care is applied, or will something different happen. Who are the exceptions to the rule if you will.
- Identify costs that are outliers to our expectations or to benchmarks listed above. For example, if a population of patients who have renal function problems is costing an average of $2500 per year more than the benchmark, then questions might be is this being caused by a pure cost problem, a diagnosis/treatment issue, or something specific to the identified population such as their ethnic background being more prone to kidney problems. Once the problem is further defined, better direction can occur to make changes. This is the essence of what healthcare business intelligence should tell us.
- Drive better consumer engagement from information garnered from healthcare business intelligence. If data can be taken from multiple sources, normalized, and presented to the provider and the patient in a format they can use, then the provider and patient can make an agreement having used business intelligence tools to create real cost reduction and quality improvement. To make this work “normalized in a format they can use” means trended in graphs showing how the patient is doing relative to what goals are and then having the ability to see it change through a Personalize Health Record or through coaching with the provider. Business intelligence can be an excellent tool to help drive different behavior.
What’s Next For Healthcare Business Intelligence?
Technological advances for better outcomes
One might suggest Business Intelligence (BI) for healthcare and Population Health Management (PHM) are both in phase 1.0 of their existence. Many other industries have advanced beyond this level and healthcare will too. In BI/PHM 1.0, we get to know our own organizations better, we figure out how to effectively group patients together by disease or health condition and then we can use evidenced based medicine against that grouping to find and fill the gaps in care so that we are consistently treating patients with similar diseases with the best information we have. If that is what you get in 1.0 version of BI/PHM then what do we want in 2.0?
The evolution of the tool sets used in BI and PHM will head in the direction of individual guidelines for patient care within a population grouped by disease or health status. As an example, for any given patient that has a morbidity of Cardiovascular disease and a co-morbidity of diabetes, we will first analyze the gaps in their care and then look at the benefits they would actually realize in reduced risk and costs dependent on which treatments are added to their specific situation. It may not seem significant to be able to tell a provider that the actual risk reduction would be 22% if they added an ace inhibitor or that plus a compounded hypertension drug to a patient’s treatment, but it is. By specifying benefits derived from actual treatments, physicians can choose what to do, and what not do, based on protocols that will help the individual patient rather than prescribing a treatment scaled for an entire population.
How do we advance from 1.0 to 2.0? The technology requires algorithms and deep mathematics computed over a large population of patients to establish baselines and actual predictive models of how patients will react if you treat them individually. Moving from 1.0 to 2.0 will also enable further distillation of characteristics to aid in treatment. Example: a patient that has blood pressure slightly below the threshold of 140 but is obese and has elevated LDL’s wouldn’t necessarily be identified, or be singled out by a 1.0 population health system, as someone who needs a hypertensive drug because evidenced based medicine may not catch them. Years of patient treatments stored in EMRs and minded by mathematicians at healthcare institutions is now becoming available and will further help identify gaps in care, as well as specific benefits tied to specific actions, all leading to healthcare providing the best quality and cost situations.
Expediting and anticipating care
In addition to the sophistication of BI/PHM 2.0 tools, expediting the care process to deal with the advantages of timing in caring for chronic and more progressive diseases is upcoming. Watch for this in the near future. Today we do calculations overnight, or more practically with claims we do them 120 days after the fact. As business intelligence turns into remote monitoring and bio-surveillance, we should see new technologies come on-board that will allow us to better predict the likelihood of care needs and prevent events such as heart attacks more effectively. There is a big victory in being able to stop heart attacks before they happen as a heart attack can cost $50,000 in care and treatment in just the first year. We are on track to see some really helpful decision support tools for population health management and soon to see some even more advanced PHM 2.0 tools that will really move the cost and quality needle.
Recommendations for Organizations
Business intelligence for healthcare is not just a software solution to plug in and expect auto-populated answers. It requires strategy and planning. So what are the recommendations for healthcare organizations thinking about implementing a business intelligence solution?
Make sure you get the data under control first, don’t ignore small data sources because they could be all you need for certain patients/populations.
Make sure you have a working user interface that allows consumers to engage with their provider in a way that is understandable.
If you would like to know more about healthcare business intelligence, you can contact us at Orchestrate Healthcare where our KLAS award-winning team can make the difference in your strategy.