Simple Stratification Strategies for Success in Risk Contracting



One of the biggest challenges for any provider organization taking on risk is knowing where to focus your time and resources.

You will never have enough resources to cover all of your patients, so the trick is to have simple and effective means to transform and leverage your data to focus on the patients where you can have the most impact from both a quality and cost perspective.


We have compiled a few simple, yet very effective stratification strategies that are used by some of the most successful provider organizations in value-based care.



1.  Stratification of Patients by Gaps in Care Based on Those That Are Highest Risk of Hospitalization



It’s easy to get carried away trying to close gaps in care for all of your patients for your quality measures. This is understandable given that in most value-based contracts/programs quality performance can impact how much of your savings you get to keep.


However, for some measures, while you still want to close gaps for all patients, it makes sense to stratify first by those that are at highest risk for hospitalization if the gap is not closed. The CQM for Flu Vaccine is a perfect example.


During the 2018-2019 flu season, roughly 93 percent of patients hospitalized for flu-related complications had an underlying condition (National Foundation for Infectious Diseases). According to the CDC, some of the underlying conditions that typically cause serious complications include:


·        Asthma

·        Neurologic and neurodevelopment conditions

·        Blood disorders (such as sickle cell disease)

·        Chronic lung disease (such as chronic obstructive pulmonary disease [COPD] and cystic fibrosis)

·        Endocrine disorders (such as diabetes mellitus)

·        Heart disease (such as congenital heart disease, congestive heart failure and coronary artery disease)

·        Kidney disorders

·        Liver disorders


When you factor in the fact that the cost for the flu vaccine is typically less than $40 and the average cost of a hospital stay in the US is more than $22,000 (Lancet Public Health), it is a no-brainer for any provider organization to first focus on patients that are at highest risk for hospitalizations when closing care gaps for the flu measure.


Here is an example of how you do this in the Clinigence Health Platform:



Patient Stratification Strategies


In this example, a user goes into our procedure report, selects the PCPs and their assigned patients within the report, and then they open up side filters that allow for more granular filters to be applied. From the side filters, the user then selects the Disease Groups and/or DRGs that they want to filter by, as well as the procedure (via procedure category, subcategory or code). In this case, the procedure is the Flu Vaccine, but any procedure could be specified. The user can then elect to show patient claims for those Disease Groups that have either had that procedure or only show those that have not had it (by clicking the button to invert their filter criteria).


The Clinigence Health Platform also empowers users with reports to be able to filter by different diagnostic clusters. See below:



Here, the user selects the expanded diagnostic clusters (EDCs) that they want to apply (which are also labeled according to clinical & financial impact), then the PCPs, and then their patients/members. Not surprisingly, the predicted admission risk for these patients is already very high even without factoring in whether or not they have had (or not had) the flu vaccine.



       2.  Seeing Chronic Disease Patients Once Per Quarter


Another very easy strategy is simply to ensure all of your chronic disease patients are seen by their PCP at least once per quarter.

This accomplishes many goals at the same time, including:

·       Forges a stronger relationship between the PCP & the patient, increasing adherence to treatment plans

·       Allows for PCP to educate patient on how to manage their chronic conditions and when they should (and should not) go to the ER.

·       Ensures proper coding of new & existing chronic conditions

·       Gives PCPs ample opportunities for gap closure

·       AWVs will be performed during one of these visits too


At the end of the day, healthcare is about the relationship between providers and patients. And there is usually a strong correlation between a patient’s adherence to their treatment plan and how strong their relationship is with their PCP.

This ultimately keeps the patient out of the hospital, which is not only good for the quality of care delivered to that patient, but also for the bottom line of any provider organization taking risk – remember a primary care visit is usually around $150, but a hospital stay may be $22,000 (or more)!


See below how this is done in the Clinigence Health Platform:


Stratification of Patients by Last Office Visit


In this example, the user stratifies all of their patients and applies filters to show those that have two or more chronic conditions AND have not been seen in the last three months. Care coordinators would work directly off of this list to schedule appointments or export for use outside of the platform.

Additional patient-level detail may also be pulled from the analytics platform and shared with the provider in advance of their office visit, i.e., what gaps in care they have, what coding gaps they have, last ER visit, etc. and/or the data can be visualized within the EMR workflow at the point and time of care.


The Clinigence Health Platform brings these strategies to life. Click below to schedule a consultation with an expert from our team.


       3.  Using Predictive Modeling


Despite what any data analytics firm will tell you, there is no predictive model that eliminates all guesswork and gives you a perfect vision into what will happen within your population in the future.

However, predictive models can be very helpful tools in the stratification tool chest if used properly. Check out the example below from the Clinigence Health Platform:


Patient Stratification Using Predictive Modeling


Specifically, in our platform, we have integrated the Johns Hopkins ACGs. We leverage the ACGs to produce both a current and future risk score that is normalized to your population. The natural inclination may be to stratify by future predicted risk and focus on those at the very top, but it has been proven that those patients at the very top of the risk spectrum are often not “impactable”, meaning that the allocation of additional care coordination or care management resources may provide very little benefit from the perspective of both cost reduction and patient outcomes.


Instead, the user has selected to filter the patients by only those in the “Medium” risk group, which are ones where the current and future predicted risk scores fall somewhere in the middle of the pack. These are patients that are predicted to be pretty sick in the next 12 months, but are also likely to still be “impactable”, i.e., a reduction in cost and improvement in patient outcome & quality of care is likely to result from the allocation of additional care management resources.


This method of using a predictive model focused exclusively on the clinical perspective; however, another way to use this type of predictive model is to view it from a lens of the predicted cost and/or resource utilization.


Here is a screenshot of this in the Clinigence Health Platform:


Patient Stratification Using Predictive Modeling


This leverages the output of the predictive model (Johns Hopkins ACGs) to stratify by projected cost & resource utilization instead of purely on how sick the patient is clinically. Of course, how sick a patient is and how expensive their care is are generally correlative, but you still need to look at the complete picture to understand which patients are actually “impactable”.

This could be done many ways depending on your analytic capabilities, but in the above example, the user has selected the top resource (cost) bands and then we see all of the patients within those predicted resource bands, along with their predicted admission risk and likelihood of care coordination risk. The latter is calculated based on how complex the care is for that patient, i.e., how many different conditions they have, the number of medications they take, how many specialists they see, etc. This is an especially helpful data point because, as discussed earlier, there are never enough care coordination resources for every patient, and many patients may not benefit from additional resource allocation, so every data point that can be used to sharpen your focus is critical in the stratification process.



       4.  Medication Adherence


According to “Adherence and Health Care Costs”, published by Aurel Luga & Maura McGuire, medication nonadherence rates in the US range from 25% to 50%, depending on disease, patient characteristics and insurance coverage.


Furthermore, as cited by Luga & McGuire, an estimated 10% of hospitalizations in older adults may be caused by medication nonadherence. With an average Admission Rate Per Thousand for Medicare beneficiaries generally between 20-30%, having good medication adherence within your population can literally save you millions in your risk contract.


For example, in a Medicare risk contract with 10,000 lives, you would see around 2,500 admissions per year assuming an Admit Per Thousand rate of 25%. This means that potentially 250 of those admissions per year (10%) are the result of medication nonadherence. You can never eliminate 100% of medication nonadherence, so to be conservative, let us say you are only able to cut those 250 in half to 125. If your average paid per admission was only $10,000, that is a cost reduction of $1.25 million just from increasing medication adherence!


A good approach to executing on this strategy would be to borrow some from example #1 above, i.e., focus on those patients that would be at highest risk of hospitalization if they do not take their medications. And to understand who is not taking their medications, you will need analytic tools to mine your data to find medication gaps, preferably with the ability to drill down by chronic condition and drug class. See here from the Clinigence Health Platform:


Stratification of Patients by Pharmacy Gaps


Note that you won’t be able to calculate this type of data if you do not have pill counts in the claims files. This is a current limitation in Medicare CCLF claims feeds. If you cannot get medication gap data due to “gaps” in your claims data, then take a different approach and look for patients with chronic conditions that require routine medication (e.g., Diabetes) and stratify by those that have little or no spend on Pharmacy. It should go without saying, but they may not be filling the prescriptions because they don’t have pharmacy coverage. If this is true, help them get coverage – it will be better for the patient and better for your bottom line.


All of these strategies are fairly simple to implement and have been proven to be successful, but they only scratch the surface of the possibilities.

The Clinigence Health Platform brings these strategies to life. Click below to schedule a consultation with an expert from our team.