Discharge Communication of Dementia-Related Neuropsychiatric Symptoms and Care Management Strategies during Hospital to Skilled Nursing Facility Transitions

Abstract

Provided the complexity of managing dementia-related Neuropsychiatric Symptoms (NPS), accurate communication about these symptoms at hospital discharge is critical to facilitating safe and effective transitions, particularly transitions from hospitals to Skilled Nursing Facilities (SNF), which are often poorly managed. SNF providers have cited under-communication regarding NPS as a major challenge that contributes to poor outcomes including re-hospitalization. This multi-site retrospective cohort study identified omission rates for NPS and associated management strategies in discharge communication as compared to medical record documentation in the 72 hours preceding discharge among hospitalized patients with dementia. High rates of omission were found across NPS and management strategies: anxiety (94%), agitation/aggression (77%), hallucinations (85%), 1:1 supervision (90%), high fall risk (89%), use of restraints (91%). Omission rate for new or modified antipsychotic medication was 12.9%. Findings underscore the need for additional research on cross-setting communication regarding care needs of patients with dementia—who often cannot communicate these needs on their own — in facilitating high quality transitions.

INTRODUCTION

People living with dementia experience frequent transitions between and within settings of health care.¹ People with dementia are particularly vulnerable during these periods of transition, partly due to their diminished ability to advocate for their care needs and compensate for gaps in communication during transitions.^{1, 2} For patients with dementia, transitions into a Skilled Nursing Facility (SNF) after hospitalization are particularly common given many are residing in a SNF prior to hospitalization (more than half of SNF patients have dementia) and compared to cognitively unimpaired persons, patients with dementia are more likely to be discharged to a SNF following hospitalization.^{3, 4} Dementia-related neuropsychiatric symptoms (NPS) often complicate these transition events, as they are often exacerbated during acute illness and hospitalization due to the loss of familiar environments and routines.^5–7

Common and persistent non-cognitive symptoms in dementia, NPS are broadly understood to have personal, caregiving, and environmental determinants.⁸ Examples of NPS include agitation, aggression, care resistance, wandering, and apathy.⁶ These complex symptoms often pose specific safety risks (i.e. wandering could lead to elopement or aggression could lead to physical harm) and necessitate specialized management (i.e. the use of individualized interventions) to reduce symptom burden or disruptiveness. Management of NPS in hospital settings often includes the use of chemical restraints, primarily antipsychotic medications, which have received a United States Food and Drug Administration (FDA) Black Box warning for their use among persons with dementia attributable to increased risk for cerebrovascular events and death.^{9, 10} Other commonly used strategies in response to disruptive or aggressive NPS include direct 1:1 observation and in some cases physical restraints despite extensive guidelines disincentivizing the use of physical restraints.^{11, 12} Because efforts to prevent negative sequelae that accompany NPS are often hindered by difficulty identifying individual NPS and their underlying causes, alerting the next setting of care to specific NPS and existing management strategies is a critical component of high quality care coordination for this population.

Despite their distinct transitional care needs and heightened vulnerability, emerging evidence documents that people with dementia receive suboptimal care coordination.^{13, 14} The consequences of suboptimal care are reflected by substantial disparities in the use of acute and post-acute care.¹³ Specifically, people with dementia experience disproportionately higher rates of acute and post-acute care utilization, including hospitalizations, emergency department visits, 30-day re-hospitalization events, repeat emergency department visits, and entry into post-acute care settings.^{3, 15–20} People with dementia are also more likely to experience other adverse events because of these acute care and post-acute utilization patterns, including falls, delirium, hastened cognitive decline and permanent institutionalization.^{19, 21, 22} Collectively, frequent acute and post-acute care utilization among persons with dementia is considered to reflect an over-reliance on reactive management strategies, as many of these transitions are unplanned and potentially avoidable.²³ Despite the importance of NPS management for patients with dementia, their management surrounding acute and post-acute illness has not been previously examined.

Recent studies detailing the perspectives of hospital and SNF providers and a review examining perspectives of family caregivers identify inadequate transfer of information regarding NPS as a major factor in poorly orchestrated transitions and contributor to unnecessary patient, caregiver and staff stress, as well as critical events such as avoidable death or re-hospitalization.^24–26 A recent systematic review identified specific areas of breakdown in communication between hospital and SNF providers, highlighting the omission of key information about NPS and related care management strategies such as use of 1:1 supervision, restraints, or antipsychotic medications.²⁷ The systematic review further reported that studies linked the under-reporting of key care management strategies to post-discharge placement concerns.²⁷ In cases where communication on NPS and care management strategies is present, SNF providers have cited that discharge communication is often inaccurate, conflicting or missing key details.²⁴ Of particular note, SNF providers report challenges admitting persons with dementia requiring antipsychotics to manage NPS and describe routinely receiving incomplete medication histories regarding antipsychotic utilization during hospitalization.^{24, 28} In addition, SNF providers highlight abrupt withdrawal of antipsychotics prior to discharge as a common problem.^{24, 28} Given attention to antipsychotic use among residents with dementia as a quality indicator in SNF settings, its consideration as a type of care management strategy is particularly relevant.^{29, 30} Under-communication and omission of information about NPS and care management strategies may significantly limit the nursing facilities’ ability to develop a safe, individualized, person-centered care plan and subsequently increase risk for re-hospitalization.^{24, 25}

SNF clinicians–often nurses who are responsible for managing the transition of patients from the hospital and developing and communicating plans of care to allied health providers–and non-licensed staff alike rely heavily on the written discharge summary as the primary source for relevant patient information to adequately care for the patient.²⁴ Written discharge summaries are relied on for admitting orders in this context as they have historically been, and continue to be, the only mandated form of communication between providers following hospitalization in the United States.³¹ Accurate, thorough discharge communication is particularly important for SNF patients as they are not required to be seen by a physician for up to 30 days post discharge.³²

Not only is accurate and thorough discharge summary communication critical for SNF staff to adequately manage transitions for people with dementia, but it is an important source of information for family members.²⁶ Family members are integral stakeholders in decision-making surrounding transitions into SNF settings; however they may be unable to make informed decisions without detailed, accurate knowledge of the patient’s hospital stay.^{33, 34} A recent study found that in up to 20% of transitions, family members and caregivers played a primary role in decisions to re-hospitalizations for patients with dementia.³⁵ In these situations, premature discharge, unresolved clinical issues, poor discharge planning, and perception of poor SNF care or medication management, were identified as factors influencing decision-making regarding re-hospitalization. While inadequate and inaccurate written discharge communication has been cited as exacerbating existing poor communication surrounding transitions into SNFs for patients with dementia, the type and extent of written discharge communication regarding NPS and care management strategies has not been previously examined outside of qualitative reports. The objective of this study was to assess written discharge communication regarding NPS and associated care management strategies during hospital-to-SNF transitions and to determine rates of omissions in communication about target NPS and management strategies.

MATERIALS AND METHODS

Sample/Setting and Data Sources

We conducted a retrospective cohort study to examine documentation of NPS and management strategies leading up to hospital discharge, and to determine rates of omission about these specific care needs in written discharge summaries. The sample included all non-hospice Medicare beneficiaries with a diagnosis of dementia who were hospitalized due to stroke or hip/femur fracture and discharged to SNF settings from one of two urban hospitals in the Midwest between years 2003–2009. Discharge to SNF settings was identified by evaluating documented discharge disposition which is mandatorily recorded by hospital staff and was verified by presence of a SNF claim within 2 days of hospital discharge. For the purposes of this study, a SNF was denoted as a post-acute care site that received payments from Centers for Medicare & Medicaid Services (CMS) for Skilled Nursing services, with patients in the study sample spanning those receiving short-term rehabilitative care and those previously receiving long-term care who were newly eligible for skilled nursing care upon discharge. Both hospitals were non-profit, with one being an academic and the other being a community hospital.

The analytic sample for this study was derived from an initial cohort of patients hospitalized with hip/femur fracture and stroke discharged to SNF settings. These conditions were selected as they constitute a high proportion of SNF admissions and are at high risk for adverse outcomes.^36–38 Medicare claims obtained from the Chronic Conditions Data Warehouse were utilized to confirm primary diagnosis of hip fracture and stroke (hip fracture: ICD-9 codes 805.6, 805.7, 806.6, 806.7, 808, and 820; stroke: ICD-9 codes 431, 432, 434, and 436). This resulted in an initial overall sample of 1,449 patients across both hospitals. Subsequently, medical records were reviewed for other inclusion criteria and patients were excluded if they were not discharged to a SNF or were discharged with hospice (n=295), were determined to have a miscoded primary diagnosis (most commonly subdural hematoma coded as stroke; n=66), did not have a discharge summary (n=13), or were under the age of 18 (n=5.) To derive the analytic sample for the present study, patients with pre-hospitalization diagnosis of dementia were identified using a validated algorithm developed by Taylor and colleagues (ICD-9 codes 331.0, 331.1, 331.2, 331.7, 290.0, 290.1, 290.10, 290.11, 290.12, 290.13, 290.20, 290.21, 290.3, 290.40, 290.41, 290.42, 290.43, 294.0, 294.1, 294.8, and 797).³⁹ This yielded a final analytic sample of 298 patients for evaluation of NPS, fall risk, restraints and 1:1 supervision, and 296 for evaluation of antipsychotic medications as 2 patients had missing medication administration record data in their EHR. Medicare data were then linked to abstracted clinical data from patients’ electronic health records (EHR) and discharge summaries by aligning Medicare identification number, patient age and sex, eligible hospital stay, and dates of hospitalization.

Research Ethics and Patient Consent

The study was approved by the University of Wisconsin-Madison Institutional Review Board as a Minimal Risk IRB in 2012 and has been under continuous review, with the last approval in 2019 (Approval Number: 2012–0577-CR007). Because this is a retrospective chart review with minimal risk to patients, written and oral consent was waived by the IRB.

Data Collection and Analysis

Four medical record abstractors, with Bachelor’s and Master’s-level degrees and concurrent or prior training in nursing and medicine, independently reviewed and extracted data on NPS and care management strategies from each patient’s EHR. Specifically, medical record abstractors reviewed clinical documentation from the last 72 hours of hospitalization, including all available flowsheets and clinical notes, and the written discharge summary, to identify clinical documentation on NPS and care management strategies. Abstraction of antipsychotic medications focused specifically on identifying administration and communication regarding new or modified (as opposed to long-standing) antipsychotic medications. All abstractors received detailed training with abstraction manuals and tools, which included demonstrated and supervised piloting of standardized methods for accessing medical records and abstracting study variables.

Abstraction procedures were performed in a standardized manner using abstraction manuals and tools with training and guidance for medical record review based on prior work.^{40, 41} Guidance on identifying NPS within clinical documentation contained within abstraction manuals and tools was informed by the Neuropsychiatric Inventory.⁴² Specific variables included anxiety, agitation/aggression, hallucinations, depression, disinhibition, irritability or lability and aberrant motor behavior (Supplementary Material 1). We identified documentation of associated care management strategies including use of new or modified antipsychotic medications, use of 1:1 direct supervision, designation of high fall risk status, and use of restraints of any kind (Supplementary Material 1). To further characterize the sample, presence of other relevant medications during the index stay (i.e. anxiolytics) were also identified.

Frequency of clinical documentation of NPS or care management strategies constituted discrete instances of clinical documentation of distinct NPS or care management strategy events. For example, if a clinician described a singular symptom in one assessment note early in the shift this was counted as one NPS documentation event; a separate note from the end of the shift would constitute a separate, distinct NPS documentation event unless it was explicit in the documentation that the author was referring to the same event based upon recorded details (e.g. time, presentation, intervention) in their charting. This approach was adopted to provide a conservative and event-centered estimate of NPS documentation frequency, rather than for example counting the number of times descriptors such as “agitation” are mentioned which would serve as a less sensitive approximation for NPS events and introduce additional bias. Documentation of psychiatric diagnoses did not constitute clinical documentation of an NPS event but were identified through a review of H&P and diagnosis lists to inform sample demographics.

Interrater reliability and agreement assessment were evaluated for a randomly selected 10% of all patient records using Cohen’s κ, with disagreements on specific extraction and categorization of clinical documentation resolved by a third team member. Following arbitration, the circumstances regarding specific disagreements and the data in question were discussed at study team meetings to foster greater reliability in abstraction. Disagreements between medical record abstractors most frequently arose when reviewers differentially categorize it as a specific yet similar variable, such as labelling data as documentation of aggression or agitation. Across all abstracted variables, overall percent agreement was 95%, and the Cohen’s κ average was 0.89.

Analyses were performed in STATA version 15. Clinical documentation regarding specific NPS and care management strategies were evaluated using descriptive statistics to determine prevalence and frequency of clinical documentation in the medical record for each day in the 72-hour time period leading up to hospital discharge, and in the written discharge summary. Omission rates were only calculated for patients who had clinical documentation for a particular symptom or care management strategy and were calculated at a patient level. Omission rates denote lack of communication about a specific symptom or care management strategy in the discharge summary that was documented in clinical documentation during the 72 hours prior to discharge.

RESULTS

Sample Characteristics

Characteristics of the 298 patients with dementia, 81.2% of whom had a hip/femur fracture and 18.8% who were hospitalized for stroke are detailed in Table 1. Utilization measures (e.g. length of stay), pre-existing psychiatric diagnoses, and chronic conditions are also reported in Table 1 with additional detail in Supplementary Material 2. Medicare claims indicated 30.5% (n=91) received nursing home care in the 365 days leading up to the index hospitalization. A number of patients within the sample had pre-existing psychiatric diagnoses which were communicated in their clinical documentation: 32.6% of patients (n=97) had a depression diagnosis, 3.0% (n=9) schizophrenia, 2.7% (n=8) substance use disorder, 1.7% (n=5) bipolar disorder, and 0.7% (n=2) a personality disorder. The majority of patients had multiple chronic conditions, with 42.3% experiencing 1–3 chronic conditions and 37.6% experiencing 4–6 chronic conditions (Table 1; Supplementary Material 2). Nearly 15% of patients had use of an anxiolytic medication documented during the final 72 hours of hospitalization, and 45.3% had use of an antidepressant medication documented during this time period (specific medications listed in Supplementary Material 1).

Table 1.

Sample Characteristics (N = 298)

Average Age in Years at Discharge, mean (SD)	84.60 (6.71)
Age, n (%)
<65 years	2 (0.7)
65–69 years	4 (1.3)
70–74 years	11 (3.7)
75–79 years	47 (15.8)
80–84 years	73 (24.5)
≥85 years	161 (54.0)
Female, n (%)	219 (73.5)
Medicaid, n (%)	31 (10.4)
Race Non-Caucasian, n (%)	5 (1.7)
Nursing Home Claim Present During the 365 Days Prior to Index Hospitalization	91 (30.5)
Index Hospital Length of Stay in Days, mean (SD)	6.1 (3.8)
Length of Stay, n (%)
1–4 days	100 (33.6)
5 days	69 (23.2)
6–7 days	76 (25.5)
>8 days	53 (17.8)
Primary Diagnosis, n (%)
Stroke	56 (18.8)
Hip Fracture	242 (81.2)
Pre-Existing Psychiatric Diagnosis, n (%)
Bipolar Disorder	5 (1.7)
Depression	97 (32.6)
Personality Disorder	2 (0.7)
Schizophrenia	9 (3.0)
Substance Use Disorder	8 (2.7)
Chronic Conditions, n (%)
0	17 (5.7)
1 to 3	126 (42.3)
4 to 6	112 (37.6)
7 to 9	31 (10.4)
10+	12 (4.0)
Outcomes, n (%)
Re-hospitalization Within 30 Days	29 (9.7)
Death Within 30 Days	22 (7.4)
30 Day Re-hospitalization and/or Death (Combined Outcome)	45 (15.1)

Clinical Documentation of Neuropsychiatric Symptoms and Care Management Strategies

Overall, clinical documentation of any NPS and care management strategy in the last 72 hours of hospitalization was present for the majority of the sample, with documentation on NPS present for 79.9% of patients (n=238) and care management strategies for 91.9% of patients (n=274) (Table 2, Figures 1 and 2). For NPS, clinical documentation on agitation/aggression was most common, present for 51.0% of patients (n=152). Documentation on anxiety was also common, present for 23.2% patients (n=69). Across patients with hip/femur fracture and stroke, documentation on NPS was relatively consistent, though patients with hip/femur fracture had more documentation on anxiety and agitation/aggression while patients with a stroke had more documentation on irritability/lability (Table 2). The average frequency of documentation on NPS remained relatively consistent in the 72 hours leading up to discharge for the 238 patients who had any clinical documentation of NPS (Table 3). The most commonly documented management strategy was high fall risk designation, present for 89.3% of patients (n=266). Across patients with hip/femur fracture and stroke, care management strategies were documented in relatively proportionate number, though patients with hip/femur fracture had more documentation indicating use of 1:1 supervision (Table 2). Other less frequently documented NPS and management strategies are illustrated in Table 2, and Figures 1 and 2. Antipsychotic medication use during the last 72 hours of hospitalization was recorded for 23.6%, (n=70).

Table 2.

Patterns of Documentation of Neuropsychiatric Symptoms and Care Management Strategies during Hospitalization and Communication Upon Discharge by Condition

Neuropsychiatric Symptoms
	Presence of clinical documentation during the last 72 hours of hospitalization			Presence of communication upon discharge
	Overall (%) (n=298)	Hip fracture (%) (n=242)	Stroke (%) (n=56)	Overall (%) (n=298)	Hip fracture (%) (n=242)	Stroke (%) (n=56)
Anxiety	69 (23.2%)	66 (27.3%)	3 (5.4%)	4 (1.3%)	3 (1.2%)	1 (1.8%)
Agitation/ Aggression	152 (51.0%)	131(54.1%)	21 (37.5%)	35 (11.7%)	28 (11.6%)	7 (12.5%)
Hallucinations	13 (4.4%)	11 (4.6%)	2 (3.6%)	2 (0.7%)	1 (0.4%)	1 (1.8%)
Depression	5 (1.7%)	3 (1.2%)	2 (3.6%)	3 (1.0%)	1 (0.4%)	2 (3.6%)
Disinhibition	12 (4.0%)	9 (3.7%)	3 (5.4%)	0 (0%)	0 (0%)	0 (0%)
Irritability/ Lability	21 (7.1%)	11 (4.6%)	10 (17.9%)	1 (0.3%)	0 (0%)	1 (1.8%)
Aberrant Motor Behavior	27 (9.1%)	25 (10.3%)	2 (3.6%)	4 (1.3%)	3 (1.2%)	1 (1.8%)
Care Management Strategies
	Presence of clinical documentation during the last 72 hours of hospitalization			Presence of communication upon discharge
	Overall (%) (n=298)	Hip fracture (%) (n=242)	Stroke (%) (n=56)	Overall (%) (n=298)	Hip fracture (%) (n=242)	Stroke (%) (n=56)
1:1 Supervision	88 (29.5%)	83 (34.3%)	5 (8.9%)	9 (3.0%)	8 (3.3%)	1 (1.8%)
Fall Risk	266 (89.3%)	213 (88.0%)	53 (94.6%)	29 (9.7%)	22 (9.1%)	7 (12.5%)
Restraints	55 (18.5%)	42 (17.4%)	13 (23.2%)	5 (1.7%)	4 (1.7%)	1 (1.8%)
New of Modified Antipsychotic Medications
	Presence of clinical documentation during the last 72 hours of hospitalization			Presence of communication upon discharge
	Overall (%) (n=296)	Hip fracture (%) (n=241)	Stroke (%) (n=55)	Overall (%) (n=296)	Hip fracture (%) (n=241)	Stroke (%) (n=55)
Any Antipsychotic	70 (23.6%)	64 (26.6%)	6 (10.9%)	61 (20.6%)	55 (22.8%)	6 (10.9%)
Aripiprazole	2 (0.7%)	2 (0.8%)	0 (0.0%)	2 (0.7%)	2 (0.8%)	0 (0.0%)
Chlorpromazin	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)
Clozapine	1 (0.3%)	0 (0.0%)	1 (0.4%)	0 (0.0%)	0 (0.0%)	0 (0.0%)
Haloperidol	11 (3.7%)	10 (4.2%)	1 (1.8%)	5 (1.7%)	5 (2.1%)	0 (0.0%)
Olanzapine	16 (5.4%)	16 (6.6%)	0 (0.0%)	12 (4.1%)	11 (4.6%)	1 (1.8%)
Quetiapine	34 (11.5%)	28 (11.6%)	6 (10.9%)	27 (9.1%)	22 (9.1%)	5 (9.1%)
Risperidone	13 (4.4%)	0 (0.0%)	13 (5.4%)	12 (4.1%)	0 (0.0%)	12 (5.0%)
Ziprasidone	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)

Figure 1.

Rates of omission for neuropsychiatric symptoms between clinical documentation and discharge summary (N=298).

Figure 2.

Rates of omission for care management strategies between clinical documentation and discharge summary.

^a Antipsychotic medication use N=296

^b 1:1 Supervision, Fall Risk, and Restraints, N=298

Table 3.

Number of Discrete Instances of Documentation on Neuropsychiatric Symptoms by Day in the 72 Hours Prior to Hospital Discharge

	Day of discharge a	One day prior to discharge b	Two days prior to discharge c
Anxiety	1.3	1.2	1.3
Agitation/Aggression	1.5	1.8	2.1
Hallucinations	1.0	1.4	1.2
Depression	1.3	1.0	1.7
Disinhibition	1.3	1.1	1.0
Irritability/Lability	1.3	1.4	1.6
Aberrant Motor Behavior	1.3	1.1	1.5

^a‘Day of discharge’ denotes clinical documentation that was record in the chart in 0–24 hours prior to discharge

^b‘One day prior to discharge’ denotes clinical documentation that was recorded in the chart 24–48 hours prior to discharge, and

^c‘Two days prior to discharge’ denotes clinical documentation that was recorded in the chart 48–72 hours prior to discharge

Discharge Communication on Neuropsychiatric Symptoms and Care Management Strategies

Inclusion of information in discharge summaries regarding NPS and management strategies was relatively low across domains, with 13.8% of patients (n=41) having any mention of NPS and 13.1% of patients (n=39) any mention of management strategies communication upon discharge. Specifically, discharge summary communication on agitation/aggression was present for 35 (11.7%) patients and anxiety for 4 (1.3%) patients, with less frequently communicated NPS. Discharge summary communication of fall risk was present for 9.7% of patients (n=29). There were no notable descriptive differences in discharge communication of NPS and care management strategies by hip/femur fracture or stroke diagnosis. For approximately 20% of patients (n=61), discharge communication of antipsychotic medication use was present.

Omission Rates for Neuropsychiatric Symptoms and Care Management Strategies

Six of seven NPS examined had an omission rate higher than 75%, with 100% omission for disinhibition, 95.2% for irritability/lability, 94.2% for anxiety, 85.2% for aberrant motor behaviors, and 84.6% for hallucinations (Figure 1). For agitation/aggression, the most commonly documented NPS during the 72 final hours of hospitalization, omission rates from the discharge summary were 77.0%. Omission of discharge communication of depression was comparatively low at 40.0%. Discharge communication on three of four management strategies during hospitalization was omitted at similar rates: 89.8% for 1:1 supervision, 89.1% for fall risk, and 90.9% for use of restraints (Figure 2). In contrast, omission rates for antipsychotic medication use were comparatively low at 12.9%.

DISCUSSION

Findings from this multi-site retrospective cohort study are consistent with prior qualitative reports documenting clinician and family member perceptions that information about NPS and associated management strategies are under-communicated at the point of hospital discharge.²⁴ Although the written discharge summary plays a critical role in helping providers at the next site of care establish an appropriate, safe, and individualized care plan, discharge documentation within this study omitted essential information for informing safe care for persons with dementia at remarkably high rates with many NPS and care management strategies being omitted more than 75% of the time.⁴³ Discharge communication regarding new or modified antipsychotic medications was omitted at a relatively lower rate, 12.9% of the time.

Altogether, these omission rates are significant as behavioral care needs are challenging to prepare for, and a lack of preparation can have serious safety implications for persons with dementia who may be at high risk for adverse events such as falling or elopement, and even re-hospitalization.²⁴ Similarly, any omission of communication of a new or modified antipsychotic medication is potentially clinically significant given the high-risk nature of antipsychotic medication for patients with dementia, compounded by the vulnerability patients with dementia experience during care transitions and high risk for medication discrepancies.^{9, 10, 24, 28} Comprehensive medication reconciliation and management post-discharge is particularly important among older adults who are at greater risk for mediation-related adverse events.⁴⁴ Additionally, communication around new or modified use of antipsychotic medications during hospitalization may provide subsequent caregivers with critical knowledge on the therapeutic benefits, or lack thereof, for the patient with dementia which can inform later care management strategies.^{30, 45}

Minimally, discharge summaries should accurately document and communicate a patient’s diagnostic findings, hospital management, and planned follow-up to the post-hospital care team.^{46, 47} These steps are critical to preventing re-hospitalization events for people with dementia. However, ultimately addressing improved continuity for persons with dementia may require more longitudinal plans of care and transition documents that travel with patients across settings and are less fragmented and siloed within specific institutions.⁴⁸

To our knowledge, this is the first study to investigate and quantitatively document NPS and related dementia care information during hospital to SNF transitions, a highly vulnerable point in the care continuum. The high rate of omission could reflect several practice patterns and issues related to structured requirements within written discharge summaries. First, most discharge summaries are authored by physicians – or in some situations by another prescribing provider.³¹ Prior work has considered the potential consequences of limited inclusion of clinical data generated by other disciplines (i.e. nursing, physical therapy) in the discharge summary.⁴⁹ Authoring providers may not consider these data elements as important for inclusion in the discharge summary.

Similarly, information specific to preexisting or secondary diagnoses, such as dementia, are more frequently omitted in discharge summaries, potentially because providers may believe this information to already be included in the patient’s health record.^{50, 51} It is also possible that, because information about NPS or management strategies are not mandated components of written discharge summaries, authoring providers may not consider or think to include them in written discharge communication. In fact, prior studies have found that hospital providers reported being completely unaware that written discharge summaries were used for and served as admitting orders for patients transitioning to SNF setting.^{52, 53} This reflects ongoing concerns that a lack of familiarity regarding the capabilities, needs, and practices in SNF settings among hospital providers may underlie ongoing gaps in communication between these settings – and different perceptions regarding what should be required to be communicated.^{24, 25, 52, 54}

In the absence of clear documentation related to pre-admission diagnoses during an acute care stay, it may also be difficult for hospital providers to determine if NPS are relatively new or otherwise transient such that the expectation is that those symptoms will resolve as the person gets healthier and is able to return to their baseline level of functioning, a distinction which may impact providers’ perceived need to communicate NPS upon discharge. However, evidence suggests that NPS are both common and persistent in moderate to advanced disease stages.^{6, 55} Hence, there may also be a lack of continuity in understanding prior functioning and symptomatology that may subsequently hinder hospital providers’ true understanding of symptom history and thus communication regarding NPS.

Prior research also highlights that the decision to discharge older adults to a SNF setting is often made without a standardized system or framework, and is often used as a “safety net” for older adults.⁵³ Additionally, SNFs have been reported to refuse admission to patients who have recently required intensive care management strategies and antipsychotics, limiting hospital providers’ ability to rely on SNFs as a necessary “safety net.”⁵³ As such, providers may differentially document and communicate information regarding care management strategies in order to avoid discharge delays and increase likelihood of placement in a SNF.

There is growing interest in centering clinical improvement efforts on the priorities and needs of persons with dementia and their family caregivers.^{56, 57} The report from the first National Research Summit on Care, Services, and Supports for Persons with Dementia and Their Caregivers calls specifically for more intentional engagement of persons living with dementia and their caregivers in research, and for additional research focused on facilitating transitions between settings of care.⁵⁸ Findings from this study suggest that discharge communication practices during hospital to SNF transitions are not reflective of or built around the priorities and needs of persons with dementia, their family members, or clinicians orchestrating entry into the next setting of care and should be a priority area for future research. Additionally, there remains a largely overlooked opportunity to activate and engage family caregivers, when available, as informed advocates and partners in care surrounding hospital-to-SNF transitions. Prior transitional care interventions focused on transitions from hospital to home environments for cognitively vulnerable older adults that emphasize activation of informal caregiving networks have demonstrated significant reductions in medication discrepancies and re-hospitalization events.^{59, 60} However, these approaches have not as of yet been applied to transitions from hospitals to SNFs but may serve to address several important priorities and outcomes – including more intentional inclusion of care partners.

In moving towards more comprehensive discharge communication, a likely concern is increased documentation burden and potential inaccuracy in information communicated under the current model of authoring discharge summaries, which often relies on one provider to author the discharge communication.³¹ In response, approaches have been developed to streamline integration of updated clinical data from other clinical disciplines into the discharge summary using algorithms built into the electronic health record.⁴⁹ These approaches could be integrated into communication during hospital to SNF transitions for patients with dementia. However, a major consideration regarding practical implementation of these approaches is the fact that behavioral symptoms in particular are under-recognized and under-documented throughout hospitalization, meaning that any strategy designed to improve inter-setting communication may also need to necessitate improvements in recognition of condition-specific neuropsychiatric symptoms and management strategies in the acute care setting.⁶¹

This study has limitations, particularly given it examined discharge communication in a limited sample of patients from two hospital settings from the years 2003–2009. There has been a great deal of work towards establishing evidence-based transitional care models since 2009, including progress in dissemination of several transitional care models including the Transitional Care Model (TCM) and Interventions to Reduce Acute Care Transfers (INTERACT^®) program. Though TCM and INTERACT^® target different points in the care continuum, both programs have been shown to improve transitional care outcomes for various older adult populations.^62–65 The TCM is a nurse-led post discharge intervention that has been tested with chronically ill older adults and utilizes in home follow-up visits. INTERACT^® on the other hand focuses on reducing hospitalization among nursing home residents including those receiving Skilled Nursing care by focusing on management of acute changes in condition. The INTERACT^® program has evolved considerably in recent years with testing of INTERACT II^® and includes four basic types of tools to improve the identification, evaluation, and communication about changes in nursing home resident status. While communication quality surrounding transitions or care coordination specific to NPS or behavioral care needs is not an explicit focus of either of these models, their implementation may naturally shift communication patterns toward more patient-centered domains. However, it is also important to note that several evidence-based transitional care models have excluded older adults with cognitive impairment in their initial testing, and thus their impact on care specific to persons with dementia remains unclear.⁶⁶ Additional research is needed to clarify whether broader adoption of these interventions has engendered new practices and norms around inter-setting communication for populations with dementia, and specific to coordination and management of NPS.

Adoption and meaningful use of electronic health records have also evolved following implementation of the 2009 Affordable Care Act (ACA) and the American Recovery and Reinvestment Act (ARRA) of 2009.⁶⁷ The ACA included several provisions intended to incentivize improved coordination of care, in part through the Hospital Readmissions Reduction Program (HRRP) which penalizes hospitals with higher than average 30 day recidivism rates.⁶⁸ The ARRA introduced the Health Information Technology Economic and Clinical Health (HITECH) Act, which incentivized broader adoption of EHRs.⁶⁹ While no studies have evaluated the impact of these policies on dementia-specific discharge communication components, recent reports indicate that interoperability and Health Information Exchange via EHRs between acute and post-acute care settings remains limited.^{68, 70, 71} Even when Health Information Exchange is available during hospital to post-acute care transfers it is underutilized.⁷² Absent further evaluation of discharge communication completeness and accuracy, it remains unknown whether these policies have had a spillover effect on discharge communication practices broadly, and if they have positively affected communication specific to NPS among patients with dementia. Notably, the Joint Commission has not updated guidelines regarding content or timeliness of written discharge summaries, and several studies indicate that inclusion of even mandated discharge summary components remains low post-ACA and ARRA absent targeted intervention.^{51, 73}

This study evaluated omissions of information in discharge communication as compared to clinical documentation in the EHR, and as such, likely under-documents the true prevalence of dementia-specific symptoms. Despite the use of trained medical record abstractors who reviewed clinical documentation for indications of NPS in a manner consistent with the NPI, absent prospective data collection there is a risk for bias toward under capturing NPS events, as NPS that are documented in the EHR may constitute be more severe or disruptive symptoms. Additionally, although we were able to ascertain that 30% of patients received nursing home care in the year leading up to their index hospitalization and were thus likely newly eligible for SNF benefits in continuation of pre-existing long-term care, we were unable to confirm long vs short-term designation via longitudinal SNF claims data. However, it is likely that individuals with dementia are generally over-represented among those using long-term care, and consequentially findings may reflect a dementia population with more advanced disease.¹⁶ There may be other patient factors, like comorbid psychiatric conditions or delirium, that impact occurrence and communication of NPS and care management strategies that are not controlled for in calculating rates of omission. Although there is limited evidence regarding NPS assessment and documentation practices of NPS in general acute care practice, recent research indicates that that formal assessment and documentation of dementia symptoms is limited.^{61, 74} With prospective data collection to confirm and validate presence and role of other conditions and specific features of neuropsychiatric symptomatology, future research could explore whether occurrence and documentation of NPS differ by associated conditions and diagnoses such as delirium, along with variable NPS profiles.

This study only examined four care management strategies – antipsychotic medication use, fall risk, restraint use, and 1:1 supervision. Future research should consider a wider range of care management strategies such as relevant consultative teams that are now in broader use including psychiatry or acute care for elder (ACE) consultation, or implementation of delirium protocols, along with the role of other medications that may play a role in NPS management. For example, antidepressant or anxiolytic use may be important to explore. Additionally, future research might examine provider decisions around communication in the discharge summary and factors that might play a role in the type and amount of information communication. For example, a provider’s long-standing relationship with a SNF or specific patient may play a role in their level of communication, yet accurate communication of the hospital stay is still vital to care continuity as symptoms, needs and interventions may change future care plans.

CONCLUSIONS

In this study omissions in reporting NPS care management strategies experienced by patients with dementia in the last 72 hours of hospitalization were frequent in the discharge summary, a critical tool for managing care in subsequent care settings. Provided the extensive reports of substantial disparities, bounce-backs, and stressful transitions that persons with dementia experience after hospitalization, these study findings lend support to the need for additional evaluation of the contribution of poor discharge communication to negative post-acute outcomes and events. Future research should investigate perceptions regarding important clinical data to be communicated for transitions into SNF settings, as well as barriers to communication regarding NPS.