TY - JOUR
T1 - Nursing Home Staffing Variability and Quality
T2 - Assessing Alternative Measures of Daily Staffing Level Variation
AU - Brunt, Christopher S.
AU - Bowblis, John R.
N1 - Publisher Copyright:
© 2025 Post-Acute and Long-Term Care Medical Association
PY - 2025/5
Y1 - 2025/5
N2 - Objectives: Variability in daily nursing staffing levels is an important factor associated with nursing home quality and has led to calls to incorporate it in the Centers for Medicare and Medicaid's staffing star rating calculation. This study assessed 6 distinct daily staffing level variation measures for their validity in explaining quality outcomes to determine an optimal variation measure for inclusion in the CMS staffing star rating. Design: Analysis of secondary data. Setting and Participants: The study analyzed data from the CMS Care Compare archives and Payroll-Based Journal for 2022Q1 to 2023Q3 and included 42,228 to 62,180 nursing home–quarter observations. Methods: Six measures of daily staffing level variation, including the coefficient of variation (CV), total outlier days (TOD), low outlier days (LOD), coefficient of quartile variation (CQV), the mean coefficient of dispersion of the median absolute deviation (mean-CDMAD), and median coefficient of dispersion of the median absolute deviation (median-CDMAD) were each evaluated for statistical validity based on desirable properties using linear regression to evaluate their relationship with 9 patient-based nursing home quality outcomes. Performance was analyzed by assessing statistical significance and direction of correlation, goodness of fit, and monotonicity. Results: The measure median-CDMAD demonstrated superior statistical properties, including the best goodness of fit in 56% to 67% of quality outcomes and consistency with the predicted direction in 51% to 89% of cases. Median-CDMAD, mean-CDMAD, and CV also demonstrated superior performance for monotonicity, indicating they are more robust against outliers and greater resistance to potential gaming by nursing homes than the other staffing level variation measures evaluated. Conclusion and Implications: Median-based variation measures, particularly median-CDMAD, exhibit the greatest validity relative to nursing home quality outcomes, making them suitable for CMS's staffing star rating system. Incorporating daily staffing level variation can improve public information on staffing consistency and incentivize quality improvement in nursing homes.
AB - Objectives: Variability in daily nursing staffing levels is an important factor associated with nursing home quality and has led to calls to incorporate it in the Centers for Medicare and Medicaid's staffing star rating calculation. This study assessed 6 distinct daily staffing level variation measures for their validity in explaining quality outcomes to determine an optimal variation measure for inclusion in the CMS staffing star rating. Design: Analysis of secondary data. Setting and Participants: The study analyzed data from the CMS Care Compare archives and Payroll-Based Journal for 2022Q1 to 2023Q3 and included 42,228 to 62,180 nursing home–quarter observations. Methods: Six measures of daily staffing level variation, including the coefficient of variation (CV), total outlier days (TOD), low outlier days (LOD), coefficient of quartile variation (CQV), the mean coefficient of dispersion of the median absolute deviation (mean-CDMAD), and median coefficient of dispersion of the median absolute deviation (median-CDMAD) were each evaluated for statistical validity based on desirable properties using linear regression to evaluate their relationship with 9 patient-based nursing home quality outcomes. Performance was analyzed by assessing statistical significance and direction of correlation, goodness of fit, and monotonicity. Results: The measure median-CDMAD demonstrated superior statistical properties, including the best goodness of fit in 56% to 67% of quality outcomes and consistency with the predicted direction in 51% to 89% of cases. Median-CDMAD, mean-CDMAD, and CV also demonstrated superior performance for monotonicity, indicating they are more robust against outliers and greater resistance to potential gaming by nursing homes than the other staffing level variation measures evaluated. Conclusion and Implications: Median-based variation measures, particularly median-CDMAD, exhibit the greatest validity relative to nursing home quality outcomes, making them suitable for CMS's staffing star rating system. Incorporating daily staffing level variation can improve public information on staffing consistency and incentivize quality improvement in nursing homes.
KW - Quality of care
KW - five-star quality rating
KW - nursing homes
KW - staffing
KW - variation
UR - http://www.scopus.com/inward/record.url?scp=86000579284&partnerID=8YFLogxK
U2 - 10.1016/j.jamda.2025.105532
DO - 10.1016/j.jamda.2025.105532
M3 - Article
C2 - 40024614
AN - SCOPUS:86000579284
SN - 1525-8610
VL - 26
JO - Journal of the American Medical Directors Association
JF - Journal of the American Medical Directors Association
IS - 5
M1 - 105532
ER -
