Targeted HCC Risk Coding Investment Based on Calculated ROI of Gap Closure

There are many steps in the calculation of gross shared savings. Some of them are fairly complicated, but the last step is not:

Gross Shared Savings = Updated Benchmark – Performance Year Expenditures

Most ACOs focus only on the right side of that equation, lowering performance year (PY) expenditures, even though it is (literally) only half of the equation. Oftentimes people treat the updated benchmark as if it is set in stone. It isn’t, though. It is the updated benchmark, and it gets updated using data from the PY. There are two updates made using PY data: risk adjustment, and market trend adjustment. You can’t really do anything about the market trends, but you can in fact have control over the risk adjustment portion of the updated benchmark.

How is the Updated Benchmark Calculated?

Before the PY data is all in, you will have what we call the Historical Benchmark. This quantity is set in stone by the time the PY starts. But to go from the Historical Benchmark to the Updated Benchmark, two adjustments based on PY data are made:

Updated Benchmark = Historical Benchmark * Market Trend Factor * Risk Ratio

While not always true, for the sake of simplicity we’re going to treat the market factor as out of your control. The only thing under your control is the risk ratio then. The calculation of the risk ratio is simple:

Risk Ratio = PY Risk Score / BY3 Risk Score

The updated benchmark, then, is impacted directly by the PY risk score.

Caveat on “PY” Risk Scores

Despite being called the PY risk score, events that occur during the performance year will not actually have an impact on the PY risk score.

Risk scores for a given PY are always calculated using claims from the year prior to the PY. For example, 2021 final risk scores will be calculated using claims from 2020. Therefore, when making an effort to improve your risk scores, it is important to remember that any changes implemented now will benefit the ACO in the next PY.

How are Risk Scores Calculated?

There are tons of different filters and conditions that are applied to claims to determine what exactly counts towards risk scores, but they are out of the scope of this article.

The most important thing to know here is that there are approximately 10,000 HCC eligible ICD-10 codes, and coding any of them will result in a beneficiary getting one HCC that the ICD-10 code gets mapped to. For example, there are very many codes that relate to diabetes, but all of them (that are HCC eligible) will result in the beneficiary either receiving the “Diabetes without complications” or “Diabetes with complications” HCC.

These HCCs are then used in combination with demographic factors to calculate a risk score.

A Simple and High ROI Method for Increasing Risk Scores

Increasing the updated benchmark boils down to increasing your PY risk score, which boils down to having more HCCs get coded.

Perhaps the simplest and most effective way to do this with the least possible effort is to close chronic HCC gaps.

A “gap” refers to HCCs that have been coded in prior years and has not been coded in the current year, and “chronic” refers to conditions that are… chronic.

For example, say we have a beneficiary that has HCC 111 (COPD) in 2021 (coded some time in 2020). That means we know that this beneficiary should also have HCC 111 in 2022 (coded in 2021) because COPD is chronic.

This sounds simple, but from the data we’ve seen, a lot of these chronic conditions are left uncoded year to year. Making sure these HCCs actually get coded is the simplest way we’ve found to improve risk scores.

Prioritizing High Impact Beneficiaries

The impact of any given beneficiary’s chronic HCC gaps on the total ACO risk score can vary wildly. Thus it is important to identify those with the biggest potential impact to the ACO risk score.

We do this by calculating what risk scores would be given current claims data, identifying the chronic HCC gaps, and then recalculating the risk scores as if those gaps had all been closed. This allows us to see, at a beneficiary level, how much of an impact gaps are having on the risk scores.

In our experience, closing the chronic gaps for a small fraction (usually 10-20% or so) of actionable beneficiaries when prioritized as outlined above can lead to a risk score improvement big enough that the ACO can hit the risk ratio cap (1.03 or 3%, e.g. BY3 risk score is 1.0 and PY is 1.03).

Example Using on Real Data

Below is a simplified chart made from data from real data.

Patient Category Risk Score Lift After Closing Gaps % of Theoretical Maximum Lift % of Maximum Actionable Lift
All patients 15.1% 100% N/A
Patients with visit in 2020 who haven’t passed away (actionable patients) 15.1% 100% 100%
Top 100 actionable patients 10.51% 69.6% 100%
Top 200 actionable patients 2.8% 18.5% 26.5%
Top 300 actionable patients 3.7% 24.5% 35.2%
Top 400 actionable patients 4.5% 29.8% 42.8%
Top 500 actionable patients 5.2% 34.3% 49.3%
Top 600 actionable patients 5.8% 38.3% 55.1%
Top 700 actionable patients 6.3% 42% 60.3%
Top 800 actionable patients 6.8% 45.3% 65%
Top 900 actionable patients 17.3% 48.4% 69.6%
Top 1000 actionable patients 7.7% 51.1% 73.4%
Top 1980 actionable patients 10.51% 69.6% 100%

Out of 1,980 patients we classified as actionable, just closing the gaps for the top 200 (~10% of actionable beneficiaries, <5% of total beneficiaries) achieved a lift in the risk score of 2.8%.

Most of the time, risk ratios aren’t terribly far from 1.0, as risk in patient populations often stays relatively steady, as does coding practices. In the case of a 1.0 risk ratio by default, a 2.8% increase in the (next) PY risk score of 2.8% means a risk ratio of 1.028, which means a 2.8% increase in the updated benchmark. A 2.8% increase for putting a bit of extra care into the coding for just 200 beneficiaries.

Opportunities for Provider Education

In many cases, there are obvious patterns in which HCCs are being inconsistently coded. Below is another simple chart made from actual data:

HCC Code HCC Name Total Increase Percentage of total Cumulative Percentage of Total
hcc58 Major Depressive, Bipolar, and Paranoid Disorders 363 18% 18%
hcc18 Diabetes with Chronic Complications 289 14% 32%
hcc85 Congestive Heart Failure 214 10% 42%
hcc111 Chronic Obstructive Pulmonary Disease 210 10% 52%
hcc40 Rheumatoid Arthritis & Inflammatory Connective Tissue Disease 194 9% 62%

We can see from the chart that over 60% of the missed risk score opportunity is left out due to failure to code just 5 HCCs. This may be a sign that these issues are simply not being brought up during visits with providers, and can be an excellent educational opportunity to ensure these conditions are being addressed during visits. This helps the patient and the ACO.