The variation in observed costs not predicted by the case-mix model is substantial. To the extent that some facilities cannot respond appropriately to the incentives in the bundled system to reduce costs without compromising patient outcomes, these facilities may face material financial risk.
University of Michigan Kidney Epidemiology and Cost Center (KECC) Report, September 20091
The dialysis community as a whole is concerned about the potential negative effects of the proposed new payment rules on bundling; these concerns are especially acute among the smaller dialysis providers. Given the lesser economies of scale available to these providers, any misalignment of the payment system and consequent underfunding for patient care will have a more serious impact on these providers. They also have a lesser ability to spread risk, so they are vulnerable to any variation in payment across the patient population; accordingly, among small providers under the newly proposed prospective payment system (PPS), the chances of having a significant number of patients whose characteristics lead in the aggregate to a significantly higher cost of providing care but inadequate payment are clearly higher. A consequence of the proposed payment system may therefore be further consolidation of an already consolidated provider environment.
A major concern among providers is the manner in which the payment model has been constructed. The choice of case-mix adjustments entered into the model appears at best both arbitrary and perverse. The KECC report, in describing the methodology used to design the PPS, states:1
As we developed predictive models suitable for implementation in a case-mix adjusted prospective payment system, we considered whether specific patient characteristics should be included in the model, based on both magnitude and statistical significance of relationships between cost and the characteristic. In addition, the potential for creation of adverse incentives or social inequity by inclusion of a patient characteristic was considered. Subsequent research informed CMS [Centers for Medicare & Medicaid Services] decisions about which patient characteristics to include in payment models. Case-mix measures were reviewed for accuracy and objectivity of diagnostic criteria, temporal relationship between comorbidity appearance and cost, and model parsimony.
Nowhere in this list is clinical plausibility or relevance taken into account. Among parameters excluded is myocardial infarction, since according to the KECC report, “[d]iagnostic criteria are vague and addition of this comorbidity to a payment model would likely result in increased coding.”1 It is surprising that the diagnostic criteria for myocardial infarction are considered vague, and it is unclear why “increased” (and presumably more accurate) coding should be considered undesirable.
The tenor of this language is illustrative of an adversarial approach widely present in this document. An example is the decision to exclude low hematocrit, which predicts higher subsequent costs. The only reason given for excluding low hematocrit is the concern that including it would “effectively reward facilities achieving lower hematocrit.” So hematocrit is excluded from the models, even though “adding a measure of the hematocrit six to eight months prior to the current month to the case-mix adjustment model raised the R-squared of preliminary models by approximately five per cent.”1 This exclusion has a major effect, since the final model has an R2 value of only 39%.
On the other hand, there are modifiers whose entry into the model may in fact reward poor patient care, including septicemia, which often is associated with the use of central venous catheters for vascular access, and pneumonia, which may be associated with failure to ensure that patients receive pneumococcal vaccine appropriately. Pericarditis is especially bothersome as a modifier, since the most common cause of this in the dialysis population is inadequate dialysis.
Leaving aside the question of the doubtful validity of the payment model and its potential consequences, there are other special concerns for small providers, as detailed in the remainder of the editorial.
First, the complexities of the case-mix adjusters in the final model make implementation difficult for small providers who may well lack the personnel resources and the information technology systems necessary to gather this data and accurately bill according to the appropriate case-mix adjusters.
Second, the inclusion into the bundle of all laboratory tests ordered by the MCP (monthly capitation payment) physician, wherever ordered and for whatever indication, not only puts the dialysis provider at risk for all these tests, but introduces a potential adversarial relationship between physician and provider. If there is a difference of opinion between provider and nephrologist as to whether a given test should be performed for a given patient or for a given indication, how is this to be resolved? Moreover, the facility will in many cases only find out that a test has been ordered when it receives the bill from the laboratory.
Third, the inclusion of all oral drugs, whether they have an intravenous equivalent or not, is of major concern to smaller providers. There is general agreement in the dialysis community that this proposed rule, which abruptly adds Part D drugs to the bundle, is greatly underfunded and includes at least one expensive drug which has no substitute (cinacalcet). Facilities also must either dispense drugs to patients or contract with a pharmacy to do so. Compliance with state pharmacy rules will be especially difficult for small providers.
Fourth, while some attempt is made in the proposed rule to allow increased payments to smaller facilities, the definition of a small facility (<3,000 treatments per year) is too rigid. This would correspond to a census of 20 patients, and there are few dialysis facilities with so few patients.
Finally, CMS has not established a process to monitor the results of this major change in the payment system and its effect on providers. The Medicare Improvements for Patients and Providers Act of 2008 mandates that the US Government Accountability Office submit to Congress a report on the implementation of the payment system by March 1, 2013, but the specifics requested relate to utilization of erythropoiesis-stimulating agents and to the effect on low volume and rural providers. The legislative language includes only a vague requirement for “[a]ny other information or recommendations for legislative and administrative actions determined appropriate by the Comptroller General.”2 Given the radical nature of the change to the payment system, and since it is a true leap into the dark, a close real-time monitoring of the effects, particularly as it affects the viability of providers, is mandatory.
FULL TEXT
