One of the biggest goals of population health is to improve health outcomes and contain costs. So one of the best ways to do that is to proactively figure out which patients need what, different patient subgroups need different things. Defining those different patient subgroups is called risk stratification. So in this module, we're going to talk about the different kinds of stratification of patients. We're going to map those different levels of risk stratification onto different types of prevention. Then we're going to discuss a few examples of risk stratification for population health management. So what is risk stratification? A lot of people think of it as an ongoing process, and it's a process to categorize patients risk. It's based on a lot of different factors. Here are five of them that are typically used. Certainly the vital signs because of that is central to somebody's health, their medical history because that's an indication of where they are on the health continuum or risk continuum. Information about lifestyle, family, history, and even core things like age. When we talk about risk stratification, typically there's two different ways to represent it. One is categorical, so somebody falls into one of four different buckets. Either they're low risk, intermediate, risk is rising, or maybe they're high risk or it can be on a continuous scale and those scales can change based on what you really want to indicate. They can be 1-5 scales, 1-10 scales or even probabilities like, 1-100. There's many different ways, but these are a couple of the ways that people actually use in practice. Risk stratification, if it's going to work well, has to map onto some level of protection or action. So what I have represented in this slide is the relationship between the risk population, healthy population, a low-risk, a rising-risk or a high-risk subgroup. Then the ways in which it maps on to either primary prevention, secondary prevention, or the tertiary prevention's that really is disease management and being ahead of what we think of as what could be pretty difficult health events that somebody might run into because of their disease. By and large, this is a good framework for any kind of chronic disease care prevention. A lot of it maps on to other diseases. But for many of the stratified populations, they have foundations in health education, primary care, really looking at prevention programs and self-management and so this is a pretty generic map that can be used for many different chronic diseases. One of the things that most of us don't think about is the fact that the clinical screening recommendations are risk stratified, and they are risk stratified mostly by age and by sex. Here's an example from the US Preventive Services Task Force. Eight different recommendations around breast cancer, cervical cancer, colon cancer, lipid screening, vaccination for influenza or tetanus or pneumonia, and then diabetes screening. Each one of these has its own recommendation based on what we think of their risk stratification. The cancers in particular are very age and sex stratified. The vaccinations tend to be more uniformly recommended, doesn't matter what age or sex you are. Then the diabetes screening, which we've focused so much in this course so far, really starts early and should be progressing throughout somebody's life stages. What we've talked about, risk stratification basically in one dimension at a time, they can be combined. So a really great thing to do is to understand how a patient's risk score basically, are they very healthy or are they very sick, can interact with another risk score. In this case, it would be a surgical risk. This is a way to avoid peri-operative complications or respiratory fear of very significant things that can come because somebody is already sick and they're going into a high-risk situation. So being able to map out the surgical and the patient risk scores simultaneously allow plans to be put into place as to whether or not people can just be given information or maybe they need a phone screening, or maybe they actually need a pre-surgery visit. Bring the person in and really walk through the what if scenarios and what support do they need? A lot of the examples of patient risk that matter in these situations has to do with the comorbidities. Does somebody have asthma or hypertension or hyperlipidemia? In other cases, it's much more about their age or their weight and we want to keep into our mindset about risk stratification that it can be from many different angles and it should be to the best of our abilities, representing things that are important to the operations. Let's say it's patient population health improvement plan. We want to be able to put that in the context of somebody's already existing health. With a lot of the improvements in technology and electronic medical records, risk stratification is also becoming a lot more sophisticated. So what it means to be high risk can be based on what we've just been describing in terms of patient's past history. It's coming from the electronic medical record, but it could also be based on the utilization. So have they been coming into the emergency room or urgent care at a relatively high frequency. Another way would be to identify key and very severe health outcomes like unstable behavioral health or active cancer that put somebody up into the high-risk because of a single factor. If you think about the ways in which we've got all this data at hand, it means that we actually can be very fine tuned about the way that we do risk stratification based on what our goals or vision are for its use. For example, where we might have thought before about, "Oh, this person is low, moderate, or high" We can now have levels inside of each one of these risk stratifications and it can be mapped onto particular things like maybe it's the incorporation of information about where somebody lives or how they are utilizing the health care system, because that becomes an important part of the consideration when thinking about their care. A few things that we want you to remember as you think about risk stratification are it should be based on data. It should be something that anybody can look at and can see that the risk is associated with this patient because of this information, and a recognition that your risk score is only as good as the data that you have. So if somebody is a new patient and you don't have a lot of data, your risk stratification is probably not going to be as accurate as somebody that you've got a long established relationship with and you had a lot of information on them. So by and large, we know that we are becoming much more automated and the possibility of complex risk stratification is increasing. I just want to put in a plug that, "Wow, there's some great things about it" Being able to uncouple and to look inside can also be really important. As I've gone through risk stratification in this module, one of the things I didn't really cover was incorporation of social and environmental factors into risk stratification. I know it's a growing trend. People are trying to think about how to incorporate social determinants of health into risk assessment inside of healthcare systems. I just don't know any good examples right now about the way in which that's been done. That really represented a way to increase health equity. Often the incorporation of social determinants of health into risk strata simply follows or reinforces stereotypes. It's not really based on data, it's based on what people think that they know about the risk associated with different groups of people in society. I'm sure that there are a lot of health systems that are trying to work responsibly on incorporating social determinants of health that will ultimately increase health outcomes. I just don't know any right now. If you do, please let me know. Or this could really be a very important bright spot in the evolution of population health. It's so important that health care actually understand and incorporate the social determinants of health so that they really can help people heal and actually get the support that they need, both inside and outside of the clinic doors.
