The Formula for Optimizing the Value-Based Healthcare Equation
Adapt or die. As far back as Darwin, the principle of evolution has been applied to organizations. With changes in the business environment, those who adapt to change will thrive and survive while others die off. With the move toward value-based care, organizations need to formulate a new equation in order to provide the best care and remain viable. Optimizing care for patient populations has always been part of the equation, but another variable– receiving the right dollars to care for the population – often is ignored. To impact care and outcomes of the population in this new environment, the entire equation needs to be optimized.
Receiving the Right Dollars Is Key to the Value-Based Care Equation
If value-based care is an apple tree, the low hanging fruit is in getting the right dollars to care for the population. When the 2014 results for the Medicare Shared Savings Plan Accountable Care Organizations (MSSP ACOs) were released in 2015, a number of facilities examined their results to determine where things could have gone better – where the low hanging fruit was. In my conversations with a number of these organizations, the vast majority concluded that the easy pickings were in getting the right dollars in the door to care for the population in the first place. As part of their retrospective for the year, each organization identified a large number of patients that were not appropriately risk adjusted going into the year. As such, the organization did not receive the dollars to care for those patients that were more complex than anticipated. One organization estimated that 90 percent of the low hanging fruit for its ACO was to be found by appropriately identifying the risk of the beneficiaries.
Assessing the Risk of the Population
To appropriately assess the risk of the population, we need a new paradigm where we focus on documenting diagnoses instead of procedures. In a very simplified form, value-based care contracts determine the dollars per beneficiary by looking at demographics and underlying conditions of the beneficiaries. CMS determines the dollars for MSSP ACOs by the demographics of the covered population and its Hierarchical Condition Category (HCC) scoring system. The HCC score is the proxy for the underlying conditions of each beneficiary. If the HCC scores for individual beneficiaries are inaccurate, then the dollars provided to care for them do not reflect the care they need.
Would any organization be surprised to find out that roughly 20 percent of its patients have missing or inaccurate diagnoses that would affect this risk score? Studies that review charts of Medicare and Veterans Administration patients have conservatively shown this to be true.
Though there are certainly aberrations and mistakes where “up coding” makes patients appear to have more conditions or risks than they actually do (and consequently provides too many dollars to the ACO to care for that individual), the fee-for-service world we have been living in for decades hasn’t trained clinicians to capture and bill for diagnosis very accurately.
Fee-for-service reimbursement is based almost exclusively on procedures and care delivered, which is why it is crucial to ensure that these interventions are accurately documented. Though DRGs and other bundled payments have moved us down the payment continuum from fee-for-service toward value-based care, value-based contracts go further, allowing care organizations to take on a larger part or all of the risk for the beneficiary. Because the health status of the population is crucial to determining the payment to the care organization, the focus shifts from accuracy of documentation for all procedures and interventions, to accuracy of documentation of all diagnoses. Missing diagnoses result in inaccurate pricing of how much care a beneficiary would optimally require during the life of the contract.
4 Common Methods for Addressing Inaccurate Diagnoses Coding
1. High-Risk Gaps over Time
Most value-based care contracts review past diagnoses going back 1 to 2 years when calculating the risk score used to adjust the dollars-per-beneficiary paid to the care organization. If, during a visit, the clinician does not document a diagnosis, then over time that diagnosis may fall off of the adjustment calculation. That’s great if it’s not a persistent diagnosis, but if it is a chronic condition that is not regularly documented, the risk score for the patient may be artificially low.
One of the best ways to quickly identify beneficiaries that may have missing diagnoses is to compare historical HCC scores to the current HCC score. If the past score is high, but the current HCC score is low, then that patient may need to be evaluated so her past diagnoses that persist can be documented. Tools that enable comparison of past diagnoses with current diagnoses ensure the current picture is accurate and simplify this evaluation.
2. Persistent Diagnosis Tracking
A more granular use case involves persistent diagnoses that have fallen off or have not been evaluated and cared for in the current year. For example, a patient with a limb amputation documented years ago is in all likelihood still missing the limb. A worklist of beneficiaries who have not been evaluated in the last year with historical diagnoses of persistent conditions ensures they are seen and treated. The right risk is then considered when determining how many dollars the care organization receives to care for beneficiaries and their persistent conditions.
3. Code Adequacy Identification
Given two choices in a process, most people will pick the shortest or easiest one, even if this choice is inadvertent.
The same is true when choosing diagnosis codes in an EHR. One organization I have worked with in the past identified that its physicians were frequently choosing the diagnosis code for uncomplicated diabetes simply because that diagnosis showed up first in the EHR, when they should have been choosing the code for diabetes with complications. This inaccurate coding alone impacted payments to their organization on the order of six figures. After some training and adjustments to their system, the distribution of the diagnoses more accurately reflected the true risk of their population.
4. Identifying Likely Diagnoses
Believe it or not, a large percentage of patients being treated do not have the proper diagnosis documented. One way to identify these individuals is to leverage rules-based engines that evaluate clinical data to identify probable diagnoses. The classic example is identifying beneficiaries using Betaseron without a diagnosis of MS; or identifying regular use of albuterol inhalers and/or Singular without a diagnosis of asthma. A great example of a practical use of these types of tools is an ACO in the western U.S. that examined its diabetic population in 2015. By looking for patients who had A1c results or fasting blood sugars above a certain level, it identified roughly 10 percent of its diabetic ACO population as not having a diabetes diagnosis recorded when it should have. The ACO leadership reached out to the primary care physicians with information about patients with the likely diabetes diagnosis. Several false positives were discovered, but for the vast majority of patients, the clinicians either didn’t know that a diagnosis had not been recorded or found that the patient was diabetic and hadn’t been diagnosed or treated. Within one month, many of the patients were seen and had a diagnosis recorded.
Right Dollars to Care for the Right Conditions
Not many of the provider organizations negotiating value-based contracts focus on (or have the clout to negotiate) dollars specific to a diagnosis or population, but eventually, this too will become more important to the equation of profitable contracts. Even with bundled payments, I’m consistently surprised by how many organizations have little insight into their true costs associated with a DRG or bundled payment. As treatment protocols evolve, optimal care costs will change. The organizations that take time to evaluate the true cost of providing care for various diagnoses will ultimately be more prepared for this shift.
An ACO Case in Point
Partners Healthcare, the largest integrated healthcare delivery system and ACO in New England, understood the need to have access to information about the full scope of services provided to its patients, including cost and outcomes of care. Under the Bundled Payments for Care Improvement Initiative Model 2, an episode of care includes all of the services a patient receives for a certain health event, beginning with a qualifying inpatient admission and ending 30, 60 or 90 days after discharge. Partners needed a way to reliably capture all of this data to identify promising opportunities to improve care delivery and outcomes, while reducing cost and waste.
Using Health Catalyst’s Late-Binding™ Data Warehouse Platform, Key Process Analysis tool, and Bundled Payments analytics tool, Partners has been able to identify cost-driving clinical areas and then evaluate the cost and variation associated with care delivery for patients. Partners has developed on-demand access to 48 standardized episodes of care, with costs and coefficient of variation. This has allowed them to compare themselves to other providers and to communicate effectively with payers. And it has enabled patient care and service performance improvement initiatives based on their analytic insights.
Adding It All Up
Evolving toward value-based care requires a shift in thinking. In addition to optimizing the care for the patient population, accurately defining the risk of the population is, in many cases, an effective and easy step toward securing the needed resources to care for beneficiaries. Organizations can do this by leveraging tools that identify high-risk gaps, persistent diagnoses, adequacy of codes, and likely diagnoses. Ultimately, this leads to better patient outcomes and viable value-based care organizations.
Would you like to use or share these concepts? Download this quality improvement presentation highlighting the key main points.