A Primer on CMS’ Hierarchical Condition Category (HCC) Coding Physician Practice Roundtable A Primer on CMS’ Hierarchical Condition Category (HCC) Coding How risk adjustment affects our bottom line Board Update from the Physician Practice Roundtable Updated June 2019 The focus of today’s presentation is to demystify what Hierarchical Condition Categories, or HCCs, are. We’ll explain why HCCs matter for our organization and how advancing HCC coding proficiencies across our medical group can help us to: Receive the appropriate compensation for treating high-risk beneficiaries, and Ensure we’re not held against unreachable benchmarks in shared savings arrangements
Accurate risk adjustment critical for success How risk adjustment impacts payment Demographic Factors Disease Burden Risk adjustment in brief Used to predict health care costs based on the relative actuarial risk of patients Relies on medical record documentation and diagnosis coding Applied to providers to ensure performance-based payments adequately reflect patient complexity and risk Applied to health plans to mitigate the impacts of adverse selection and to stabilize premiums Determined by HCCs (conditions coded, mapped from ICDs) Risk Adjustment Factor (RAF) Score Impacts payment under Medicare risk contracts Supports care management activities Adjusts PMPM payments to MA plans and adjusts target benchmark for savings under MSSP, NGACO Offers clearer and fuller picture of patient health to enable better targeting of care management efforts Source: Financial Leadership Council interviews and analysis. In Medicare, CMS uses a Risk Adjustment Factor, or RAF, score to predict health care costs based on the risk of patients. While CMS has primarily used RAF scores to determine levels of health plan reimbursement in the past, as payers have shifted risk onto providers, RAF scores start to impact provider reimbursement and shared savings potentials. Risk adjustment is a way to ensure that you are reimbursed fairly based on the risk profile of your population. Essentially, if you care for a very sick population, your reimbursement should be higher than the reimbursement of someone who cares for a healthy one. CMS looks at two things to calculate RAF scores: the demographic profile of a patient and their disease burden. “HCCs”, or Hierarchical Condition Categories, map to groups of ICD-10 codes, and are used to determine disease burden. That is why medical record documentation and accurate diagnostic coding are so important. Across all risk contracts, HCC capture is important– if providers don’t code appropriately and to the highest degree of specificity, they risk lowering the financial benchmarks and payments from each contract. In short, sub-par coding efforts can hurt both our patients and our bottom line. I’ll get into this in detail in the slides to come, but first,
Three steps CMS uses to calculate payment using HCCs Calculate individual risk scores Determine plan average risk score Set corresponding adjustments, benchmarks1 Key Inputs: Disease burden (i.e. HCCs coded, mapped from ICDs) Disease interactions Demographics (e.g. age, sex, disability, Medicare status) Risk scores are aggregated across beneficiaries Risk scores are prospective (prior year risk scores used for future payments, benchmarks) In Medicare Advantage (MA): plans paid each month for HCC risk-adjusted beneficiaries In MSSP2, Next Generation ACOs: HCCs are used to risk- adjust financial benchmarks HCC coding impact If providers don’t code appropriately and to the highest degree of specificity, aggregated HCC codes will not capture the full risk burden and expected costs of beneficiaries If disease burden is under represented, risk adjustment factors (RAFs), financial benchmarks, and per member per month (PMPM) payments will all be lower A lower benchmark means it is more difficult to achieve savings in shared savings programs ! CMS-HCC risk adjustment is also used to determine reimbursement for the Hospital Value-Based Purchasing program. Medicare Shared Savings Program. Source: “Risk Adjustment Methodology Overview,” CMS, May 2012, available at: www.cms.gov; Advisory Board interviews and analysis. Let me explain the three steps CMS uses to determine payments and benchmarks under the HCC model. The first step is calculating an individual patient’s risk, which comes from their disease burden, their disease interactions (i.e. how a combination of diseases may adversely affect a patient), and finally, from their demographic information. This includes age, sex, disability, and Medicare and Medicaid status. Next, CMS determines a plan’s average risk score by aggregating scores across that plan’s beneficiaries. One important thing to note is that these risk scores are prospective in nature, meaning that risk scores determined for a plan’s beneficiaries in one year is actually applied to payment for the following year or years. This average risk score then impacts both the payments and benchmarks for participants in risk-based payment models. Under MA, plans are paid per member per month (PMPM) based on HCC risk adjustment from the previous year. For shared savings models, HCCs are used to risk-adjust financial benchmarks on which future savings potential relies.
Impact of accurate coding depends on program Improvement opportunity varies by magnitude and frequency Triennial Annually Other Benefits to Enhancing Documentation MA Shared Risk Increases total annual expenditure target, and therefore annual savings opportunity1 Sets a foundation for population health initiatives MSSP Used to trend past expenditures to benchmark year in re- basing at start of 3-year term Offers helpful insight into the health of your patient panel Magnitude (on per-beneficiary basis) NGACO² Change in RAF score compared to 2014 can drive up to 3% increase in benchmark Provides advantage for engaging commercial payers in risk Frequency Presumes contracts based on percent of premium cost target. Next Generation ACO model. Source: Financial Leadership Council interviews and analysis. We assessed the opportunity of each risk contract by looking at both the magnitude of HCCs in contract’s benchmark calculations, and how frequently these benchmarks are recalculated. MA wins in terms of magnitude-- Under MA, plans are paid per member per month (PMPM) based on HCC risk adjustment from the previous year. Improving HCC capture can increase the risk-adjusted capitated payments from CMS to the health plan, and providers may share in upside if they have risk-based MA (or capture full upside with a wholly owned plan). Remember, CMS calculates a plan’s average risk score by aggregating the risk scores of all the beneficiaries in the plan. But these risk scores are prospective—the risk scores determined for a plan’s beneficiaries in one year is actually applied to payment for the following year or years (the length of time depends on the contract). The more frequently a benchmark used to set payments in a plan is set, the more quickly you’ll see an impact from improved documentation and HCC capture. Under MA, the PMPM is recalculated annually, but adjusted every 6 months. So, in terms of frequency, MA wins again. But beyond Medicare payment, enhancing documentation compliance is a critical competency for succeeding in population health. It can offer greater insight into our patient panel and provide a comprehensive look across care settings. Commercial payers are also working to expand their risk contracts, offering better terms to providers who have proven value through risk-based models. And because commercial payers’ payment under MA plans is dictated by HCCs, these payers have skin in the game to incentivize physicians to capture patient complexity. And history tells us that where CMS goes, commercial payers eventually follow. Now is the time to get our physicians on board and perfect the process.
CMS-HCC hierarchies determined based on ICD-10 HCC diagnostic classification system Thousands of ICD-10 codes Map to 70 HCC codes 70 HCC codes fall into one of 25 categories Associated codes for HCC category: “diabetes body system” Diabetes coding hierarchy example HCC 17 - Diabetes with Acute Complications HCC 18 - Diabetes with Chronic Complications HCC 18 - Diabetes with Chronic Complications HCC 19 - Diabetes without Complications HCC 19 - Diabetes without Complications ! Only one code per HCC category is assigned; highest severity code used when both could apply Source: “Evaluation of the CMS-HCC Risk Adjustment Model,” CMS, March 2011, available at: www.cms.gov; Advisory Board interviews and analysis. Now that we know what’s at stake, let’s take a minute to review the basics of what the HCC code actually is. The CMS-HCC risk model has 70 distinct HCC codes that we can assign to a patient, each of which has a score correlated to the relative risk. These codes are sorted into 25 disease categories. HCC codes in the same HCC category are mutually exclusive, so when more than one code from a category applies, we only use the code for the most severe manifestation from that category. See the example on the bottom of the slide here for details on that: Over the course of a year, various providers may submit claims on a patient with multiple ICD codes that fall within the same HCC Category, such as the “Diabetes Body System” seen here. That patient may have been coded, for example, both HCC 18 - Diabetes with Chronic Complications, and HCC 19 - Diabetes without Complication. Since HCC 18 is a higher severity than HCC 19, the patient will only be assigned HCC 18 at the end of the year. And the patient's risk score will be calculated only from the regression score of HCC 18.
Minor HCCs coding inaccuracies largely impact RAFs Hierarchical Condition Categories (HCC) ICD-10-CM codes are arranged into clinically related categories ICD-10 codes map to a specific HCC, which is used in risk-adjusting Medicare capitation payment Risk Adjusted Factors (RAF) Patient’s RAF is calculated by the sum of the HCC weights added to the age and sex of the patient 85% of codes that drive the RAF score are generated by primary care providers HCC# ICD-10 code Description Risk factor 19 E11.9 Type 2 diabetes mellitus without complications 0.118 18 E136.28 Other specified diabetes mellitus with other skin complications 0.368 Example of material difference between codes HCC 18 vs HCC 19: 0.368-0.118= 0.250* $800 (PMPM1)= $200 Per member per month. For purposes of this exercise, this is an illustrative example of a PMPM value. Source: Financial Leadership Council interviews and analysis. So, what does this mean for a patient’s RAF score and provider payments? If a patient were to be tagged to both HCC 18 (Diabetes with Chronic Complications) and HCC 19 (Diabetes without Complications) over the course of a 12-month period, only HCC 18 would factor into their “clinical risk score”, as HCC 18 is the more severe diabetic condition. Patients can be tagged with multiple HCCs. As long as they fall into different categories there is no limit on the number of HCCs a single patient can be assigned. The sum of the weights for the HCCs coded over the course of a 12-month period are then rolled up into a patient’s RAF score. The RAF score is calculated based on the sum of HCC weights, along with patient demographic information. In the example on the bottom of the slide you can see that coding someone with diabetes with complications vs without complications has a major impact on the risk score, translating to a $200 dollar difference in expected monthly cost.
HCC and RAF score calculations in practice CMS-HCC expenditure prediction and risk score Hypothetical example of a female, 76 with several chronic and acute conditions1 Risk marker Incremental prediction Relative risk factor 1 Female, age 75-79 $3,409 0.457 2 Acute myocardial infarction (HCC 81) $2,681 0.359 3 Angina pectoris (HCC 83)* $0 - 4 COPD (HCC 108) $2,975 0.399 5 Renal failure (HCC 131) $2,745 0.368 6 Chest pain (HCC 166)** 7 Ankle sprain (HCC 162)** 8 TOTAL $11,810 1.583 Predicted values for demographic factors of age and sex Acute myocardial infarction and angina pectoris fall within the same hierarchy, so only the more severe code is counted Both COPD2 and renal failure are distinct HCC categories so they both get counted Chest pain and ankle sprain are excluded from the HCC payment model3 Total predicted expenditures are the sum of the individual increments. The total risk score is the sum of the individual relative factors. Predicted dollar values are from the Version 12 Aged-Disabled, Community Continuing Enrollee CMS-HCC model, as estimated using 2004 diagnostic data and 2005 spending data. Chronic obstructive pulmonary disease. Chest pain is excluded because it is associated with a variety of medical conditions; ankle sprain is excluded because it is typically transitory. Source: “Evaluation of the CMS-HCC Risk Adjustment Model,” CMS, March 2011, available at: www.cms.gov; Advisory Board interviews and analysis. Here is an example to show how an overall RAF score for a patient with multiple chronic conditions is calculated. The table represents the predicted expenditures and risk score for a woman, age 76, with multiple chronic and acute conditions. First, this woman receives a predicted risk factor and incremental payment based on her demographics, as you see in the first row of the table. As we can see in rows two and three, although this female receives codes for both AMI and angina pectoris, they both fall into the same HCC category. Therefore, she receives no extra payment for angina pectoris because AMI is the more severe manifestation of coronary artery disease. Rows four and five include HCCs from distinct categories (COPD and renal failure) and therefore also contribute additively to this person’s risk profile. And finally in rows 6 and 7, we see that the HCCs for major symptoms and other injuries are excluded from the payment calculation. This includes chest pain, which is a symptom associated with a variety of medical conditions ranging from minor to serious, and ankle sprain, which is typically transitory and has minimal implications for next year’s cost. Her total risk score is therefore the sum of the individual relative factors, or 1.583 as we see in the final row. Now, just imagine the different payment potential of this patient given her complexity if her chronic conditions were not documented.
Coding and documentation errors to look out for Coding errors Errors associated with chronic, active conditions Medical record does not contain legible signature, authentication Discrepancy exists between diagnosis codes and written description Does not indicate that diagnosis is being monitored, evaluated, addressed, or treated (MEAT) Highest degree of specificity was not assigned the most precise ICD-10 codes to fully explain the narrative description of the symptom or diagnosis in chart “History of” coding used when the condition is still active Chronic or coexisting conditions not documented or left out of clinical documentation A link or causal relationship is missing with failure to report a mandatory manifestation code Example documentation errors Documenting depressive disorder (311) but coding major depressive affective disorder (296.20) Documenting “very obese”, not “morbidly obese” Example coding errors Coding “Asthma” (493.90), not “Chronic Obstructive Asthma (492.20) Coding for “history of” COPD”, not “COPD controlled w/Advair” Example chronic, active conditions that should be coded once per year Diabetes HIV Amputations CHF Transplants COPD Source: “Risk Adjustment Documentation and Coding,” Asuris Northwest Health, 2014, available at: www.auris.com; “Medicare Risk-Adjustment & Correct Coding 101.” Advantage, Oct. 2014, available at: www.advantagecdn.com; Advisory Board interviews and analysis. Now that we walked through why HCCs matter and what they are, let’s talk about common coding pitfalls because coding can get pretty complicated. The role of individual physicians is to provide appropriate documentation and ICD coding specificity at the point of care to optimize overall HCC capture and RAF scores. This slide lays out some common HCC coding errors that we should keep an eye out for. First, many errors are linked to improper documentation. This can be something as simple as not having proper authentication in the medical record. Often, documentation errors happen when there is a discrepancy between the diagnosis code given and the written description. For example, if you document “depressive disorder” but code for “major depressive affective disorder” that code is actually unsubstantiated in the documentation, because those are separate diagnoses and carry different risk factors. Specificity and accuracy is key. We also need to watch out for coding errors. Not coding to the highest degree of specificity to fully explain the narrative can undermine your coding efforts. For example, coding for “asthma” instead of “chronic obstructive asthma” undermines the severity of the condition and carries a lower, inappropriate risk score. Using codes for “history of” conditions when the condition is still active similarly undermines the condition the patient is dealing with. Instead, if the condition is controlled, that should be noted along with how it is being controlled. So, for example, you would code for “COPD controlled with Advair” not “history of COPD”. Finally, many sources of coding error arise with improper coding of chronic or active conditions. Often, these are left out of clinical documentation or causal relationships between conditions are not noted. Providers should not code for “diabetes” and “neuropathy”, for example, if the patient has “diabetic neuropathy”. An easy win and of the upmost importance is to ensure chronic and active conditions are coded every 12 months. If this is not done, CMS will downgrade that patient’s risk score for the following year, even if the condition still exists. This includes ensuring to code chronic conditions such as diabetes, CHF, and COPD every year, as well as active conditions such as amputations, transplants, and HIV. Focusing on these “open conditions” as we call them, can be an easy win, and is a frequent misstep in HCC coding. Less than a quarter of patients’ chronic diseases are coded the year after an initial encounter—and even fewer the following year—negatively impacting both risk scores and patient care.
Four Ways to Improve HCC Documentation and Coding Prioritize Patient Problem Lists 1 Keeping problem lists up-to-date and comprehensive for every patient is imperative. This includes re- documenting all chronic conditions every 12 months and diagnosis codes for each patient encounter. Accurate problem lists support care, documentation, billing, and HCC credit for conditions that impact risk-adjusted payments while eliminating the need for retrospective chart reviews. Gather Baseline Data on HCC Capture 2 Review previous year’s billing data and problem lists for patients, keeping a close eye on discrepancies between billing data and documentation, and potential gaps in HCC coding. Gathering baseline data will enable medical group leaders to quantify their HCC opportunity and be better positioned to improve HCC complexity, engaging stakeholders as necessary. Launch Provider Engagement Initiative 3 Provider engagement is essential to capturing a patient’s full complexity, as many clinicians don’t fully understand the HCC system of coding and reimbursement. Medical groups should proactively educate clinicians around HCCs, highlighting how comprehensive problem lists facilitate more effective care plans and appropriate reimbursement. Embed HCC Management Tools into Workflows 4 Appropriate HCC management and support tools are critical, even when providers are engaged. These tools should be accessible to providers at the point of care, allowing them to make more informed decisions without adding “clicks”, aiding clinician productivity while realizing gains. Source: “Four ways to better track your patients’ complexity – and get paid for it,” The Advisory Board, June 3, 2015, available at: www.advisory.com; Advisory Board Company interviews and analysis. Now that we’ve seen how HCC coding works and some common documentation and coding errors, let’s look at four ways we can simplify and improve coding across our practice. While the HCC system may seem complex, it really boils down to providers doing one thing: keeping the medical chart up-to-date and comprehensive. Prioritizing patient problem lists for every patient is imperative. This means if a patient had a heart attack five years ago, it should be on the problem list of that patient’s medical chart. Accurate problem lists not only support our patient care but aid in our documentation, billing, and coding efforts. Next, we should take a baseline reading of our HCC capture. We should review previous year’s billing data and problem lists for patients; keep a close eye on any discrepancies between billing data and documentation, as well as potential gaps in HCC coding. For example: Are there items with HCC value in the billing data that aren’t on the problem list for particular patients? Are there items on the problem list but not in the billing data? Where do documentation and billing process challenges tend to recur? And is there clinical data in the EHR that suggests HCC-relevant conditions that haven’t been addressed? By gathering this data and having a baseline understanding, we can go on to fully quantify our HCC opportunity, and be better positioned to engage internal stakeholders and rally support. The next step is to launch a provider engagement initiative. The majority of clinicians don’t fully understand the HCC system of coding and reimbursement, so engagement is critical to capturing a patient’s full complexity. We should proactively educate clinicians around HCCs, which can include creating an FAQ document on what is relevant for them to know and framing the work in clinical terms. Messaging that demonstrates how a comprehensive problem list facilitates both an effective care plan and reimbursement that reflects true patient complexity can be particularly useful. Finally, we should think about how to embed HCC management tools in our workflows. Even if our staff and providers are engaged, they will be ineffective without appropriate support tools. Without these workflow tools, providers can lack effective prompting to consider a patient’s history, or take an action that conforms to a care protocol. Critical information should be accessible to providers at the point of care, allowing them to make more informed decisions without adding "clicks." Organizations that find innovative ways to do this without damaging clinician productivity will realize the biggest gains.