Benchmarking the nation’s HCC recapture
ACOs and Medicare Advantage plans are very accustomed to the annual requirement to recapture HCCs—hierarchical condition category codes—which are used for risk-adjusting their population and which determine, to a substantial degree, the budgets they are allocated to care for the patients they serve.
HCCs must be recaptured every year, even for conditions—like loss of limb—that necessarily persist over time. What we see in the organizations we work with, and in the country’s Medicare data overall, is that diagnoses are often not recaptured completely, which can cause underperforming coding organizations to be penalized even if the quality of care they deliver is otherwise high.
Understanding HCC capture across a nationwide population of digital twins
At HDAI, our team of data analysts works with de-identified patient data on Medicare’s Virtual Research Data Center, or VRDC, to understand the state of American healthcare. One way we look at HCC capture is by finding cohorts of patients who share historical recapture—so for any patient, rolled up to any ACO or group, we find patients who had the same pattern of capture or non-capture of the code in the previous three years.
Patients with the condition coded each of the last three years
Patients with the condition coded two and three years ago, but not coded this past year
Patients without the condition coded in any of the last three years
For most conditions, the most common pattern is 0 / 0 / 0: that is, the patient never had the condition. Most of those patients, as you would expect, don’t have the code captured in the following year, either. 3.7% of patients who have not had a diabetes code in the previous three years will have an uncomplicated diabetes code this year; fewer than 0.1% of patients without an HIV code over that period will have one this year.
For any given condition, the patients who are most likely to have the diagnosis this year are those who had it in the previous three years. For diabetes, the cohort with uncomplicated diabetes the past three years, 91.9% will have it diagnosed again this year; for HIV, it’s 98.2%.
Evolving capture rates throughout the year
For ACOs and providers evaluating their capture rates, it’s important to consider these cohorts of similar twins—don’t expect high diabetes rates for patients without a diabetes history, but also don’t expect 100% recapture among those with one. It’s also useful to understand how your peers perform at recapture throughout the year. For each of the cohorts of patients—those with and without each code in each of the previous three years—for every HCC, we look at capture rates by week.
The shapes differ depending on the code, but this data can help providers and population health leaders evaluate, at any point of the year, how they are performing relative to expectations. In general, capture is front-loaded—so if you expect to make up the difference late in the year, it will require a concerted effort.
In the graphs that follow, you can see the evolution of capture rates for any HCC you choose over the course of the last full calendar year.
Recapture rates across a full calendar year
Each line represents capture of this HCC as of the given point in the year, for a different population of patients based on whether the HCC was coded in the prior three years or not.
Click a combination in the legend to hide or show it.
Most codes are recaptured earlier than you might expect
For most conditions—at least for the patients who have a consistent history of having the condition—capture is about halfway complete by the spring, and 75% complete by summer. This makes sense: these patients are generally sicker and have providers who are managing these conditions, and they are more likely to have visits early in the year that address them. But what that means is that providers who haven’t recaptured these codes by summer are behind their peers—they shouldn’t bet that these HCC gaps will get closed late in the year.
When each combination reaches its full-year capture
The month each cohort crosses 25%, 50%, 75%, and 90% of its eventual full-year capture rate. Recently-coded conditions hit their milestones months earlier.
See your ACO’s capture performance for the year
We can provide performance by HCC for all the conditions your organization is coding (or missing)—no data required from you.
Even “forever” conditions are not always recaptured
One striking observation as we’ve looked at the data: recapture rates are not 100%, and sometimes are not especially close to 100%, even for conditions that do not resolve. Consider transplant status, cystic fibrosis, and Huntington’s—conditions that are both permanent and require ongoing health management. Recapture rates even for patients who had those codes in each of the past three years leave up to 5% of patients without a qualifying diagnosis in the next year.
That means at least one of two things, both worrying: either some of these patients are not getting management of these conditions, or a failure to code them means their providers are not being reimbursed in accordance to the costs they incur.
Highest recapture rates for persistently coded conditions
Full-year capture rate for every HCC, given it was documented in each of the previous three years.