15 Oct

Business Intelligence is the Origination, not the Destination

The entire spectrum of healthcare has embraced the concept of healthcare data analytics and business intelligence. Reports and surveys have been published in large volume indicating that this nascent capability within healthcare is increasingly more important to providers and payers alike; not to mention all of the intermediaries that are focused on quality or cost of care.  The new prevailing question is, “When we are able to transform data into information, what do we do with it all?”

We must remind ourselves that business intelligence is two-dimensional unless the user is empowered to act upon it. The third dimension becomes taking the information and using it to affect positive change.  In other words, we must evolve from measuring the needle to moving the needle in a desired direction.  Once we have accomplished this action-taking, the effects of our action can be measured and refined.  Through this initiative we are able to attain business knowledge.  If we are fortunate enough to then seek replication and further improvement; a fourth dimension is introduced.  That final factor goes by the name of “wisdom.”

So let us presume that wisdom is the ultimate goal of any healthcare organization. The first step in the journey, like any good recovery program, is to admit that there is a problem.  We must rid ourselves of the thought that intelligence is the end; instead embracing intelligence as the origination.  The catalyst in this process is the ability to take action.

Before we can embark on the journey to intelligence, we must first complete our navigation of the ocean called analytics. The foundation for our trek revolves around data.  It must be acquired from multiple sources; it must be evaluated and validated, cleansed, and normalized.  This is without a doubt the least glamorous part of our transformation.  By nature, healthcare data is dirty and incomplete.  It takes a skilled eye to find the data and determine how its quality will affect the outcomes reported within analytics.  This step is essential prior to generating and sharing intelligence.

Once we have managed to corral all of the data and make some sense of it; it must be placed within a data model that is scalable, extensible, and flexible. Quality can be addressed from clinical data and cost can be tackled from administrative data.  However, the true power of analytics is unleashed when clinical, claims, financial, and operational data are all placed into a single data model.  This 360-degree view of the healthcare business forms the building blocks for creation of insights which then lead to outcomes-driven intelligence.

Once you are able to see the direct relationship between gaps in care on the clinical side and excess cost on the claims side; a platform is established for action. It is in moving to an aggressive position from the passive norm that intelligence takes on life.  Decisions can be made and measured.  The results lead to non-stop refinement of interventions and an evolving decision support system.  In other words, we attain knowledge.  Once we can replicate the results of our knowledge; we reach a state of wisdom.

All journeys begin with a single step and they never commence at the destination. Yes, keep your eye on business intelligence; but realize that harnessing the data must come first.  A journey also seems to take less time and be more enjoyable with a companion.  Do not fear the thought of a technology partner as your traveling partner.  Enjoy the journey, learn from your experiences, and realize that the goal is well worth the effort.  In doing so, wisdom will be your reward.

09 Oct

Jump Start Your Healthcare Analytics Initiative

We see them by the side of the road all of the time; those poor souls whose automobile battery has come to its inevitable mortal demise. Some chose to keep driving; somehow thinking that not making eye contact will make that desperate person disappear.  Others will come to a stop beside the disabled vehicle, turn on their hazard lights, pop the hood open, and take on the role of the Good Samaritan!  I guess if you know how you would react to this scenario; it speaks volumes to who you are as a member of the humanity pool.  I suppose that is a topic for quite a different type of blog!

The same analogy can be extended to the Healthcare Information Highway. Many of us start of in perfectly good “IT Cars.”  Over time, newer models come into the showroom; but we prefer to keep driving “old reliable!”  Eventually, they add new lanes to that highway and the speed limit is increased.

Suddenly, our trustworthy IT vehicle no longer functions as well, in spite of the preventive maintenance investment we might have made. Eventually the car will no longer even start up and we find ourselves on the side of the highway; jumper cables in hand!  The problem is that all of the other IT cars are barely chugging along themselves, so they are unlikely to stop to lend a hand.  Every now and them, a high performance IT car will come by whizzing by; but they rarely see your distressed situation or have not yet figured out where the brakes are.

Analytics and business intelligence has quickly become the high-powered sports car on the Healthcare Information Highway. The overarching problem is that there are a lot of dealerships selling their own brand of analytics.  The body style is nice; but usually the engine is underpowered.  In some cases, the battery under the hood has never been charged and you are left holding the jumper cables once again!

In order to jump start the battery for your healthcare analytics, it is essential that you have a set of predefined measures in place that are built upon trusted data sources. The gas for your engine is that data; but that is the topic of many other posts in this blog.  For now we will focus on jump starting the engine.  Gas is essential, but it is worthless if the motor will not turn over!

There is a plethora of standard measures, created by recognized and respected national bodies, that will measure quality (clinical) and administrative (claims) measures. We are inundated with well-meaning measures from NQF, HEDIS, HRSA, USPTF, HDC, and many others.  Some of these measures are relatively easy to compute; while others read like Cold War-era rocket science and require rooms of analysts and programmers to code.  For the most part, all of the answers that you seek from your data reside within these established measures.

Most analytics vendors (the car dealers) will sell you a framework wherein you can create your analytic measures (the car with the dead battery). They leave it to you to do all of the work (charge your own battery).  If the measures do not work or are not to be trusted; that is your problem (you charged the car, not them).  Even if you are able to build your measures; are you just getting numerators and denominators of who is in compliance?  If so, your car battery just became an AAA power cell.  It will power most toys, but it does not help a car engine start!

If you are looking for the maximum cranking power and cold amps for your analytics engine to fire up; your battery has to be so much more than what most vendors provide. A true analytics partner (the reputable car dealer) will give you a solution for you battery and engine needs.  Yes, you will get numerators and denominator; but you will see both who is compliant and who is not.  The latter surely will have gaps in care.  Beyond the top and bottom number; the best battery will also give you drill-down capability to the individual patient and provider.  Now you have actionable intelligence to fill those gaps in care!

If you choose the right IT car, with the right analytics engine under the hood, ready to fire up with a predefined set of measures; I assure you that you will be the master of the Health Information Highway. In fact, you may find yourself as the only vehicle in the passing lane.  To get the best performance you will need high octane fuel (cleansed, normalized, validated, trusted data); but we can talk about that the next time you visit.  In the meantime, feel free to sell those jumper cables!

02 Oct

The Analytics Elephant is Sitting in the Waiting Room

Whether you are an ambulatory practice, an urgent care center, a hospital, an insurer, an integrated delivery system, an accountable care organization, a health information exchange, a provider hospital organization, or whatever healthcare delivery or administration acronym we have managed to dream up now; you share a very common trait with all of your other healthcare brethren. You have an analytics elephant sitting just outside your office!

There is an old joke that poses the question, “How do you eat an elephant?” The punch line is, “One bite at a time!”  I will give you a moment to pause in reflection and perhaps chuckle quietly to yourself!  This piece of humor is truly applicable to analytics in healthcare; there is indeed an elephant out there that each of us needs to consume.

Before we even think of how we are going to ingest this pachyderm; let us look more closely at our dinnertime “catch of the day.” Our elephant’s anatomy is made up of four components. First, we have a multitude of data sources, of varying quality, and scattered across what seems like an endless series of disparate silos.  The beast has multiple categories of measurement; including clinical quality, administrative cost, financial, and operation key performance indicators.  For each of the categories, there exist a set of measure families that have been generated by national standards bodies or that have been created to suit regional initiatives.  Finally, our pachyderm supports an entire community of would-be information consumers.

If we are to eat the analytics elephant one bite at a time; we wish to assure that the entire healthcare village gets fed well. This goal mandates that every anatomical portion of the analytics elephant must be addressed.  The first challenge is to find all of the data that make up the core of the elephant.  This information is nestled in multiple disparate sources and will require serious heavy lifting to extract.  Since the village is best fed from a common pachyderm; we wish to place all of our data in a single community data foundation.  Once we normalize and cleanse this data, it is fair to say that we now have a single version of the truth across the 360-degree continuum of healthcare.

Different consumers of the elephant will have varying tastes as to what part of the beast they prefer to ingest. For this reason we need to acknowledge that providers, payers, quality organizations, government, and all of the other guests at the banquet will each request a different cut of the meat.  Our elephant must satisfy the palate with analytics that addresses quality, cost, finance, operations, utilization, and access of care.  There will also be those consumers who will come back for seconds; more than likely having a change in taste along the way!  Answers always beget new questions, right?

In order to assure that everyone requesting a certain cut of the elephant gets the same quality and portion; it is critical to harmonize our measures. The entire village needs to be eating off the same platter; that is to say, nationally accepted standards must be used and the determination of how the measure is calculated must remain constant.  These measures must meet the specific needs of a group, with enough flexibility to be customized.

Finally, we need a way to share our analytics elephant bounty with the entire village. We can cook the beast to perfection; but it serves dubious purpose if there is no serving platter to bring it from hearth to table.  In any analytics effort there are three principal measures of success; (1) did we identify gaps or places for improvement, (2) can we surface the opportunity to the right person, and (3) can that person take action based upon this knowledge?  If you cannot surface and share what you have learned from the analytics elephant; better to have ordered delivery pizza!

Assuming you can access all of your data, measure each piece, utilize the correct measures, identify gaps, and act upon what you have surfaced; the entirety of the analytics elephant can become too much to consume, even for a large and prosperous village. If you try to address everything from everywhere for everyone, you will simply choke your analytics initiative.  The answer is that we must all eat the elephant one bite at a time.  For our case of the elephant, these bite-size pieces are:

  • Limit the initial number of data sources – Select several sources that perhaps use a common EHR/HIS system, or a common protocol. These sources should be prequalified with respect to the quality of their data
  • Limit the measure domains (categories) that you will examine – Do not try to meet the needs of providers, payers, quality organizations, and government simultaneously. Restrain yourself to just one of these domains initially.
  • Limit the measure families for initial implementation – You will simply not be able to handle the volume of information that comes from a solid analytics platform. Select a population subset such as diabetics, opt for actionable measures such as A1c percentage reduction, and focus on identifying gaps and taking action.
  • Limit the number of consumers for analytics – This is most easily accomplished by data-marting results back out to the data sources originally used. In this way, immediate benefit can be shown and trust of the results reported will occur.

Most healthcare organizations are in no position to build an analytics solution. Third party vendors abound in the marketplace. Two types are worthy of closer examination. The first is driven through a consulting model, loaded with front end services, and eventually delivering a customized analytics solution. This is usually an extremely large elephant, that tends to be very expensive, takes forever to cook, and is served in its entirety. The second vendor type delivers a product that is integrated to data sources, preloaded with measure sets, and is accelerated through services. This elephant is still very large, but it is cooked and served in bite-size pieces, making it less expensive and easier to digest.

As you move forward with identifying your partner for analytics, ask yourself a very important question. “Does this vendor understand the elephant in my room?” If the answer is yes, then there is only one more question to ask. “Will my organization be forced to choke on the entire elephant, or will my vendor help me eat it one piece at a time?”

25 Sep

The Triple Aim is Right on Target when Supplemented by Analytics

There are historical indications that the three-legged stool dates back to ancient Byzantium. This type of seating was made popular due to its low cost, ease of portability, lasting construction, and universal appeal.   In healthcare we use the three-legged stool analogy quite frequently.  A shining example would be the Triple Aim™ as defined and developed by the Institute for Healthcare Improvement.

The Triple Aim is usually depicted as a pyramid or triangle; but in this case we could also use a three-legged stool. The three legs of our stool are (1) improving the patient experience of health, (2) enhancing the health of populations, and (3) reducing the per capita cost of healthcare.  These three legs would then support a seating surface that we can call “optimized healthcare!”  Admittedly, there was an urge to use the phrase “sit on this” as a descriptor; but a higher sense of decorum won out.

In the 17th Century, the largely unchanged three-legged stool underwent a transformation.  By joining the three uprights of the stool near the bottom, using three horizontal slats; the stool became significantly stronger and a lesser thickness of wood was required for the legs.  This joined design has pervaded into the modern age, as a sign of higher quality stool construction.

Returning to the Triple Aim, let us look more closely at the stool as originally built. Each leg has its own merits and challenges:

  • Improving the Patient Experience of Health – This includes such things as quality of care, patient satisfaction, and access. This leg relies on both the empowerment of the patient and the collaboration of care team for its strength.
  • Enhancing the Health of Populations – Improvement of the health status of the general population is at the surface of this leg. As we dig deeper, we also see that select populations (demographic, morbidity, or otherwise segmented) are targeted for improvement.
  • Reducing the Per Capita Cost of Healthcare – Financial motivators such as fee for value, bundled outcome payments, and penalties for preventable events make up this leg; which relies on money as a behavioral motivator.

Rather than attempt to paraphrase the approach endorsed within the Triple Aim; it is best to go to the source and share the description of approach as set forth by the IHI:

In most health care settings today, no one is accountable for all three dimensions of the IHI Triple Aim. For the health of our communities, for the health of our school systems, and for the health of all our patients, we need to address all three of the Triple Aim dimensions at the same time.

Because the IHI Triple Aim entails ambitious improvement at all levels of the system, we advocate a systematic approach to change. Based on six phases of pilot testing with over 100 organizations around the world, IHI recommends a change process that includes: identification of target populations; definition of system aims and measures; development of a portfolio of project work that is sufficiently strong to move system-level results and rapid testing and scale up that is adapted to local needs and conditions.

IHI believes that to do this work effectively, it’s important to harness a range of community determinants of health, empower individuals and families, substantially broaden the role and impact of primary care and other community based services, and assure a seamless journey through the whole system of care throughout a person’s life.

In the US environment many areas of health reform can be furthered and strengthened by Triple Aim thinking, including: accountable care organizations (ACOs), bundled payments, and other innovative financing approaches; new models of primary care, such as patient-centered medical homes; sanctions levied for avoidable events, such as hospital readmissions or infections; and the integration of information technology.

It is the very last statement that contains the greatest promise for maximization of the Triple Aim. By harnessing the power of IT; all three legs of the stool become stronger.  Analytics holds the key to being able to effectively create a better three-legged stool.  There are already frameworks on the marketplace that are designed to furnish analysis for any one given leg.  The problem becomes trying to synthesize these differing information sources into a single view.

In order to have trusted information to effect change in all three aspects of the Triple Aim; it is necessary to find an analytics solution that addresses all three Aims. To achieve that goal requires an application that can gather data from a single, trusted, 360-degree view of healthcare.  It also means that all data (clinical quality as well as payer cost) must be extracted, normalized, cleansed, and put into a reliable and trusted data model.

So, have you found your joined three-legged stool yet?

18 Sep

The Lowest Common Denominator of Prescriptive Analytics

Prescriptive analytics automatically synthesizes big data, mathematical sciences, business rules, and machine learning to make predictions and then suggests decision options to take advantage of the predictions. Prescriptive analytics is the third phase of business analytics (BA) which includes descriptive, predictive and prescriptive analytics.

The most traditional of business analytics is descriptive analytics, and it accounts for the majority of all business analytics today. Descriptive analytics looks at past performance and understands that performance by mining historical data to look for the reasons behind past success or failure. Almost all management reporting within healthcare (such as quality, clinical, administrative, operations, and finance) uses this type of retrospective   analysis.

The next evolutionary phase of BA is predictive analytics. This is when historical performance data is combined with rules, algorithms, and occasionally external data to determine the probable future outcome of an event or a likelihood of a situation occurring. In short, we look to the past to determine what the future will be.

The final phase is prescriptive analytics. Prescriptive analytics goes beyond predicting future outcomes by also suggesting actions to benefit from the predictions and showing the decision maker the implications of each decision option.  Prescriptive analytics not only anticipates what will happen and when it will happen, but also why it will happen. Further, prescriptive analytics can suggest decision options on how to take advantage of a future opportunity or mitigate a future risk and illustrate the implication of each decision option. In practice, prescriptive analytics can continually and automatically process new data to improve prediction accuracy and provide better decision options.

Prescriptive analytics has been in existence since about 2003. The technology behind prescriptive analytics synergistically combines data, business rules, and mathematical models. The data inputs to prescriptive analytics may come from multiple sources: internal, such as inside a corporation; and external, also known as environmental data. The data may also be structured, which includes numerical and categorical data, as well as unstructured data, such as text, images, audio, and video data, including big data. Business rules define the business process and include constraints, preferences, policies, best practices, and boundaries. Mathematical models are techniques derived from mathematical sciences and related disciplines including applied statistics, machine learning, operations research, and natural language processing.

Multiple factors are driving healthcare providers to dramatically improve business processes and operations as the United States healthcare industry embarks on the necessary migration from a largely fee-for service, volume-based system to a fee-for-outcome, value-based system. Prescriptive analytics will play a key role to help improve the performance in a number of healthcare areas.

Prescriptive analytics can benefit healthcare strategic planning by using analytics to leverage operational and usage data combined with data of external factors such as economic data, population demographic trends and population health trends, to more accurately plan for future capital investments such as new facilities and equipment utilization as well as understand the trade-offs between adding additional beds and expanding an existing facility versus building a new one.

In provider-payer negotiations, providers can improve their negotiating position with health insurers by developing a robust understanding of future service utilization. By accurately predicting utilization, providers can also better allocate personnel.

The greatest benefit to the healthcare industry will come in terms of bending the cost curve down, while simultaneously bending the quality curve up. The beneficiary will always be the patient/member in that they will receive the right care, at the right time, in the right setting, from the right provider.  In the process, the care delivered will assure that quality is enhanced parallel to an enhancement in financial and operational efficiency and an enhancement to cost containment.

While all of this talk of prescriptive analytics is seductive; there is a fundamental flaw that we have not discussed. Prescriptive analytics is built on top of predictive analytics.  In turn, predictive analytics is built upon descriptive analytics.  Finally, descriptive analytics is built upon a foundation of data.  If that data is incomplete, tainted, unstructured, or otherwise suspect in quality; we launch a domino effect.  Bad data quality begets poor descriptive analytics begets poorer predictive analytics begets poorest prescriptive analytics.

Are you ready to predict the future from a cloudy crystal ball or is it time to address your data access and quality issues today?

11 Sep

AP2CD™: The Last Healthcare Acronym You Will Ever Need

Put a freshly grilled steak between two starving dogs and it is inevitable that both canines will fight to their last breath to gain ownership of the morsel. In healthcare, that perfect T-bone steak presents itself as the comprehensive and trusted intersection of quality and administrative data.  Unfortunately, we also have two dogs fighting for possession of the healthcare analytics steak; payers and providers!

As veterans of healthcare, we are keenly aware that quality cannot truly be enhanced through claims (payer) data. We are also quick to realize that clinical (provider) data is a poor indicator of cost and access.  But what if we could somehow magically combine all of the provider data silos with all of the payer data silos? The All Provider and Payer Care Database™ (AP2CD™) promises to deliver the up until now elusive 360-degree view of the patient/member experience across the continuum of healthcare; fusing quality and administrative data in a single database.

The success of the AP2CD is dependent on being able to access all available data sources, penetrating the siloes they are stored within.  The least effort will come by extracting quality data from clinical HIEs and retrieving administrative data from current APCDs.  Effectiveness of analytics performed against these sources would have to assume that the HIE and APCD are capturing all data.  Knowing this not to be true; we must be prepared to mine information directly out of provider EHR and HIS systems, as well as from claims payment systems.  This becomes the “heavy lifting” effort that forms a solid data foundation.

Presuming we can tap into every available data source, our AP2CD is still flawed until such time as we can standardize data.  We must be able to take textual information and transform it into structured formats.  Maps must be established and maintained to accommodate all of the different local conventions used by payers and providers.  In short, every single piece of data needs to be put on common ground and in contextual concept.

Next we need to determine the measures that will be applied against the AP2CD information.  Naturally, we all wish to use established clinical quality indicators such as NQF, USPTF, and PCMH, to name a few.  We also wish to look at claims indicators such as HEDIS, cost, and access.  Then it will become necessary to venture into virgin territory; formulating queries that examine cause and effect across both the quality and administrative data.  Only then can we appreciate the 360-view of the healthcare experience for any patient/member, provider, and payer.

We could cite an infinite number of examples to illustrate the efficacy of the AP2CD.  For now, let us examine one illustrative scenario.  Our cost for the diabetic population in one benefit plan has risen due to a spike in Emergency Department visits.  Whether the visit was warranted or not will be excluded from this cursory analysis.  Rather, let us ask ourselves why the visit occurred.

Was it a question of access and the patient presented due to severe symptoms and had nowhere else to go for care? Was our benefit plan design flawed and there were not enough copay disincentives to have the member wait until the next day, when they could visit their PCP?  Were the additional diabetic patients that visited the ED attributable to one provider or one practice?  Were these patients chronically poorly controlled in their A1c measurements?  For that matter, had they been given an A1c test in the past year?  Were these patients suffering additional comorbidities that contributed to the ED visit?  Had the patients been counseled on preventive lifestyle changes that could have lessened the severity of symptoms; allowing them to avoid the ED altogether?  For that matter, were each of these patients also diagnosed as depressive, predisposing them to be more likely to seek immediate care; but their health plan had a weak mental wellness benefit, so they were poorly managed?

The above scenario is simplistic in nature. We can all attest to the complexity of any healthcare episode.  To understand this complexity requires the ability to examine quality and administrative data in a homogenous data foundation.   If we return to the backyard called healthcare, and look at analytics as the sizzling steak; there are still two hungry dogs circling the T-bone.  If the dogs cooperate, they can each share in the benefits of the steak.  If they fight over the steak, one of them has to lose.

By implementing the AP2CD™ (The All Provider and Payer Care Database™), we assure that all of the dogs are fed.  More importantly, we finally arrive at a 360-degree view of healthcare data that assures we can measure cause and effect with respect to quality enhancement and cost reduction efforts.  In that scenario, no dog goes hungry!

04 Sep

Common and Standardized Terminology Delivers Trusted Data for Trusted Analytics

Most of us are familiar with the Biblical story of the Tower of Babel.  In trying to reach the heavens, a group of people began erection of a massive tower.  Unfortunately they became arrogant about the endeavor and were cursed by God with the affliction of each speaking a different language.  Unable to communicate with one another; the builders abandoned the project and scattered to the ends of the earth.

The same can be said for healthcare as we build our own Tower of Accountable Care.  We are attempting to reach a heaven filled with higher quality and lower costs; yet we are still plagued with the fact that we speak different languages.

Regardless of your role in healthcare, the use of information technology and how well it leverages clinical and administrative terminologies affects you.  This holds true for providers, health information exchanges, payers, and Accountable Care Organizations.

The expectation for healthcare applications to pass, interpret and act upon medical record information continues to rise and simply having an interface connection between two systems is not enough. The information passed between systems must be “machine readable” and accurately represent the original clinical context.

In other words, it’s not enough to map which fields correlate between systems (allergy → allergy). The actual data received must be translated into the terminology code sets used by the receiving system (allergy A → allergy A). Otherwise, it is like having a conversation with someone in a language you do not understand.

Given the vast amounts of data and unstructured information stored in existing applications the industry cannot continue to manage terminologies as it has to date. It helps that a number of terminology sets have been designated as required communication “standards”, but they are not easily integrated into existing software solutions or helpful in understanding historical records. The industry needs a solution that enables everyone in healthcare to take control of terminologies and achieve semantic interoperability between any system that can contribute information valuable to patient care and business operations.

Providers – The way hospitals, health systems and physicians deliver care and manage their businesses is being molded by a variety of incentives, penalties and expanded reporting requirements. Much, if not all of these initiatives require unprecedented use and interpretation of the patient data stored in disparate software applications using a mix of standard, proprietary and home-grown terminologies. Increasingly success is defined by those who take control of how they use terminologies to capture, transmit, aggregate and understand patient data.

  • The expected benefits of common terminology include:
  •  Establishing and Maintaining Semantic Interoperability
  • Improvement in Quality Reporting Processes
  • Increased Operational Efficiency
  • Understanding Aggregated Clinical and Administrative Data

HIEs – Health Information Exchanges play a crucial role in supplying access to medical record information. Initially a portal solution for displaying referential patient information, HIEs are increasingly being asked to deliver actionable data to a wide variety of clinical applications. While HIEs make an ideal “central switch” for patient record data, there are significant challenges to maintaining semantic continuity as data passes to and from various software applications, each using a mix of standard, proprietary and local home-grown terminologies, or free text. Overcoming these challenges allows HIEs to deliver enhanced value, create new offerings and generate sustainable revenue.

The expected benefits of common terminology include:

  •  Delivery of Semantic Interoperability
  • Improved Operational Efficiency
  • Added Value to Quality Reporting
  • Interpretation of Aggregated Claims and Clinical Data
  • Establishment of Sustainability

Payers and ACOs – Improving care, controlling cost, and assessing risk requires access to accurate and normalized information.  There is no lack of data and more is becoming available as electronic medical record systems, all payer claims databases, and health information exchanges come online. For payers and accountable care organizations, success moving forward will be defined by their ability to make sense of the vast amount of data available.

The expected benefits of common terminology include:

  • Interpret Aggregated Clinical and Claims Data for Improved Analytics
  • Improve Operational Efficiency
  • Make Informed Decisions from Trusted Insights

Recap – As we can see, regardless of which side of the healthcare continuum you reside in; there is quantifiable benefit to the use of terminology services.  Standardization and normalization are your foundation for trusted data, which in turn leads to trusted analytics.  Maybe there is still hope for the Tower of Accountable Care!

29 Aug

Heavy Lifting of Data as a Precursor to Successful Healthcare Analytics

May 29th of this past year marked the sixtieth anniversary of a notable historical accomplishment.  In 1953, a British expedition, led by John Hunt arrived in Nepal.  Hunt selected two climbing pairs to attempt to reach the summit of Mount Everest. The first pair (Tom Bourdillon and Charles Evans) came within 100 meters of the summit on 26 May 1953; but turned back after running into oxygen problems. As planned, their work in route finding and breaking trail and their caches of extra oxygen were of great aid to the following pair. Two days later, the expedition made its second and final assault on the summit with its second climbing pair, the New Zealander Edmund Hillary and Tenzing Norgay, a Sherpa climber from Darjeeling, India. They reached the summit at 11:30 am local time on 29 May 1953.

At first glance, this anecdotal story has nothing to do with your healthcare analytics.  However, if we look at an analytics endeavor in light of a climbing expedition; the similarities begin to come to light!  The success of any climb comes from two critical factors; planning and heavy lifting.  To reach a pinnacle of 29,029 feet requires a methodical and well-thought out strategy.  The ascent is made in stages, with multiple base camps being established along the way.  Sixty years ago, it took over 40 climbers and 300 Sherpas to put two men on top of Everest.

By any other name, a Sherpa is a member of an indigenous Nepal people who are well-known as being the porters for climbing expeditions.  Able to function in high altitudes, incredibly nimble in their mountaineering skills, and confident in their abilities; Sherpas can carry packs weighing as much as they do.  NPR reported a study that showed this was possible through lowered metabolism and acclimation to steep inclines, all coupled with an inherent understanding of biomechanics!

Now picture your analytics solution as a mountain; perhaps even one in the Himalayas, depending on how tall your challenge is.  Any vendor can sell you a framework to accommodate your data.  Think of it as an empty backpack, waiting to be filled with your data.  Some of these vendors can sell you sets of measures as well.  Think of this as a walking stick, to help offset the weight of your data.  But these tools ring empty if you do not have somebody to find the data, load it into the backpack, and haul it up a mountainside!  How will that person find the scattered healthcare data? How will they assure its validity and integrity?  Will they know how to pack it correctly?  Finally, will they have the strength and experience to make a successful ascent?

The success of any healthcare analytics endeavor resides within your data.  Getting that data into your framework and deriving actionable knowledge is the critical factor.  Otherwise, what should be an investment merely becomes another expense.  Seek out a vendor who is willing to be more than a climbing companion; rather, find yourself a Sherpa!  Your vendor needs to be a “heavy lifter” who can dig data out of disparate sources, who can normalize it, validate it, properly load it, and carry it into your data warehouse.  A highly experienced Sherpa knows every nuance of the mountain.  The same holds true for your analytics vendor; they need to come out of the healthcare domain, preferably having operational experience on top of IT prowess.  Finally, your Sherpa needs to be able to help you plant the flag at the summit; surfacing knowledge from your analytics that is trustworthy, representing a 360-degree view of the healthcare continuum.

The life of the Sherpa is not easy and a trusted heavy lifter is difficult to find.  However, selecting the right healthcare analytics vendor can make a vast difference.  Would your organization rather come within 100 meters of the summit or reach the healthcare knowledge pinnacle?   A Sherpa can make a world of difference!

25 Aug

Data Marts as a Sustainability Strategy

Grocery chains all across this great land stumbled upon a quirk about human nature that has led to massive profits over time.  We all have a need to be presented with organization; call it a human frailty if you will.  Sections of supermarkets have been redesigned to resemble marts, if you will.  Traditionally, the mart was a place where sellers and buyers of specialized goods convened to do business.

The local supermarket offers virtually everything you might need in terms of food and household goods.  Since the 1950’s, a level of compartmentalization has transpired.  Now you walk in and head to your desired section; dairy, meats, bakery, and deli, to name a few.  It is almost like having a specialty store within a store; where you can dart in, get just what you need, and dart back out.

A data warehouse is not much different from that supermarket.  Assuming proper data extraction and cleansing; it has virtually every kind of healthcare data you would ever want.  (Please note that the previous sentence presumes you retrieved clinical, claims, operational, and financial data.)  A more sophisticated data warehouse owner is going to create marts of specialty data that appeal to the individual shopper.  It is a way of giving your user that store within a store.

A data mart is the access layer of the data warehouse environment that is used to get data out to the users. The data mart is a subset of the data warehouse that is usually oriented to a specific business line or team. In some deployments, each department or business unit is considered the owner of its data mart including all the hardware, software and data. This enables each department to use, manipulate and develop their data any way they see fit; without altering information inside other data marts or the data warehouse. In other deployments where conformed dimensions are used, this business unit ownership will not hold true for shared dimensions like customer, product, etc.

The primary use for a data mart is business intelligence (BI) applications. BI is used to gather, store, access and analyze data. The data mart can be used by smaller business units to utilize the data they have accumulated. A data mart can be less expensive than implementing a data warehouse, thus making it more cost effective for the small user. A data mart can also be set up in much less time than a data warehouse.  These facts make data analytics more attractive to the smaller organization.  An example of how this scenario would work is to assume that a larger organization (such as an HIE) builds and operates the data warehouse (our supermarket).  They then create a series of data marts for smaller organizations (IDN, Payer, PCMH, etc.) with specialized offerings of information pertinent to each of those organizations (dairy, bakery, meats, etc.).

There is value to healthcare analytics, much the same as there is value to that gallon of milk.  Unless you wish to become best friends with the store security staff; you are going to have to pay for that container of 2% moo juice!  The very same concept holds true for the owner of the data analytics infrastructure.  There is value in the data and analytics shared with constituents; so it is perfectly fair to charge for furnishing that information.

Therein lies a sustainability model for the HIE or central data aggregator.  Faced with diminishing funding options from grants and government programs; sustainability has to be addressed through recurring revenue; in this case from the sale of complete, cleansed, measured, and actionable information back to data contributors.  So, have you set up your deli section yet?

17 Aug

Your Organization Needs to Get SMART about Analytics

SMART is a mnemonic to guide people or organizations when they set objectives.  This Key Performance Indicator (KPI) approach is especially beneficial for project management, staff performance management, and personal development.  The approach also comes in handy when determining, applying and learning from your data through analytics.

SMART stands for:

  • Specific
  • Measureable
  • Achievable
  • Realistic
  • Timely

Each of these categories merits closer examination and a more precise definition.

Specific – This first term stresses the need for a specific goal over and against a more general one. This means the goal is clear and unambiguous; without vagaries and platitudes. To make goals specific, they must tell a team exactly what is expected, why is it important, who’s involved, where is it going to happen and which attributes are important.

A specific goal will usually answer the five “W” questions:

  • What: What do I want to accomplish?
  • Why: Specific reasons, purpose or benefits of accomplishing the goal.
  • Who: Who is involved?
  • Where: Identify a location.
  • Which: Identify requirements and constraints.

Measurable – This second term stresses the need for concrete criteria for measuring progress toward the attainment of the goal. The thought behind this is that if a goal is not measurable, it is not possible to know whether a team is making progress toward successful completion. Measuring progress is supposed to help a team stay on track, reach its target dates, and experience the exhilaration of achievement that spurs it on to continued effort required to reach the ultimate goal.

A measurable goal will usually answer questions such as:

  • How much?
  • How many?
  • How will I know when it is accomplished?

Attainable – This third term stresses the importance of goals that are realistic and attainable. While an attainable goal may stretch a team in order to achieve it, the goal is not extreme. That is, the goals are neither out of reach nor below standard performance, as these may be considered meaningless. When you identify goals that are most important to you, you begin to figure out ways you can make them come true. You develop the attitudes, abilities, skills, and financial capacity to reach them. The theory states that an attainable goal may cause goal-setters to identify previously overlooked opportunities to bring themselves closer to the achievement of their goals.

An attainable goal will usually answer the question:

  • How: How will the goal be accomplished?

Relevant – This fourth term stresses the importance of choosing goals that matter. A bank manager’s goal to “Make 50 peanut butter and jelly sandwiches by 2:00pm” may be Specific, Measurable, Attainable, and Time-Bound, but lacks Relevance. Many times you will need support to accomplish a goal: resources, a champion voice, someone to knock down obstacles. Goals that are relevant to your boss, your team, your organization will receive that needed support.  Relevant goals (when met) drive the team, department, and organization forward. A goal that supports or is in alignment with other goals would be considered a relevant goal.

A relevant goal can answer yes to these questions:

  • Does this seem worthwhile?
  • Is this the right time?
  • Does this match our other efforts/needs?
  • Are you the right person?
  • Is this acceptable for correction?

Timely – This fifth term stresses the importance of grounding goals within a time frame, giving them a target date. A commitment to a deadline helps a team focus their efforts on completion of the goal on or before the due date. This part of the S.M.A.R.T. goal criteria is intended to prevent goals from being overtaken by the day-to-day crises that invariably arise in an organization. A time-bound goal is intended to establish a sense of urgency.

A time-bound goal will usually answer the question:

  • When?
  • What can I do 6 months from now?
  • What can I do 6 weeks from now?
  • What can I do today?

Analytics has become a golden child within the healthcare industry.  Conferences spring up almost overnight to discuss “Healthcare Big Data” and “Predictive Models.”  Expensive seminars are held to proclaim the value of patient and provider behavior modification through analytics.  The public speaking tour is littered with cost containment and utilization management gurus, using analytics as the basis for their expertise.  Yet most healthcare organizations’ data quality and completeness is so suspect that analytics run against it will paint a false story, at best!

Apply the SMART method that we have just discussed in detail to your analytics approach and you will be assured of a successful outcome:

  • Specific – Decide exactly what you want to analyze and what benefits will be derived. You cannot analyze and act upon the vast universe of data that can be captured.
  • Measureable – Assure the highest data quality, know your sources, and work on your sickest populations and individuals first.
  • Achievable – You cannot predict the future if you do not understand the present. Assure your measures are well-defined, your data is trustworthy, and your results are actionable.
  •  Realistic – Your analytics approach has to recognize what can be surfaced and what can be acted upon. Information is just that, unless you have the resources to act upon it.
  • Timely – A phased approach always works best. Perform your population health analysis to determine what population groups need the most attention; then dig into the gaps in care within individuals in that group to transform information into knowledge through action