Quality Gap in Long-Stay Antipsychotic Quality Measure Performance Widens Over the Pandemic,Reversing Past Gains

Abstract

The Centers for Medicare & Medicaid Services (CMS) grades nursing home performance in antipsychotic prescribing quarterly, publishing findings as a quality measure. While scores have improved since 2011, marked performance variation between facilities persists. To assess quality gap changes between best- and worst-performing deciles, we compared quarterly prescribing changes between these groups pre-pandemic (April 2011 to March 2020) and during the pandemic (April 2020 to March 2022). Antipsychotic quality measure scores, improving pre-pandemic, deteriorated during the pandemic. The pre-pandemic quality gap between the best- and worst-performing deciles narrowed as the worst-performing decile improved faster than the best-performing decile. During the pandemic, the quality gap widened as the worst-performing decile relapsed more than the best-performing decile (p < .0001). The pandemic disrupted quality performance gains and compounded disparities between facilities. A better understanding of the factors allowing high performers to weather pandemic stressors better than poor performers may reveal opportunities to improve nursing home quality and equity for all residents.

Introduction

Since 2012, the Centers for Medicare & Medicaid Services (CMS)—responsible for the oversight of care provided to the vulnerable residents of nursing homes (NHs)—has publicly graded the “Percent of Residents who Received an Antipsychotic Medication” as a NH quality measure (QM). This long-stay antipsychotic QM is a key contributor to CMS’ aggregated NH facility star rating. Transparent antipsychotic QM reporting is one of CMS’ approaches to curbing the inappropriate overutilization of risky antipsychotic medications among long-stay residents, a primary focus of the National Partnership to Improve Dementia Care in Nursing Homes (Nursing Home Compare, 2022; Social Security Act, 2016).

While antipsychotic QM scores have improved over the last decade since the measure’s introduction, significant variation persists between the best- and the worst-performing NHs. The reasons for this enduring inequity in quality remain unclear. It also is increasingly evident that the COVID-19 pandemic has had a pronounced detrimental effect on quality in all locations of care, including NHs; the pandemic’s impact on equity in quality, however, is less understood (Fashaw et al., 2020; Long-Term Trends of Psychotropic Drug Use in Nursing Homes, 2022; National Healthcare Quality and Disparities Report, 2022; Winter et al., 2022).

The COVID-19 pandemic served as a stress test with potential to expose the subtle but meaningful forces and factors differentiating best- and worst-performing NHs and, thus, presents a unique opportunity to explore the elusive drivers of equity divergence. The antipsychotic QM is a foundational measure in NH assessment and a convenient lens for providing a snapshot of global NH performance quality (Long-Term Trends of Psychotropic Drug Use in Nursing Homes, 2022). This focused study set out to compare the performance trajectories of the best- and worst-performing NH deciles in antipsychotic QM scores with an emphasis on the inflection point of the pandemic. A better understanding of quality gap trends in this keystone measure of long-stay performance may offer insight into the opaque domain of quality equity in NHs.

Methods

Quarterly long stay antipsychotic QM ratings for all NHs in the United States, publicly available at a facility-level from CMS, were sorted into deciles of performance and trended from the second quarter of 2011 to the first quarter of 2022 (Antipsychotic Medication Use Data Report, 2022). The antipsychotic QM reports each NH’s quarterly rate of antipsychotic prescribing (excluding residents with diagnoses of schizophrenia, Tourette’s syndrome, and Huntington’s disease). The months of January, February, and March represented the first calendar quarter; April, May, and June, the second; July, August, and September, the third; and October, November, and December, the fourth (Nursing Home Compare, 2022). The pre-pandemic period was defined as the period from quarter 2 of 2011 to quarter 1 of 2020; and the pandemic period was defined as the period from quarter 2 of 2020 to quarter 1 of 2022 (also the last quarter of publicly available data).

To measure the difference in changes in quality between the best- and worst-performing deciles, defined as the 10th percentile and 90th percentile respectively, we used an OLS linear regression model. The regression used quarterly antipsychotic QM scores for the best- and worst-performing deciles and incorporated a one-degree spline with a knot at quarter two of 2020 to model trends before and during the pandemic. The model is specified as

$$\begin{array}{l}{\beta }_0+{\beta }_1*APQMScore+{\beta }_2*APQMScore*I\left({10}^{th}Decile\right)\\ +{\beta }_3*APQMScore*I(Pandemic)\\ +{\beta }_4*APQMScore*I\left({90}^{th}Decile\right)*I(Pandemic)\end{array}$$

where $I(*)$ is the indicator function. Rates of change for the best- and worst-performing deciles are reported pre-pandemic $\left({\beta }_1,{\beta }_2\right)$ and during the pandemic $\left({\beta }_1+{\beta }_3,{\beta }_2+{\beta }_4\right)$ , as well as p-values associated with differences in rates. All statistical computation was done using SAS Version 9.4 Statistical Software (SAS Institute, Cary, NC). This study was approved by the Virginia Commonwealth University Institutional Review Board (#HM20025382).

Results

Across the entire study interval, an average of 13,837 NHs reported the quarterly “Percent of Residents who Received an Antipsychotic Medication” long-stay QM (range: 13,301–14,342). Over this 11-year period, all deciles of performance showed antipsychotic reductions. During the pre-pandemic period, median antipsychotic prescribing improved from a rate of 22.2% to a nadir of 12.5%. Median antipsychotic prescribing deteriorated, however, over the pandemic period from 12.5% to 13.0%. The worst-performing decile improved performance from 37.5% to 24.1% over the pre-pandemic period, relapsing to 25.9% over the pandemic. The best-performing decile’s pre-pandemic performance improved from 11.1% to 3.9%, and then continued to improve over the pandemic to 3.4% (see Figure 1).

Figure 1.

Best and worst-performing deciles in the “Percentage of residents receiving antipsychotic medication” long-stay quality measure: Q2 2011 to Q1 2022 linear regression model.

Change in performance between the best and worst deciles was significantly different both in the pre-pandemic period and over the pandemic. Fitting the regression model, the best-performing antipsychotic QM decile had a modeled starting value of 10.8% and decreased 0.9% annually pre-pandemic. The worst-performing antipsychotic QM decile had a modeled starting value of 37.4% and decreased 1.7% annually pre-pandemic. The difference between the rates of decrease was statistically significant (F = 194.23, df = (1, 82), p < .0001), an indication that the worst-performing decile was improving faster in antipsychotic QM performance than the best-performing decile. During the pandemic, the best-performing antipsychotic QM decile had a modeled increase of 0.5% annually, while the worst-performing decile had a modeled increase of 1.8% annually. These two rates were also statistically different (F = 36.16, df = (1, 82), p < .0001), indicating the best-performing decile was much more stable in maintaining its level of antipsychotic QM performance than the worst-performing decile during the pandemic.

Discussion

Though there has been progress in measured quality in all deciles of antipsychotic QM performance, our brief reveals that the best and worst-performing deciles are on different performance trajectories, both before the pandemic and during the pandemic period. The underlying processes driving divergence are likely manifold and remain uncertain. Despite the critical importance of ensuring equity in quality of care in US nursing homes, trends in quality performance equity, and the factors driving it, are less understood than trends in quality performance (National Healthcare Quality and Disparities Report, 2022). That said, it appears they were influenced by the pandemic which appears to have amplified existing inequities in care quality across all locations of care (Grabowski & Mor, 2020). Whether greater staffing reserves, facility resources, or community and environmental supports provided already best-performing deciles a protective advantage over this unprecedented period of disorder is speculative; that the negative impacts of the pandemic seem to have disproportionately concentrated in the lowest-performing deciles is not.

This was a limited pilot-level study, describing the response of the distribution of NHs to the pandemic. Only one long-stay QM was assessed. Since only facility-level data is publicly available, we were unable to adjust for resident-level demographic factors including age, race, sex, gender, or for other social determinants of health. Nor did we control for facility characteristics. A stratified, resident-level analysis of all quality measures, adjusting for regional, facility, and individual factors would an ideal next step. That said, antipsychotic QM performance is a fundamental measure of NH quality, and this is just a first step. This is also one of the first studies to compare changes in measured quality between the best- and worst-performing NH deciles over time, an approach which may provide insight into the changing landscape of all care quality in NHs especially in the context of mitigating harm associated with the COVID-19 pandemic.

While focused in scope, these findings highlight two key opportunities to improve quality of care in nursing homes. First, given that the best and worst-performing deciles responded differently to the pandemic, the pandemic disruption presents a unique laboratory to better understand the factors influencing equity in quality, a domain rich in conjecture but deficient in data. Second, since already poor-performing deciles were the most negatively impacted by the pandemic, this group may require focused efforts and support to restore lost gains in quality and right the improvement trajectory back towards equity and a shrinking gap in quality.

Conclusion

The pandemic disrupted equity in quality performance in addition to quality performance with respect to antipsychotic prescribing. Notwithstanding, even at its most equitable, the gap in antipsychotic QM performance was far too wide, hinting at ongoing opportunities for improvement in all aspects of NH quality performance. For these potential gains to be realized, a greater understanding of equity drivers is essential. As CMS increasingly emphasizes the importance of measured quality, equity in quality deserves attention too—even as quality improves. Quality improvement policies may also need to be modified to better enable and facilitate quality gains in low performers. Accomplishing this, while continuing to incentivize high standards and ongoing improvement overall, will be challenging; yet, despite the obstacles, this ideal is worthy of pursuit. Currently, the quality gap between the best and worst performers is widening and inequity in quality is increasing. This should sound an alarm, but also highlight an improvement opportunity. A better understanding of the factors shaping changes in long-stay equity may reveal new and better approaches to facilitate and enable ongoing progress in all care quality.