Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 May 1.
Published in final edited form as: J Am Geriatr Soc. 2024 Mar 28;72(5):1442–1452. doi: 10.1111/jgs.18890

Trends in Observation Stays for Medicare Beneficiaries With and Without Alzheimer’s Disease and Related Dementias (AD/ADRD)

Laura G Burke a,b, Ryan C Burke a, Ciara E Duggan b, Jose F Figueroa b, Marie Boltz c, Donna Fick c, E John Orav d, Edward R Marcantonio e
PMCID: PMC11090746  NIHMSID: NIHMS1978854  PMID: 38546202

Abstract

Background:

There has been a marked rise in the use of observation care for Medicare beneficiaries visiting the emergency department (ED) in recent years. Whether trends in observation use differ for people with Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (AD/ADRD) is unknown.

Methods:

Using a national 20% sample of Medicare beneficiaries ages 68+ from 2012–2018, we compared trends in ED visits and observation stays by AD/ADRD status for beneficiaries visiting the ED. We then examined the degree to which trends differed by nursing home (NH) residency status, assigning beneficiaries to four groups: AD/ADRD residing in NH (AD/ADRD+ NH+), AD/ADRD not residing in NH (AD/ADRD+ NH−), no AD/ADRD residing in NH (AD/ADRD− NH+) and no AD/ADRD not residing in NH (AD/ADRD− NH−).

Results:

Of 7,489,780 unique beneficiaries, 18.6% had an AD/ADRD diagnosis. Beneficiaries with AD/ADRD had more than double the number of ED visits per 1,000 in all years compared to those without AD/ADRD and saw a faster adjusted increase over time (+26.7 vs. +8.2 visits/year; p<.001 for interaction). The annual increase in the adjusted proportion of ED visits ending in observation was also greater among people with AD/ADRD (+0.78%/year, 95% CI 0.77–0.80%) compared to those without AD/ADRD (+0.63%/year, 95% CI 0.59–0.66%; p<.001 for interaction). Observation utilization was greatest for the AD/ADRD+ NH+ population and lowest for the AD/ADRD− NH− population, but the AD/ADRD+ NH− group saw the greatest increase in observation stays over time (+15.4 stays per 1,000 people per year, 95% CI 15.0–15.7).

Conclusions:

Medicare beneficiaries with AD/ADRD have seen a disproportionate increase in observation utilization in recent years, driven by both an increase in ED visits and an increase in the proportion of ED visits ending in observation.

Keywords: Emergency medicine, Alzheimer’s disease, dementia, Medicare, observation

INTRODUCTION

Dementia is a leading cause of morbidity and mortality in the United States and a key driver of healthcare spending, on par with other leading causes of death such as cancer and heart disease.1–3 By 2050, the prevalence of dementia is expected to more than double,4 and therefore raises the importance of efforts to study and improve care delivery for this population. Acute care delivery represents a pivotal opportunity to intervene and maximize outcomes for this population. Emergency department (ED) visits are common among older adults and often the initial point of contact for acute care episodes.5 Such encounters signal a potentially serious illness but also carry their own inherent risks, such as delirium,6 and are associated with an accelerated trajectory of functional decline.7 These risks are thought to be even greater for people with Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (AD/ADRD),8 and thus maximizing the quality of ED care for this population is essential. One key driver of emergency care quality is the disposition decision. Unnecessarily admitting a patient increases costs and exposes the patient to nosocomial risks, whereas inappropriate discharge risks worsening of the acute illness, further clinical deterioration and even death.9 Improving decision-making around admissions and reducing variation has been a focus of improving emergency care quality and value in recent years. 10,11,12

This focus on improving decision-making around ED admissions has occurred alongside an evolution in acute care delivery toward an outpatient model of care, with declining rates of inpatient stays per capita.13 However, there has also been a rise in the use of observation care,14,15 which is when the patient is kept in the hospital (either in the ED, a dedicated observation unit or a traditional hospital ward), but with outpatient status for ongoing care, ideally for less than 48 hours. This trend is thought to be due to several factors. The growth of alternative payment models has incentivized the use of outpatient care and limiting more expensive inpatient admissions whenever possible.16 Additionally, Medicare audits penalizing hospitals for billing for short inpatient stays have further incentivized hospitals to use observation care when there is any doubt as to whether the patient will qualify for a full inpatient stay.17 While there has been substantial debate about the impact of the growing use of observation care for the broader Medicare population,18–21 there has been little study of whether these trends have uniquely impacted those with AD/ADRD, who account for a disproportionate share of acute care utilization and are likely more vulnerable to differences in acute care quality. Additionally, whether patterns of observation care utilization vary for adults with AD/ADRD residing in nursing homes (NH) versus those not living in NHs is unknown. One might hypothesize that rates of observation stays may be lower among NH residents, since NHs have more resources to prevent and treat acute episodes compared to other settings. Additionally, among patients who are seen in the ED, ED physicians may feel more comfortable discharging patients not clearly in need of an inpatient stay back to a NH where they can be monitored, rather than keeping them in observation status for ongoing care. Yet, empirical evidence testing these hypotheses is lacking.

To address this gap in the evidence, we used national Medicare data from 2012 to 2018 to answer the following questions. How has the volume of observation care changed in recent years for Medicare beneficiaries with AD/ADRD and how do these trends compare to those without AD/ADRD? Second, among beneficiaries who come to the ED, how has their site of disposition (home, observation, or inpatient hospitalization) changed over time and does this vary by AD/ADRD status? Finally, do patterns of observation utilization for beneficiaries with and without AD/ADRD vary by beneficiary nursing home residency status?

METHODS

Patients

We used a random 20% national sample of traditional, fee-for-service Medicare beneficiaries ages 68 and older from 2012–2018. For each beneficiary, we identified 26 chronic conditions (including AD/ADRD) from the Chronic Conditions Warehouse File (CCW) file and used the end-of-year indicator from the previous calendar to determine the presence of condition. We used age 68 as the age minimum because the AD/ADRD chronic condition variable uses a 3-year lookback. We identified those with and without AD/ADRD yearly (i.e., some individuals are lost when they die, and others enter the cohort when newly diagnosed with AD/ADRD). We identified beneficiary age, sex, race, and Medicaid eligibility from the Master Beneficiary Summary File. We determined if each beneficiary was a long-term NH resident using a previously validated approach 22 (determined yearly using the prior year of claims). Furthermore, we calculated a frailty score for each beneficiary using previously validated methods.23,24

ED Visits and Observation Stays

We identified all ED visits to acute care, non-federal hospitals in the 50 United States and the District of Columbia among eligible beneficiaries and determined the disposition (inpatient admission, observation, transfer, discharge or died in the ED). Observation stays were defined by the following revenue center codes: 0760, 0761, 0762, or 0769.25 Observation stays that were later converted to an inpatient admission were assigned a disposition of observation for the associated ED visit. We identified the principal diagnosis for each encounter, classified according to the Healthcare Utilization Project Clinical Classifications Software (HCUP-CCS) single level diagnosis categories.

Outcomes

Our primary outcome was observation stays per 1,000 beneficiaries. However, to put this finding in broader context, we also examined rates of ED visits and inpatient stays per 1,000 beneficiaries. Additionally, we examined ED disposition as an outcome (ED discharge, inpatient and observation stays), examining the change in the proportion of ED visits ending in each disposition.

Statistical Analysis

First, to compare how overall ED utilization has changed in recent years for people with and without AD/ADRD, we compared population trends using linear regression models with the beneficiary as the unit of analysis and the number of ED visits (regardless of final visit disposition) during the year as the outcome. The primary predictors were year and AD/ADRD diagnosis, as well as an interaction between year and AD/ADRD diagnosis. We calculated yearly mean visit rates per 1,000 eligible beneficiaries for each population in a model that incorporated year as a categorical indicator, hospital referral region (HRR) fixed effects, AD/ADRD diagnosis and an interaction between AD/ADRD diagnosis and year. We then compared adjusted time trends for the two groups incorporating year as a continuous predictor, HRR fixed effects, AD/ADRD diagnosis, an interaction between year and AD/ADRD diagnosis as well as beneficiary age and sex. We repeated these models for the outcomes of observation and inpatient stays.

Next, to determine the degree to which any population trends were driven by changes in disposition decisions made by emergency physicians, we examined trends in the proportion of all ED visits ending in observation, inpatient admission, and ED discharge, using a linear probability regression model with ED visits as the unit of analysis. For each ED visit, we used disposition as the outcome and year and AD/ADRD status as the primary predictors, along with a year-AD/ADRD interaction. We adjusted models for hospital fixed effects, principal visit diagnosis (International Classification of Diseases, 9th Revision [ICD-9] and 10th Revision (ICD-10) diagnosis codes aggregated into HCUP-CCS diagnosis categories), and beneficiary characteristics (age, sex, race, Medicaid eligibility, frailty score, and chronic conditions). Each chronic condition was included as a separate indicator in the model.

To examine the degree to which population trends varied by beneficiary NH status, we created the following four beneficiary categories: those with AD/ADRD who were NH residents (AD/ADRD+, NH+), those with AD/ADRD who did not reside in NHs (AD/ADRD+, NH−), those without AD/ADRD who were NH residents (AD/ADRD-, NH+) and those without AD/ADRD who did not reside in NHs (AD/ADRD−, NH−). We repeated our linear models for ED visits as well as inpatient and observation stays per 1,000 beneficiaries. These models used four beneficiary categories and year as the primary predictors as well as interactions between year and beneficiary categories. We calculated adjusted time trends for each group, incorporating the same HRR fixed effects as well as beneficiary age and sex.

Sensitivity Analysis

Prior work has noted that after the transition from the ICD-9 to the ICD-10 diagnosis codes, the CCW variable for AD/ADRD incorporated several ICD-10 codes that are not specific to AD/ADRD.26 We thus reclassified beneficiaries whose AD/ADRD status resulted solely from one of these nonspecific codes (Supplementary Methods S1) as not having AD/ADRD. We repeated our analyses examining trends in ED visits and observation stays by AD/ADRD status and for the four groups of AD/ADRD and nursing home residency status using this alternative AD/ADRD classification.

RESULTS

Sample Size and Characteristics

Our sample consisted of 7,489,780 million unique beneficiaries from 2012–2018, of whom 18.6% had AD/ADRD. Compared to those without AD/ADRD, people with AD/ADRD were older (mean age 83.6 vs. 76.4 in 2018, respectively; Table 1), had a higher mean frailty score (0.27 vs. 0.16 in 2018) and a greater proportion were nursing home residents (24.5% vs. 1.0% in 2018) and were Medicaid eligible (33.2% vs. 10.0%). A higher proportion of people with AD/ADRD were Black (9.2 vs. 6.7% in 2018) and Hispanic (2.1% vs. 1.4% in 2018) and fewer were White (84.2% vs. 85.9%).

Table 1.

Characteristics of Medicare Beneficiaries by Alzheimer’s Disease and Related Dementias (AD/ADRD) in 2012 vs. 2018

2012 2018
No AD/ADRD
(N=4,040,223; 88.4%)		 AD/ADRD
(N=529,692; 11.6%)		 No AD/ADRD
(N=4,181,405; 88.8%)		 AD/ADRD
(N=525,021; 11.2%)
Number ED visits 2,112,729 585,973 2,279,454 643,752
Age, mean (SD) in years at end of reference year 77.0 (7.1) 84.0 (7.5) 76.4 (6.9) 83.6 (7.9)
Mean Frailty Score (SD) 0.16 (0.06) 0.28 (0.09) 0.16 (0.06) 0.27 (0.09)
Nursing home resident (N, %) No 3,876,150 (95.9%) 363,984 (68.7%) 4,012,766 (96.0%) 385,221 (73.4%)
Yes 63,718 (1.6%) 157,492 (29.7%) 43,753 (1.0%) 128,564 (24.5%)
Missing 100,355 (2.5%) 8,216 (1.6%) 124,886 (3.0%) 11,236 (2.1%)
Age range (years), % 68–74 1,798,973 (44.5%) 67,628 (12.8%) 2,017,641 (48.3%) 80,583 (15.4%)
75–79 904,421 (22.4%) 80,687 (15.2%) 953,318 (22.8%) 85,526 (16.3%)
80–84 673,258 (16.7%) 117,384 (22.2%) 617,044 (14.8%) 108,570 (20.7%)
85 and older 663,571 (16.4%) 263,993 (49.8%) 593,402 (14.2%) 250,342 (47.7%)
Gender Female 2,283,788 (56.5%) 355,142 (67.0%) 2,316,031 (55.4%) 337,515 (64.3%)
Male 1,756,435 (43.5%) 174,550 (33.0%) 1,865,374 (44.6%) 187,506 (35.7%)
Race/Ethnicity, N (%) White 3,532,277 (87.4%) 446,799 (84.4%) 3,593,261 (85.9%) 441,829 (84.2%)
Black 287,498 (7.1%) 52,069 (9.8%) 280,412 (6.7%) 48,298 (9.2%)
Hispanic 61,297 (1.5%) 13,569 (2.6%) 58,891 (1.4%) 11,231 (2.1%)
Asian 73,473 (1.8%) 9,410 (1.8%) 81,639 (2.0%) 11,073 (2.1%)
Other 62,639 (1.6%) 5,092 (1.0%) 74,931 (1.8%) 7,117 (1.4%)
Unknown 6,187 (0.2%) 815 (0.2%) 72,732 (1.7%) 2,962 (0.6%)
NorthAm Native 16,852 (0.4%) 1,938 (0.4%) 19,539 (0.5%) 2,511 (0.5%)
Medicaid Eligible, % No 3,542,458 (87.7%) 329,249 (62.2%) 3,765,335 (90.1%) 350,502 (66.8%)
Yes 496,038 (12.3%) 200,399 (37.8%) 416,070 (10.0%) 174,519 (33.2%)
Missing 1,727 (0.04%) 44 (0.01%) 0 0
Open in a new tab

Visits among a random 20% sample of beneficiaries of traditional, fee-for-service Medicare in 2018.

Trends in per-population acute care utilization by AD/ADRD Status

Beneficiaries with AD/ADRD had more than double the rate of ED utilization in all study years compared to those without AD/ADRD (1,280 vs. 554 visits per 1,000 in 2012; Figure 1). The adjusted increase in ED visits (further incorporating beneficiary age and sex) was greater for people with AD/ADRD (+26.7, 95% CI 25.7 to 27.6, visits per year per 1,000 beneficiaries; Supplementary Table S1) compared to those without AD/ADRD (+8.2, 95% CI 8.0 to 8.5, visits per year per 1,000) with p<.001 for the year-AD/ADRD interaction. We saw a similar pattern for observation stays. In 2012, beneficiaries with AD/ADRD had 304 observation stays per 1,000 compared to 93 per 1,000 among beneficiaries without AD/ADRD (Figure 2). Beneficiaries with AD/ADRD also saw a faster adjusted increase in observation stays per year per 1,000 compared to those without AD/ADRD [+14.3 (95% CI 13.9 to 14.6) vs. +4.2 (95% CI 4.1 to 4.3; year-AD/ADRD interaction p <.001). We found that people with AD/ADRD had more than double the number of emergency inpatient stays per 1,000 population compared to those without AD/ADRD in all study years (587 vs. 193 in 2012; Supplementary Figure S1). Additionally, while both groups saw a decline in adjusted per-beneficiary rates of inpatient stays, this decline was less for those with AD/ADRD compared to those without AD/ADRD (−1.9 vs. −3.1 stays per year per 1,000; interaction p<.001; Supplementary Table S1)

Figure 1. Emergency Department Visits per 1,000 Among Medicare Beneficiaries With and Without AD/ADRD Ages 68 and Older, 2012–2018.

Figure 1.

Open in a new tab

Visits among a 20% sample of fee-for-service Medicare beneficiaries ages 68 and older who presented to EDs in the 50 United States and the District of Columbia. Linear regression model incorporates hospital referral region fixed effects, year, beneficiary AD/ADRD diagnosis, and an interaction between year and AD/ADRD diagnosis. Beneficiary AD/ADRD diagnosis was obtained from the Chronic Conditions Warehouse File.

Figure 2. Observation Stays per 1,000 Among Medicare Beneficiaries With and Without AD/ADRD Ages 68 and Older, 2012–2018.

Figure 2.

Open in a new tab

Observation stays originating in the ED among a 20% sample of fee-for-service Medicare beneficiaries ages 68 and older who presented to EDs in the 50 United States and the District of Columbia. Linear regression model incorporates hospital referral region fixed effects, year, beneficiary AD/ADRD diagnosis, and an interaction between year and AD/ADRD diagnosis. Beneficiary AD/ADRD diagnosis was obtained from the Chronic Conditions Warehouse File.

Trends in ED Visit Disposition by AD/ADRD Diagnosis

When we examined disposition trends at the ED visit level, we found that the adjusted proportion of ED visits ending in observation increased to a greater degree for people with AD/ADRD (+0.78% absolute percentage points per years; Supplementary Table S2) compared those without AD/ADRD (+0.63% points; interaction p<.001). While people with AD/ADRD had a lower adjusted proportion of ED visits ending in observation in 2012 compared to those without AD/ADRD (11.4% vs. 12.1%), this pattern was reversed in 2018 (16.4% vs. 16.2%; Figure 3). The decline in the proportion of ED visits ending in admission was similar for people with and without AD/ADRD (−1.01% vs. −1.00%; p=0.25 for year-AD/ADRD interaction; Supplementary Table S2). The proportion of ED visits ending in discharge for those without AD/ADRD increased by 0.33% absolute percentage points per year compared to +0.18% points per year for those with AD/ADRD, with a significant year-AD/ADRD interaction (p<.001). Taken together, these results suggest that while both groups have seen a decline in admission once they present to an ED, people without AD/ADRD are increasingly being discharged while those with AD/ADRD are being kept for observation (Supplementary Table S2).

Figure 3. Adjusted Trends in Disposition for Emergency Department (ED) Visits Among Traditional Medicare Beneficiaries With and Without AD/ADRD from 2012–2018.

Figure 3.

Open in a new tab

Estimates were generated from a linear probability model with visit year and patient AD/ADRD diagnosis as the primary predictors as well as an interaction between year and AD/ADRD status. Model also incorporates hospital random effects, principal visit diagnosis as well as beneficiary age, sex, Medicaid-eligibility, race and beneficiary chronic conditions (Medicare Chronic Conditions Warehouse file chronic conditions using the end-of-year indicator from the prior year).

Interaction Between Nursing Home Residency and AD/ADRD Status

When we examined trends in per-beneficiary acute care utilization by AD/ADRD and NH residency status, we found several patterns emerge. First, we found that ED utilization was highest in all study years for the AD/ADRD+ NH− group (1,291 ED visits per 1,000 beneficiaries in 2012). Rates of ED visits were similar for both groups of NH residents (ADRD− NH+ and ADRD+ NH+; Supplementary Figure S2). The ADRD− NH− group had less than half the rate of ED utilization in all seven years compared to the other three groups. All four groups saw an adjusted increase in ED visits (Supplementary Table S3) but this increase was greatest for the AD/ADRD+ NH− group and smallest for the AD/ADRD− NH− group (+28.3 vs. +8.5 visits per year per 1,000, respectively).

Observation utilization was highest among the AD/ADRD+ NH+ group (370 observations stays per 1,000 in 2012; Figure 4). The AD/ADRD+ NH− population had an observation utilization rate similar to the AD/ADRD− NH+ group (275 vs. 292 stays per 1,000 people in 2012), whereas the ADRD− NH− group had the lowest observation utilization (90 per 1,000 in 2012.) The AD/ADRD+ NH− group saw the greatest adjusted increase in observation stays over time (+15.4 stays per year per 1,000, 95% CI 15.0 to 15.7 per 1,000; Supplementary Table S3) and the AD/ADRD− NH− beneficiaries had the smallest increase in observation utilization over time (+4.2, 95% CI 4.1 to 4.3, stays per year per 1,000). Rates of inpatient stays were greatest among the AD/ADRD+ NH+ group in all study years (671 per 1,000 in 2012; Supplementary Figure S3) and lowest among AD/ADRD− NH− beneficiaries (+187 stays per year per 1,000 in 2012). All groups saw a significant adjusted decline in inpatient stays per 1,000 beneficiaries across the study period except the AD/ADRD− NH+ group (−1.0 stays per year per 1,000, 95% CI −2.1 to 0.2; Supplementary Table S3).

Figure 4. Trends in Observation Stays per 1,000 Population from 2012–2018 by Beneficiary AD/ADRD Diagnosis and Nursing Home Residency (NH) Status.

Figure 4.

Open in a new tab

Observation stays originating in the emergency department (ED) among a random 20% sample of beneficiaries of traditional Medicare ages 68 and older presenting to EDs in the 50 United States and the District of Columbia. Beneficiary AD/ADRD diagnosis was obtained from the Chronic Conditions Warehouse File and beneficiary nursing home residency status was determined using a method previously described by Yun et al (Health Serv Outc Res Methodol. 2010;10:100–10., 2010. 10: p. 100–110.)

Sensitivity Analyses

When we used an alternative classification for AD/ADRD diagnoses, a total of 51,798 (6.4%) beneficiaries in 2016–2018 with 115,002 ED visits (6.1%) in those years were reclassified as not having AD/ADRD (Supplementary Figure S4). The prevalence of AD/ADRD in 2018 was 10.2% (compared to 11.2% in the primary analysis using the CCW variable). The trends in ED visits and observation stays by beneficiary AD/ADRD and NH residency status were qualitatively similar to the primary analysis. We again found that the adjusted increase in observation stays was greater for people with AD/ADRD (+13.9 visits per year per 1,000, 95% CI 13.5, 14.2; p<.001) compared to those without AD/ADRD (+4.6 per year per 1,000, 4.5 to 4.7) with interaction p<.001.

DISCUSSION

In this analysis of over 19 million ED visits among nearly 7.5 million Medicare beneficiaries from 2012–2018, we found a marked rise in observation stays among Medicare beneficiaries with AD/ADRD, driven by both an increase in the number of people seeking care in the ED as well as an increase in the proportion of ED visits ending in an observation stay. People without AD/ADRD also saw an increase in observation utilization but of lesser magnitude. Moreover, within these groups, there was variation by beneficiary NH residency status, with the greatest observation utilization for those with AD/ADRD living in nursing homes and lowest for people with neither AD/ADRD nor nursing home residency. These differences widened over time, suggesting a differential impact of the evolving acute care landscape on vulnerable subpopulations of older adults. In particular, people with AD/ADRD experiencing an acute illness have had greater exposure to observation care compared to people without AD/ADRD and this trend has accelerated over time.

The reasons why people with AD/ADRD, particularly those who do not reside in nursing homes, have seen a greater increase in observation care utilization in recent years is likely multifactorial. We found that this population has more than double the rate of ED visits compared to people without dementia, likely reflecting a combination of poorer health status and impaired access to effective care. Prior work has shown that people with AD/ADRD in the community, but not NH settings, have higher use of the ED and inpatient care, but this work largely preceded the rapid rise in observation stays.27 Our work suggests people with AD/ADRD in NHs have the highest exposure to observation care, whereas people with AD/ADRD who do not reside in NHs have rates of observation utilization similar to nursing home residents without AD/ADRD. It is possible that people with AD/ADRD may have more nonspecific illness presentations (e.g., delirium) and thus do not “meet criteria” for a full inpatient stay initially in the ED. The disproportionate rise in observation stays in the non-NH AD/ADRD population may reflect the need for more time to coordinate care and/or arrange home supports than is feasible over the course of a typical ED visit.

Whatever the underlying causes of greater observation care use for people with AD/ADRD, the impact on their wellbeing deserves further scrutiny. The evidence regarding the impact of the growth of observation stays on patient outcomes has been mixed. The rate of return visits to the ED and hospital setting is high after observation stays.28,29 Some work suggests that the Hospital Readmissions Reduction Program may have been associated with higher mortality as hospitals began to preferentially use ED discharge and observation stays over inpatient hospitalization.30 Other work shows no excess mortality related to observation stays.28 Observation stays may also expose Medicare beneficiaries to higher out-of-pocket costs,31 which has led to greater financial concerns for lower-income Medicare beneficiaries.32 Additionally, it has also been hypothesized that the growth in observation stays may be limiting access to post-acute rehabilitation, as Medicare payment policy requires a three-day inpatient stay to qualify for the post-discharge skilled nursing facility (SNF) benefit.33 While this rule was suspended during the COVID-19 public health emergency,34 it was reinstated in May 2023, again generating concern that patients who would benefit from a SNF stay will either lack access or incur substantial out-of-pocket costs after an observation stay in the hospital. Reduced access to the SNF benefit is likely particularly detrimental for the population with AD/ADRD, who likely have greater post-discharge needs following an acute hospitalization in addition to greater health and social needs at baseline. All of these potential adverse consequences of observation stays for people with AD/ADRD have implications for health equity, given prior work showing that Black patients and people residing in communities with greater social disadvantage have both greater exposure to observation 21,29 as well as higher prevalence of AD/ADRD.35,36 37 Thus, there are likely multiple reinforcing mechanisms underlying racial and socioeconomic disparities in acute care delivery and outcomes, as has been noted for hospital readmissions in the population of Medicare beneficiaries with dementia. 38

This study builds on a growing body of work demonstrating that emergency and acute care is evolving away from inpatient stays toward an outpatient model of care and that evolution of emergency medicine practice plays a key role. Reducing admissions from the ED has been identified as a key mechanism for emergency physicians to improve value. 10 This trend has produced overall healthcare savings per episode, as inpatient stays are much more costly than an ED discharge or observation stay.39 This notion that emergency physician admission decisions are a key driver in moderating acute care spending is the basis of proposed alternative payment models involving emergency physicians in the shared risk and savings. However, any efforts to further drive down admission rates for the AD/ADRD population must consider factors that are not traditionally available in claims data (e.g., social supports, functional status).40 While it is reassuring that outcomes have improved in recent years for emergency care for Medicare beneficiaries overall41 and with AD/ADRD42, evidence specifically examining the impact of the apparent substitution of observation stays for inpatient care is needed in the AD/ADRD population. Our findings are also consistent with other work suggesting that community-dwelling people with AD/ADRD are a distinct population with high rates of acute care utilization that differ from those in long-term care settings.27,43 While fragmented care is associated with greater ED and hospital utilization for this population,44 whether a comprehensive community-based dementia care program can reduce ED visits and hospitalizations is less clear.45 Of note, the Centers for Medicare & Medicaid Services recently announced a new model—the Guiding an Improved Dementia Experience (GUIDE) Model—which will aim to test the impact of such a program on hospital and ED use among people with dementia beginning in 2024.46

This study has a number of limitations. First, it is limited to traditional Medicare beneficiaries and a growing number of Medicare beneficiaries have enrolled in Medicare Advantage (MA) in recent years. Future work should also evaluate the association of MA and observation care among those with AD/ADRD, especially given that prior work has found that higher use of observation care among those in MA than those in TM in general.47 Second, it is likely that some beneficiaries with AD/ADRD have not yet received a diagnosis. However, this is likely to lead to an underestimation of the differential increase in observation stays among people with AD/ADRD, given prior work showing that people with undiagnosed dementia have increased utilization of acute care compared to those without dementia.48 Furthermore, our dataset precedes the COVID-19 pandemic, but the growth in ED crowding and the reinstatement of the three-day inpatient rule to qualify for a SNF benefit suggests that the same forces impacting clinician decisions are still relevant currently.

In conclusion, we found that the rise in observation utilization for Medicare beneficiaries visiting the ED has disproportionately affected those with AD/ADRD, who have seen greater increases in ED visits and observation stays per visit than the non-AD/ADRD population. Our findings suggest that these trends are driven by rising ED utilization as well as evolving acute care practice patterns within the ED and hospital settings. These patterns vary by beneficiary residency status, with more rapid increases in ED and observation utilization among people with AD/ADRD who do not reside in NHs, who may have high health and social needs but limited supports relative to people without AD/ADRD or those with equivalent illness burden residing in NHs. The impact of greater observation care use for the AD/ADRD population, including subsequent lack of access to post-acute care services, on patient and caregiver wellbeing requires additional study.

Supplementary Material

Supinfo

Key Points

  • Among Medicare beneficiaries visiting the emergency department (ED) from 2012–2018, beneficiaries with Alzheimer’s Disease and Alzheimer’s Disease Related Dementias (AD/ADRD) experienced a disproportionate rise in observation care compared to those without AD/ADRD.

  • This rise in observation stays was driven by both an increase in the number of people seeking care in the ED as well as an increase in the proportion of ED visits ending in an observation stay.

Why does this matter?

  • Older Medicare beneficiaries with AD/ADRD account for a disproportionate share of acute care utilization and are likely more vulnerable to differences in acute care quality. These findings suggest that evolving patterns in emergency and acute care are having a differential impact on vulnerable subpopulations of older adults.

Funding Statement:

This work was funded by a grant from the National Institutes of Health National Institute on Aging (grant R56AG075017).

Sponsor’s Role

The sponsor of this work had no role in the design, methods, subject recruitment, data collections, analysis, and preparation of paper.

Footnotes

Conflict of Interest

Dr. Figueroa reported grants from the National Institute on Aging, the Department of Veterans Affairs, the Commonwealth Fund, Arnold Ventures, and the Episcopal Health Foundation, as well as consulting fees from Humana, Inc. and the InterAmerican Development Bank. No other disclosures were reported.

REFERENCES

  • 1.2022 Alzheimer’s disease facts and figures. Alzheimers Dement. Apr 2022;18(4):700–789. doi: 10.1002/alz.12638 [DOI] [PubMed] [Google Scholar]
  • 2.Kelley AS, McGarry K, Gorges R, Skinner JS. The burden of health care costs for patients with dementia in the last 5 years of life. Ann Intern Med. Nov 17 2015;163(10):729–36. doi: 10.7326/M15-0381 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Hurd MD, Martorell P, Delavande A, Mullen KJ, Langa KM. Monetary costs of dementia in the United States. N Engl J Med. Apr 04 2013;368(14):1326–34. doi: 10.1056/NEJMsa1204629 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Hebert LE, Weuve J, Scherr PA, Evans DA. Alzheimer disease in the United States (2010–2050) estimated using the 2010 census. Neurology. May 7 2013;80(19):1778–83. doi: 10.1212/WNL.0b013e31828726f5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Venkatesh AK, Mei H, Shuling L, et al. Cross-sectional Analysis of Emergency Department and Acute Care Utilization Among Medicare Beneficiaries. Acad Emerg Med. Jul 2020;27(7):570–579. doi: 10.1111/acem.13971 [DOI] [PubMed] [Google Scholar]
  • 6.Elder NM, Mumma BE, Maeda MY, Tancredi DJ, Tyler KR. Emergency Department Length of Stay Is Associated with Delirium in Older Adults. West J Emerg Med. May 3 2023;24(3):532–537. doi: 10.5811/westjem.59383 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Nagurney JM, Fleischman W, Han L, Leo-Summers L, Allore HG, Gill TM. Emergency Department Visits Without Hospitalization Are Associated With Functional Decline in Older Persons. Ann Emerg Med. Apr 2017;69(4):426–433. doi: 10.1016/j.annemergmed.2016.09.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Hill JD, Schmucker AM, Siman N, et al. Emergency and post-emergency care of older adults with Alzheimer’s disease/Alzheimer’s disease related dementias. J Am Geriatr Soc. Sep 2022;70(9):2582–2591. doi: 10.1111/jgs.17833 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Obermeyer Z, Cohn B, Wilson M, Jena AB, Cutler DM. Early death after discharge from emergency departments: analysis of national US insurance claims data. BMJ. Feb 1 2017;356:j239. doi: 10.1136/bmj.j239 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Smulowitz PB, Honigman L, Landon BE. A novel approach to identifying targets for cost reduction in the emergency department. Ann Emerg Med. Mar 2013;61(3):293–300. doi: 10.1016/j.annemergmed.2012.05.042 [DOI] [PubMed] [Google Scholar]
  • 11.Sabbatini AK, Nallamothu BK, Kocher KE. Reducing variation in hospital admissions from the emergency department for low-mortality conditions may produce savings. Health Aff (Millwood). Sep 2014;33(9):1655–63. doi: 10.1377/hlthaff.2013.1318 [DOI] [PubMed] [Google Scholar]
  • 12.Glober N, LaShell A, Montelauro N, et al. Impact of interhospital transfer on patients with Alzheimer’s disease and other related dementias. Alzheimers Dement (Amst). Jul-Sep 2023;15(3):e12469. doi: 10.1002/dad2.12469 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.McWilliams JM, Barnett ML, Roberts ET, Hamed P, Mehrotra A. Did Hospital Readmissions Fall Because Per Capita Admission Rates Fell? Health Aff (Millwood). Nov 2019;38(11):1840–1844. doi: 10.1377/hlthaff.2019.00411 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Feng Z, Wright B, Mor V. Sharp rise in Medicare enrollees being held in hospitals for observation raises concerns about causes and consequences. Health Aff (Millwood). Jun 2012;31(6):1251–9. doi: 10.1377/hlthaff.2012.0129 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Nuckols TK, Fingar KR, Barrett M, Steiner CA, Stocks C, Owens PL. The Shifting Landscape in Utilization of Inpatient, Observation, and Emergency Department Services Across Payers. J Hosp Med. Jun 2017;12(6):443–446. doi: 10.12788/jhm.2751 [DOI] [PubMed] [Google Scholar]
  • 16.Figueroa JF, Burke LG, Horneffer KE, Zheng J, John Orav E, Jha AK. Avoidable Hospitalizations And Observation Stays: Shifts In Racial Disparities. Health Aff (Millwood). Jun 2020;39(6):1065–1071. doi: 10.1377/hlthaff.2019.01019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Wright B, Parrish C, Basu A, Joynt Maddox KE, Liao JM, Sabbatini AK. Medicare’s hospital readmissions reduction program and the rise in observation stays. Health Serv Res. Jun 2023;58(3):554–559. doi: 10.1111/1475-6773.14142 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Wadhera RK, Joynt Maddox KE, Kazi DS, Shen C, Yeh RW. Hospital revisits within 30 days after discharge for medical conditions targeted by the Hospital Readmissions Reduction Program in the United States: national retrospective analysis. BMJ. Aug 12 2019;366:l4563. doi: 10.1136/bmj.l4563 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Sheehy AM, Graf B, Gangireddy S, et al. Hospitalized but not admitted: characteristics of patients with “observation status” at an academic medical center. JAMA Intern Med. Nov 25 2013;173(21):1991–8. doi: 10.1001/jamainternmed.2013.8185 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Sheehy AM, Graf BK, Gangireddy S, Formisano R, Jacobs EA. “Observation status” for hospitalized patients: implications of a proposed Medicare rules change. JAMA Intern Med. Nov 25 2013;173(21):2004–6. doi: 10.1001/jamainternmed.2013.9382 [DOI] [PubMed] [Google Scholar]
  • 21.Sheehy AM, Powell WR, Kaiksow FA, et al. Thirty-Day Re-observation, Chronic Re-observation, and Neighborhood Disadvantage. Mayo Clin Proc. Dec 2020;95(12):2644–2654. doi: 10.1016/j.mayocp.2020.06.059 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Yun H KM, Curtis JR, Delzell E, Gary LC, Saag KG, Morrisey MA, Becker D, Matthews R, Smith W, Locher JL. Identifying types of nursing facility stays using Medicare claims data: an algorithm and validation. Health Serv Outc Res Methodol 2010;10:100–10. 2010;10:100–110. doi:DOI 10.1007/s10742-010-0060-4 [DOI] [Google Scholar]
  • 23.Kim DH, Schneeweiss S, Glynn RJ, Lipsitz LA, Rockwood K, Avorn J. Measuring Frailty in Medicare Data: Development and Validation of a Claims-Based Frailty Index. J Gerontol A Biol Sci Med Sci. Jun 14 2018;73(7):980–987. doi: 10.1093/gerona/glx229 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Kim DH, Glynn RJ, Avorn J, et al. Validation of a Claims-Based Frailty Index Against Physical Performance and Adverse Health Outcomes in the Health and Retirement Study. J Gerontol A Biol Sci Med Sci. Jul 12 2019;74(8):1271–1276. doi: 10.1093/gerona/gly197 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Sheehy AM, Shi F, Kind AJH. Identifying Observation Stays In Medicare Data: Policy Implications of a Definition. J Hosp Med. Feb 1 2019;14(2):96–100. doi: 10.12788/jhm.3038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Moura L, Festa N, Price M, et al. Identifying Medicare beneficiaries with dementia. J Am Geriatr Soc. Aug 2021;69(8):2240–2251. doi: 10.1111/jgs.17183 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Feng Z, Coots LA, Kaganova Y, Wiener JM. Hospital and ED use among Medicare beneficiaries with dementia varies by setting and proximity to death. Health Aff (Millwood). Apr 2014;33(4):683–90. doi: 10.1377/hlthaff.2013.1179 [DOI] [PubMed] [Google Scholar]
  • 28.Dharmarajan K, Qin L, Bierlein M, et al. Outcomes after observation stays among older adult Medicare beneficiaries in the USA: retrospective cohort study. BMJ. Jun 20 2017;357:j2616. doi: 10.1136/bmj.j2616 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Wright B, Zhang X, Rahman M, Abir M, Ayyagari P, Kocher KE. Evidence of Racial and Geographic Disparities in the Use of Medicare Observation Stays and Subsequent Patient Outcomes Relative to Short-Stay Hospitalizations. Health Equity. Sep 2018;2(1):45–54. doi: 10.1089/heq.2017.0055 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Wadhera RK, Joynt Maddox KE, Wasfy JH, Haneuse S, Shen C, Yeh RW. Association of the Hospital Readmissions Reduction Program With Mortality Among Medicare Beneficiaries Hospitalized for Heart Failure, Acute Myocardial Infarction, and Pneumonia. JAMA. Dec 25 2018;320(24):2542–2552. doi: 10.1001/jama.2018.19232 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Kangovi S, Cafardi SG, Smith RA, Kulkarni R, Grande D. Patient financial responsibility for observation care. J Hosp Med. Nov 2015;10(11):718–23. doi: 10.1002/jhm.2436 [DOI] [PubMed] [Google Scholar]
  • 32.Goldstein JN, Schwartz JS, McGraw P, Hicks LS. “Implications of cost-sharing for observation care among Medicare beneficiaries: a pilot survey”. BMC Health Serv Res. Mar 7 2019;19(1):149. doi: 10.1186/s12913-019-3982-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Goldstein JN, Schwartz JS, McGraw P, Banks TL, Hicks LS. The Unmet Need for Postacute Rehabilitation Among Medicare Observation Patients: A Single-Center Study. J Hosp Med. Mar 2017;12(3):168–172. doi: 10.12788/jhm.2700 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Sheehy AM, Locke CF, Kaiksow FA, Powell WR, Bykovskyi AG, Kind AJ. Improving Healthcare Value: COVID-19 Emergency Regulatory Relief and Implications for Post-Acute Skilled Nursing Facility Care. J Hosp Med. Aug 2020;15(8):495–497. doi: 10.12788/jhm.3482 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Aranda MP, Kremer IN, Hinton L, et al. Impact of dementia: Health disparities, population trends, care interventions, and economic costs. J Am Geriatr Soc. Jul 2021;69(7):1774–1783. doi: 10.1111/jgs.17345 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Mayeda ER, Glymour MM, Quesenberry CP, Whitmer RA. Inequalities in dementia incidence between six racial and ethnic groups over 14 years. Alzheimers Dement. Mar 2016;12(3):216–24. doi: 10.1016/j.jalz.2015.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Dintica CS, Bahorik A, Xia F, Kind A, Yaffe K. Dementia Risk and Disadvantaged Neighborhoods. JAMA Neurol. Jul 19 2023;doi: 10.1001/jamaneurol.2023.2120 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Gilmore-Bykovskyi A, Zuelsdorff M, Block L, et al. Disparities in 30-day readmission rates among Medicare enrollees with dementia. J Am Geriatr Soc. Jul 2023;71(7):2194–2207. doi: 10.1111/jgs.18311 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Burke LG, Burke RC, Epstein SK, Orav EJ, Jha AK. Trends in Costs of Care for Medicare Beneficiaries Treated in the Emergency Department From 2011 to 2016. JAMA Netw Open. Aug 3 2020;3(8):e208229. doi: 10.1001/jamanetworkopen.2020.8229 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Smulowitz PB, Weinreb G, McWilliams JM, O’Malley AJ, Landon BE. Association of Functional Status, Cognition, Social Support, and Geriatric Syndrome With Admission From the Emergency Department. JAMA Intern Med. Aug 1 2023;183(8):784–792. doi: 10.1001/jamainternmed.2023.2149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Burke LG, Epstein SK, Burke RC, Orav EJ, Jha AK. Trends in Mortality for Medicare Beneficiaries Treated in the Emergency Department From 2009 to 2016. JAMA Intern Med. Nov 4 2019:1–9. doi: 10.1001/jamainternmed.2019.4866 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Burke LG, Burke RC, Duggan CE, Figueroa JF, Orav EJ, Marcantonio ER. Trends in healthy days at home for Medicare beneficiaries using the emergency department. J Am Geriatr Soc. Jun 10 2023;doi: 10.1111/jgs.18464 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Anderson TS, Marcantonio ER, McCarthy EP, Ngo L, Schonberg MA, Herzig SJ. Association of Diagnosed Dementia with Post-discharge Mortality and Readmission Among Hospitalized Medicare Beneficiaries. J Gen Intern Med. Dec 2022;37(16):4062–4070. doi: 10.1007/s11606-022-07549-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Amjad H, Carmichael D, Austin AM, Chang CH, Bynum JP. Continuity of Care and Health Care Utilization in Older Adults With Dementia in Fee-for-Service Medicare. JAMA Intern Med. Sep 1 2016;176(9):1371–8. doi: 10.1001/jamainternmed.2016.3553 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Jennings LA, Laffan AM, Schlissel AC, et al. Health Care Utilization and Cost Outcomes of a Comprehensive Dementia Care Program for Medicare Beneficiaries. JAMA Intern Med. Feb 1 2019;179(2):161–166. doi: 10.1001/jamainternmed.2018.5579 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Centers for Medicare & Medicaid Services (CMS). Guiding an Improved Dementia Experience (GUIDE) Model. CMS. https://www.cms.gov/priorities/innovation/innovation-models/guide
  • 47.Beckman AL, Frakt AB, Duggan C, et al. Evaluation of Potentially Avoidable Acute Care Utilization Among Patients Insured by Medicare Advantage vs Traditional Medicare. JAMA Health Forum. Feb 3 2023;4(2):e225530. doi: 10.1001/jamahealthforum.2022.5530 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Amjad H, Samus QM, Huang J, et al. Acute care utilization risk among older adults living undiagnosed or unaware of dementia. J Am Geriatr Soc. Feb 2022;70(2):470–480. doi: 10.1111/jgs.17550 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supinfo
NIHMS1978854-supplement-Supinfo.pdf (194.6KB, pdf)

RESOURCES