Chapter 36 — Quiz

26 questions: multiple choice and short answer. The answer key is in the collapsed block at the bottom. Every coefficient, benchmark, and dollar figure is a constructed teaching figure — risk models, mappings, and coefficients are revised annually and the version in force differs by contract year.


1. Under a risk-adjusted capitated arrangement, a health plan's payment for a member is driven primarily by:

  • A. The services furnished to that member during the month
  • B. The diagnoses documented and reported for that member during the previous year
  • C. The member's premium
  • D. The number of claims submitted

2. The primary purpose of risk adjustment is to:

  • A. Reward organizations for coding completely
  • B. Remove the financial incentive to avoid enrolling sick people
  • C. Replace fee-for-service payment
  • D. Measure quality of care

3. A risk adjustment factor of 1.0 means the member is expected to cost:

  • A. One thousand dollars per month
  • B. The same as an average member
  • C. Nothing, until a service is furnished
  • D. The plan's benchmark, exactly

4. Short answer: state the two things a risk model predicts and does not predict, in the form "the model predicts , not ."

5. The word "hierarchical" in hierarchical condition category refers to:

  • A. The ranking of plans by size
  • B. The ranking of categories within one disease family, so a member is counted once at the most severe
  • C. The order in which diagnoses appear on the claim
  • D. The seniority of the reviewing coder

6. A chart contains fourteen accurate ICD-10-CM codes and only two map to a condition category. This indicates:

  • A. A coding error
  • B. A mapping failure in the software
  • C. Normal operation — most ICD-10-CM codes map to no category in any risk model
  • D. That the encounter was not face-to-face

7. A record reports both diabetes without complications and diabetes with chronic complications for the same member in the same collection year. The model counts:

  • A. Both
  • B. The average of the two coefficients
  • C. One — the more severe
  • D. Neither, and flags the record

8. Reporting only the least severe version of a documented condition is:

  • A. The conservative and safe choice
  • B. Inaccurate, and the hierarchy guarantees it costs, because the model cannot promote a category that was never reached
  • C. Required when the coder is uncertain
  • D. Acceptable if the claim pays

9. An interaction term in a risk model is:

  • A. A code appended to the claim
  • B. A modifier
  • C. Computed by the model when two underlying conditions are each documented and reported
  • D. Assigned by the plan's medical director

10. Compute: a member's demographic component is 0.346 and her surviving disease coefficients are 0.328 and 0.302. There is no interaction. Her RAF is:

  • A. 0.630
  • B. 0.976
  • C. 1.278
  • D. 0.674

11. Compute: a member's RAF is 1.28 and the benchmark is \$1,050.00 per member per month. The payment for the year is:

  • A. \$1,344.00
  • B. \$12,600.00
  • C. \$16,128.00
  • D. \$13,440.00

12. Under the annual reset, a chronic condition documented in detail last year and not documented this year:

  • A. Carries forward automatically for permanent conditions
  • B. Is gone for next year's payment
  • C. Is carried forward for three years
  • D. Triggers a query from the plan

13. Short answer: which year's diagnoses set which year's payment?

14. Which of the following recaptures a chronic condition for the collection year?

  • A. An entry on the reviewed problem list
  • B. A laboratory result documenting the abnormality
  • C. A face-to-face encounter at which a diagnosing provider assesses the condition and states a plan
  • D. A medication refill authorized without an encounter

15. G0439 reports:

  • A. A subsequent annual wellness visit
  • B. A risk adjustment data validation audit
  • C. A quality measure numerator
  • D. A prolonged service

16. The MEAT criteria stand for:

  • A. Monitor, Evaluate, Assess/Address, Treat
  • B. Measure, Evaluate, Attest, Transmit
  • C. Medicare, Employer, Advantage, TRICARE
  • D. Monitor, Escalate, Audit, Track

17. How many of the four MEAT criteria must a note satisfy before a condition is reportable as addressed?

  • A. All four
  • B. Three
  • C. Two
  • D. Any one

18. Short answer: MEAT is not a federal regulation. What is the binding authority the mnemonic operationalizes?

19. A note reads "History of CHF." Reporting active chronic heart failure from that phrase is:

  • A. Correct, because the condition is chronic
  • B. Incorrect — "history of" is read by the classification as a past condition, and inferring an active one is a clinical judgment
  • C. Correct if the patient is on a diuretic
  • D. Correct only in the inpatient setting

20. The single design feature that separates a legitimate prospective chart review from an improper one is whether the prompt asks the clinician to:

  • A. Sign the note
  • B. Evaluate rather than affirm
  • C. Complete the visit within a time standard
  • D. Use the practice's preferred template

21. A retrospective chart review program has reviewed thousands of charts across several years and has never deleted a diagnosis. The most useful description of that program is:

  • A. Highly accurate
  • B. A one-way ratchet, and the first thing an external reviewer will ask about
  • C. Compliant, since deletions are optional
  • D. Understaffed

22. CMS's mechanism for auditing the diagnoses submitted for risk adjustment in Medicare Advantage is:

  • A. CERT
  • B. RADV — risk adjustment data validation
  • C. The Outpatient Code Editor
  • D. The MS-DRG Definitions Manual

23. CMS's contract-level RADV final rule, published in January 2023, provides that audit findings:

  • A. Will not be extrapolated
  • B. Will be extrapolated, beginning with payment year 2018, with no fee-for-service adjuster
  • C. Apply only to hospitals
  • D. Are advisory only

24. CPT Category II codes such as 0001F and 3044F:

  • A. Carry relative values and are paid at a reduced rate
  • B. Carry no relative value, are not paid, and exist to report quality measures
  • C. Replace the ICD-10-CM diagnosis on the claim
  • D. Are assigned only in the inpatient setting

25. 3044F reports a hemoglobin A1c below 7.0%. As a measure component, this is:

  • A. A process measure — an action taken
  • B. An outcome measure — a result achieved
  • C. A structural measure
  • D. A balancing measure

26. Short answer: in a shared-savings arrangement, why does more complete diagnosis coding usually not manufacture savings the way it raises revenue in a capitated arrangement?


Answer key 1. **B** — the diagnoses collected in the prior year build the score that pays every month of the following year. No claim in that system pays for a diagnosis. 2. **B** — a fixed per-member payment pays an organization to avoid sick people; adjusting for expected cost is the countermeasure. It is not a quality program and it is not a reward for coding. 3. **B** — the score is relative and the model is normalized so the average member sits near 1.0. 4. **The model predicts COST, not SICKNESS.** Conditions earn coefficients because people who have them cost more on average, not because they are medically grave. 5. **B** — the hierarchy is within a disease family; the most severe category trumps the rest. 6. **C** — a risk model is a cost-prediction instrument that selects a minority of diagnoses. Symptom codes, most musculoskeletal conditions, and uncomplicated hypertension generally map to nothing. 7. **C** — trumping. The lower category in the same family is zeroed. 8. **B** — under-specifying is not conservative. The asymmetry is deliberate: over-reporting gains nothing and under-reporting loses everything. 9. **C** — there is no code for an interaction; it is computed from two conditions that must each be documented and reported independently. 10. **B** — 0.346 + 0.328 = 0.674; 0.674 + 0.302 = **0.976**. 11. **C** — 1.28 × 1,050.00 = 1,344.00 per month; 1,344.00 × 12 = **\$16,128.00**. 12. **B** — every condition resets to zero each collection year, however permanent it is. 13. **The collection year's diagnoses set the following payment year's score.** Year 1's encounters pay every month of year 2. 14. **C** — a documented, face-to-face encounter with a diagnosing provider who assessed the condition. A problem list, a lab, and a refill are not encounters. 15. **A** — subsequent annual wellness visit; frequently the scheduling anchor for recapture, with the caution that a preventive visit is not automatically a problem-oriented one. 16. **A** — Monitor, Evaluate, Assess/Address, Treat. 17. **D** — any one suffices. MEAT is a list of ways a note can demonstrate engagement, not a checklist requiring all four. 18. **Section IV of the ICD-10-CM Official Guidelines** (report conditions that coexist and require or affect patient care, treatment, or management), together with the general documentation standards and the payment rules' face-to-face encounter requirement. 19. **B** — Chapter 12 §12.9's history codes. The phrase is ambiguous in clinical usage and unambiguous in the classification. 20. **B** — evaluate, not affirm. Chapter 10's Case Study 2 is the fully worked failure. 21. **B** — the one-way ratchet. The deletion count is the number that makes a review program credible, and an auditor asks for it first. 22. **B** — RADV. 23. **B** — extrapolation beginning with payment year 2018 and no fee-for-service adjuster. Litigation followed; verify the current status of both the rule and the challenge. 24. **B** — five characters ending in F, no relative value, not paid, reported to support quality measurement. 25. **B** — an outcome: a value the patient's body produced. 0001F, by contrast, reports that an assessment occurred, which is a process. 26. **Because the shared-savings benchmark is itself risk-adjusted.** Better coding moves both sides of the comparison rather than only the revenue side — and programs including the Medicare Shared Savings Program cap how much an organization's risk score may grow between periods, precisely so documentation change cannot be mistaken for cost performance.