CRC (Other exams) cheat sheet

Risk adjustment predicts future healthcare cost from demographic factors and documented diagnoses; higher predicted cost produces higher capitated payment.

The risk adjustment factor (RAF) equals demographic coefficient plus HCC coefficients plus interaction coefficients, then adjusted by model-specific normalization; average risk equals 1.0.

A risk score of 1.25 predicts 25% higher cost than average, but it is not a dollar amount; payment is calculated by applying the score to a benchmark or bid.

Prospective models use prior-year diagnoses to predict next-year cost; concurrent models use current-year diagnoses for current-year payment.

CMS-HCC applies to Medicare Advantage, HHS-HCC to individual and small-group markets, and RxHCC to Medicare Part D drug spending.

Coding intensity adjustments and phased model blends reduce payment differences caused by variation in diagnosis capture rather than true illness burden.

Budget neutrality calibrates aggregate risk-adjusted payments so that increases in risk scores do not raise total program spending.

Risk scores are relative measures normalized to a national average of 1.0, allowing comparison across populations and payment years.

An HCC groups ICD-10-CM diagnoses with similar expected cost; multiple HCCs can be captured, but hierarchy limits payment within a condition family.

CMS-HCC risk scores combine age, sex, original Medicare or institutional status, Medicaid eligibility, disability, HCCs, and disease interactions.

HHS-HCC uses age, sex, metal tier, enrollment duration, and diagnosis-based HCCs for individual and small-group market risk adjustment.

RxHCC is a separate Part D model with its own diagnosis groupings, hierarchies, and coefficients for predicting drug costs.

HCC coefficients are regression-derived estimates of incremental predicted cost associated with a diagnosis group.

Hierarchy examples include metastatic cancer superseding primary cancer and diabetes with complications superseding uncomplicated diabetes within the same family.

Diagnoses that are nonspecific, historical, or symptom-based often do not map to a payment HCC and do not increase the risk score.

Model versions, mappings, and coefficients are updated annually; a diagnosis must map to an HCC in the payment-year model to count.

Assign the most specific ICD-10-CM code supported by documentation, including site, laterality, type, severity, and complication status.

Only conditions that are active, current, and documented as evaluated or treated in the encounter year are reportable for prospective risk adjustment.

Combination codes capture an underlying disease and its manifestation in one code, such as diabetes with retinopathy or CKD.

Chronic conditions must be documented and supported in each payment year to be recaptured; a prior-year code alone is insufficient.

Suspected, probable, or rule-out diagnoses are not coded; the provider must clarify the diagnosis before code assignment.

Signs and symptoms integral to a confirmed diagnosis are not coded separately when the diagnosis fully explains them.

Most history, status, and aftercare Z codes indicate resolved or administrative circumstances and generally do not map to payment HCCs.

Sequencing follows the reason for the encounter first when required, but all documented conditions that affect management may be coded.

Abstraction captures the diagnosis, date of service, provider identification, and clinical indicators supporting the condition.

Valid review sources include progress notes, discharge summaries, consultation notes, and laboratory or imaging reports; a problem list alone is usually insufficient.

The date of service must fall within the data collection or payment year specified by the risk adjustment model.

The documenting provider must be eligible for the model and the entry must be authenticated, dated, and legible.

Cloned, copied, or templated text must be validated against the current encounter and cannot be accepted solely because it appears in the record.

A compliant query is written, non-leading, cites the conflicting documentation, and requests clarification without suggesting a diagnosis.

Audit trails should identify the record source, page or location, reviewer, review date, and rationale for the abstraction decision.

Abstraction accuracy is assessed against a validated gold standard and through inter-rater reliability review.

MEAT stands for Monitor, Evaluate, Assess, or Treat; at least one element in the encounter supports the condition as active.

Monitor examples include tracking signs or symptoms, disease progression, or ordering tests; Evaluate includes interpreting test results or medication response.

Assess includes documenting the diagnosis or clinical status; Treat includes prescribing medication, therapy, or performing a procedure.

Chronic conditions require annual documentation and MEAT support to remain eligible in prospective risk adjustment models.

Hierarchy logic counts only the most severe HCC in a condition family, even when multiple related diagnoses are documented.

A suspect condition flagged by analytics cannot be coded until a provider documents and confirms it in the medical record.

Resolved or inactive conditions should be coded as history rather than current disease unless the provider documents ongoing activity.

Complication codes require causal linkage, such as documentation connecting kidney disease to diabetes, to support the combination diagnosis.

RADV is CMS Risk Adjustment Data Validation, which audits Medicare Advantage HCC diagnoses against medical records.

Unsupported diagnoses identified in RADV audits can result in repayment, error-rate calculation, and extrapolation to the audited population.

Knowingly submitting unsupported diagnoses or severity levels to increase payment can create False Claims Act liability.

An effective compliance program includes written policies, training, auditing and monitoring, reporting mechanisms, enforcement, and corrective action.

First-tier, downstream, and related entities must be monitored because their documentation and data can affect risk-adjustment submissions.

Medical records must not be altered to add diagnoses; any late amendment must be dated, attributed, and supported by the clinical facts.

Chart review can confirm existing documentation but cannot create a diagnosis that the treating provider did not document.

Program integrity includes self-monitoring, error correction, and appropriate disclosure of identified overpayments.

Diabetes mellitus is chronic hyperglycemia from insulin deficiency or resistance and can cause microvascular and macrovascular complications.

Chronic kidney disease is staged by glomerular filtration rate: stage 1 is at least 90, stage 2 is 60-89, stage 3a is 45-59, stage 3b is 30-44, stage 4 is 15-29, and stage 5 is below 15.

Heart failure is classified by ejection fraction as reduced, mildly reduced, or preserved, and documentation should identify type and acuity.

COPD is chronic airflow limitation; coding distinguishes exacerbation and obstruction when documented.

Coronary artery disease, angina, and myocardial infarction describe related but distinct ischemic conditions requiring specific documentation.

Stroke is classified as ischemic or hemorrhagic, and sequelae are coded differently from a current acute event.

Neoplasms require identification of primary versus secondary sites, active treatment status, and whether the condition is current or historical.

Sepsis, severe sepsis, and septic shock are distinct clinical states; documentation of organ dysfunction and causative organism affects code selection.