HIM 360 Module 6 Risk Adjustment Short Paper Example

Sicker on Paper or Healthier on Paper? Risk Adjustment Coding in Juniper Health's Clinics

[Student Name]

Southern New Hampshire University

HIM 360: Coding and Classifications Systems II

Module Six Short Paper

[Instructor Name]

[Date]

The organization, setting and figures below are a composite written as a model document. No real employer, client, colleague or patient is described.

Sicker on Paper or Healthier on Paper? Risk Adjustment Coding in Juniper Health's Clinics

Every January, the chronic conditions of Juniper Health's Medicare Advantage and accountable care patients disappear from the payment record. A patient with heart failure, diabetic kidney disease and a history of stroke counts as healthy until a clinician documents those conditions again in a qualifying visit that year. Our clinics recapture only 71% of chronic conditions annually, which means nearly three in ten patients look healthier to payers than they are. This paper explains how hierarchical condition category (HCC) coding works, codes five cases from our clinics and describes the line between complete documentation and inflated risk scores.

How HCC Risk Adjustment Works

Medicare pays Advantage plans a monthly amount per enrollee, adjusted by a risk score that predicts how costly that person's care will be. The CMS-HCC model builds the score from demographic factors such as age, sex and Medicaid eligibility, plus a coefficient for each payment HCC the enrollee's diagnoses fall into, plus interaction terms for combinations such as diabetes with heart failure. A score of 1.0 represents an average beneficiary. Accountable care organizations in the Medicare Shared Savings Program also use HCC scores when their spending benchmarks are set, so the same coding affects Juniper's ACO contract.

Only certain diagnoses count. They must come from a face-to-face or qualifying telehealth encounter with an acceptable provider type, and diagnoses taken from diagnostic radiology reports are excluded. Many ICD-10-CM codes map to no payment HCC at all, and the 2024 revision of the model, known as V28, regrouped or removed a number of conditions, so coders should confirm mappings against the current year's tables rather than memorized lists.

Hierarchies and the Annual Reset

HCCs are arranged in hierarchies so that only the most severe condition in a family counts. Kautter et al. (2014), describing the closely related HHS-HCC model built for the individual insurance market, explained that hierarchies keep a patient from being paid twice for different severity levels of the same disease. A patient coded for diabetes with a chronic complication receives the complication category, not both it and uncomplicated diabetes.

Because scores are recalculated from each year's claims, a condition documented in March 2025 supports nothing in 2026. For stable chronic conditions such as an amputation status or a transplant, this means someone must still assess and record the condition in a visit every year. The reset is a design choice: it keeps scores current, but it also punishes organizations whose clinicians list conditions without addressing them.

What a Note Must Show

A diagnosis that appears only on the problem list or in a copied-forward history does not support an HCC. Auditors look for evidence that the provider monitored, evaluated, assessed or treated the condition at that visit, a convention the industry abbreviates as MEAT. A sentence such as "CKD stage 3a, eGFR 52 today, stable, continue lisinopril and recheck in six months" supports the code. The word "CKD" in a list of past diagnoses does not. Juniper's coders apply this test to every chronic condition before a claim is released.

Case 1: Diabetes With Kidney Disease

A 72-year-old woman's note documents type 2 diabetes and chronic kidney disease stage 3a, with a current eGFR and a medication change. Because the Alphabetic Index places kidney disease beneath the word "with" under diabetes, ICD-10-CM assumes the two conditions are related unless the provider says otherwise, so the coder assigns E11.22, which captures both the diabetes and its kidney complication, and adds N18.31 to report the stage. Coding E11.9 and N18.31 separately would understate the diabetes and ignore the guideline.

Case 2: Heart Failure on the Problem List Only

A 68-year-old man seen for a sinus infection has "CHF" on his problem list. The visit note discusses only the sinus symptoms and an antibiotic. No weight, edema check, medication review or plan mentions heart failure. The coder assigns J01.90 for acute sinusitis and does not code heart failure from this visit. Instead, the documentation team flags the patient for his annual wellness visit, where the physician can evaluate the heart failure and document its type and status if it is still present.

Case 3: Lung Cancer History

A 75-year-old woman had a lobectomy for lung cancer three years ago and completed treatment. Today's note says there is no evidence of recurrence and she remains on surveillance imaging. Guidelines direct coders to use a personal history code when a malignancy has been excised or eradicated, no further treatment is directed to it and there is no evidence of existing disease. The correct code is Z85.118, not a C34 code. Coding active cancer here would raise the risk score without clinical support, which is precisely the error that audits recover.

Case 4: Depression and Specificity

A 66-year-old man's note says "depression, on sertraline, doing better." Without further detail, the code is F32.A, depression unspecified. When the psychiatrist's note from the same year documents major depressive disorder, recurrent, moderate, the code becomes F33.1. The two codes carry different weight in the model. The difference comes entirely from what the clinician wrote, so the fix, if the clinical picture supports it, is a compliant query or education on specificity, never a coder's inference.

Case 5: Obesity and BMI

A nurse records a BMI of 42.3 for a 70-year-old woman, and the physician's assessment says "morbid obesity, discussed diet and referral to weight management." Guidelines allow BMI to be taken from other clinicians' documentation, but the associated diagnosis must come from the provider. The coder assigns E66.01 and Z68.41. If the physician had not documented obesity, the BMI code alone could not be reported.

Summary of Cases

Table 1 summarizes each case, its codes and the rule that decided it.

Table 1. Risk Adjustment Coding Cases

Note. Codes reflect the ICD-10-CM code set in use at the time of writing.

Coding Intensity and Its Limits

Risk adjustment creates an incentive to document more conditions, and research shows the incentive works. Kronick and Welch (2014) found that risk scores of Medicare Advantage enrollees grew faster than those of comparable fee-for-service beneficiaries, a pattern they attributed to coding intensity rather than sicker patients. Geruso and Layton (2020) followed beneficiaries who switched into Advantage plans and found their risk scores rose after the switch, even though the people themselves had not changed. Jacobs and Kronick (2018) compared plan risk scores with other measures of health, such as mortality, and found the scores implied more illness than those measures supported. Congress requires CMS to apply a coding pattern adjustment of at least 5.9% to Advantage scores partly for this reason, and CMS audits plans through risk adjustment data validation.

Where Juniper Should Stand

These studies do not mean that capturing chronic conditions is wrong. A patient whose heart failure goes undocumented makes Juniper look overpaid for a healthy population and leaves care managers blind to real needs. The line is simple to state: every coded condition must be present, addressed at a qualifying visit and documented by the provider that year. Chart reviews should look for unsupported codes as hard as they look for missing ones, and any improvement goal should measure accuracy in both directions.

Conclusion

HCC coding turns clinic notes into payment, which gives outpatient coders responsibility once reserved for inpatient work. The five cases show that the outcome depends on conventions, guideline rules and, above all, what the clinician addressed at the visit. Juniper's 71% recapture rate is a documentation problem worth fixing, and the research on coding intensity is a reminder that the fix must add only what the record supports.

References

Geruso, M., & Layton, T. (2020). Upcoding: Evidence from Medicare on squishy risk adjustment. Journal of Political Economy, 128(3), 984-1026. https://doi.org/10.1086/704756

Jacobs, P. D., & Kronick, R. (2018). Getting what we pay for: How do risk-based payments to Medicare Advantage plans compare with alternative measures of beneficiary health risk? Health Services Research, 53(6), 4997-5015. https://doi.org/10.1111/1475-6773.12977

Kautter, J., Pope, G. C., Ingber, M., Freeman, S., Patterson, L., Cohen, M., & Keenan, P. (2014). The HHS-HCC risk adjustment model for individual and small group markets under the Affordable Care Act. Medicare & Medicaid Research Review, 4(3), E1-E46. https://doi.org/10.5600/mmrr.004.03.a03

Kronick, R., & Welch, W. P. (2014). Measuring coding intensity in the Medicare Advantage program. Medicare & Medicaid Research Review, 4(2), E1-E19. https://doi.org/10.5600/mmrr.004.02.sa06