Algorithm for Identifying Nursing Home Days Using Medicare Claims and Minimum Data Set Assessment Data.

Area of Science:

• Gerontology
• Health Services Research
• Data Science in Healthcare

Background:

• Lack of standardized methods for measuring nursing home (NH) length-of-stay (LOS) for Medicare beneficiaries.
• Difficulty in distinguishing between long-stay (≥101 days) and short-stay (<101 days) NH residents.
• Need for precise resident classification for care planning and resource allocation.

Purpose of the Study:

• To develop and validate an algorithm for measuring NH days of stay.
• To differentiate between long-stay and short-stay NH residents.
• To compare the performance of the new algorithm against Minimum Data Set (MDS) alone and Medicare claims data.

Main Methods:

• Linked 2006-2009 MDS assessments with Medicare Part A skilled nursing facility (SNF) data.
• Developed an algorithm to determine daily NH stay evidence using MDS and SNF dates.
• Compared resident classifications and LOS derived from the MDS/SNF algorithm with MDS-only and Medicare Parts A & B data.

Main Results:

• The MDS/SNF algorithm identified 276,844 residents, with 40.8% classified as long-stay.
• Long-stay residents were generally older, male, white, unmarried, low-income, with more comorbidities and higher mortality.
• The MDS/SNF algorithm showed higher concordance with MDS-only classification (98.9-100%) compared to Medicare claims (95.0% and 67.1%).

Conclusions:

• The developed MDS/SNF algorithm effectively differentiates long-stay and short-stay NH residents.
• This algorithm provides a more precise NH resident grouping compared to using Medicare claims data exclusively.
• The findings support the use of the MDS/SNF algorithm for improved accuracy in NH resident categorization.