Health,Social,and Functional Characteristics of Older Adults With Continuing Care Needs: Implications for Integrated Care

Abstract

Objective: To identify older adults who could benefit from integrated care, we examined (a) health, social, and functional characteristics of older, hospitalized adults who required continuing care on discharge and (b) associations between these characteristics and potentially unnecessary health care use. Method: Personal characteristics were extracted from patient charts (N = 214) and examined in relation to three outcomes: discharge to institutional care, unnecessary hospital stay (alternative level of care), and long hospital stay. Results: Twenty-nine percent of the sample was discharged to an institution, 32.7% was coded as alternate level of care, and 27.6% had a long length of stay. Independent predictors of potentially avoidable health care use were mental and behavioral issues, living alone, functional status, and preadmission concerns about the patient managing in the community. Discussion: High users of health care services were identifiable prior to hospital admission, supporting the use of community-based integrated care approaches.

Keywords

health services social services chronic illness hospitalization community care

Background and Objectives

Addressing the health service needs of older adults is a societal priority as the population ages with increasingly high rates of chronic disease and multimorbidity (World Health Organization [WHO], 2015). It is estimated that up to 90% of community-dwelling older adults are living with chronic disease or disability (Denton & Spencer, 2010), resulting in high health and social care needs. Older adults are the heaviest users of inpatient hospital services (Canadian Institute for Health Information [CIHI], 2011), with consequences at both the patient and system levels. These issues include high occurrences of adverse events following hospital discharge (Forster, Murff, Peterson, Gandhi, & Bates, 2003; Forster, Murff, Peterson, & Gandhi Bates, 2005), preventable hospital readmissions (CIHI, 2012), postdischarge emergency department visits (CIHI, 2012; Rising, White, Fernandez, & Boutwell, 2013), hospital stays that extend beyond medical need due to deficits in community-based care (CIHI, 2011), and uncoordinated transitions of care between the hospital and other care environments (LaMantia, Scheunemann, Viera, Busby-Whitehead, & Hanson, 2010).

Care models for older adults are evolving to address these issues. Integrated care is a promising approach to meet the complex needs of older adults with chronic conditions (Hollander & Prince, 2007). The broadest conceptualizations of integrated care include integration of social and health services, as well as, integration of primary, secondary, and tertiary care (Hollander & Prince, 2007). The desired outcome of integrated care is more continuous and higher quality of care while maintaining cost-effectiveness by substituting hospital and residential care with less expensive home and community-based care (Chappell & Hollander, 2013). Therefore, a major goal of integrated care is to reduce potentially avoidable health care use, including transitions in and out of institutional care settings (Hollander & Prince, 2007).

Integrated care approaches are believed to be most effective when applied to populations most in need. To ensure cost-effectiveness, health care programs are working to target the small, high-risk, complex group of patients who have high service needs and health care cost consumption (Sansosi, Grootemaat, Habibur Seraji, Blanchard, & Snoek, 2015). In fact, there is evidence that the inability to target the right population for integrated care may be more of a barrier to improving clinical outcomes than the intensity of interventions (Threapleton et al., 2017).

Although the need for integrated health care has been identified, and conceptual models have been developed (Chappell & Hollander, 2013; WHO, 2015), it is not well known which older adults should be targeted for integrated care interventions. The research focus to date has been on developing easy to implement risk prediction tools for the general older adult population (Sansosi et al., 2015), rather than exploring the complexity inherent in older adult populations with known continuing care needs to inform integrated care approaches. Therefore, we do not have a full understanding of the characteristics of the complex older adult population with high hospital and community health care needs (Harrison et al., 2017; Kansagara et al., 2011).

Systematic reviews of hospital outcomes for older adults suggest that our understanding of the predictors of adverse hospital outcomes such as institutionalization postdischarge and readmissions are limited due to the narrow scope of variables that have been studied (Harrison et al., 2017; Kansagara et al., 2011). For example, discharge to institutional care seems to be driven by functional dependency, dementia, and being female, but social and clinical variables such as informal caregiving and continence are often not included (Harrison et al., 2017). The situation is similar for the outcome of length of stay. Certain medical conditions (deBuyser, Petrovic, Taes, Vetrano, & Onder, 2014), preadmission medications and falls (deBuyser et al., 2014), malnutrition (Lim et al., 2012), walking speed (deBuyser et al., 2014), pain (deBuyser et al., 2014), delirium/dementia (Fick, Steis, Waller, & Inouye, 2013; Zekry, 2012), and other mental health conditions (Bressi Nath & Marcus, 2012; Prina et al., 2013) are associated with long lengths of stay in studies, but there has been minimal study of social and clinical variables.

Given this gap in the literature, the overall aim of this study was to develop an understanding of older hospitalized adults with continuing care needs. We focused on these individuals, as they would likely benefit the most from integrated care interventions to reduce potentially avoidable health care use. The specific objectives were to examine (a) the health, social, and functional characteristics of older hospitalized adults who required continuing care upon discharge from hospital and (b) relationships between these characteristics and potentially avoidable health care use, including discharge to institutional care, unnecessary hospital use (alternate level of care, ALC), and long hospital stay.

Research Design and Methods

We conducted a retrospective chart review of older, hospitalized adults with continuing care needs. The hospital chart review allowed for collection of both preadmission and in-hospital variables in more depth than is available from hospital administrative data (Gearing, Mian, Barber, & Ickowicz, 2006), while avoiding some of the disadvantages of primary data collection, such as recruitment or attrition issues related to cognitive impairment, or severe illness.

Setting

This sample was taken from a teaching hospital in a Canadian province. Canada has a universal health care system in which physician and hospital services are provided free of charge. Other services vary by province. The province where the research was conducted has a free-of-charge provincial home care program that provides individuals with in-home services to allow them to stay in their home for as long as possible. Access to home care is via a standardized assessment conducted by a case coordinator who, if the person is deemed eligible, also develops a care plan that may include personal care, home support, in-home health care services, or respite. Home care coordinators also facilitate transitions to nursing homes. Nursing home admission is also based on a needs assessment. The cost of the nursing home is shared between the resident and government, with a daily resident rate calculated according to the individual’s income.

Sample

This study included older adults who were under the care of the general medicine service of the study hospital. Inclusion criteria were age 65 years or above at the time of hospital admission, living at home preadmission, and either discharged home with home care, or transferred to institutional care, such as inpatient rehabilitation or nursing home. Patients who were residing in an institution prior to admission or who died during the hospitalization were excluded. Starting in December 2016, charts that met the study inclusion and exclusion criteria were pulled backward in time, until the desired sample (minimum of 200) was reached. All patients had a discharge date between January 2014 and September 2016. In the case where an individual had multiple admissions that met the study criteria, we collected data from the most recent discharge. The charts for nine patients were not included because they were not received from the hospital information services within the data collection period. The final sample was 214 individuals.

Data Sources and Collection

Data were extracted from three sources. First, a data report from the hospital’s health information services department was used to obtain basic demographic and health data. The department extracts data from charts posthospitalization in compliance with the mandatory reporting requirements for the CIHI Discharge Abstract Database (DAD; CIHI, 2015, n.d.). Variables gleaned from this source were age, length of stay, ALC, preadmission and in-hospital diagnoses, postal code to derive income, and hospital interventions.

Second, a chart review was conducted of patient records. Extraction procedures were informed by guidelines developed by Gearing and colleagues (2006) and Allison and colleagues (2000). Social, health, and functional information was extracted into a standardized data abstraction form within a secure, web-based application (REDCap) designed to support confidential data capture for research studies (Harris et al., 2009). The lead author (CB) and a research assistant, who are both licensed health care professionals, completed the data extraction. CB provided 20 hours of training to the research assistant, including joint data extraction of two charts, and side-by-side extraction of two charts to facilitate coding consistency and discussion for resolution of coding discrepancies prior to independent collection. A data abstraction procedures manual was developed that included variable definitions, time frame guidelines, and instructions for management of negative (including absent or not applicable) information, as recommended by Allison and colleagues (2000). Following training, CB and the research assistant both extracted data from the same five charts to calculate interrater reliability (IRR), which was 96.9% agreement for the variables included in this study. Data extraction fidelity was promoted by regularly updating the data abstraction procedures manual and with random checks of data quality by CB (Allison et al., 2000; Gearing et al., 2006).

All variables not included in the data report from the hospital’s health information department were extracted directly from the patient records. Sources included the admission sheets, preadmission community home care plans that are included in the chart to inform hospital care and discharge planning, medication reconciliation records, standard forms (e.g., a fall risk tool), and interprofessional progress notes (including nursing, physical therapy, occupational therapy, social work, and physician notes).

Third, neighborhood-level income was derived from public access data, the 2016 census from Statistics Canada (http://www12.statcan.gc.ca/census-recensement/2016/dp-pd/prof/index.cfm?Lang=E), as income-related information was not available in the hospital records.

This study was approved by the institution’s Ethics Board and reviewed and approved for impact by the study hospital.

Measures

Demographic and social characteristics

Sex, age at admission, and language spoken at home were included as dichotomous variables. Age was categorized into below 80 years or above 80 years of age, and language spoken at home was categorized as English or other, given that relatively few patients spoke a language other than English at home. We recorded whether or not the patient lived alone and if an informal caregiver was identified in the chart. Neighborhood income was derived from the patient’s postal code using the 2016 Statistics Canada census, and then dichotomized using a median split into high or low income using a cutoff of $24,000/year (Cdn dollars).

Preadmission variables

Home care services

We included whether or not patients were enrolled in the provincial home care program.

Health status

For a measure of comorbidities, we derived the number of body systems affected preadmission from the hospital information services preadmission diagnostic codes. Diagnoses are categorized using an enhanced version of the International Statistical Classification of Diseases and Related Health Problems, 10^th Revision, Canada (ICD-10-CA) developed by CIHI for morbidity classification in Canada (CIHI, n.d.). With this coding system, there is some repetition in diagnostic labeling. To avoid double counting conditions, we counted the number of ICD-10-CA alphabetic blocks (each alphabetic block represents a main body system), rather than using a raw count of diagnoses. For example, coding of E11.52 (diabetes mellitus [DM] with complications), E11.28 (DM with kidney complications), I25.2 (old myocardial infarction), and I50.0 (congestive heart failure) was counted as two main body systems affected. We also coded as a dichotomous variable whether or not there was a behavioral or mental diagnosis present preadmission.

We counted the number of scheduled (non-prn) prescribed medications taken preadmission using admission medication reconciliation forms and dichotomized this variable into five or less, or six or more medications. Only non-prn medications were included to minimize data collection burden.

Cognition and behavior

Preadmission cognitive impairment was considered present if one or more of the following was met: (a) standardized cognitive screening score below normal, (b) a health professional thought the patient was unsafe to be left alone or to use the stove, or (c) the patient received home care reminders for medication, a service provided when a patient requires cognitive assistance with medication adherence. Preadmission challenging behavior was considered present if the patient was recorded as being verbally, or physically abusive, sexually suggestive, or if agitation interfered with caregiving.

Incontinence and function

Preadmission incontinence was recorded as present if partial or full urinary incontinence was documented (excluding device use such as catheter). For activities of daily living (ADL), we counted the number of ADL for which the patient was dependent (of bathing, toileting, and taking medications) to derive a score of 0 to 3. For preadmission mobility, the patient was recorded as requiring assistance if they were receiving supervision or physical assistance to mobilize indoors, regardless of mobility aid use (including wheelchair). We recorded presence of preadmission vision and hearing difficulties if the patient had a visual impairment that was not corrected with regular corrective lenses, or if the patient had hearing loss.

Concerns and issues regarding managing in the community

We derived a dichotomous variable from narrative notes in either the community home care report (if present) or admission notes to indicate if there were concerns or issues with meeting the patient’s needs in the community. Issues or concerns were considered broadly as any situation that a health care professional, family member, or the patient indicated was interfering with the patient’s needs being met in their current environment. We included individual-level issues, caregiving issues, social issues, issues related to needing or waiting for new services or setting, and family concerns.

In-hospital variables

Hospital interventions

This dichotomous variable indicated if the patient had at least one of the following interventions during the hospital stay: biopsy, cardioversion, chemotherapy, dialysis, endoscopy, feeding tube, heart resuscitation, paracentesis, or ventilation.

Health status

This was derived as for preadmission status but included only the diagnoses that were recorded as being treated during the admission. As per CIHI coding guidelines, comorbidity codes are assigned only when the condition “requires treatment beyond maintenance of the pre-existing condition,” “increases the length of stay by at least 24 hours,” or “significantly affects the treatment received” (CIHI, n.d., p. 28). The medication count was derived from discharge medication reconciliation records and then dichotomized (5 or less; 6+).

Cognition and behavior

Orientation at discharge from hospital was used as a measure of cognition. Orientated times three (O × 3) indicates the patient is alert and aware of person (themself), place, and time. Orientation was coded as impaired if the score was oriented times two (O × 2) or less. Challenging behaviors in hospital was coded the same as for preadmission behavior but included only behaviors documented during the hospital stay and included agitation that put the patient at high risk of personal injury (such as falls).

Continence and function

Continence was coded as present if the last nursing recording of urinary function indicated partial or full urinary incontinence. Needing assistance with toileting was coded using the last recording of toileting capacity in nursing or occupational therapy notes (including supervision and/or physical assistance). We determined need for assistance for mobility at discharge by referring to physical therapy discharge recommendations, or if not present, the last occupational therapy or nursing note that recorded mobility assistance. Fall risk was determined using the highest Schmid fall risk score recorded during the hospitalization. The Schmid fall risk–screening tool is completed by nurses on admission and weekly thereafter, and in previous research had 83% to 99% IRR (Schmid, 1990). The tool’s recommended cutoff is 3 (for clinical intervention purposes), but due to low numbers of patients with a Schmid score of 2 or below in this sample, we used a cutoff of 5 to develop a dichotomous fall risk variable.

Potentially avoidable health care use

We used three different dichotomous outcomes to represent potentially avoidable health care use. The first outcome was discharge to an institutional setting (rehabilitation or nursing home) or discharge home with home care upon hospital discharge. The second outcome was whether or not the patient was given an ALC designation, which is an indicator of inappropriate hospital use (CIHI, 2009). Hospital ward health care staff assigns ALC status when the patient no longer requires the intensity of acute hospital services but cannot be discharged. This includes patients awaiting placement to an alternate setting such as long-term care, waiting for services to be set up for discharge, or cannot leave hospital due to social circumstances (CIHI, 2015). The final outcome was length of stay; a simple count of days from the date of hospital admission that was then dichotomized into less than 30 days, or 30 days and more. Longer lengths of hospital stay increase the likelihood of adverse effects on older adults such as functional decline, infection, negative psychological consequences, and injury from falls (Admi, Shadmi, Baruch, & Zisberg, 2015). Correlations confirmed that the three outcome variables were correlated, but not redundant with each other (Spearman rho ranged from .51-.76).

Data Analysis

Data were downloaded from REDCap into IBM SPSS Statistics Version 24. Descriptive statistics were performed of all the demographic, social, preadmission and hospital variables, stratified by the three outcomes of interest. Bivariate associations were tested for statistical significance using chi-square tests for dichotomous variables and the Student’s t tests for continuous variables. Variables with a statistically significant association with at least one of the three outcomes at a p value of .05 or less were included in multivariate analyses. We used multivariate logistic regression to identify factors associated with each of the three outcomes for both preadmission and in-hospital variables, respectively. Demographic and social variables that were statistically significant in the bivariate analyses were also included in the multivariate analyses.

Post Hoc Analysis

Given that the preadmission concerns variable was strongly associated with all outcomes, we wanted to explore these effects further. Therefore, we analyzed the concerns noted in the patient records thematically, with five major themes emerging. The themes and examples were as follows:

Difficult to manage symptoms or behaviors: Patient-level issues such as unmanaged pain or breathlessness, impulsive behaviors, and/or safety issues such as frequent falls or forgetting medication.

Informal caregiver issues: Caregiver burn-out, capacity issues for carrying out caregiving tasks, or is no longer able to provide care.

Considering different services or setting to better meet patient’s needs: There were unmet needs for which services or a different setting was being considered (e.g., assisted living, nursing home, start or increase in home care services).

Waiting for new services or setting to be in place: A formal assessment for services/setting had been completed but was not in place prior to hospital admission.

Declines health professional recommendations for services or different care setting: The patient and/or family declined services such as lifeline, home care, and initiation of paperwork for nursing home.

Each patient was categorized in terms of the presence or absence of each of these themes and may have been included in more than one concerns variable. For example, a patient waiting to be admitted from community to personal care home may also have a caregiver reporting burn-out, and therefore, would be included in two of the five variables. We then conducted chi-square analyses for each of the five dichotomous preadmission variables with each of the three outcomes.

Results

The characteristics of the overall sample and stratified by the three outcomes are provided in Table 1. Approximately half of the study sample was above 80 years old, lived in a low-income neighborhood, and lived alone. As is typical with an older adult population, women were overrepresented, with 59.3% of the sample being female. Approximately three quarters of the sample spoke English at home and had at least one identified informal care provider. Most of the patients were already known to community health services, with 90.2% enrolled in the publicly funded home care program.

Table 1.

Sample Characteristics and Bivariate Analyses.

Characteristics	All (N = 214)	Discharged to an institution (nursing home/rehabilitation vs. home with home care)			Alternate level of care			Length of hospital stay (more than or less than 30 days)
Characteristics	% (n) orM ± SD	Yes (n = 62)	No (n = 152)	p value	Yes (n = 70)	No (n = 144)	p value	Long (n = 59)	Short (n = 155)	p value
Demographic and social
Above 80 years of age	47.2 (101)	61.3 (38)	41.4 (63)	.008	60.0 (42)	41.0 (59)	.009	52.5 (31)	45.2 (70)	.33
Female	59.3 (127)	59.7 (37)	59.2 (90)	.95	57.1 (40)	60.4 (87)	.65	57.6 (34)	60.0 (93)	.75
Low income ($24,000 and below)	49.5 (106)	50.0 (31)	49.3 (75)	.93	51.4 (36)	48.6 (70)	.70	52.5 (31)	48.4 (75)	.59
Speaks English at home	76.6 (164)	82.3 (51)	74.3 (113)	.21	82.9 (58)	73.6 (106)	.13	79.7 (47)	75.5 (117)	.52
Lives alone	49.5 (106)	67.7 (42)	42.1 (64)	.001	34.3 (42)	44.4 (64)	.03	55.9 (33)	47.1 (73)	.25
Has informal care provider	74.3 (159)	69.4 (43)	76.3 (116)	.29	71.4 (50)	75.7 (109)	.50	78.0 (46)	71.1 (113)	.45
Preadmission
Enrolled in home care	90.2 (193)	85.5 (53)	92.1 (140)	.14	88.6 (62)	91.0 (131)	.58	93.2 (55)	89.0 (138)	.36
Number of body systems affected	1.07 ± 1.02	1.04 ± 1.09	1.09 ± 1.0	.81	1.21 ± 1.14	1.00 ±.95	.15	1.4 ± 1.2	.94 ± .91	.002
Preadmission mental/behavioral diagnosis	11.2 (24)	22.6 (14)	65.8 (10)	.001	22.9 (16)	5.6 (8)	<.001	27.1 (16)	5.2 (8)	<.001
Six or more medications (n = 210)	65.7 (138)	62.3 (38)	67.1 (100)	.50	63.8 (44)	66.7 (94)	.68	65.5 (38)	65.8 (100)	.97
Cognitive impairment	35.0 (75)	51.6 (32)	28.3 (43)	.001	47.1 (33)	29.3 (42)	.002	47.5 (28)	30.3 (47)	.02
Challenging behavior	9.8 (21)	19.4 (12)	5.9 (9)	.003	21.4 (15)	4.2 (6)	<.001	20.3 (12)	5.8 (9)	<.001
Incontinent	35.5 (76)	45.1 (28)	31.6 (48)	.06	47.1 (33)	29.9 (43)	.01	39.0 (23)	34.2 (53)	.51
ADL dependency (count of 3 ADL)	1.61 ± 1.05	1.5 ± 1.07	1.65 ± 1.04	.34	1.49 ± 1.06	1.67 ± 1.04	.65	1.34 ± 0.98	1.71 ± 1.06	.02
Needs assistant to mobilize indoors	15.0 (32)	17.7 (11)	13.8 (21)	.47	18.6 (13)	13.2 (19)	.30	18.6 (11)	13.5 (21)	.35
Vision or hearing difficulties	30.8 (66)	50.0 (31)	23.0 (35)	<.001	41.4 (29)	25.7 (37)	.02	44.1 (26)	25.8 (40)	.01
Concerns or issues in community	49.5 (106)	72.6 (45)	40.1 (61)	<.001	68.6 (48)	40.3 (58)	<.001	74.6 (44)	40.0 (62)	<.001
In-hospital
Flagged intervention	13.1 (28)	11.3 (7)	13.8 (2)	.62	10.0 (7)	14.6 (21)	.35	4.2 (9)	12.3 (19)	.56
Number of body systems affected	2.8 ± 1.4	2.9 ± 1.6	2.8 ± 1.3	.66	3.06 ± 1.54	2.73 ± 1.31	.32	4.1 ± 1.8	3.1 ± 1.4	<.001
New mental/behavioral diagnosis	21.0 (45)	40.3 (25)	13.2 (20)	<.001	55.7 (39)	16.0 (23)	<.001	42.4 (25)	12.9 (20)	<.001
Six or more medications at discharge	75.7 (162)	74.2 (46)	76.3 (116)	.74	72.9 (51)	77.1 (111)	.50	78.0 (46)	74.8 (116)	.63
Disorientation at discharge	30.8 (66)	58.1 (36)	19.7 (30)	<.001	55.7 (39)	18.8 (27)	<.001	62.7 (37)	18.7 (29)	<.001
Challenging behaviors	32.2 (69)	46.8 (29)	26.3 (40)	.004	44.3 (31)	26.4 (38)	.01	59.3 (35)	17.4 (27)	<.001
Incontinent at discharge	30.4 (65)	43.5 (27)	25.0 (38)	.007	52.9 (37)	112.0 (77.8)	<.001	55. 9 (33)	20.6 (32)	<.001
Needs help with toileting at discharge	35.0 (75)	64.5 (40)	23.0 (35)	<.001	62.9 (44)	31.0 (21.5)	<.001	64.4 (38)	23.9 (37)	<.001
Needs help to mobilize at discharge	37.9 (81)	69.4 (43)	25.0 (38)	<.001	64.3 (45)	25.0 (36)	<.001	62.7 (37)	28.4 (44)	<.001
Schmid fall risk score 5+	44.4 (95)	64.5 (40)	36.2 (55)	<.001	87.1 (61)	69.4 (100)	.005	78.0 (46)	31.6 (49)	<.001

Note. Associations calculated with chi-square test (χ²) for categorical variables and Student’s t test for continuous variables. Variables used in multivariate analysis are bolded (significant at p ≤ .05). ADL = activities of daily living.

In terms of the outcome variables, 29.0% of patients were discharged to an institution, 32.7% were coded as ALC, and 27.6% had a length of stay longer than 30 days. Bivariate analyses showed many commonalities in the variables associated with the three outcomes. For example, among preadmission variables, the presence of a mental or behavioral diagnosis, cognitive impairment, challenging behaviors, vision or hearing difficulties, and concerns about managing in the community were associated with poorer outcomes for all measures.

Multivariate analysis of the preadmission variables and the three outcomes are presented in Table 2. The odds of being discharged to an institution versus to home with home care was increased for those living alone, with the presence of a preadmission mental or behavioral diagnosis, having vision or hearing difficulties, and concerns about managing in the community. The variables associated with ALC were similar to those for the discharge destination outcome except that challenging behavior was also associated with ALC, whereas vision and hearing difficulties were not. Length of stay was only associated with having a preadmission behavioral or cognitive diagnosis, and concerns about the patient managing in the community.

Table 2.

Multivariate Logistic Regression of Preadmission Predictor Variables.

Characteristics	Discharged to institution		Alternate level of care		More than 30-day length of stay
Characteristics	OR	95% CI	OR	95% CI	OR	95% CI
Above 80 years of age	1.61	[0.77, 3.36]	2.03*	[1.01, 4.06]	—	—
Lives alone	4.35**	[1.99, 9.52]	2.60**	[1.28, 5.22]	—	—
Number of body systems affected	—	—	—	—	1.35	[0.93, 1.95]
Preadmission mental/behavioral diagnosis	4.04**	[1.41, 11.53]	5.05**	[1.75, 14.57]	3.78*	[1.25, 11.42]
Cognitive impairment	1.74	[0.80, 3.77]	1.08	[0.52, 2.25]	1.04	[0.48, 2.28]
Challenging behavior	2.62	[0.83, 8.23]	5.06**	[1.60, 15.97]	2.18	[0.74, 6.38]
Incontinent	1.47	[0.70, 3.09]	1.83	[0.91, 3.67]	—	—
ADL dependency	—	—	—	—	.86	[0.61, 1.23]
Vision or hearing difficulties	2.16*	[1.05, 4.47]	1.23	[0.61, 2.50]	1.94	[0.96, 3.94]
Concerns about managing in community	3.14**	[1.49, 6.62]	2.46**	[1.24, 4.89]	3.52**	[1.67, 7.45]

Note. — = was not included in multivariate modeling as was not significant in bivariate analysis. Significant results are bolded. OR = odds ratio; CI = confidence interval; ALC = alternate level of care; ADL = activities of daily living.

significant at p ≤ .05. **significant at p ≤ .01.

In the post hoc analysis of preadmission concerns, difficult to manage symptoms or behaviors, waiting for services, and declining recommendations was significantly associated with discharge to an institution. Difficult to manage symptoms or behaviors and waiting for services were also associated with ALC, and difficult to manage symptoms or behaviors, informal caregiver issues, and declining recommendations were associated with longer length of stay (see Table 3).

Table 3.

Post Hoc Analysis: Associations Between Community Concerns and Issues and Potentially Avoidable Health Care Use.

Category of concerns or issues	Total (N = 214)	Discharged to an institution			Alternate level of care			Length of hospital stay
Category of concerns or issues	% (n)	Yes (n = 62)	No (n = 152)	p value	Yes (n = 70)	No (n = 144)	p value	Long (n = 59)	Short (n = 155)	p value
Difficult to manage symptoms or behaviors	20.1 (43)	37.1 (23)	13.2 (20)	<.001	32.9 (23)	13.9 (20)	.001	37.3 (22)	13.5 (21)	<.001
Informal caregiver issues	15.9 (34)	22.6 (14)	13.2 (20)	.09	21.4 (15)	13.2 (19)	.12	25.4 (15)	12.3 (19)	.02
Considering different services or setting to better meet patient’s needs	12.1 (26)	11.3 (7)	12.5 (19)	.81	14.3 (10)	11.1 (16)	.51	16.9 (10)	10.3 (16)	.19
Waiting for new services or setting to be in place	11.2 (24)	21.0 (13)	7.2 (11)	.004	18.6 (13)	7.6 (11)	.02	15.3 (9)	9.7 (15)	.25
Declines health professional recommendations for services/different care setting	9.3 (20)	16.1 (10)	6.6 (10)	.03	14.3 (10)	6.9 (10)	.08	18.6 (11)	5.8 (9)	.004

Note. Associations calculated with chi-square statistics (χ²). Significant results are bolded.

Results for the multivariate analysis of in-hospital variables and the three outcomes are presented in Table 4. The odds of being discharged to an institution increased for those who lived alone were diagnosed with a mental/behavioral condition in hospital, were not fully oriented at discharge, and needed assistance with toileting and mobilizing. The same variables were significant in the ALC model, except that orientation at discharge was not significant. A different pattern emerged for the outcome of length of stay; the odds of having a long length of stay was higher for those with more body systems affected, with difficult behaviors, and having a higher fall risk. Similar to the other two outcomes, the need for assistance for toileting was an independent predictor of long length of stay.

Table 4.

Multivariate Logistic Regression of In-Hospital Predictor Variables.

Characteristics	Discharged to institution		Alternate level of care		More than 30-day length of stay
Characteristics	OR	95% CI	OR	95% CI	OR	95% CI
Above 80 years of age	1.06	[0.46, 2.45]	1.22	[0.56, 2.64]	—	—
Lives alone	7.98**	[3.14, 20.28]	3.49**	[1.59, 7.68]	—	—
Number of body systems affected	—	—	—	—	1.46**	[1.42, 1.88]
Mental/behavioral diagnosis	2.38	[1.14, 7.36]	2.96*	[1.22, 7.20]	1.76	[0.72, 4.28]
Disorientation at discharge	3.59*	[1.37, 9.37]	2.33	[0.98, 5.53]	2.11	[0.86, 5.18]
Challenging behaviors	1.00	[0.41, 2.48]	0.77	[0.32, 1.81]	1.80**	[1.34, 7.43]
Incontinent at discharge	0.74	[0.29, 1.88]	1.07	[0.46, 2.52]	0.340	[0.60, 3.66]
Needs help with toileting	4.45**	[1.72, 11.55]	3.72**	[1.55, 8.92]	2.86*	[1.10, 7.41]
Needs help to mobilize	4.77**	[1.89, 12.03]	2.57*	[1.11, 5.95]	1.87	[0.76, 4.58]
Fall risk	1.34	[0.53, 3.36]	2.19	[0.95, 5.03]	2.89*	[1.20, 6.97]

Note. — = was not included in multivariate modeling as was not significant in bivariate analysis. Significant results are bolded. OR = odds ratio; CI = confidence interval.

significant at p ≤ .05. **significant at p ≤ .01.

Discussion and Implications

The aim of this study was to identify older adults who could benefit from integrated care. We accomplished that goal by focusing specifically on hospitalized older adults who needed continuing care on discharge. In this way, we were able to examine a group of older adults with high health service needs to determine who may benefit from more specialized integrated services. Chart review methods allowed for examination of a breadth of variables in a vulnerable group from whom primary data collection would be challenging. Even though 90% of the older adults in this study were already enrolled in the provincial home care program prior to hospital admission, approximately 30% of the sample had the potentially avoidable outcomes of ALC (32.7%) or being discharged to an institution (29.0%), even after adjusting for medical need. Approximately 30% of the sample, therefore, may be able to avoid or reduce hospital and institutional health care use with robust community management. Even in cases where hospital use is unavoidable due to chronic conditions that require periodic acute care intervention, the high health care users identified in this study would benefit from integrated care, as one of the primary goals of this approach is to provide seamless care between different levels of health care. For these individuals, integrated care provides the benefit of ensuring that care is consistent between primary and acute care providers for the individual receiving care, and from a systems perspective, ensures that the acute care use is efficient.

A main finding of this study was the strong influence of behavioral and/or cognitive diagnoses and symptoms on the potentially avoidable health care use outcomes. The influence of delirium, dementia, and severe mental illness on length of stay and institutionalization (Carter et al., 2016; Fick et al., 2013; Harrison et al., 2017; Jackson, Wilson, Richardson, & Lord, 2016; Saravay & Iavin, 1994) has been known for some time. This study demonstrates that this association persists within a group of older adults who all have continuing care needs. Furthermore, this study reemphasizes the importance of considering mental as well as physical needs of older adults in health service design and delivery. Health care systems continue to silo acute services into psychiatric and physical services, and in the community, primary care services focus primarily on physical health (Domino et al., 2016). Mental illness decreases the capacity of older adults to self-manage physical conditions, such as learning or remembering how to use respiratory medications, or monitor heart failure symptoms. In the case of delirium, while thought of as a temporary condition, its effects can last up to 12 months (Wass, Webster, & Nair, 2008). Therefore, community services for older adults need to have capacity for managing physical and mental health in tandem.

A second main finding was that preadmission issues and concerns of health care professionals were very common in this sample and predicted potentially avoidable health care use. The post hoc analyses indicated that at-risk individuals are identifiable by health care professionals. For example, concerns expressed preadmission with difficult to manage symptoms and behaviors were associated with institutionalization, ALC, and long lengths of stay. Similarly, informal caregiver issues, such as caregiver burden were associated with long lengths of stay. These findings are important from a clinical perspective, as it suggests that the opinion of a community health professional may be just as useful as clinical indicators or specialized screening tools in determining risk.

The nature of the concerns and issues expressed by community care professionals and family elucidates the features of integrated care that would be beneficial to the older adults in this study. At a macro level, integrated care approaches emphasize the need for a psychosocial as well as medical approach to care (Brown & Menec, 2018; Chappell & Hollander, 2013; Threapleton et al., 2017). The patients in this study had psychosocial needs including mental and behavioral needs, social support needs, and need for support for informal caregivers, all of which are prioritized in an integrated care approach. Integrated care approaches also tend to work best when a single funding envelope is used to provide and move care across and between health and social services. This allows all settings to have access to flexible funding, rather than one setting being prioritized or competing against another for financial resources (Chappell & Hollander, 2013), and could minimize waiting times for service increases.

At a clinical level, integrated care approaches use case managers who provide care across settings for continuity (Brown & Menec, 2018; Chappell & Hollander, 2013; Threapleton et al., 2017). An ongoing relationship between an intensive case manager and the patient could have allowed for earlier intensive intervention implementation for patients with difficult to manage symptoms or behaviors, the development of trust with older adults and families reluctant to accept services, ongoing support for informal caregivers, and a liaison between settings for needed hospital admissions. At a clinical level, integrated care uses interprofessional teams that deliver rehabilitative and restorative care in the community to best manage problematic symptoms and behaviors and reduce the need for hospital convalescence (Brown & Menec, 2018; Threapleton et al., 2017).

Independent predictors of potentially avoidable hospital outcomes that were similar to other studies of hospitalized older adult populations were living alone (Sansosi et al., 2015), dementia and other mental conditions (Bressi Nath & Marcus, 2012; Fick et al., 2013; Harrison et al., 2017; Prina et al., 2013; Vivanco & Roberts, 2011; Zekry, 2012), cognitive impairment (Sansosi et al., 2015), and functional dependency (Harrison et al., 2017; Sansosi et al., 2015; Vivanco & Roberts, 2011). Only for the outcome of long length of stay was multimorbidity (Sansosi et al., 2015; Vivanco & Roberts, 2011) and falls also independent predictors that have been previously identified (deBuyser et al., 2014).

Independent predictors of undesirable hospital outcomes in other studies that were not replicated here were medication use (deBuyser et al., 2014; Sansosi et al., 2015), gender (Harrison et al., 2017), low income (Sansosi et al., 2015), and ethnicity (as represented by language in this study; Sansosi et al., 2015). These factors may be less relevant for patients who have high health care needs and already require continuing care.

The present study has several limitations. The results need to be generalized to other settings with caution because the sample is from a single setting. Examining the demographic, social, health, and functional profile of this sample can help to determine to what extent the findings of this study are generalizable to other contexts. An unavoidable limitation of chart review methodology is the inability to verify the information documented in charts (Gearing et al., 2006). However, multiple strategies were used to ensure that the data extracted from charts were high quality: consultation with site-specific clinicians regarding patient documentation; development of abstraction protocols and guidelines; careful selection; training and monitoring of a data abstractor; protocols for managing ambiguous, conflicting, and/or missing data; conducting a pilot study of approximately 10% of the target sample; and measurement of IRR (Gearing et al., 2006). We were not, however, able to blind abstractors to the three studied outcomes, as this information is embedded in charts in multiple locations. Finally, some of the variables in post hoc analyses may be underpowered, resulting in Type 2 errors.

Conclusion

This study identified and characterized an older adult population with high health and social needs that may benefit from specialized integrated supports to reduce hospital use, nursing home admission, or at minimum, ease care transitions between acute and nonacute care settings. Potentially high users have mental, behavioral, and social as well as physical health issues and may be identifiable via reporting of issues and concerns by community health care professionals as a proxy for risk screening.

Footnotes

Acknowledgements

Thank you to the following organizations for scholarship support of Cara Brown: the Centre on Aging at the University of Manitoba, the Canadian Occupational Therapy Foundation, and Research Manitoba. Thank you to Emily Smith who was an instrumental research assistant for this project.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by the Mary Judd Research Grant awarded by the Manitoba Society of Occupational Therapists, in partnership with the Canadian Occupational Therapy Foundation.

References

Admi

Shadmi

Baruch

Zisberg

(2015). From research to reality: Minimizing the effects of hospitalization on older adults. Frontiers in Nursing, 6(2), e0017. doi:10.5041/RMMJ.10201

Allison

J. J.

Wall

T. C.

Spettell

C. M.

Clahoun

Fargason

C. A.

Kobylinski

R. W.

. . . Kiefe

(2000). The art and science of chart review. Journal on Quality Improvement, 26, 115-136.

Bressi Nath

Marcus

S. C.

(2012). Impact of psychotic disorders on discharge dispositions of adults 65 or older after a general medical inpatient stay. Psychiatric Services, 63, 333-337. doi:10.1176/appi.ps.201100332

Brown

Menec

(2018). Integrated care approaches used for transitions from hospital to community care for older adults: A scoping review. Canadian Journal on Aging, 37(2), 145-170.

Canadian Institute for Health Information. (2009). Alternate level of care in Canada. Retrieved from https://secure.cihi.ca/free_products/ALC_AIB_FINAL.pdf

Canadian Institute for Health Information. (2011). Health care in Canada, 2011: A focus on seniors and aging. Retrieved from https://secure.cihi.ca/free_products/HCIC_2011_seniors_report_en.pdf

Canadian Institute for Health Information. (2012). All-cause readmission to acute care. Ottawa, Ontario: Author. Retrieved from https://secure.cihi.ca/estore/productFamily.htm?locale=en&pf=PFC1823

Canadian Institute for Health Information. (2015). DAD abstracting manual, 2015-2016 edition. Available from www.cihi.ca

Canadian Institute for Health Information. (n.d.). Canadian Coding Standards for Version 2015 ICD-10-CA and CCI (Standards and Data Submission). Retrieved from https://secure.cihi.ca/free_products/Coding%20standard_EN_web.pdf

10.

Carter

Reynolds

Carter

Potluri

Uppal

Chandran

Potluri

(2016). The impact of psychiatric comorbidities on the length of hospital stay in patients with heart failure. International Journal of Cardiology, 207, 292-296. doi:10.1016/j.ijcard.2016.01.132

11.

Chappell

N. L.

Hollander

M. J.

(2013). Aging in Canada. Don Mills, Ontario, Canada: Oxford University Press.

12.

deBuyser

S. L.

Petrovic

Taes

Y. E.

Vetrano

D. L.

Onder

(2014). A multicomponent approach to identify predictors of hospital outcomes in older in-patients: A multicenter, observational study. PLoS ONE, 9(12), e115413. doi:10.1371/journal.pone.0115413

13.

Denton

Spencer

(2010). Chronic health conditions: Changing prevalence in an aging population and some implications for the delivery of health care services. Canadian Journal on Aging, 29, 11-21. doi:10.1017/S0714980809990390

14.

Domino

M.E.

Jackson

Beadles

C.A.

Lichstein

J. C.

Ellis

A.R.

Farley

J. F.

. . . Dubard

C.A.

(2016). Do primary care medical homes facilitate care transitions after psychiatric discharge for patients with multiple chronic conditions? General Hospital Psychiatry, 39, 59-65. doi: 10.1016/j.genhosppscyh.2015.11.002

15.

Fick

D. M.

Steis

M. R.

Waller

J. L.

Inouye

S. K.

(2013). Delirium superimposed on dementia is associated with prolonged length of stay and poor outcomes in hospitalized older adults. Journal of Hospital Medicine, 8, 500-505. doi:10.1002/jhm.2077

16.

Forster

A. J.

Murff

H. J.

Peterson

J. F.

Gandhi

T. K.

Bates

D. W.

(2003). The incidence and severity of adverse events affecting patients after discharge from the hospital. Annals of Internal Medicine, 138, 161-167.

17.

Forster

A. J.

Murff

H. J.

Peterson

J. F.

Gandhi Bates

D. W.

(2005). Adverse drug events occurring following hospital discharge. Journal of General Internal Medicine, 20, 317-323. doi:10.1111/j.1525-1497.2005.30390.x

18.

Gearing

R. E.

Mian

I. A.

Barber

Ickowicz

(2006). A methodology for conducting retrospective chart review research in child and adolescent psychiatry. Journal of the Canadian Academy of Child and Adolescent Psychiatry, 15, 126-134.

19.

Harris

P. A.

Taylor

Thielke

Payne

Gonzalez

Conde

J. G.

(2009). Research electronic data capture (REDCap)—A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics, 42, 377-381. doi:10.1016/j.jbi.2008.08.010

20.

Harrison

J. K.

Walesby

K. E.

Hamilton

Armstrong

Starr

J. M.

Reynish

E. L.

. . . Shenkin

S. D.

(2017). Predicting discharge to institutional long-term care following acute hospitalization: A systematic review and meta-analysis. Age and Ageing, 46, 547-558. doi:10.1093/ageing/afx047

21.

Hollander

M. J.

Prince

M. J.

(2007). Organizing healthcare delivery systems for persons with ongoing care needs and their families: A best practices framework. Healthcare Quarterly, 11(1), 42-52. doi:10.12927/hcq.2013.19497

22.

Jackson

Wilson

Richardson

Lord

(2016). Predicting outcome in older hospital patients with delirium: A systematic literature review. International Journal of Geriatric Psychiatry, 31, 392-399. doi:10.1002/gps.4344

23.

Kansagara

Englander

Salanitro

Kagen

Theobald

Freeman

(2011). Risk prediction models for hospital readmission: A systematic review. Journal of the American Medical Association, 306, 1688-1698. doi:10.1001/jama.2011.1515

24.

LaMantia

M. A.

Scheunemann

L. P.

Viera

A. J.

Busby-Whitehead

Hanson

L. C.

(2010). Interventions to improve transitional care between nursing home and hospitals: A systematic review. Journal of the American Geriatrics Society, 58, 777-782. doi:10.1111/j.1532-5415.2010.02776.x

25.

Lim

Ong

Chan

Loke

Ferguson

Daniels

(2012). Malnutrition and its impact on cost of hospitalization, length of stay, readmission and 3-year mortality. Clinical Nutrition, 31, 345-350. doi:10.1016/j.clnu.2011.11.001

26.

Prina

Huisman

Yeap

Hankey

Flicker

Brayne

Almeida

(2013). Association between depression and hospital outcomes among older men. Canadian Medical Association Journal, 185, 117-123. doi:10.1503/cmaj.121171

27.

Rising

K. L.

White

L. F.

Fernandez

W. G.

Boutwell

A. E.

(2013). Emergency department visits after hospital discharge: A missing part of the equation. Annals of Emergency Medicine, 62, 145-150. doi:10.1016/j.annemergmed.2013.01.024

28.

Sansosi

Grootemaat

Habibur Seraji

Blanchard

Snoek

(2015). Targeting integrated care to those most likely to need frequent health care: A review of social and clinical risk factors. Wollongong, Australia: Centre for Health Service Development, Australian Health Services Research Institute, University of Wollongong.

29.

Saravay

S. M.

Iavin

(1994). Psychiatric comorbidity and length of stay in the general hospital: A critical review of outcome studies. Psychosomatics, 35, 233-252. doi:10.1016/S0033-3182(94)71772-2

30.

Schmid

N. A.

(1990). Reducing patient falls: A research-based comprehensive fall prevention program. Military Medicine, 155, 202-207.

31.

Threapleton

D. E.

Chung

R. Y.

Wong

S. Y. S.

Wong

Chau

Woo

. . . Yeoh

(2017). Integrated care for older populations and its implementation facilitators and barriers: A rapid scoping review. International Journal for Quality in Health Care, 29, 327-334. doi:10.1093/intqhc/mzx041

32.

Vivanco

Roberts

(2011, December). Predicting patients likely to overstay in hospitals. In Chen

Dilion

Ishbuchi

Pei

Wang

Wani

M.A.

(Eds.), 10th International Conference on Machine Learning and Applications and Workshops (ICMLA) (Vol. 2, pp. 168-171). Los Alamitos, CA: IEEE.

33.

Wass

Webster

P. J.

Nair

B. R.

(2008). Delirium in the elderly: A review. Oman Medical Journal, 23, 150-157.

34.

World Health Organization. (2015). World report on ageing and health. Retrieved from http://www.who.int/kobe_centre/mediacentre/world_report_on_ageing_and_health_eng.pdf

35.

Zekry

D. S.

(2012). Predicting adverse outcomes in hospitalized elderly: Relative contributions of cognitive, functional and nutritional status, comorbidities and biomarkers (Doctoral thesis). University of Geneva, Switzerland.