The Importance of Getting Risk Adjustment Right
Repost from March, 2022:
Background
Anyone who has taken the CRC (Certified Risk Adjustment Coder) curriculum can tell you that risk adjustment was first utilized in the mid-90’s for Medicaid purposes. The goal of risk adjustment has always been to collect data on patients so that money being allocated could plan for not only the current diagnoses, but to allow for a model where that value is increased in correlation with the increasing costs of caring for patients with manifestations or complications of those known chronic conditions while also considering all comorbidities. RISK ADJUSTMENT DOES NOT COLLECT DIAGNOSIS CODES IN ORDER TO AUDIT BACK TO A SPECIFIC ENCOUNTER, BUT INSTEAD SEEKS TO DOCUMENT ALL CURRENT DIAGNOSES THAT ARE ANNOTATED OR DOCUMENTED BY AN APPROVED PROVIDER TYPE IN CONNECTION WITH A FACE-TO-FACE ENCOUNTER. The diagnoses collected are aggregated for a whole number (risk adjustment factor or RAF) for the patient for each year. The diagnoses are not required to be addressed or treated, but they must be actual current diagnoses. The goal isn’t to account for that specific visit, but to collect all current diagnoses for each patient for each year. There aren’t additional payments for finding the diagnosis more times. Once a diagnosis is picked up, the value for the diagnosis is applied to every month of that same year of service of that encounter year because it is the overall yearly risk score that is being adjusted. It is not a risk score per encounter in the way there is an RVU (relative value unit) of work for each encounter. This is the hardest concept for some to understand. Even some elected officials called foul without understanding this new modeling concept. They accused health plans of suddenly gaming the system because ICD codes were being reported at an all-time high while not realizing the truth is that we have been under-reporting diagnoses for decades and risk adjustment was correcting this problem. I decided to share this information below because I believe in risk adjustment, and I believe that if we do it properly that we can improve the health of millions. The train is clearly running off track and it will take all of us to correct its course.
I have had the privilege of being hired by one of the very first risk adjustment vendors nationally in 2008 and subsequently was invited to the very first CMS RADV training because one of our insurance clients didn’t have a coding director yet. I have worked closely with people at both CMS and HHS. I have met people at RTI who developed the HCC models for CMS and I have worked on many cases before the DOJ and OIG. Providers do the best they can to accurately document diagnoses, but many of them are unaware of certain coding rules and are shuffled through a chaotically packed schedule day. Health plans are held accountable for getting diagnoses submitted right, but they are often only forwarding on what they themselves have received on claims. AS THE ORIGINAL CRC CURRICULUM AUTHOR, AND CREATOR OF THE TAMPER™ ACRONYM, I FEEL THE NEED TO SPEAK UP SO THAT WE CAN ALL CODE ON COMMON GROUND. THIS IS VITAL TOWARD THE SUCCESS OF RISK ADJUSTMENT’S PURPOSE.
Risk Adjustment was created because FFS cannot help us become proactive clinically
Fee-For-Service (FFS) is when a provider submits a claim for an encounter or visit that was provided in the past and they seek reimbursement for the services or procedures that were provided. Procedure codes are supported by diagnosis codes to show “why” the encounter or visit was necessary, and they need to correlate with one another in a way that the diagnosis explains the medical necessity of the service that had been provided. We have long known that this model is a disservice to the providers on the front lines as they do not gain credit for complex medical decision making with the chronically ill patient population and can only bill for the relative value units (RVUs) that were provided directly on that visit date. We also have no correlation with FFS billing and coding with quality-of-care initiatives. It is difficult to determine if a provider who sees a certain type of patient more often does so with great success or if those visits were needed due to poor clinical outcomes, patient noncompliance, or some other factor. It is important to note that 5 to 10% of all claims are still being submitted using paper superbills which are not updated timely and do not allow for robust reporting. Another variable is the selection of EMR systems as some limit the number of diagnosis codes that can be added and others populate commonly used codes, both of which impact risk adjustment. While providers are incrementally paid for each visit or encounter, the issuers or health plans themselves are not paid this way by CMS or HHS in risk modeling. This disparity is part of the problem. Health plans are the guardians overseeing the care of patients and they have often been caught off guard with diagnoses that they were not aware the patients even had, and this is directly due to the under-reporting of health data (diagnosis codes in this case). The biggest losers in this scenario are the patients themselves. Diagnoses that are not reported are left out of analytics and planning. It is nearly impossible for health plans to prepare for the financial needs of manifestations and complications of diagnoses when they are unknown. It is commonplace for health plans to run all kinds of analytics. They compare what is known for each member from year to year and watch for diagnoses that “fall off” (of claims reporting for the year) unexpectedly. They run suspecting analytics to try to estimate clinical outcomes and what may be needed for differing patient trajectories and individualized to each patient. They analyze a host of factors from age, race, gender, socioeconomic status, geographical territories, known life-long and permanent conditions, medications that are being used, common clinical progressions of certain diagnoses, and more. When diagnoses are left out of this mix, two major negative impacts occur. One is that the financial reserve to help pay for those manifestations and complications is terribly under-budgeted and the other is that patients may not get enrolled for wellness programs that aim to keep patients out of the ER and avoid unnecessary hospitalizations. There are many patients who, with appropriate proactive clinical efforts, could avoid complications and manifestations.
What about MEAT?
MEAT (Monitor, Evaluate, Assess, or Treat) is an acronym that was used long before risk adjustment was even an idea. This acronym was made by coding professionals to help providers in selection of the appropriate Evaluation and Management (E/M) code (992xx-994xx). Its origins come from the green instructional pages in front of the E/M section of our procedure code book. Because of the below excerpt, providers were told to avoid documenting diagnoses (even when true current conditions) unless they were monitored, evaluated, assessed, or treated because those would be the only qualifying factors to influence a higher level of E/M service.
“Comorbidities/underlying diseases, in and of themselves, are not considered in selecting a level of E/M services unless their presence significantly increases the complexity of the medical decision making.” (CPT® 2022, American Medical Association)
Thus, a mantra followed to teach providers to avoid coding any diagnosis unless there was “MEAT.” This acronym was made up during an era where Fee-For-Service (FFS) payments were king and all payments were dependent upon the correct level of service being chosen. In FFS coding, diagnosis codes are only used to show medical necessity of the procedure or service. We are not paid on them at the provider level. Even though we have always had official diagnosis coding guidelines to code for all coexisting conditions, we ignored those guidelines for decades because we were not being paid on diagnosis codes, and there used to be fewer positions for diagnosis codes on claim forms. MEAT HAS NO PLACE IN RISK ADJUSTMENT (other than its utilization for providers to shape up correct levels of service for E/M FFS claims purposes) AND IN FACT, IT CAN BE HARMFUL TOWARD THE GOALS OF RISK ADJUSTMENT CODING WHERE WE NEED TO KNOW ALL CONDITIONS THAT EACH PATIENT HAS SO THAT WE CAN BETTER CALCULATE THEIR CURRENT AND FUTURE MEDICAL NEEDS. We have guidance from both CMS and HHS that MEAT is not needed during RADV or HRADV audits that is shared in a few pages below. Here is an excerpt from ICD guidelines. Recall that the word “and” in ICD guidelines means “and” or “or”:
ICD-10-CM: Section IV. Diagnostic Coding and Reporting Guidelines for Outpatient Services
G. ICD-10-CM code for the diagnosis, condition, problem, or other reason for encounter/visit
List first the ICD-10-CM code for the diagnosis, condition, problem, or other reason for encounter/visit shown in the medical record to be chiefly responsible for the services provided. List additional codes that describe any coexisting conditions. In some cases the first-listed diagnosis may be a symptom when a diagnosis has not been established (confirmed) by the physician. (ICD-10-CM, 2022)
J. Code all documented conditions that coexist
Code all documented conditions that coexist at the time of the encounter/ visit and require or affect patient care treatment or management. Do not code conditions that were previously treated and no longer exist. However, history codes (categories Z80-Z87) may be used as secondary codes if the historical condition or family history has an impact on current care or influences treatment (ICD-10-CM, 2022)
Change is hard and the MEAT acronym has been applied for so many years that coders believe that it is a requirement, when it was only really a guide in the correct selection of E/M services provided and as it pertains to claims processing of encounters. A provider cannot choose a higher paid E/M code just because of many diagnoses existing, but rather should only take credit for the relative value (RVU) of the work performed for that visit, meaning that only diagnoses that were managed or treated during that visit can count toward that level of service provided for that encounter. WE HAVE ALWAYS BEEN INSTRUCTED TO CODE FOR ALL COEXISTING CONDITIONS (EVEN THOSE NOT BEING TREATED TODAY) BECAUSE THEY AFFECT MEDICAL DECISION MAKING. A PROVIDER MUST STOP AND CONSIDER COMORBID CONDITIONS AND THE PRESCRIPTIONS OR THERAPIES USED TOWARD THEM WHEN TREATING MOST ANY OTHER CONDITIONS OR PROBLEM. THERE ARE DRUG INTERACTIONS TO BE TAKEN INTO CONSIDERATION, DELAYED HEALING IN SOME PATIENTS THAT MUST BE PART OF THE CLINICIANS THOUGHTS, AND THERE ARE MANIFESTATIONS AND COMPLICATIONS THAT CAN OCCUR IF THE PROVIDER DOES NOT THINK ABOUT ALL OF THE OTHER DIAGNOSES WHILE TREATING THE DIAGNOSES ADDRESSED IN THAT VISIT. This is one reason why providers have always complained about the E/M leveling as they have often felt that it does not give them proper credit for how intricate and deeply thoughtful patient visits can be, especially for patients with multiple chronic conditions and those on many medications. Before risk adjustment, it was actually more proper or correct to code for all diagnoses that were true during the visit or encounter and to only allow those that were managed or addressed on that visit to influence E/M leveling. We never followed that properly historically and it didn’t matter before risk adjustment for the reasons noted above.
What about those high-level national audits we see?
There certainly have been some cases that have made the national news in risk adjustment. I have worked on several such cases, and I can attest that actuarial accounting firms are often used for these audits because of the money tied to the diagnosis codes for the insurance carriers. The problem with this practice is that many actuarial firms lack coding expertise, and those who do have coding expertise are often still stuck in the past with coding vendors who are largely known for Fee-For-Service coding. In these situations, once a vendor is selected and the rules are agreed to and applied, the course is often set in stone; however, this doesn’t mean that the diagnoses reported were false or fraudulent if the patient really had them. It is all our responsibility to update our government systems through knowledge sharing so that everyone is on the same page. We are obviously struggling with this because we are still seeing differing opinions on proper coding within our own ranks. MANY GOOD INTENTIONED PEOPLE ARE JUST DIGGING UP OLD E/M AND FFS RULES AND RECYCLING THEM INSTEAD OF STOPPING TO NOTICE THAT RISK ADJUSTMENT CODING IS DIFFERENT FROM FFS CODING BY DESIGN. We know this is true also because of the annual coding intensity adjustment and the FFS normalization adjustment that CMS applies to MA risk scoring that is discussed further below. They fully expected this coding difference between FFS and risk adjustment so much that it is a part of the accounting process. This means that under-reporting diagnoses is cutting the value down even more because these factors are universally applied no matter how many diagnoses are being reported. Those plans under-reporting are cutting the financial reserves that are meant to help pay for all of the diagnoses each year, to include those that are being missed. I have worked on several cases myself. I have seen fraudulent reporting, so this does happen. However, whistleblowers who state that coding the diagnosis because it wasn’t treated or addressed in the date of service or encounter do not understand the purpose of risk adjustment. It’s one problem if the providers are reporting historical diagnoses that are no longer present or continuing to document cancer in an assessment when it is no longer present, or contradict their own diagnostic statements within the same visit documentation. It’s a completely different problem if a provider is just adding diagnoses that the patient does not have at all. I have seen providers and plans fall victim to internal processes where they truly attempt to help providers within an office or a system to code correctly and then those very memorandums or PowerPoints are used against the treating providers who are just trying to create an honest coding and documentation standard for all to follow. They may be accused of ‘HCC mining’ because many of their initiatives focus on chronic conditions, which coincidentally happen to commonly be the ones carrying financial value. The media isn’t helping anything my sounding the fraud alarm when there is no real fraud to be seen and what instead happened is the rules established for the case improperly utilized FFS coding rules that audit back to the specific encounter RVU’s instead of evaluating if the patient had the condition as a true diagnosis in that year of service for risk adjustment purposes. This negative media attention and worry of fraudulent activity has negatively impacted the health plan’s ability to become proactive in disease management. Why try to use risk modeling to become proactive in treatment if you may end up accused of fraud? I have heard many argue that if a patient had a diagnosis, then surely it would be addressed with MEAT in at least one encounter within the year, but this is incomplete thinking. There are many diagnoses that are lifelong and permanent which may never have an encounter that demonstrates active treatment. Amputations may never need a follow up appointment, sans phantom pain or ulcerations, and yet that patient’s circulatory system is forever altered, and therefore some amputations status codes are in the model. There are diagnoses such as angina, that once known, are easily managed with ongoing sublingual nitroglycerin and unless there is a new episode or worsening condition, the diagnosis will always only appear in a PMH list. There are a plethora of genetic disorders that are in the model because patients with those problems are at higher risks of other clinical manifestations.
Trusting in risk adjustment model design
Those concerned with the increased diagnosis code collection in risk adjustment models should trust the model design. These models are re-calibrated each year. Old MI for example used to be a Part C valued HCC in the CMS model but after some time of higher-than-expected reporting of patients who had this diagnosis, it was demoted to a Part D only. Diagnoses are added and readjusted in the models as needed as national data is collected over time. We have seen recent re-calibrations in the CKD codes and others over the years. (While this portion is CMS focused, recall that Medicaid & HHS are modeled similarly.)
Additionally, CMS is required to make an adjustment to reflect the differences in coding patterns between Medicare Advantage (MA) who use risk adjustment modeling to coding patterns in traditional Medicare which often has lower diagnosis reporting for FFS type billing. MA plan risk scores increase faster than FFS scores due to the coding intensity of collecting all current diagnoses more specifically. The adjustment made is called a “coding intensity adjustment.” The goal of those applied coding intensity adjustments is to maintain MA risk scores at the level they would normally be if MA plans coded similarly to FFS coding. An increase in coding intensity factor causes a decrease in risk scores.
On top of the coding intensity adjustment, there is also a Fee-For-Service Normalization Adjustment to ensure that all payments are based on a population with an average risk score of 1.0. This is considered the national average. A 1.0 risk score represents the accepted average annual Medicare costs for an individual. A risk score higher than 1.0 means the patient is likely or expected to incur costs higher than average (based on those chronic conditions that they are known to have). A risk score of less than 1.0 means the individual is expected to incur costs less than average. The Deficit Reduction Act requires CMS to apply the FFS normalization factor to the risk scores. Patient risk scores are divided by the normalization factor that was established for each calendar year and there are different normalization factors applied for special populations such as PACE model, ESRD model, etc. There are also different normalization factors used for Part D (the Rx model for CMS).
The problem with lists in a medical record
There are many different types of lists of diagnoses that can be found in a medical record. They can vary by facility and even by provider. Some examples include PMH (Past Medical History), Active Problems, Ongoing Problems, Current Problems, etc. PMH lists are a remnant of paper records before we had EMRs. Without an electronic system to help remind providers of previous conditions treated, there used to be a form or sheet in the front of a paper medical record that was often called a PMH list and sometimes was called a clinical summary. Providers used this not only for historical or old diagnoses, but it was often used as a ‘diagnosis diary’ to annotate all the diagnoses that had been treated by the provider within that paper record’s notes. When EMRs came along, many wanted to carry forward these lists because providers were used to having them as a tool. Coders are taught to never code any historical diagnoses or diagnoses that have been previously treated and no longer exist, and a name like “past medical history” is foreboding to a coder. It seems risky to some because those are “historical.” The problem here is that providers often document ongoing chronic conditions in those lists by habit. The same problem can happen with a list titled “current” or “active” where the provider might accidentally add an old or historical diagnosis even though the title suggests they are current. This is why critical thinking skills, proper training, and tools are paramount for risk adjustment coders.
THIS “LIST” PROBLEM IS WHY I CREATED THE TAMPER™ ACRONYM. IT WAS NEVER INTENDED TO BE A COMPETITOR OF MEAT. RECALL THAT MEAT IS ACTUALLY HARMFUL TO THE GOALS OF RISK ADJUSTMENT. KNOWING THAT FEW MEDICAL CODERS HAVE A CLINICAL BACKGROUND, TAMPER™ WAS CREATED AS A TOOL TO APPLY AGAINST ALL TYPES OF LISTS IN GENERAL. No matter if the list is titled as historical or current/active, the acronym can be applied for any diagnoses that are within that list to help the coder determine if the listed diagnosis is still a current active diagnosis so that they can capture all current diagnoses. It was not intended as a litmus test for the entire record as it is currently being improperly applied in some organizations.
What does CMS have to say?
1. CMS issues and often updates its risk adjustment participant guide. There is a participant guide from 2003, that further highlights the approach meant for risk adjustment. This is clearly illustrated in several examples they offer.
Example 1: The patient has a broken leg and Congestive Heart Failure (CHF). The physician documents a diagnosis for the broken leg in the medical record, but does not reference the CHF in the medical record documentation on the date of service when he attended to the broken leg. In this case, the M+C organization cannot submit the code for CHF because the physician did not document the diagnosis of CHF on that date.
Example 2: The patient has a broken leg, CHF and Chronic Obstructive Pulmonary Disease
(COPD). The physician documents a diagnosis for the broken leg for the patient, and also notes that the patient has CHF and COPD in the medical record documentation on the date of service when he is attending to the broken leg. The physician, however, fails to communicate the CHF or the COPD diagnoses to the plan for the submission of this data to CMS. The plan learns that that patient did have diagnoses of CHF and COPD assigned by the physician during the visit for the broken leg visit. In this case, because the physician has noted the diagnoses in the documentation, the plan may submit the CHF and COPD diagnoses as risk adjustment data, assuming the dates of service fall within the correct data collection period.
Example 3: The patient has diabetes with complications – ophthalmic manifestations. In this case, the physician submitted to the plan a diagnosis code of diabetes without mention of complications (code 250.0). However, in the medical record documentation the physician has noted on the date of the face-to-face encounter with the patient that the patient has a diagnosis of diabetes with ophthalmic complications (250.5). The plan is made aware of the diabetes diagnosis with the higher level of specificity in the medical record documentation. In this case, the plan may correct the risk adjustment data and change the diagnosis from 250.0 to 250.5. (CMS Participant Guide, 2003)
The 2008 guide is the probably most recognized guide where CMS attempted to give more coding advice. They had such an outpouring of requests for more specific coding information, that they paused specific coding guidance and instead generically advised to follow all coding guidelines. There is an excerpt from that 2008 guidance document that has been distorted to fit the MEAT narrative. Here is the excerpt:
6.4.1 Co-Existing and Related Conditions
The instructions for risk adjustment implementation refer to the official coding guidelines for ICD-9-CM, published at www.cdc.gov/nchs/icd9.htm and in the Coding Clinic®. Physicians should code all documented conditions that co-exist at the time of the encounter/visit and require or affect patient care treatment or management. Do not code conditions that were previously treated and no longer exist. However, history codes (V10-V19 not in HCC model) may be used as secondary codes if the historical condition or family history has an impact on current care or influences treatment.
Co-existing conditions include chronic, ongoing conditions such as diabetes (250.XX, HCCs 15-19), congestive heart failure (428.0, HCC 80), atrial fibrillation (427.31, HCC 92), chronic obstructive and pulmonary disease (496, HCC 108). These diseases are generally managed by ongoing medication and have the potential for acute exacerbations if not treated properly, particularly if the patient is experiencing other acute conditions. It is likely that these diagnoses would be part of a general overview of the patient’s health when treating co-existing conditions for all but the most minor of medical encounters. Co-existing conditions also include ongoing conditions such as multiple sclerosis (340, HCC 72), hemiplegia (342.9X, HCC 100), rheumatoid arthritis (714.0, HCC 38) and Parkinson’s disease (332.0, HCC 73). (CMS Participant Guide, 2008)
Some of my well-intentioned colleagues have taken this comma-delimited listing in the last paragraph above of 8 diagnoses and promoted the idea that CMS has said that MEAT isn’t required for these 8 conditions. I often hear them referred to as “the chronic 8,” but this is not accurate. This is not an exhaustive list, but rather a list of examples of what kind of conditions are considered chronic and on-going. One might take notice that there are many permanent lifelong conditions that aren’t mentioned at all. For example: amputations, ALS, HIV, Fragile-X, and many more genetic conditions for which there is no cure. This is not a list of approved conditions that do not require MEAT, but an example from which to establish critical thinking skills on what types of diagnoses are thought of as chronic and ongoing
2. CMS issued coding guidance to its SVA (Secondary Validation Auditor), who are typically one of those actuarial firms. The guidance for reviewing RADV looked like this:
“Though official coding rules do not change based on the type of audit, the coder should be aware of the background and prospective nature of the RA payment process including its basis on chronic conditions, and dependence on validating chronic conditions for an annual payment on just the review of one record. It is imperative therefore to code all chronic conditions documented by an acceptable provider type during a face-to-face encounter with the patient, whether or not there was specific treatment mentioned in the one record submitted. Mention of EMR population of the diagnoses narrative list can be interpreted as management and care for the applicable chronic conditions of the patient once all other coding rules and checks for consistency have been applied. This is where RADV/HCC audits may differ in guidance interpretation from fee-for service, DRG audits, or others based on just the payment for one specific encounter.”
There are newer guidelines for RADV from CMS, but I did not include them here because they very confusingly reference ICD-9-CM guidelines that no longer apply.
What does HHS have to say?
There is even more support to code for all current conditions, whether or not they are currently being addressed in the specific encounter from the HHS HRADV guidance documents. See below examples:
“The ICD-10-CM Official Guidelines for Coding and Reporting specific to outpatient services provides guidance to code, “all documented conditions that coexist at the time of the encounter/visit and require or affect patient care treatment or management. Do not code conditions that were previously treated and no longer exist. History codes may be used as secondary codes if the historical condition or family history have an impact on the current care of influences treatment. Chronic diseases treated on an ongoing basis may be coded and reported as many times as the patient receives treatment and care for the condition(s).”
Coding Chronic Conditions
As previously stated, the ICD-10-CM Official Coding Guidelines for Coding and Reporting and the Protocols must be followed when considering the applicability of the underlying diagnoses for chronic conditions. For purposes of HHS-RADV, CMS considers a chronic condition as lasting for a year or more and requiring ongoing medical attention. CMS also utilized the Chronic Condition Indicator developed by the Agency for Healthcare Research and Quality (AHRQ) to develop a list of HCCs in the HHS-operated risk adjustment models that CMS considers to be chronic conditions. See Appendix C for a listing of HCCs that may be considered chronic for HHS-RADV. A chronic condition from this list may be coded for an enrollee based on a face-to-face visit documented in the medical record. The chronic conditions must be documented in a way that it is reasonable to determine that a physician is managing the patient and treating the chronic condition for the benefit year that is being audited. Coders should utilize these Protocols and ICD-10-CM coding guidelines when determining chronicity of a diagnosis.
Below are some recommended steps to follow:
Step 1: Is the condition chronic or acute? If chronic, proceed to step 2. Reference Appendix C for a listing of HCCs that may be considered chronic for the purposes of HHS-RADV.
Step 2: If chronic, is the condition still present?
Step 3: If unsure, is there documentation of current medication or other management of the condition? Examples of this would include: a current problem list or past medical history, prescription (Rx), encounter/progress notes, medical management list or medication lists. Chronic conditions affect the medical decision making of providers when considering treatment options for other presenting problems. This includes considering current medications being taken for those chronic conditions, and the risk/benefit analysis of medical decision making for the treatment options of any other comorbidities.
Step 4: Do the current provider encounter notes contradict or further specify the condition?
Step 5: If the condition is chronic and there is no documentation that contradicts or shows the condition is resolved, then the condition can be abstracted. A chronic condition is assumed to be present if it is documented in any portion of the medical record.
Problem lists (active, ongoing, current, etc.) must reside within an encounter note, as a part of the medical record and not a standalone listing, for the date of service of a face-to-face visit. CMS recognizes that problem lists may be utilized differently based on the provider and method of capturing (electronic health records or paper charts) and encourage the use of ICD-10-CM Coding Guidelines for final guidance.
Past Medical History (PMH) may also be utilized differently, as some providers use these lists to document all current and past diagnoses, while other providers attempt to use them for PMH or resolved conditions only. A diagnosis listed in PMH must be supported as either a life-long permanent condition, or have additional supporting documentation elements within the same date of service (since each date of service stands on its own), such as medication which is documented for the diagnosis or used only to treat the diagnosis in question. Such diagnoses considered to be current should be coded as a part of the enrollee’s risk profile.
“History of” statements documented in the encounter in the format of Chief Complaint (CC) or History of Present Illness (HPI) such as “Ms. ____ is here today for her history of x” should not be handled in the same manner as PMH lists. Such statements found within a MR under the CC or HPI section establish the reason for the visit by including those histories of items as a part of the review in the encounter.
RADV, HRADV, and other audits
As noted above, neither CMS nor HHS require MEAT for a RADV or HRADV. It is not part of risk adjustment at all. When issuers choose a vendor or partner to be an auditor, they should not be dictating things like “code every date of service” or “there must be MEAT.” I have even seen one case where the issuer demanded MEAT for support of the diagnosis, even when found in the assessment. I had to remind them that the “A” in MEAT stood for “assessment” but that didn’t matter. WE HAVE WANDERED SO FAR OFF OF THE PATH OF PROPER RISK ADJUSTMENT CODING THAT WE ARE DOING A TERRIBLE JOB NATIONALLY AND IT IS TIME FOR EVERYONE TO BETTER UNDERSTAND SO THAT WE CAN APPROACH RISK ADJUSTMENT CODING IN UNISON. We should be coding for all current active diagnoses, whether there is current treatment in that one visit or not, if the patient really carries the diagnoses. Additional diagnoses should not be used to beef up the E/M value but should be reported for risk adjustment purposes. For RADV and HRADV, the proof is in the diagnosis being true, not treated. Any ICD codes that carry HCC value that cannot be found in supporting documentation will result in financial penalties (pay backs) for the health plans and any funding for the expected future treatment of those conditions is essentially being returned when they are found unsupported or not current for the patient under review.
Medical coding is difficult enough without adding unintended rules
Some think risk adjustment coding is easy. They think you can just hire any certified coder to do the work and some are so convinced this is true that they utilize business models where they only hire 1099 or part time coders to get the work done. Being certified as a coder is only the minimal requirement. In fact, an entire credential the CRC (Certified Risk Adjustment Coder) was created in acknowledgement of this fact. CODERS IN RISK ADJUSTMENT NEED PRACTICE AND TRAINING. THEY MUST BE CRITICAL THINKERS. THEY ARE COLLECTING IMPORTANT INFORMATION THAT AFFECTS THE ABILITY TO CARE FOR MANIFESTATIONS AND COMPLICATIONS THAT ARISE IN THE CHRONICALLY ILL. There are difficulties in specific diagnoses with special coding rules. There are difficulties sometimes in interpreting the clinical documentation and especially so when that documentation appears contradictory. Leaving behind diagnoses just because they weren’t treated or addressed in that visit leaves behind factual data elements that affect our ability to care for those conditions. These are real diagnoses that patients do suffer from and are often being treated through pharmaceuticals or are just known to be true and do not require ongoing treatment but they are in the risk models because the patient’s overall health is affected by having the diagnoses. Someone wrote in to the coding clinic about coding for diagnoses from a list and asked if it was appropriate to code them using medication as support and the coding clinic, which is focused on inpatient coding and fee-for-service views, said no. However, risk adjustment is not about coding for diagnoses being treated in that encounter for that claim, but rather gathering accurate diagnosis data on that patient for risk adjustment purposes. I am sure that the focus on the question was purely related to a FFS design and not taking risk adjustment diagnosis coding into account. WE ARE COLLECTING DIAGNOSIS DATA TO ESTABLISH A FUTURE NEED. WE ARE NOT CODING TO A SPECIFIC ENCOUNTER DATE OR SERVICE PROVIDED IN THE PAST. Health plans have care management programs but patients cannot be enrolled in them if the health plans are unaware of these diagnoses. It truly challenges our ability to become proactive in treating conditions (risk adjustments very purpose) if we are under-reporting them.
Bringing it all together
I am hopeful that this article and information sharing will help guide those unsure about risk adjustment coding and how it differs from Fee-For-Service claims coding. Some of the diagnoses are tied back to the E/M RVU work when they are addressed in the encounter, and others are supplemental diagnoses that are part of medical decision making that do not affect the E/M RVU but are known comorbidities of that patient. Risk adjustment models are prospective by nature and the whole reason we started using them was to try to get ahead of these chronic conditions. The goal is to keep patients out of the ER, out of the hospital, and to get ahead of these diagnoses in the outpatient setting while minimizing manifestations, complications, and resulting medical problems. We will not succeed at this goal if we are purposefully leaving accurate current diagnoses behind because it came from a particular portion of a record or there wasn’t current treatment in that one visit. I have seen the internal workings and coding polices of over 6 different competitor vendors within this space over the past year and dozens of health plan policies and no one is following the same coding guidance. This will cause for a skew of result and a failing of risk adjustment’s purpose. Risk adjustment is about population health and improving population health. If one group is coding for all current diagnoses as they should, and another group is not following this same guidance, there will be a disparity not only in the dollars being allocated, but more importantly a disparity in being proactive in the management of these chronic conditions. Many do not know that diagnosis codes are aggregated and compared globally. This is one way that we compare the overall health of one country to another. In ICD-9 reporting, the United States had a huge number of diabetic patients reported but lower than other countries with diabetic manifestations. The reason? Most providers memorized the 250.00 code and just used it for all diabetics while many of those patients had neuropathy, CKD, retinopathy, and more. ICD-10 gave us new combination codes so we could report those diabetics (among others) more accurately. WE ALL MUST REMEMBER THAT RISK ADJUSTMENT CODING IS ABOUT BEING READY FOR THE CURRENT AND FUTURE MANIFESTATIONS OF ALL KNOWN CURRENT DIAGNOSES. THIS IS WHY WE COLLECT ALL CURRENT DIAGNOSES AND NOT JUST THOSE TREATED ON THE VISIT. WE ARE NOT VALIDATING DIAGNOSES THAT WERE TREATED ON A PARTICULAR VISIT OR ENCOUNTER, BUT RATHER PREPARING FOR CURRENT AND FUTURE NEEDS BASED ON ALL CURRENT DIAGNOSES THAT EACH PATIENT CARRIES AS A HEALTH PROBLEM FOR EACH CALENDAR YEAR. The diagnoses in risk adjustment models were selected because having them as a patient is supposed to matter. Knowing the patient has them is supposed to help us get ahead of them so we can keep people as healthy as possible.
