Statistics – the one discipline that almost anyone that has ever been forced through does everything in their power to forget anything they learned as soon as they can. I used to be a member of this crowd. In fact, the first statistics course I encountered in undergraduate school nearly ended my college career. The D I received in that course remains my proudest academic achievement. For a number of years, I thought my sadistical torture was over and behind me. Then I enrolled in a post graduate program at The University of Alabama at Birmingham and much to my chagrin, in order to finish a doctorate program I was going to have to complete five more statistics courses including an Applied Multivariate Statistics for Health Administration course. The icing on the cake was the requirement that my dissertation be based on statistically valid research. By the time I completed my doctorate degree, I fully understood the meaning of evidence-based leadership in healthcare administration.
What I have learned about ‘sadistics’ is that it is not nearly as complicated and intimidating as most people think. In fact, to prove my degree of neurocognitive disorder, I will admit that I now study quantitative methods for fun. They got to me at UAB. As I learn better the application of the discipline, I cannot get enough knowledge or understanding of it. For example, I am currently trying to get my head around Chua’s incredible article on Normative Decision Theory and its application to healthcare administration and strategy development. Pretty good stuff for those of you that are interested in making yourselves better at what you do.
For the most part, statisticians are interested in understanding if this thing is different from that thing or if there is some kind of association between two or more things. Some of them even try to make sense of the magnitude and frequency of the vibration (variance) of things. They spend time trying to figure out what the probability might be that they are drawing the correct conclusion from what they are seeing. Why anyone in healthcare administration would care about this concept escapes me. One of the things that statisticians are obsessed with that does have profound application in healthcare administration is what they call ‘effect’ or more precisely, isolation of effect. In other words, what the hell is happening and does the thing I think is causing what I am seeing really have anything to do with it?
As healthcare administrators, we spend a lot of (too much) time looking at data. What we really need is information. What in the data is relevant? Is the change we are seeing in the data significant or is the change within the realm of expected normal variation? Two of the worst mistakes we can make as leaders are to fail to react to data that is telling us something is not right or reacting to something when there is no legitimate reason for concern. For those of you sensing statistical ghosts, I am talking about type I and type II (alpha and beta) errors.
The CMS is increasing its focus on payment for outcome instead of payment for throughput. In this process, the risk associated with outcomes that vary (variance) from the expected outcome that will be the basis of reimbursement and that risk will be borne solely by the provider.
Those of you that follow my blog regularly know that I frequently digress and I am feeling the need to digress right now. Sometimes, my digressions devolve to rants and I am feeling one coming on. I have heard about all I care to hear about population health management and integrated delivery systems. The largest part the cost of an episode of illness in most cases is the hospitalization. Leadership teams are distracted into grandiose fantasies of managing the health of large populations. The hypocrisy of this is that the same leadership teams cannot find the courage, will or means to manage the population under their direct control. If there were ever a group of people that know how to utilize healthcare benefits, it is healthcare workers. Healthcare groups routinely utilize healthcare benefits at higher rates than their non-healthcare counter-parts. What I do not understand is why provider organization leaders I have worked with in multiple organizations go to meetings to obsess over medical homes and population health management while they fail or refuse to do anything about the inappropriate utilization that is occurring right under their noses. One leader I worked with was talking about medical homes before most people had any idea what a medical home was. In spite of the fact that the organization had all of the resources necessary to develop and implement a medical home for its employees, they did nothing to demonstrate what a medical home was or what it might accomplish while the leaders were running up and down the road selling the idea to other organizations. The message I was hearing was, “We are not doing anything about our medical costs and utilization but trust us, we will help you manage yours.” Why would anyone expect to be believed while doing this? I postulate that if you are the low cost, high quality provider, organizations needing the services you provide will be blazing a path to your door. If you are not a low cost, high quality provider, either your organization will be left behind, it will become a part of a larger organization with utilization management capabilities or someone else will be running it. I do not believe that the responsibility to achieve better outcomes and cost can be outsourced.
One of the fascinating aspects of healthcare cost is that a disproportionate amount of cost is concentrated in a relatively small number of resource intensive or catastrophic cases. It is not unusual to see over half of the entire cost of a health plan was driven by less than five percent of the covered lives. There is a lot in the press about the very high percentage of Medicare expenditures during the final two weeks of life. The CMS is currently proposing to establish reimbursement for physicians to manage palliative care. The point of this is that you do not have to manage the health of a population. You have to case manage the cost that is being driven by a very small part of the population. The question is who are the people? What are their problems? Are resources being allocated efficiently for the best benefit of the patients? In order to manage a population, we have to identify those at risk of becoming large claims and help them manage their conditions including medical costs incurred outside the hospital. The other 95% will take care of themselves.
A good friend and classmate of mine was recently appointed CEO of a metropolitan safety net hospital. Joe’s dissertation topic was on population health management so I suspicion he might know something about the topic. If there were ever a population needing help managing their health, it is the people that utilize safety net hospitals. What do you think one of the first things Joe did after arriving at the hospital? He hired a Ph. D. statistician to head up his clinical quality department. This is the difference between Joe and me. Because of my background, experience and training, I tend to focus on helping the hospital reduce its cost by operating more efficiently. Joe’s approach is to manage cost by stopping inefficient and inappropriate utilization of hospital services. If we ever get a chance to work together, we are going to set some organization on fire. We won’t make the cover of Rolling Stones but we might make the cover of Modern Healthcare.
Why would Joe do such a thing anyway? Do you think it might be to have someone in his charge that could make sense of the mountain of data every hospital accumulates and give him some information that he can respond in ways that might improve patient outcomes and hospital costs? I wonder if it might have anything to do with isolation of effects that are driving costs and outcome and focusing very limited resources in areas that might make a difference? Do you think the application of quantitative methods might have some potential benefit to help prioritize initiatives that would make a material difference quickly? Do you wonder how long it will be before Joe knows which of his physicians’ practice is too expensive for the hospital to afford? Too often, leadership teams get drawn off their focus by trivial distractions that have little if any potential to make a lasting difference in the organization while issues that are costing thousands of dollars per day are ignored. If you really want to get fancy, you can start thinking about variance and the analysis of variance or ANOVA as a means to understand the significance of the effect that is being observed.
A leadership team has a limited ability to focus because of a phenomenon that I describe as bandwidth constraint or not enough of us to go around. The more items the leadership team takes up, its ability to provide concentrated focus on any particular item is degraded. I have been frustrated by my failed efforts to get leadership teams to agree on what is important and resolve themselves to focus on a few high priority items that everyone agrees will make a significant difference if implemented while ignoring distractions. A year later, the performance of the organization has failed to improve and most of the strategic initiatives remain incomplete and no one can remember what most of the time was spent on during the year. Anyone that has ever hunted birds knows what usually happens when a rabbit runs between the dogs and the birds. Where do you think this term ‘chasing rabbits’ came from anyway?
The world is changing quickly. Our ability as leaders to make sense of what is happening around us and to discern how to respond effectively will define our potential for success. One of the tools that can dramatically enhance this ability is to replace old skill sets that have little future potential with the skill sets necessary for survival in a different world. Among these is quantitative methods. The best part of this is that you do not have to know statistics but you need experts who do know statistics and who know how to mine ‘big data.’ If you are an occupant of most of the C-suite roles, you have become a generalist. We do not need the high technical knowledge required of many of those that report to us. But you do have to have the correct resources in the organization if you plan to survive.
Throughout this article, I have interchanged the proper use of the word ‘statistics’ with a form that resonates with anyone that has ever been tortured by study of the subject; ‘sadistics’, a term coined by another classmate, Dr. Jim Burkhart that is currently the CEO of a large, municipal safety net hospital.