As most of our readers and members know, on October 1, 2019, the Patient-Driven Payment Model (PDPM) replaced the Resource Utilization Groups Version IV (RUG-IV) reimbursement system in skilled nursing facilities (SNFs). The intention was to implement a budget-neutral system that promoted patient-centered care. The Minimum Data Set (MDS) 3.0 and Resident Assessment Instrument User’s Manual v 2.201, and its prior iterations, constitute the document and guidelines used for determining reimbursement.

The change to PDPM was implemented in an organizational context that contained two obstacles to achieving the promise of PDPM. The first concerns common practices associated with the MDS/RAI. It is a dynamic and complex clinical assessment and documentation process. Yet, it does not require any additional education or certification for clinical staff to complete. While F-tag 642 states: A registered nurse (RN) must conduct or coordinate each assessment (i.e., MDS), with the appropriate participation of health professionals,” the RN signature only serves to certify the completion of the assessment, rather than its accuracy. In practice, a licensed vocational nurse (LVN) typically completes most of the form, along with other members of the interdisciplinary team. It is possible that some LVNs provide competent assessments. However, it is beyond the scope of practice for an LVN licensed in California to independently perform a comprehensive assessment.¹ The second and longstanding concern is a lack of transparency in financial cost reports of SNFs in California and other states.²

These two obstacles have been exploited by some in ways that have been noticed by the Centers for Medicare and Medicaid Services (CMS) and health services researchers. As reported in McKnights Long Term Care News (April 2, 2026). “As PDPM has matured, CMS has continued to monitor case-mix trends to ensure that payment remains aligned with actual patient acuity rather than changes in coding practices.” One mechanism advertised within the SNF industry is to automate MDS with AI as a means of “revolutionizing SNF reimbursement.” The current process of completing MDSs is advertised as being “outdated, costly, (resulting in) errors, missed documentation, staff burnout, and lost revenue.”³ Researchers conducted secondary analysis of 100% traditional Medicare claims (2018-2021) and found that “PDPM was associated with increased coding intensity across multiple measures-and more so in for-profit SNFs – highlighting the need to further evaluate whether SNFs are accurately documenting or falsely inflating clinical complexity”.⁴

To the extent that these practices are occurring in SNFs where medical and nursing directors practice, it is problematic. Directors of nursing and MDS coordinators may be pressured to up-code on the MDS and use AI to capture codes benefiting the SNF financially, but irrelevant to a resident’s clinical needs and services. The additional revenue may not be used to enhance nurse staffing levels or other clinical services of potential benefit to residents. Lack of transparency in cost reports may make this practice possible. Accurate completion of the facility assessment may present a challenge for medical directors and DONs who knowingly participate in such problematic approaches to upcoding and use of AI to complete the MDS. 

1. Dellefied ME. Implementation of the RAI/MDS in the nursing home as organization: implications for quality improvement in RN clinical assessment. Geriatric Nursing, 2007;28(6):377-86.
2. Where do the billions of dollars go- a look at NH related party transactions. 2023. The National Consumer Voice for Quality Long-Term Care.
3. Rohan Handa (2 May 2025). Beyond excel sheets: how top SNFs are automating MDS with AI in 2025. Nanonets Health.
4. Amaravaadi H, Prusynski RA, Fishman PA, Leland NE, Mroz TM (2026). Too sick to be true? Evaluating potentially problematic diagnosis coding practices in Medicare’s Patient-Driven Payment Model. Health Services Research, 2026; 61:e70084 (https://doi.org/10.1111/1475-6773.70084)