Clinical Architecture Healthcare IT Blog

Recent News >

May 09, 2019

This year’s slogan, 4 Million Reasons to Celebrate, is a reminder of how diverse the field of..

Read more »

Closure of Documentation Gaps to Optimize HCC Capture

Risk adjustment is a process that is used to estimate future health spending for payers with Medicare Advantage or Affordable Care Act Marketplace health insurance plans, providers in Medicare Shared Savings Program accountable care organization contracts, and a variety of other value-based payment arrangements. While there are important differences in risk adjustment models across these contracts, in general they use ICD-10-CM codes that roll up to certain Hierarchical Condition Category (HCC) codes. These HCC codes represent chronic conditions, each of which are associated with different risk adjustment factor (RAF) scores. Individual RAF scores based on chronic condition HCC codes, in combination with demographic criteria, are summed up to yield a total RAF score that can be converted into a per-member dollar amount representing expected annual health expenditures.

To accurately estimate health expenditures, it often requires a coordinated effort across physicians, coding professionals, and/or management teams. Physicians document conditions and diagnoses for their patients, coding professionals ensure that the appropriate code systems and codes are used in conjunction with supporting clinical notes, and management teams implement educational interventions and clinical documentation improvement efforts to optimize HCC capture. While over-representation of HCC codes can be interpreted as fraud, the focus of this discussion is under-representation of HCC codes which translates into under-reimbursement for expenditures related to providing health care services to members with chronic conditions. For this discussion, I’ll provide reimbursement figures based on a Medicare Advantage Community, NonDual, Aged model, recognizing that dollar amounts would vary depending on patient demographics and other factors.

Consider a 68-year-old male patient who is being seen in a clinic after receiving lab results which showed markedly elevated fasting glucose and hemoglobin A1c. The physician performs a diabetic foot exam, refers the patient to ophthalmology for retinopathy screening, and starts the patient on metformin for glycemic management. The physician documents in a note the details of the assessment, evaluation, treatment plan, and follow-up for continued monitoring. In the electronic medical record, the physician enters a problem code for impaired glucose tolerance. While not technically incorrect, impaired glucose tolerance is not associated with an HCC code and would not factor in to the overall RAF score. On the other hand, a diagnosis of type 2 diabetes mellitus would have been a more specific and accurate description of the patient’s diagnosis and would also roll up to HCC 19 (diabetes without complications) which would result in additional reimbursement of $992.96/year which reflects the expected annual expenditures to manage uncomplicated diabetes.

While these errors may be intercepted by coding professionals, and documentation may be improved by offering physician education, structuring physician incentives, and implementing other approaches to improve HCC capture, consider how the use of technology-based solutions can rapidly perform similar assessments for an entire population. By performing rules-based logical reasoning of structured clinical data, the following additional types of gaps in clinical documentation can be found:

1) Undocumented Diagnoses. Several scenarios can lead to the discovery of undocumented diagnoses. In one scenario, a patient might receive care at multiple locations, and it is not until data are aggregated that we see the complete patient record and discover, for example, that a patient has an elevated hemoglobin A1c (indicating that the patient is diabetic) from a care facility other than where primary care services are normally obtained. In another scenario, a physician might be treating a patient for diabetes and documenting as such in clinical notes but erroneously omitting the diagnosis from the problem list. Using scalable technology to search for medications, lab results, and observations, Clinical Architecture helped one of the largest health systems in the U.S. identify approximately 3,500 patients with undocumented type 2 diabetes mellitus, 1,300 patients with undocumented heart failure, and 11 patients with undocumented COPD. These conditions have associated HCC codes that translate into RAF scores and revenue that is needed to care for patients with these chronic conditions.

2) Coding Specificity. Physicians are encouraged to code to the highest level of specificity. However, code systems like ICD-10-CM contain a large number of diagnoses that are sometimes deeply nested in its hierarchies. Consider a patient with type 2 diabetes and proteinuria. Entering a diagnosis of “Type 2 diabetes mellitus” (ICD-10-CM, E11) would translate into HCC 19 (Diabetes without Complication) and payment of $992.96/year (Community, NonDual, Aged model). However, a more specific diagnosis of “Type 2 diabetes mellitus with diabetic chronic kidney disease” (ICD-10-CM, E11.22) would translate into HCC 18 (Diabetes with Chronic Complications) and appropriate reimbursement of $2,875.83/year.

3) Annual Re-validation. Chronic conditions must be documented every calendar year so that chronic conditions can be factored into risk adjustment calculations. Reimbursement is not provided if patients are lost to follow-up, and the same scenario applies if patients are seen but not properly documented. Consider a patient who was seen last year and had a documented diagnosis of “Major depressive disorder, recurrent, mild” (ICD-10-CM, F33.0) which rolls up to HCC 59 (Major Depressive, Bipolar, and Paranoid Disorders). If a patient is seen in the current calendar year and has no depressive episodes on the current antidepressant regimen, one might not think to code any specific diagnoses. However, a diagnosis of “Major depressive disorder, recurrent, in full remission” (ICD-10-CM, F33.42), along with appropriate clinical notes, would indicate that the patient’s depression is still being actively managed and monitored and would result in appropriate reimbursement of $3,306.73/year. Additionally, patients with chronic conditions who have not had a visit in that calendar year and do not have any future appointments can be flagged so that annual wellness visits can be scheduled before year’s end.

4) Code System Verification. In many cases, physicians code problem lists using interface terminologies such as SNOMED CT because they contain terms that more closely reflect common clinical vernacular as compared with reference terminologies such as ICD-10-CM that are often used for research and statistical purposes in addition to HCC capture. However, interface terminologies do not roll up to HCC codes and therefore must be identified and translated into ICD-10-CM terms, often by professional coders. One might apply rules-based reasoning to identify diagnoses such as “Type 2 diabetes mellitus with hyperosmolar coma” (SNOMED CT, 190331003) to ensure that “Type 2 diabetes mellitus with hyperosmolarity with coma” (ICD-10-CM, E11.01) will be included on a claim.

In summary, there are many opportunities to identify undercoding and miscoding in structured clinical documentation. The closure of documentation gaps can translate into the proper payments that enable payers and providers to be accountable for the health and well-being of the populations they serve.

Related Posts:

  1. Understanding ICD-10-CM - Part III - A Terminology by the Book
  2. Understanding ICD-10-CM - Part II - What's In A Code?
  3. Understanding ICD-10-CM - Part I - Origin Story
  4. The Road to Precision Medicine - Part II - The First Pillar