Risk adjustment is taking some heat.
Approaches pursued by Medicare Advantage Organizations (MAOs) receive criticism voiced by several actors, not the least of which are in governmental oversight. One of the concerns voiced is that there is a significant difference in the amount of diagnostic codes associated with members of MA health plans in comparison with Medicare beneficiaries still in Original Medicare. So we need to be mindful of this point and ensure that we are on firm ground with our HCC coding as an industry lest we leave ourselves open to an unwelcome backlash
The way these issues are getting played out is reflected in several legal cases that have set precedents over recent years, as well as the latest spate of suits brought by the Department of Justice against Humana. When you look at these regulatory and legal activities, you will see several different actors on the part of the government. They include the regular CMS folks conducting national and targeted RADV audits, of course. Also, there is the OIG of HHS, now equipped with the ability to conduct their own RADV audits, independent of the regular CMS RADV audits. Then there is the Department of Justice pursuing actions along the lines of the False Claims Act. We also have commentary and recommendations on the part of the General Accounting Office (GAO), citing “high error programs” for correction in areas of government that are targeted for “improper payment”, which they estimate to amount to 9.5% ($124 B) in Medicare Advantage Part C and another $2.1 B in Part D (projected for 2014 on https://paymentaccuracy.gov).
These figures pronounced by the GAO are based on the assumption that Fee for Service is the benchmark by which Medicare Advantage can be gauged to be guilty of excessive coding, and hence, generating overpayment by the government. This argument we believe to be an erroneous assumption. It appears in a January 2012 GAO report to Congress regarding Medicare Advantage risk scoring (GAO-12-51). In this report, the GAO diligently attempted to normalize comparative populations in order to conduct an apple-to-apples comparison between Medicare Advantage and Original Medicare beneficiaries. However, our theory is that the claims for beneficiaries under the FFS Medicare program are actually under-coded for diagnostic accuracy and that health conditions of members of Medicare Advantage Organizations might be better described by the coding captured through their private health plans.
Our theory is based upon one simple but important factor. Under FFS payment procedures, the physicians’ offices are submitting CMS-1500 forms for processing payment through the Medicare carriers. Their payments are based off procedure codes, to which their billing staff and billing services pay significant attention. If the CPT or HCPC codes are not nearly perfect, the claims will reject. By comparison, however, the ICD-9 diagnostic codes associated with the claims are not rejected if they are not comprehensive and highly accurate, only if they are not minimally in agreement with the procedure codes. Thus all the attention and effort is placed on procedure code accuracy, not on diagnostic accuracy.
By contrast, when hospitals bill the Fiscal Intermediaries, their DRG payments are entirely keyed off the ICD-9 codes and, hence, the level of diagnostic code capture is robust and achieves high levels of data quality. But since there is no financial incentive for physicians to code their diagnoses so optimally, the theory is that they will not invest the effort or expense necessary to do so.
Indeed, it is common for a doctor’s office to maintain a short list of the most common diagnoses at the lowest level of specificity in order to streamline the doctor’s administrative efforts. There is no incremental payment for doing a better job on diagnostic coding and it falls by the wayside.
There are barriers to comprehensive code capture built into Fee for Service billing. The claim only provides information on a specific presenting complaint at the time of the encounter, not a sweep of all the concurrent conditions. There are opportunities in the annual wellness visit to do more comprehensive diagnosis capture, of course, but there is no real reason to go to that level of documentation under FFS, despite the request of the ICD-9 guidelines to do so. Further, many billing system limitations only allow at most four diagnostic codes per claim. In order to submit more codes, additional diagnoses must be gathered in future visits, which may or may not happen. In order to make up for this deficit, retrospective chart reviews are conducted by Medicare Advantage plans to file supplemental information, but there is no reason to do so under FFS.
In contrast, CMS risk adjustment guidance to MAOs, as stated in the 2006 manual: risk scores measure individual beneficiaries’ relative risk and risk scores are used to adjust payments for each beneficiary’s expected expenditures. It is not for the purpose of reimbursing a particular encounter today, but rather to project forward in time with the best information available. Additionally, MAOs are required to get these diagnoses submitted during every calendar year within a limited period of elapsed time. To my knowledge, no one ever asked providers to assist Medicare in its efforts to gather comprehensive and highly precise diagnoses to predict future expenditures under Original Medicare. But CMS requires this of MAOs using ICD-9 guidelines for coding appropriateness, and which are used along with medical charts to grade coding accuracy in RADV audits of MAOs.
If this theory has merit, the benchmark for diagnostic coding accuracy is not FFS Medicare but probably some place more like the Medicare Advantage levels. Simply put, the same Medicare beneficiary with objectively identical health conditions would be coded differently under the Original FFS Medicare program versus Medicare Advantage.