The Impact of Dementia Special Care Units on Quality of Care: An Instrumental Variables Analysis

Abstract

Objective: To compare the quality of care following admission to a nursing home (NH) with and without a dementia special care unit (SCU) for residents with dementia.

Data sources/study setting: National resident-level minimum dataset assessments (MDS) 2005-2010 merged with Medicare claims and provider-level data from the Online Survey, Certification, and Reporting database.

Study design: We employ an instrumental variable approach to address the endogeneity of selection into an SCU facility controlling for a range of individual-level covariates. We use "differential distance" to a nursing home with and without an SCU as our instrument.

Data collection/extraction methods: Minimum dataset assessments performed at NH admission and every quarter thereafter.

Principal findings: Admission to a facility with an SCU led to a reduction in inappropriate antipsychotics (-9.7 percent), physical restraints (-9.6 percent), pressure ulcers (-3.3 percent), feeding tubes (-8.3 percent), and hospitalizations (-14.7 percent). We found no impact on the use of indwelling urinary catheters. Results held in sensitivity analyses that accounted for the share of SCU beds and the facilities' overall quality.

Conclusions: Facilities with an SCU provide better quality of care as measured by several validated quality indicators. Given the aging population, policies to promote the expansion and use of dementia SCUs may be warranted.

Keywords: Nursing home; dementia; quality; special care unit.

Figure 1

Scatter Plot of the Adjusted Predicted Probability of Admission to a Facility with a Dementia Special Care Unit (SCU) and Mean Number of Comorbidities by the Differential Distance to a Facility with versus without a Dementia SCU

1. Notes: Area of the bubble represents relative sample size at each measure of differential distance rounded to the nearest 0.1 km. Comorbidities include severe cognitive impairment, age greater than 85, hallucinations, delusions, severe behavior as measured on the Resource Utilization Group Scale (RUGS) III, bipolar disorder, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease, stroke, diabetes, and hypertension. Probabilities are adjusted for covariates reported in Table 1, as well as state, year, and period fixed effects. Standard errors are clustered at the level of the facility (number of facilities = 15,743).