As a doctor, I can tell you that if you try to teach doctors risk-adjustment coding by using phrases like “Code to the greatest degree of specificity”, such messages will likely go in one ear and out the other as we do not throw around “coding vernacular” on a day to day basis.  You may be much more successful if you gently remind them that there are ICD-9 codes to use, for example, for their patients with peripheral vascular disease. Remind them to screen their diabetic patients for peripheral vascular disease and use the 250.7 range if appropriate.  Our ICD-9 books remind us to “Use additional code to identify manifestation” – in this case, also submit – 443.81 Peripheral Angiopathy In Diseases Classified Elsewhere. Coding this manifestation, as directed by coding guidelines, yields HCC 105 and RxHCC 106 – these in addition to the Diabetes HCC 15 that you captured with the 250.7 code.

As a primary care doctor, I know that diastolic heart failure often goes unreported in claims data.  Many primary care docs are unaware that diastolic heart failure may account for as many as 1/3 of the patients with heart failure symptoms.  In the absence of an established diagnosis of CHF (based on an echo with a low ejection fraction) physicians will often code such a patient’s recurrent encounters as “edema” or “shortness of breath”.  In actuality though, these patients may have normal systolic function,  but have “diastolic” heart failure- i.e. a stiff ventricle with impaired relaxation  as seen on a 2D echocardiogram, and thus will need diuretics and also to keep their heart rate down to promote better heart filling.  Diastolic heart failure is one of those 1,500+ ICD-9 codes that actually have Part C and Part D monetary value, namely it is mapped to HCC 80 as well as RxHCC 91. So, not being aware of how to appropriately use codes in the CMS-HCC model can lead to a double penalty for MA-PD plans.

This is yet another prime example of how provider education in the CMS-HCC model requires teaching doctors to think about how to use more appropriate diagnoses for patients.

CMS very much prefers that qualified providers amend their documentation of face-to-face encounters as lab and pathology data becomes available in the days after a clinic visit.  If a patient presented to the clinic with abdominal pain, and the labs return the next day showing elevated amylase and lipase consistent with acute pancreatitis, then the provider should amend the record to reflect acute pancreatitis which carries HCC value, instead of acute abdominal pain which carries zero value. 

Risk adjustment requires a whole new way of thinking.  Physicians should be taught that the goal of submitting risk-related ICD-9 codes is not for the purpose of justifying their CPT codes or E/M office visit.  Rather, risk-related ICD-9 codes are submitted to help CMS assess the costs of taking care of the disease burden among the beneficiaries in a particular Medicare Advantage plan and to adjust their payments to the plan accordingly.

 The 3,000 or so ICD-9 codes in the CMS-HCC model were chosen quite specifically for their ability to be a simplified reporting model of the more common conditions that increase health care expenditures.  The intent of the CMS version of the HCC risk adjustment model was not to list every single condition that was associated with expense, but rather to create a simplified reporting system using the most representative codes.   

Physicians should have confidence in submitting the applicable ICD-9 codes in the CMS-HCC model when, in their clinical opinion, the patient meets the criteria needed to use a particular ICD-9 code.  A good example of this is mild malnutrition.  This does not necessarily have to be a disease condition that the submitting physician has solely managed, rather, it should be a disease condition or comorbidity that either influences their medical decision making for that face-to-face encounter or that simply signifies that the patient’s condition will require the plan to outlay additional expenditures for the care of that condition.  While a nutritionist may have done the lion’s share of “the work” involved in confirming the diagnosis of mild malnutrition and in making a treatment plan, the physician should submit to plans ICD-9 codes for malnutrition when appropriate as the doctor likely had originally ordered the nutrition consult and because the state of malnutrition entered in their medical decision making, most probably in that it necessitated further lab work and clinical visits to monitor the patient’s status.

 So, submitting ICD-9 codes for risk adjustment is not about justifying a CPT code, rather it is to report to Medicare that plans and providers are caring for sicker patients.

Addendum 3/20/08:  It is, however, very important that the provider not just write down a diagnosis from a nutritionist (or any other non-qualifying provider) without documenting the pertinent exam, lab or clinical findings to support that diagnosis. If a provider includes a diagnosis of mild malnutrition, there better be other parts of the note that support that ICD-9, such as mention in the assessment that the patient has lost > 20% body weight or has BMI <18 , or mention that there were lab values consistent with malnutrition such as low serum albumin, or exam findings that show any atrophy or temporal wasting.  I say this because plans can only submit the “one best medical encounter” that supports this diagnosis.  Notes from the dietitian and lab reports cannot be submitted to support this diagnosis.  A coder should be able to see the diagnosis in the assessment section and supporting clinical commentary or other features of the note that support this diagnosis.  In other words, the “one best medical record” to validate this HCC must be self-contained and sufficient to justify the ICD-9 code underlying the HCC code. CMS auditors will not go beyond the face to face encounter record to validate an HCC code.

The more astute, business-savvy physicians currently in practice, especially larger group practices and IPAs, will soon discover that first-hand knowledge and understanding of the CMS-HCC risk adjustment model will bring them considerable bargaining power when it comes time to renew their contracts with Medicare Advantage Plans.  Obviously, payers cannot reward physicians directly for submitting certain ICD-9 codes in the disease model, as such might be interpreted as influencing a doctor’s decision on a diagnosis.  However, plans can and will likely reward those physician groups that can demonstrate an understanding of the CMS-HCC risk adjustment model.  Increasing payments to doctors for the more complex E/M office visits such as 99214 will encourage physicians to spend more time with their patients to discuss additional diseases.  Plans that reward groups well who have learned the model and the unique coding-style that it requires, will likely find that it is money well spent.

Plans then have to decide whether to consume enormous resources of their own educating their providers on the model, or should they instead encourage physician groups to learn about the model at the expense of the practices? Again, I believe the more savvy providers will have already invested time into their own education on CMS-HCC risk adjustment.  Doctors that do not embrace this model, and similarly, plans that do not fully understand the financial ramifications of the model, will surely be missing out on enormous financial opportunity.

Health care providers, physicians especially, have historically been poor coders.  Rushed doctors have the primary goal of submitting just enough ICD-9 codes to justify getting the office visit paid for.  There is little consideration given to a health plan’s need to receive more ICD-9 codes so that they can enhance their disease management data-mining or justify greater (i.e. more appropriate cost-based) payments from CMS.

Plans that excel at provider education will realize that it will take more than just replaying the same old, tired messages to providers about “coding to the greatest degree of specificity” and “submitting as many 5 digit ICD-9 codes as appropriate”.  Doctors may eventually be taught to code “more correctly”, but if they just keep submitting the same 13,000 ICD-9 codes that are not specifically in the CMS-HCC disease model, then education efforts are a waste of time and money. 

 Plans should put their in-house doctors and clinically-oriented staff to work analysing the 3,000+ ICD-9 codes that are mapped to HCC codes with monetary value to determine how providers can be specifically educated on the need to submit the ICD-9 codes that are in the disease model (assuming the face-to-face visit with the patient substantiates that those codes are appropriate indicators of their individual disease burden).

The best example of this is 250.40 Diabetes with Renal Manifestations.  Your goal is to remind doctors of what clinical and lab data are required to support this specific ICD-9 code that is worth $4,000 a year to a plan.  Specifically that “renal manifestations” may include microalbuminuria (microalbumin to creatinine ratio > 30) with diminished kidney function, such as GFR or filtration rate < 60 which is 50% of normal.  One specific “renal manifestation” is diabetic nephropathy which is clinically suspected in longstanding diabetics with progressively worsening proteinuria who have either a biopsy showing diabetic kidney damage or who have retinopathy. 

Incidentally, if the kidney function is < 50% normal, doctors should also be told to  consider if 585.3 Chronic Kidney Disease, Stage III, is appropriate, which also carries monetary value.

 Inattention to the specific ICD-9 codes in the CMS-HCC model and their relationships to other ICD-9 codes means that providers and plans will do a poor job of accurately portraying to CMS the overall disease burden of their beneficiaries and thus can only look forward to lower RAF scores and lower payments in the future.

 Additionally, failure of plans to identify patients at risk of developing more costly diseases leads to greater expense in the future.  A good example is patients with a serum creatinine > 2.0 or GFR < 50% who face future kidney failure and should be educated on dietary changes and how to protect their kidneys to delay the progression to fulminant disease.

 The fact that 585.3 Chronic Kidney Disease, Stage III, is in the CMS-HCC model is a perfect example of how coding for risk adjustment purposes is so closely tied to disease management.

The guidance from the CMS training manual is very skimpy on what to tell your providers about the importance of risk-related coding.  It simply suggests that MA organizations should educate their physicians that proper coding is important for an MAO to maintain viability and to ensure adequate payments to providers.  That’s about it.

The real answer is that proper and complete ICD-9 code submission is for now the best way for a plan to get an account of their beneficiaries’ overall disease burden.  Without proper claims data, plans have a more difficult challenge in determining which patients would best benefit from disease management intervention.  Rushed PCPs who habitually submit 250.02 for all their uncontrolled type 2 diabetics, lose out on that chance to let plans know that some of their diabetics actually have renal manifestations, 250.42, as defined by the presence of macroalbuminuria or microalbuminuria with diabetic retinopathy (American Journal of Kidney Disease 49(2) Suppl 2 Feb 2007) .  Without proper ICD-9 data, plans face a greater challenge in identifying members who need a targeted invention, in this case, an ace-inhibitor to reduce the proteinuria and stave off future kidney failure.  So, incomplete coding leads to losing opportunities for intervention.  Proper coding simply is the right thing to do for the patient.