Dementia diagnosis in the hospital and outcomes among patients with advanced dementia documented in the Minimum Dataset

Abstract

Background:

Individuals with dementia do not always have a diagnosis of dementia noted on their hospital claims. Whether this lack of documentation is associated with patient outcomes is unknown. We examined the association between a dementia diagnosis listed on a hospital claim and patient outcomes among individuals with a Minimum Data Set assessment.

Methods:

Retrospective cohort study using administrative claims data and nursing home Minimum Data Set (MDS) assessments. Hospitalized patients aged 66 and older with advanced dementia noted on an MDS assessment completed within 120 days prior to their first hospitalization in 2017. Advanced dementia was defined based on an MDS diagnosis of dementia, dependency in four or more Activities of Daily Living, and a Cognitive Function Scale score indicative of moderate to severe impairment. Multilevel regression with a random intercept at the hospital level was used to examine the relationship between documentation of dementia in inpatient hospital Medicare claims and the following patient outcomes after adjusting for patient and hospital characteristics: invasive mechanical ventilation (IMV) use, intensive care unit or coronary care unit (ICU/CCU) use, 30-day mortality, and hospital length of stay (LOS).

Results:

In 2017, among 120,989 patients with advanced dementia and a nursing home stay, 90.57% had a dementia diagnosis on their hospital claims. In adjusted models, documentation of a dementia diagnosis was associated with lower use of the ICU/CCU (Adjusted Odds Ratio (AOR):0.78 [95% confidence interval 0.74, 0.81]), use of IMV (AOR:0.50 [0.47, 0.54]) and 30-day mortality (AOR:0.81 [0.77, 0.85]). Patients with a dementia diagnosis had a shorter LOS.

Conclusions:

Among patients with advanced dementia, those whose dementia diagnosis was documented on their inpatient hospital Medicare claim experienced lower use of ICU/CCU, use of IMV, lower 30-day mortality, and shorter LOS than those whose diagnosis was not documented.

INTRODUCTION

Medicare claims’ measures of dementia lack sensitivity.^1–4 Previous work found that documentation of dementia improved since 2012.⁵ Nonetheless, 14% of inpatient hospital Medicare claims from patients with a previous clinical assessment documenting dementia did not have the diagnosis listed in 2017.⁵

Advanced dementia is a progressive, fatal illness characterized by a high probability of pneumonia, eating problems, and pain.⁶ Increased recognition of dementia in the hospital setting may decrease the use of burdensome interventions such as invasive mechanical ventilation (IMV) and stays in the intensive care unit (ICU), which prolong suffering without a clear survival benefit.^7–10 Individuals who visit the ICU and who receive IMV are at greater risk for delirium and emotional distress,^9–10 and over 80% of individuals with advanced dementia who receive IMV die within 1 year.¹¹ Similarly, longer hospital stays are associated with increased functional loss and adverse hospital events among individuals with advanced dementia;¹² recognition of dementia may help prompt physicians to not use life-sustaining measures that prolong length of stay.¹³

An accurate diagnosis and documentation of dementia in the hospital may reduce the use of potentially burdensome treatments. Unlike hospital Medicare claims, the Minimum Data Set (MDS) 3.0 contains a validated, clinical measure of cognitive functioning.¹⁴ The purpose of this study was to examine whether a diagnosis of dementia in the inpatient hospital Medicare claims was associated with healthcare utilization among patients with advanced dementia as indicated in the MDS.

METHODS

Sample

We utilized a national repository of MDS assessments linked with Medicare Provider Analysis and Review hospital claims data from 2017. We gathered demographic data and date of death from the Medicare Beneficiary Enrollment file. Hospital market characteristics came from the Dartmouth Atlas of Health Care.¹⁵

For our sample, we identified patients with at least one hospitalization and an MDS assessment completed between 0 and 120 days prior to the index hospitalization. In cases of multiple eligible hospitalizations, we utilized information from the first in 2017.

We restricted our sample to Medicare Beneficiaries aged 66 and older who had advanced dementia noted in the MDS and who were not comatose (N=127,081). Missing data were listwise deleted (5%) for a total analytic sample size of 120,989. Advanced dementia was based on a person having a diagnosis of dementia noted on any previous MDS assessment (i.e., MDS items I4200 indicating Alzheimer’s disease and/or I4800 Non-Alzheimer’s dementia), four or more dependencies in the Activity of Daily Living (ADLs), and a Cognitive Function Scale score (CFS) (range 1–4) of at least three, indicative of moderate or severe impairment. The CFS is composed of two validated scales to assess cognitive impairment: the CPS and the BIMS.^16,17 Additional staff assessment is performed if residents cannot complete the BIMS.¹² The literature suggests that the CFS, the BIMS, and the CPS are well correlated.¹⁴

Study Measures

We examined 4 outcomes: use of the intensive care unit or cardiac care unit (ICU/CCU), use of IMV, 30-day mortality, and hospital length of stay (LOS). IMV use was identified from Medicare claims data using ICD-9-CM and ICD-10-CM procedure codes 96.7x/5A1935Z, 5A1945Z, and 5A1955Z).¹⁸ Our primary independent variable was whether the patient’s hospital claim noted a diagnosis of dementia. For a list of ICD-10 diagnosis codes used to define dementia, please see the supplemental material (Table S1).

Age, race, and sex were derived from the Medicare Beneficiary Summary File and were included because dementia documentation in the hospital is higher among individuals who are older, female, and white.⁵ We controlled for marital status using data from the MDS given some research suggests that underdiagnosis of dementia is more common among unmarried people.¹⁹ We also created indicator measures depicting whether an admission was surgical to control for potential differences in diagnosis patterns across specialties. We controlled for whether the patient was dually eligible for Medicare and Medicaid at the time of hospital admission because dually enrolled individuals are less likely to be diagnosed with dementia.²⁰ We controlled for Medicare Advantage plan enrollment because of incentives to code for disease severity.²¹

Previous research suggests that documentation of dementia is lower among individuals with worse health.^19,20 Using 17 clinical items and age groups from MDS assessments, we utilized an existing validated model of mortality (the MDS 3.0 Mortality Risk Score).²² Other characteristics, medical diagnoses and medical circumstances of the person prior to hospitalization were based on an MDS assessment completed within 120 days prior to that hospitalization. Chronic conditions included chronic lung disease, diabetes, heart failure, hip fracture, stroke, and serious mental illness (schizophrenia or bipolar disorder). We also included an index of ADL impairments derived from the MDS. The scale ranged from 0–28 with the maximum score indicating total ADL dependence.²³ We also included an indicator depicting whether an individual was bedfast.

We controlled for moderate cognitive impairment by creating an indicator of whether the patient had a CFS score of 3 compared to 4. To measure dementia-related behaviors we used the Agitated and Reactive Behavior Scale, a composite measure that ranges from 0–12 with the maximum score depicting a behavior occurring daily on all four items (i.e., physical behavioral symptoms directed at other people, verbal behaviors directed at other people, other behaviors not directed at other people, and rejection of necessary care).²⁴ Indicators were created for feeding tube use and use of a mechanical soft diet.

Given that markets vary geographically in the amount of care they provide for individuals at the end of life, which may also be related to care practices such as diagnosis patterns, we incorporated a hospital care intensity index (HCI) from the Dartmouth Atlas of HealthCare. The index measures the propensity of hospitals within the hospital’s ZIP code to use acute care in the hospital to manage chronic illness.¹⁵ Values range from 0.4–1.8, with higher numbers indicating more aggressive markets.

Statistical Analyses

To examine the relationship between documentation of dementia in the hospital claim and patient outcomes, we estimated multilevel models with random intercepts at the hospital level adjusting for patient and hospital characteristics. For the binary outcomes of ICU/CCU use, IMV use, and 30-day mortality, we used a multilevel logistic regression modeling approach. For the length of stay outcome, we estimated a multilevel linear regression model.

Measurement error could result if residents had delirium and were incorrectly classified as having dementia in the MDS. As a sensitivity analysis, we repeated the analysis using only patients whose CFS score stayed the same or increased since their least recent MDS assessment within the 120 day study window, indicating a progressive disease trajectory. We also examined the extent to which the variation in documentation patterns of dementia in the hospital was explained by variations in hospital practices. To do this, we examined an empty multilevel logistic regression model with a random intercept at the hospital level and a documentation of dementia in the hospital claim as the outcome to calculate the intraclass correlation coefficient. Stata 16.0 (StataCorp LLC, College Station, TX) was used for analysis.

RESULTS

Sample Description

Table 1 presents descriptive statistics of the sample, overall and by diagnosis of dementia in the hospital. A total of 120,989 patients aged 66 and over with advanced dementia experienced at least one nursing home transition and one hospital stay in 2017. Approximately half of the cohort was aged 85 or older; 63.7% were women, 17% were Black, and 30% were married. Approximately 20.6% of the sample was bedfast and 40.4% had a CFS score indicating the most severe dementia. The average ADL dependency score was 21.1. In our sample, 90.6% had a dementia diagnosis on their hospital claims. Individuals with dementia documented in the hospital were more likely to be younger, to be less cognitively impaired, to be a non-white race, and to have chronic conditions such as diabetes and stroke.

Table 1.

Descriptive statistics of patients aged 66+ with advanced dementia in 2017 with a Minimum Data Set Assessment, by diagnosis of dementia on the hospital claim (n=120,989)

Characteristic	Full sample	Not documented	Documented
Age 75–84, n(%)	44,558 (36.8%)	4,056 (35.5%)	40,502 (37.0%)
Age 85+, n(%)	60,443 (50.0%)	4,863 (42.6%)	55,580 (50.7%)
Women, n(%)	77,341 (63.9%)	6,725(59.0%)	70,616 (64.4%)
Black, n(%)	20,553 (17.0%)	2,235(19.6%)	18,318 (16.7%)
Hispanic, n(%)	3,771 (3.1%)	439 (3.9%)	3,332 (3.0%)
Asian, n(%)	2,829 (2.3%)	361 (3.2%)	2,468 (2.3%)
Other, n(%)	2,193 (1.8%)	285 (2.5%)	1,908 (1.7%)
Married, n(%)	36,423 (30.1%)	3,479(30.5%)	32,944 (30.1%)
Medicare Advantage, n(%)	25,830 (21.4%)	2,329(20.4%)	23,501(21.5%)
Dually eligible, n(%)	75,463(62.4%)	7,161 (62.8%)	68,302 (62.3%)
Surgical admission, n(%)	14,219 (11.8%)	1,303 (11.4%)	12,916 (11.8%)
MRS, (mean, SD)a	6.5 (1.4)	6.7 (3.0)	6.5 (2.8)
ABS, (mean, SD)b	0.6 (1.4)	0.3(1.0)	0.6(1.4)
Feeding tube, n(%)	10,209 (8.4%)	1,599(14.0%)	8,610(7.9%)
Mechanically soft diet, n(%)	62,481 (51.6%)	5,627 (49.3%)	56,854 (51.9%)
CFS score 4, n(%)	48,914 (40.4%)	3,926 (34.4%)	44,988(41.1%)
ADL score, (mean, SD)c	21.1 (3.2)	21.3(3.3)	21.1(3.2)
Bedfast, n(%)	24,911 (20.6%)	2,852 (25.0%)	22,059 (20.1%)
Lung disease, n(%)	23,602 (19.5%)	2,887 (25.3%)	20,715 (18.9%)
Diabetes, n(%)	40,107 (33.2%)	4,775 (41.9%)	35,332 (32.2%)
Heart failure, n(%)	25,117 (20.8%)	3,147 (27.6%)	21,970 (20.1%)
Hip fracture, n(%)	6,375(5.3%)	493 (4.3%)	5,882(5.4%)
PVD, n(%)	12,621 (10.4%)	1,508 (13.2%)	11,113(10.1%)
Serious mental illness, n(%)	8,821(7.3%)	1,013 (8.9%)	7,808(7.1%)
Stroke, n(%)	14,974(12.4%)	2,283 (20.0%)	12,691(11.6%)
HII (mean, SD)d,e	1.1 (0.30)	1.1 (0.3)	1.1 (0.3)

Abbreviation: MRS, Mortality risk score; ABS, agitated behavior scale; CFS, cognitive function scale, ADL, activities of daily living; PVD, peripheral vascular disease’ HII, hospital intensity index.

^a.range: 1–24

^b.range: 0–12

^c.range: 0–28

^d.range: 0.4–1.8.

^e.Propensity of the ZIP code to rely on acute care in the hospital for managing chronic illness

Outcomes After a Diagnosis of Dementia in the Hospital

Figure 1 shows the unadjusted outcomes among patients with and without a diagnosis of dementia noted in the hospital claim. Patients with a diagnosis of dementia in the hospital were less likely than those without a diagnosis of dementia to receive care in the ICU or CCU (39.6% vs. 46.3%). Patients with a diagnosis of dementia on the hospital record were also less likely to receive IMV (4.3% vs. 10.0%). Rates of 30-day mortality were lower when dementia was documented on the hospital record; 26.4% of patients with a dementia diagnosis died in 30 days compared to 28.7% of patients without a dementia diagnosis. Hospital lengths of stay were shorter when dementia was noted on the hospital record; patients with a diagnosis stayed an average of 5.9 days compared to patients without a diagnosis who stayed an average of 6.4 days.

Figure 1.

Unadjusted outcomes among patients aged 66+ with advanced dementia with a Minimum Data Set assessment, with and without dementia documented on the hospital claim

Abbreviations:ICU/CCU intensive care unit or coronary care unit, IMV invasive mechanical ventilation

Notes: Hospitalizations occurred during the year 2017. Advanced dementia was defined as an MDS diagnosis of dementia, four or more Activities of Daily Living dependencies, and a Cognitive Function Scale score of three or four.

Table 2 displays the adjusted odds ratios (AOR) and confidence intervals assessing the relationship between a diagnosis of dementia noted in the hospital record and the study outcomes. Coefficients can be found in the supplementary material (Tables S2 and S3). After adjustment for socioeconomic and health characteristics, a diagnosis of dementia in the hospital claim was associated with lower odds of ICU/CCU utilization (AOR: 0.71 [95% confidence interval 0.68, 0.74]). The use of IMV was also lower among patients who had a dementia diagnosis on their hospital claim (AOR:0.48 [0.44, 0.51]), as were rates of 30-day mortality (AOR:0.74 [0.70, 0.77]). Hospital lengths of stay were lower among patients with a diagnosis of dementia documented in the hospital. Only 1.3% of our sample experienced an improvement in their CFS score between the least recent and most recent MDS assessment. When excluding these patients from our analysis, results were unchanged. Our multilevel logistic regression model examining how diagnosis patterns of dementia varied across hospitals had an ICC of 0.06, indicating that only 6% of the variation in diagnosis patterns was explained by between-hospital practice differences.

Table 2.

Association between hospital diagnosis of dementia and healthcare utilization outcomes in 2017 among adults aged 66+ with a Minimum Data Set diagnosis of dementia

Analysis	OR or Coefficient (95% CI) Year 2017
	Unadjusted	Adjusteda
Intensive care unit or coronary care unit use	0.71 (0.68, 0.75)	0.78 (0.74, 0.81)
Invasive mechanical ventilation use	0.41 (0.38, 0.44)	0.50 (0.47, 0.54)
30-day mortalityb	0.87 (0.84, 0.91)	0.81 (0.77, 0.85)
Hospital length of stayc	−0.47 (−0.58, −0.36)	−0.32 (−0.43, −0.21)

^a.The models were adjusted for all characteristics in Table 1, which include demographic characteristics, hospital admission characteristics, cognitive and functional impairment, comorbid conditions, treatments, and the hospital’s market characteristics.

^b.The coefficient represents an odds ratio from a mixed effects logistic regression model with a random intercept at the hospital level.

^c.The coefficient is from a mixed effects linear regression model with a random intercept at the hospital level

DISCUSSION

Previous research has demonstrated that Medicare claims measures of dementia lack sensitivity.^1–5 However, limited work has examined the implications of a lack of documentation of dementia in inpatient hospital Medicare claims on patient outcomes such as intensive hospital procedures and mortality. Our study found that the majority (91%) of hospital claims of individuals with advanced dementia noted on an MDS assessment prior to that hospitalization had a diagnosis of dementia documented. However, failure to document a dementia diagnosis in the hospital claim when noted on an MDS assessment prior to that hospitalization was associated with higher ICU/CCU, use of IMV, and 30-day mortality. Among these patients with dementia, a lack of documentation of dementia diagnosis in the hospital claim was also associated with a longer hospital stay than among those whose dementia diagnosis was documented.

Dementia is a progressive, neurological disease; this disease trajectory should shape decisions to use life sustaining treatments with minimal benefits regardless of admitting diagnosis. The most common primary diagnosis in our sample was pneumonia/septicemia (25%). Previous research indicates that increases in the use of IMV are not associated with decreased mortality among individuals with advanced dementia,¹¹ many who have pneumonia/septicemia at the time of their death.¹⁸ By including patients with moderate to severe cognitive impairment and 4 or more ADL impairments, we restricted our sample to patients with the most advanced disease. We were unable to assess with certainty the reason for a lack of documentation of dementia in 9% of hospital claims in our study. A lack of recognition of dementia may play a role, as clinicians often fail to recognize and document dementia in the hospital setting^5,25 which may be related to insufficient time available to spend with patients and a lack of a systematic method of assessing dementia.^26,27

Patients with dementia who reside in nursing homes rely on nursing home staff to communicate their health status to hospital physicians; however, this documentation often lacks information on mental status.²⁸ Adequate documentation of dementia in the records provided from the nursing home to the hospital can help the hospital team understand the condition of the patient. Another potential solution for increased recognition of dementia includes increased assessment of dementia in the emergency department to identify need for further assessment.^29,30

Our study has some limitations. Primarily, our results reflect patients with advanced dementia who had a nursing home stay within 120 days of a hospitalization. Therefore, our results may not be generalizable to patients without a nursing home stay. The MDS does not collect information on preferences, advance directives, or orders to forgo life sustaining treatment. Individuals with a diagnosis of dementia in the hospital were less likely to have chronic conditions such as stroke. We adjusted for disease severity in our models using measures available from the MDS and hospital records; however, there could be additional unmeasured confounding which we are unable to include in our analyses.

Further, we used the MDS CFS scale as our reference standard for dementia. Although the scale is a validated measure of cognitive function,¹⁴ there may be remaining measurement error if individuals were falsely classified as having dementia when they were experiencing delirium. Our analysis indicated that only 1.3% of our sample experienced an improvement in cognitive functioning since their least recent CFS assessment, indicating that there are likely not many of these cases in our sample.

Documentation of dementia in hospital records may improve patient outcomes. Additional research is needed to determine the explanation for a lack of documentation of dementia in the hospital. Potentially, our study points to a need to improve communication between nursing homes and hospitals, and to screen older adults for dementia in the hospital setting as steps to better target patient care for individuals who would benefit from comfort care.

Supplementary Material

supinfo

Table S1. ICD-10 diagnosis codes used to identify dementia in the hospital claim

Table S2. Multilevel logistic regression of the relationship between documentation of dementia in the hospital and healthcare outcomes among adults aged 66+ with a Minimum Data Set diagnosis of dementia

Table S3. Multilevel linear regression of the relationship between documentation of dementia in the hospital and hospital length of stay among adults aged 66+ with a Minimum Data Set diagnosis of dementia

Key Points.

• Documentation of advanced dementia in Medicare claims is associated with lower use of the intensive care unit, less use of invasive mechanical ventilation, and shorter lengths of stay in the hospital.

• Documentation was associated with lower rates of 30-day mortality.

Why does this paper matter?

Increased screening of dementia in the hospital setting may be warranted.