When the Centers for Medicare & Medicaid Services (CMS) implemented the Patient Driven Payment Model (PDPM) as the new reimbursement method for skilled nursing facilities (SNF), some of us in the coding and auditing world thought it likely to result in inaccurate payments.
The old model, the Resource Utilization Group—Version IV (RUGS-IV), calculated SNF reimbursement based on the number of minutes of therapy a patient received. CMS found the old model might have incentivized the overuse of therapy, while ignoring the clinical differences that could impact care.
The new model, however, bases payment on the patient’s clinical picture, as conveyed by conditions and treatments checked off on the Minimum Data Set (MDS) and diagnoses entered on the MDS with ICD-10 codes. The MDS is a 43-page assessment tool completed by a registered nurse that covers every aspect of the patient’s status: functional, medical, psychological, and social. These inputs from the MDS go through the PDPM grouper to produce a Health Insurance Prospective Payment System (HIPPS) code on which the payment for a case is based, much the way that diagnoses and procedures on an acute care inpatient claim go through the MSDRG grouping logic to produce a DRG code.
What Could Go Wrong?
The logic of the model itself promotes inaccuracy. For example, the facility is required to report the “primary” diagnosis with an ICD-10-CM code, but since many of the diagnoses that might truly be driving the need for the SNF admission will cause the claim to be returned to provider, the facility is forced to pick something else to get the claim through. The model also contains multiple pathways to enhanced reimbursement that are frankly at odds with coding rules and norms, such as using symptoms instead of established diagnoses or using conditions or treatments that may no longer be active in the SNF but were reported on the immediately preceding acute care stay.
Another issue is that PDPM puts the job of coding on a nurse, who may not have access to code books or an encoder, or more than a passing exposure to the guidelines, conventions, instructions, and definitions that govern the use of the code set. Auditors know that even seasoned, credentialed coders make mistakes. A coder must recognize (and secure if necessary) adequate physician documentation and then translate that narrative into the language of the code set. CMS seems to think it has provided SNFs with adequate instruction, but that is a serious underestimation of what coding entails.
Now CMS is proposing to cut rates to SNFs because the first year of PDPM, which was supposed to be budgeting neutral, ended up costing taxpayers $1.7 billion more than expected. What drove the massive increase, during the very same time that SNF admissions were hitting all-time lows? Since an active COVID infection could prompt the need for isolation, which puts the patient in one of the highest paying categories for the nursing component of the HIPPS code, the report attributed the overall increase to COVID. CMS reviewed the same data and found that, even with the COVID cases removed, the payments were still higher than expected.
Physician Documentation is Key
Since the PDPM HIPPS code calculation does not use the diagnoses reported on the claim (except in one very specific instance), analyzing SNF claims data is not going to reveal why PDPM has cost so much. The only way to see that is to validate the HIPPS codes against the documentation in the medical record. Our team at Penstock has been performing these audits and the error rate is astounding. With very few exceptions, the physician documentation we are seeing is not where it needs to be for a diagnosis-centered model.
Two main issues significantly increasing reimbursement are: reporting conditions that are not “active” and/or not documented by a physician (as required by both the RAI Manual and the Official Coding Guidelines) and failing to submit an Interim Payment Assessment (IPA) when a significant HIPPS-impacting change occurs.
The IPA is “optional,” in the sense that CMS does not require one, but that fact does not protect the facility if an audit finds the HIPPS code billed to be inaccurate. Isolation is a prime example. If a patient tested positive for COVID upon entry and was truly in isolation (as defined in the RAI manual), then the extensive services nursing category is correct. Once the patient no longer has an active infection or positive test, however, an IPA should be filed. It clearly is inaccurate to report isolation for the entire stay when it might have only applied to the first week or two.
Auditors Who Want to Help
Ignoring problems with the model and the way it has been operationalized and slashing payment rates instead is going to leave us with struggling SNFs and unreliable data. Penstock is working on several different fronts to try to help make PDPM work better for all stakeholders. In that spirit, we have created a guide to help facilities understand what an auditor looks for and how to avoid having money taken back. This free guide includes an overview of rules for reporting with ICD-10-CM codes versus MDS checkboxes, physician documentation clarification requests, and IPA scenarios, along with hints to avoid leaving money on the table. A patient-centered model is a step forward, but we need a much larger discussion about the coding accuracy and documentation integrity the new model requires.
For More Information: Pdpm why the rise in inaccurate payments