Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

School of Aging Studies

Major Professor

Ross Andel, Ph.D.

Co-Major Professor

Victor Molinari, Ph.D.

Committee Member

John R. Bowblis, Ph.D.

Committee Member

Debra Dobbs, Ph.D.

Committee Member

Janice C. Zgibor, Ph.D.


health citations, Medicaid, nurse staffing, severe mental illness, star ratings


Objective: This dissertation is made of two academic papers followed by a chapter discussing policy implications of the findings. The objective of this dissertation is to better understand if high-serious mental illness (SMI) nursing homes (NHs) provide poorer quality of care (e.g., operationalized as staffing levels, NH Compare quality star ratings, and health deficiencies) and whether quality in high-SMI NHs is confounded by payer mix (e.g., the proportion of residents who are Medicaid-reimbursed). A modified Donabedian Structure, Process, Outcome (SPO) Model was used as the main theoretical framework.

Methods: Data came from the Certification and Survey Provider Enhanced Reports (CASPER) from 2016 for Chapter 2 and from 2014 to 2017 for Chapter 3. Quality star ratings from the NH Compare website were extracted. Multistep linear models were used for Chapter 2 to measure the effect of high-SMI status on staffing levels and NH Compare quality star ratings. Generalized linear mixed effects models (negative binomial) and linear mixed effects models were used for Chapter 3 to measure the effect of high-SMI status on the number of deficiencies, a weighted deficiency score, and the likelihood of deficiencies by scope and severity.

Results: Chapter 2 found that Structure proxies such as the proportion of Medicaid-reimbursed residents confounded the relationship between high-SMI status and Process (e.g., staffing levels) and Bottom-Up (e.g., quality star ratings) proxies of quality. High-SMI NHs with a large proportion of Medicaid-reimbursed residents provided lower nurse staffing levels of all types and lower star ratings on overall stars, health inspection stars, overall staffing stars, and registered nurse staffing stars in comparison to high-SMI NHs with a low proportion of Medicaid-reimbursed residents. High-SMI NHs received more deficiencies and a greater deficiency score per inspection. However, these Top-Down proxies of quality were attenuated by the inclusion of Structure proxies such as payer mix. Deficiencies given to high-SMI NHs were associated with wider scope, especially Pattern and Widespread. In an exploratory aim, high-SMI NHs were more likely to be cited for resident abuse, neglect, or involuntary seclusion and the policies that prohibit and monitor for risk of abuse and neglect in comparison to low-SMI NHs when looking exclusively at deficiencies cited for Actual Harm or Immediate Jeopardy.

Conclusion: High-SMI NHs are unlike “typical” NHs as a majority of the residents have unique behavioral and mental healthcare needs. Findings from this dissertation suggest that quality in these facilities – measured via Top-Down (health deficiencies), Bottom-Up (NH Compare quality star ratings), or Process proxies (staffing levels) – is lower on average after controlling for a variety of confounding variables. One confounding factor that attenuates but does not fully explain differences in quality is a Structure proxy, payer mix. Specifically, the proportion of residents reimbursed through Medicaid affects the staffing levels, quality star ratings, and health deficiencies given to high-SMI NHs. A potential solution to improve quality in these facilities may be increasing Medicaid reimbursement rates, while tying the increased rates to specific staffing-related expenditures. In conclusion, high-SMI NHs provide poorer quality of care on average, though part of this association is explained by the lower margins provided by Medicaid.