Social Engagement After Nursing Home Admission: Racial and Ethnic Disparities and Risk Factors

Abstract

Older adults admitted to nursing homes (NHs) are at risk for low social engagement, which has associations with medical, psychological, and social well-being. Minorities may be at a disadvantage for social engagement because of their racial or ethnic group identity. This study assessed whether there were racial/ethnic disparities in social engagement among older adults (N = 15,927) at 1 year after their NH admission using multi-level predictors. No racial or ethnic-based disparities in social engagement were found; hence, an analysis of risk factors at NH admission that predicted low social engagement at 1 year for all residents was conducted. Significant risk factors for low social engagement were low social engagement at admission, deficits in activities in daily living and cognition, problems with vision and communication, and residing in an NH in an urban community. Results highlight the importance of initiating interventions to increase social engagement at the time of NH admission.

Keywords

Social engagement quality of life disparity nursing home

Promoting social engagement of older adults in nursing homes (NHs) is an important focus of care. Social engagement refers to having connections or relationships with others and being involved in groups or social activities (de Guzman et al., 2012; Park, 2009). Social engagement encompasses the ability, ease, and willingness to communicate and interact with others, perform social roles, and experience a sense of comfort and satisfaction from these interactions (de Guzman et al., 2012; Kiely & Flacker, 2003; Mor, Zinn, Angelelli, Teno, & Miller, 2004; Schroll, Jonsson, Mor, Berg, & Sherwood, 1997). Low social engagement can lead to feelings of isolation (Brownie & Horstmanshof, 2011; de Guzman et al., 2012), low life satisfaction (Jang, Mortimer, Haley, & Graves, 2004; Park, 2009), and loneliness (de Guzman et al., 2012; Lucas, Knowles, Gardner, Molden, & Jefferis, 2010). Low social engagement is correlated with a higher mortality rate among NH residents, an effect that persists even after adjusting for other mortality-related risk factors (Kiely & Flacker, 2003; Kiely, Simon, Jones, & Morris, 2000). However, a recreational intervention for NH residents with dementia, designed to enhance social interaction, significantly increased positive affect (Schreiner, Yamamoto, & Shiotani, 2005). Quality of life is positively correlated with social engagement (Degenholtz, Kane, Kane, Bershadsky, & Kling, 2006). In some research, social engagement is included as a dimension of quality of life (Degenholtz et al., 2006; Golden, Conroy, & Lawlor, 2009; Schreiner et al., 2005; Verbeek et al., 2009; Won et al., 2006). Other researchers have utilized the Social Engagement Scale as a direct measure of quality of life (DuBeau, Simon, & Morris, 2006; Schroll et al., 1997).

NH residents are at risk for low social engagement as it has wide-ranging associations with medical, psychological, and social well-being. Low social engagement has been associated with cognitive decline (Bassuk, Glass, & Berkman, 1999), depression (Kang, 2012; Lou, Chi, Kwan, & Leung, 2013; van Beek, Frijters, Wagner, Groenewegen, & Ribbe, 2011), visual and/or communication impairment (Kang, 2012; Resnick, Fries, & Verbrugge, 1997), worsening urinary incontinence (DuBeau et al., 2006; Schreiner et al., 2005), limitations in activities of daily living (Achterberg et al., 2003; Schroll et al., 1997), and cognitive impairment or dementia (Cordner, Blass, Rabins, & Black, 2010; Schroll et al., 1997). Alternatively, high social engagement has been associated with higher levels of cognition and physical functioning (Schroll et al., 1997).

In addition to characteristics of individual residents, factors at the NH and community levels have been examined for their association with social engagement of NH residents in a few studies. Consideration of group-level (i.e., NH) or community-level information has potential to lead to a more complete understanding of a psychosocial outcome such as social engagement (Diez-Roux, 1998; Krieger, Chen, Waterman, Rehkopf, & Subramanian, 2005). In exploring the effect of depression on social engagement of NH residents, Achterberg et al. (2003) included the type of NH ward (somatic, psychogeriatric, or rehabilitation) among the factors they modeled. Similarly, in examining the association between urinary incontinence and social engagement, DuBeau controlled for the ownership of the NH (for-profit, not-for-profit, government). Neither of these factors was statistically significant however. In a cross-sectional examination of disparities in social engagement in NHs, Li et al. reported that NH (e.g., number of residents, percentage receiving Medicaid, ownership, chain affiliation) and community-level (e.g., rural vs. non-rural) factors improved the explanation of some differences in social engagement among racial and ethnic groups of residents.

Because an NH has a social context, there is the possibility that minority groups may be at a disadvantage in experiencing social engagement because of their racial or ethnic group identity. In a large sample of NH residents, Li and Cai (2014) reported that prevalent social engagement scores were significantly lower for a cross-section of Blacks, Hispanics, and Other combined groups of minorities compared with Whites. Social engagement group differences remained significant after adjusting for demographic, NH, and county/state variables. There is limited information about disparities in social engagement in NH residents by racial and ethnic group and few studies have examined predictors of social engagement after 1 year in an NH. Knowledge of predictors of low social engagement has the potential to guide NH staff to improve the social well-being and health of older individuals during their NH stay.

The purpose of this study was to assess whether there were racial and ethnic disparities in social engagement among older NH residents at 1 year after their admission. Predictors at the individual resident, NH, and community level were examined. If no racial/ethnic disparities were found, an analysis of risk factors present at NH admission that predict low social engagement at 1 year for all residents was planned.

Method

Design, Data Sources, and Cohort Selection

This study had an observational cohort design. Existing data at the individual resident and NH/community levels collected at the time of a residents’ NH admission were analyzed in assessing disparities in social engagement and predicting low social engagement at 1 year after admission. Three data sets were linked and analyzed: (a) Minimum Data Set (MDS) records for all individuals admitted to a national chain of for-profit NHs during the years 2000 to 2002, (b) Online Survey Certification and Reporting (OSCAR) records of the NHs for the years 2000 to 2002, and (c) U.S. Census data (year 2000) for the tract in which each NH was located. MDS records provided data about the outcome of social engagement and demographic and clinical characteristics of NH admissions. OSCAR records provided information about the NH environment, staffing, and deficiencies in quality of care. The U.S. Census provided standardized data about the socioeconomic and sociodemographic status of the community in the Census tract of each NH. The Minnesota Population Center at the University of Minnesota, Minneapolis, MN, mapped NHs to the appropriate Census tract. The study was granted exempt status by the University of Minnesota Institutional Review Board as data were de-identified.

The cohort was defined as individuals aged ≥65 years with a full/admission MDS record and a social engagement score at admission and 1 year later. Race and ethnicity used MDS-defined categories: American Indian/Alaskan Native (AIAN), Asian/Pacific Islander (API), Black Not Hispanic (Black), White Not Hispanic (White), and Hispanic.

Outcome and Predictor Variables

The outcome variable was social engagement at 1 year after NH admission. Social engagement was operationally defined by the Social Engagement Index, a 7-point (0-6), additive scale using items on the MDS record (Mor et al., 1995). Validity and reliability of the Social Engagement Index has been established previously (Mor et al., 1995). Potential predictors of social engagement at the individual and NH/community levels were identified from a review of the literature and the expertise of three clinical consultants and the research team. Established composite scales with strong psychometric properties were used when possible as multiple items of the data bases could define a predictor variable. In some cases, a composite scale was constructed using previously reported methods if no scale existed (Savik, Fan, Bliss & Harms, 2005). Individual-level and composite variables were scaled so that a higher score meant a worse condition level with the exception of social engagement, spirituality, and positive relationships for which a higher score indicated a better level. All predictors were measured at NH admission and were screened for inclusion in the statistical models.

Potential predictors at the individual level included demographic and physical characteristics of the residents such as Activities of dailiy living (ADL) deficits (Morris, Fries, & Morris, 1999), age, body mass index (BMI), a comorbidity index (Charlson Index; Charlson, Pompei, Ales, & MacKenzie, 1987), gender, high school education or greater, incontinence of any type, mortality risk (Changes in Health, End-Stage. Disease, Signs, and Symptoms (CHESS) scale; Hirdes, Frijters, & Teare, 2003), number of medications per week, oxygenation problems, perfusion problems, poor nutrition, highest stage of pressure ulcers, restraint use, social engagement at admission (Mor et al., 1995), tube feeding, and vision problems. They also included cognitive and emotional resident characteristics such as cognitive deficits using the (Minimum Data Set Cognition Scale (MDS-COGS) scale; Hartmaier, Sloane, Guess, & Koch, 1994), communication difficulty, delirium (Minimum Data Set Confusion Assessment Method (MDS-CAM); Dosa, Intrator, McNicoll, Cang, & Teno, 2007), depression (Burrows, Morris, Simon, Hirdes, & Phillips, 2000), discomfort behaviors (Stevenson, Brown, Dahl, Ward, & Brown, 2006), mood (Health Services Advisory Group, n.d.), positive relationships, social engagement at admission, and spirituality.

For the NH-level predictors, composite predictor variables were created for NH personnel. These included all NH staff (e.g., activity staff, certified nursing assistants/medication aides [CNA], licensed nurses [LNs, including registered nurse and licensed practical nurse], dieticians, nursing administrators/director of nursing, physicians/physician extenders, therapists). Total LN and CNA staffing full-time equivalents (FTEs) per resident were calculated by dividing the total FTEs of the type of staff reported for a 2-week period (including full-time, part-time, and contract positions) by the total number of residents in an NH. Another NH-level variable was the percentage of residents receiving Medicaid. In addition, composite variables for deficiencies in NH care quality were constructed in four areas by summarizing the scope and severity levels of the respective deficiencies for an NH in OSCAR records. The four areas were resident behavior–facility, practices–dignity, quality of care, and resident assessment–nursing services. A fifth composite variable included the total number of these deficiencies by NH.

Potential predictors at the Census tract/community level included seven Census-level variables in their original form and 16 variables that were converted into proportions of the Census tract population, which described the socioeconomic and sociodemographic status of the community around each NH. The community-level predictors included the following: the percentage/proportion of the Census tract community that was American Indians, Asians, or Pacific Islanders, Black non-Hispanics, Hispanics, White non-Hispanics; working class; below poverty level, at <50%, 50% to 99%, or ≥100% of poverty; or residing in an urban area, urban cluster, or rural area; with 1 to 8 years or 9 to 16 years of education; with males or females aged <65 years old and 65+ years; as well as the median home value and poverty rank of the tract population, and the Census division of the NH.

Statistical Analyses

Data were summarized using descriptive statistics appropriate to their level, but differences in characteristics among groups at NH admission were not formally tested for significance as the large sample size renders small differences as statistically significant; hence no p values were generated. Variables were screened for inclusion in statistical models using bivariate associations with the outcome of social engagement and those with an association at p < .05 were considered candidates. Potential colinearity between variables was also assessed using bivariate associations. If a resident-level and an NH/community-level variable were highly correlated, the resident-level variable was included in the model as it was more specific.

Analyses of social engagement at 1 year in the various mixed race NHs used the Peters–Belson method (described below) and included the following predictors: admission scores for social engagement (Mor et al., 1995), deficits in ADLs (Morris et al., 1999), or cognition (Hartmaier et al., 1994), communication difficulties, positive relationships, spirituality, having any type of incontinence, number of vision problem indicators, percentage of NH residents that were on Medicaid, proportion of NH admissions who were White, proportion of the Census tract community of an NH that was below poverty level, and the Census division in which an NH was located.

The Peters–Belson method for linear regression analysis was used to assess disparities in social engagement 1 year after NH admission. The Peters–Belson method tests whether observed outcomes of a presumed disadvantaged group (in this study, the racial or ethnic minority groups) differ from their expected outcomes had they been part of a presumed advantaged group (i.e., White residents; Graubard, Sowmya Rao, & Gastwirth, 2005; Rao, Graubard, Breen, & Gastwirth, 2004). Social engagement was modeled first for White residents using individual-level factors then adding NH/community-level factors. The regression coefficients from the White’s model were then applied to each minority group’s data separately to determine their expected social engagement at 1 year. Because the cohort is clustered within NHs, unmeasured NH confounders were controlled for by insuring that residents of each minority group of interest were in the same NHs of Whites whose coefficients were applied to their racial and ethnic group in modeling. These NHs may have included other minority groups and are hence referred to as mixed race NHs. A one-sample two-sided z test compared the difference between observed and expected mean social engagement of each minority group. A significant difference (at p < .05) indicated a disparity based on race or ethnicity.

As no racial or ethnic disparity was found in the Peters–Belson analyses, a general linear mixed model was used to evaluate risk factors for social engagement at 1 year after NH admission for all residents. The screening of variables to include in the general linear mixed model was done independently of that for the analysis of racial and ethnic disparities and included the following factors: gender, admission scores for social engagement (Mor et al., 1995), deficits in ADL (Morris et al., 1999) or cognition (Hartmaier et al., 1994), communication difficulties, mood (Health Services Advisory Group, n.d.), positive relationships, spirituality, having any type of incontinence, number of vision problem indicators, percentage of NH residents who were on Medicaid, deficiency in quality of NH care index, and proportion of the Census tract community of an NH that was below poverty level or in an urban area. NH-specific random effects were included in the model to account for between-NH variability in social engagement due to unmeasured NH factors, thereby allowing each NH to have its own mean social engagement at 1 year. NH-specific random slopes across values of social engagement at admission were also included to allow each NH to have its own linear relation between social engagement at 1 year and social engagement at admission. The correlations between intercepts and slopes were set to unstructured.

Data management and descriptive statistics were conducted using SPSS v. 21 (Chicago, IL), SAS v. 9.3, (SAS Institute Inc., Cary, NC), or R v. 2.14. The Peters–Belson analysis used R v. 2.14. The general linear mixed model analysis was performed with SAS v. 9.3. Results were considered statistically significant if p < .05.

Results

Cohort Characteristics

The demographic and physical characteristics of the cohort at NH admission (N = 15,927) are presented in Table 1. The average age of each racial or ethnic group in this admission cohort was approximately 80 years. The percentage of females ranged from 63% to 73%, with the highest percentage among Whites. More than 40% of Whites and APIs had a high school education or greater, which was approximately twice that of the other race groups.

Table 1.

Demographic and Physical Characteristics of Older NH Admissions by Race and Ethnicity.

Variable^a/scale	American Indian/Alaskan Native	Asian/Pacific Islander	Black	Hispanic	White
No. of admissions	n = 91	n = 249	n = 1,853	n = 301	n = 13,433
M (SD)
Age at admission	80.3 (8.3)	83.4 (7.6)	81.2 (8.3)	81.2 (8.0)	82.7 (7.5)
BMI	25.2 (5.2)	22.8 (3.5)	25.1 (5.1)	25.5 (5.1)	25.1 (5.0)
Comorbidity index	2.0 (1.3)	2.3 (1.7)	2.2 (1.6)	2.4 (1.7)	1.8 (1.5)
Charlson index
Medications	5.1 (6.2)	2.8 (4.6)	5.1 (5.7)	5.0 (5.9)	7.0 (6.3)
Number per week
Mortality risk	1.3 (1.1)	1.9 (1.0)	1.4 (1.0)	1.3 (1.0)	1.6 (1.1)
CHESS Scale
n (%)
Activities of Daily Living Deficit Scale	13.2 (8.8)	19.2 (6.6)	16.7 (8.3)	16.4 (8.4)	14.3 (8.0)
Female gender	59 (64.8)	163 (65.5)	1,251 (67.5)	188 (62.5)	9,732 (72.4)
High school education or more	19 (20.9)	116 (46.6)	528 (28.4)	79 (26.2)	7,541 (56.1)
Incontinence	50 (54.9)	184 (73.9)	1,223 (66)	199 (66.1)	7,169 (53.4)
Any type
Oxygenation problems
Number of indicators
1	8 (8.8)	16 (6.4)	98 (5.3)	18 (6.0)	1,251 (9.3)
2	5 (5.5)	10 (4)	62 (3.3)	9 (3)	696 (5.2)
≥3	1 (1.1)	5 (2)	28 (1.5)	3 (1)	494 (3.7)
Perfusion problems
Number of indicators
1	24 (26.4)	69 (27.7)	429 (23.2)	47 (15.6)	3,435 (25.6)
≥2	0 (0)	5 (2)	29 (1.6)	8 (2.7)	324 (2.4)
Poor nutrition
Number of indicators
1	32 (35.2)	105 (42.2)	670 (36.1)	98 (32.6)	5,561 (41.4)
2	3 (3.3)	30 (12	79 (4.3)	20 (6.6)	814 (6.1)
Pressure ulcer highest stage = 2, 3, or 4	9 (9.9)	36 (14.5)	325 (17.5)	43 (14.3)	1,392 (10.4)
Restraint use	5 (5.5)	22 (8.8)	107 (5.8)	17 (5.6)	598 (4.5)
Any use
Tube feeding	4 (4.4)	49 (19.7)	203 (10.8)	31 (10.3)	540 (4)
Vision problems
Number of indicators
1	18 (19.8)	70 (28.1)	543 (29.3)	75 (24.9)	3,283 (24.4)
2	15 (16.5)	42 (16.9)	235 (12.7)	30 (10.0)	1,261 (9.4)
≥3	12 (13.2)	20 (8.0)	231 (12.5)	36 (12.0)	938 (7.0)

Note. NH = nursing home; BMI = body mass index; MDS = Minimum Data Set.

Operational definitions of MDS items for the variables and references for the scales used can be found in Supplemental Table 1.

The comorbidity index (Charlson et al., 1987) of residents was fairly low among all racial and ethnic groups, with average scores <2.5 of a possible score of 30 in all racial and ethnic groups. Whites took more medications on average than other groups. The average mortality risk score (Hirdes et al., 2003) ranged from 1.3 in AIANs to 2.0 in APIs on a 6-point scale. Deficits in ADLs (Morris et al., 1999) were moderate across racial and ethnic groups with slightly higher average scores (i.e., worse function) in APIs and the lowest in AIANs. More than half of each group of residents was incontinent with the highest percentage in APIs (74%). Approximately 10% (Hispanic and Blacks) to 17% (Whites) of each racial and ethnic group had an indicator of an oxygenation problem. Perfusion problems were present in approximately 20% to 30% of all groups and somewhat less common in Hispanics. Poor nutrition was present in more than one third of each racial or ethnic group. The average BMI was approximately 25 for all groups except APIs whose average BMI was approximately 23; more APIs received tube feeding compared with other groups. The percentage of residents with a pressure ulcer (stage of 2, 3, or 4) ranged from approximately 10% in AIANs and Whites to 17.5% in Blacks. Use of restraints was <10% in all racial/ethnic groups. Vision impairment was common among all groups of residents with the highest number of indicators present more often in AIANs (13%), Blacks (12%), and Hispanics (12%).

The cognitive and emotional characteristics of the cohort at NH admission are presented in Table 2. The level of cognitive deficits was moderate among all groups, with average MDS-COGS scores (Hartmaier et al., 1994) between 3 and 5 of a possible score of 10. APIs experienced the greatest communication difficulty and had the lowest level of positive relationship activities. Discomfort behaviors (Stevenson et al., 2006) were also low across groups; within that low range, AIANs had the highest level and APIs had the lowest level. Poor mood (Health Services Advisory Group, n.d.) was not common in any group, with scores near 1 on an 8-point scale. At NH admission, social engagement was fairly low among all groups (<2.5 on a 0- to 6-point scale) with Hispanics having the lowest score on average (Mor et al., 1995). Delirium (Dosa et al., 2007) was more severe among Whites and APIs. Depressive symptoms (Burrows et al., 2000) were most common among AIANs and Whites. More Blacks had strong spirituality.

Table 2.

Cognitive and Emotional Characteristics of Older NH Admissions by Race and Ethnicity.

Variable^a/scale	American Indian/Alaskan Native	Asian/Pacific Islander	Black	Hispanic	White
No. of admissions	n = 91	n = 249	n = 1,856	n = 302	n = 13,463
M (SD)
Cognitive deficits	4.3 (3.1)	4.5 (2.8)	4.6 (2.9)	4.7 (2.9)	3.8 (2.9)
MDS-COGS Scale
Communication difficulty score	2.1 (1.9)	2.5 (1.8)	1.6 (1.7)	2.0 (1.8)	1.4 (1.5)
Discomfort Behavior Scale	7.0 (11.0)	3.2 (6.3)	4.2 (9.0)	4.8 (8.2)	5.2 (9. 6)
MDS Mood Scale	1.0 (1.3)	1.0 (1.1)	0.7 (1.1)	0.9 (1.23)	1.1 (1.4)
Relationships	8.8 (3.2)	6.3 (2.9)	8.8 (3.1)	8.6 (3.1)	9.4 (3.2)
Number of positive activities
Social Engagement Index	2.2 (1.6)	2.1 (1.4)	2.0 (1.4)	1.7 (1.4)	2.2 (1.5)
n (%)
Delirium
MDS-CAM
Subsyndromal delirium level 1	23 (25.3)	45 (18.1)	264 (14.2)	60 (19.9)	2,042 (15.2)
Subsyndromal delirium level 2	7 (7.7)	29 (11.6)	127 (6.8)	18 (6.0)	1,466 (10.9)
Full delirium	0 (0.0)	1 (0.4)	7 (0.4)	1 (0.3)	129 (1)
Depression	31 (34.1)	37 (14.9)	392 (21.1)	71 (23.5)	4,258 (31.6)
Any indicator
Spirituality
Mild	29 (31.9)	65 (26.1)	542 (29.2)	97 (32.5)	3,922 (29.2)
Strong	19 (20.9)	33 (13.3)	566 (30.5)	65 (21.5)	3,318 (24.7)

Note. NH = nursing home; MDS = Minimum Data Set.

Operational definitions of MDS items for the variables and references for the scales used can be found in Supplemental Table 1.

Residents were in 443 NHs located in 27 states in all nine Census divisions. Characteristics of the NHs are presented in Table 3. Across all NHs, the average percent of residents receiving Medicaid was approximately 73%. Staffing by LNs was 1.08 hr/resident/day and by CNAs it was 2.15 hr/resident/day. The total number of deficiencies in quality of NH care was approximately 7.4 on average. Characteristics of the communities/census tracts surrounding the NHs are presented in Table 4. The communities were largely working class within urban areas. The communities had a variety of racial and ethnic groups of small percentages. For example, approximately 4% of the NHs in our cohort were located in Census tracts whose communities were 50% to <75% Black.

Table 3.

Characteristics of NHs.

Licensed nurses (FTE/resident)	Licensed nurses (hours/resident/day)	CNA(FTE/resident)	CNA (hours/resident/day)	Number of deficiencies^a	Quality of care index^b	% residents on Medicaid
Licensed nurses (FTE/resident)	Licensed nurses (hours/resident/day)	CNA(FTE/resident)	M (SD)	Number of deficiencies^a	Quality of care index^b	% residents on Medicaid
0.2 (0.1)	1.1 (0.5)	0.4 (0.4)	2.2 (2.1)	3.8 (2.4)	7.5 (6.3)	73.6 (15.5)

Note. NH = nursing home; FTE = full-time equivalent; CNA = certified nursing assistants/medication aides.

Selected deficiencies per “Method” section.

Index used scope/severity of quality of care deficiencies relevant to outcome.

Table 4.

Characteristics of Communities Around the NHs.

% of census tract/community	<25%	25% to <50%	50% to <75%	≥75%
n (%) of NHs
American Indians, Asians, Pacific Islanders^a	429 (97.5)^b	8 (1.8)	0	2 (0.5)
Black non-Hispanics	382 (87.0)	33 (7.5)	17 (3.9)	7 (1.6)
Hispanics	401 (91.3)	27 (6.2)	8 (1.8)	3 (0.7)
White non-Hispanics	16 (3.6)	40 (9.1)	73 (16.6)	310 (70.6)
Below poverty	408 (91.9)	36 (8.2)	2 (0.5)	0
Working class	2 (0.5)	45 (10.3)	326 (74.3)	60 (15.0)
Residing in an urban area	209 (47.1)	1 (0.2)	13 (3.0)	218 (49.7)
Residing in a rural area	341 (76.8)	36 (8.2)	9 (2.1)	57 (13.0)

Note. NH = nursing home.

Racial and ethnic categories are according to U.S. Census.

An example interpretation of the results is 97.5% of 443 NHs in our cohort were located in Census tracts with communities consisting of <25% American Indians, Asians, or Pacific Islanders.

Social Engagement and Racial and Ethnic Disparities

Figure 1 presents the average observed and predicted social engagement scores (Mor et al., 1995) for minority groups of NH residents and the observed scores for White residents at 1 year after NH admission. There was no significant difference between the social engagement scores that were observed for each minority group and those expected had they been part of the White group, indicating that there were no disparities in social engagement based on race and ethnicity. Models of Whites in mixed race NHs used in the Peters–Belson analysis showing no significant racial or ethnic disparities in social engagement at 1 year after NH admission are available for up to 1 year after publication by contacting the corresponding author. The average observed social engagement scores for all groups at 1 year after admission were fairly low, approximately 2, which were similar to those at admission.

Figure 1.

Average social engagement scores at 1 year after admission.

Predictors of Social Engagement at 1 Year After NH Admission

As no significant racial or ethnic disparities in social engagement were found, all residents of all racial and ethnic groups (including Whites) were included in one model to identify risk factors at NH admission of low social engagement at 1 year after admission. Results are presented in Table 5. Having a low score for social engagement at NH admission was the strongest predictor of low social engagement 1 year later. Involvement in more positive relationship activities at NH admission was associated with higher social engagement at 1 year. Residents’ physical and emotional/cognitive status at NH admission was also a significant predictor of social engagement levels. Having more limitations in ADLs, vision problems, communication difficulties, and cognitive deficits predicted lower social engagement. The significant predictor of low social engagement at the community level was residing in an NH located in a community with a higher proportion of its population living an urban area.

Table 5.

Predictors of Social Engagement of Older NH Admissions at 1 Year After Admission.

Covariate^a	Estimate	SE	p
Individual level
Female gender	0.03	0.02	.18
High school education or more	−0.003	0.02	.89
Activities of Daily Living Deficit Scale	−0.01	0.002	<.001
Comorbidity Index	0.003	0.01	.70
Charlson Index
Incontinence–Any type	−0.04	0.03	.15
Vision problems	−0.05	0.01	<.001
Number of indicators
Cognitive deficits	−0.09	0.005	<.001
MDS-COGS Scale
Communication difficulty score	−0.03	0.01	<.001
MDS Mood Scale	−0.01	0.01	.43
Relationships	0.04	0.004	<.001
Number of positive activities
Spirituality Scale	0.03	0.01	.06
Social Engagement Index	0.33	0.01	<.001
NH level
Quality of Care Deficiency Index	−0.001	0.003	.84
Percentage of residents on Medicaid	−0.002	0.001	.19
Community level
Percentage of Census tract below poverty level	−0.005	0.003	.14
Percentage of Census tract residing in an urban area	−0.12	0.06	.03

Note. NH = nursing home; MDS = Minimum Data Set.

Operational definitions of MDS items for the variables and references for the scales used can be found in Supplemental Table 1.

Discussion

The results of this study add new insight about social engagement in NHs. No racial or ethnic-based disparities in social engagement at 1 year were found. This is a positive outcome suggesting that NHs provide equitable experiences and opportunities for social engagement for older residents of all races and ethnicities. The findings of low social engagement of all groups at NH admission support the observations of Fraher and Coffey (2011) who reported that social isolation is one of the reasons prompting older adults to seek NH care, especially after a deterioration in their health. In addition to their physical and mental limitations, older adults may experience a decrease in social engagement due to death of a spouse or friends, living at a distance from family, being reluctant to burden relatives with requests for help, and inability to drive to activities (Fraher & Coffey, 2011). Our results indicate that the low social engagement at admission changes little after 1 year in the NH. Admission limitations in physical and social functioning, such as ADL deficits, communication difficulties, vision problems, and participating in social relationships and interactions appear to continue to lessen residents’ ability for social engagement during their NH stay. The significant associations of these resident characteristics with social engagement are supported by observations of cross-sectional and qualitative studies (Kang, 2012; Li & Cai, 2014).

This study also examined a number of NH- and community-level variables for their association with social engagement. An NH in a more urban community was a significant risk factor for low social engagement. Although no studies have examined the role of NH location on social engagement, there is research that has considered urban versus rural location as a factor for psychosocial outcomes in other groups. Some investigators have found that older community-living people living in an urban area experience greater loneliness than those in rural areas (Fraser, 2009), possibly due to the higher percentage of younger people in an urban area population (Havens, Hall, Sylvestre, & Jivan, 2004). Havens et al. (2004) showed that that there is social isolation and loneliness among older adults in both rural and urban areas but the risk factors for these outcomes differ by locale.

The results of our study suggest that efforts to increase social engagement should begin at the time of NH admission. They highlight the potential social as well as functional benefit of improving ADLs. Findings underscore the importance of assessing a resident’s communication and vision abilities in relation to planning social activities and making adjustments that may be needed to optimize their participation. As an example, Resnick et al. (1997) discussed numerous steps focused on hearing that might improve communication and social interactions of NH residents ranging from regular checks for cerumen removal to environmental adaptations of installing carpeting and window treatments, especially in common areas, to reduce distracting noise. In contrast to DuBeau et al. (2006) who reported that worsening urinary incontinence for more than 3 months was associated with lower social engagement, we found that merely having incontinence of any type at admission did not predict low social engagement at 1 year. Results suggest the potential social importance of preventing worsening of incontinence among NH admissions.

A variety of interventions have been examined to increase social engagement of NH residents by providing opportunities for increasing conversation, interactions, and relationship building. These include a program of indoor gardening (Tse, 2010), visits by companion animals (Prosser, Townsend, & Staiger, 2008), group reminiscence therapy (Chiang et al., 2010), integrating cultural traditions and heritage into occupational therapy sessions (Hersch et al., 2012), and videoconferencing with family members (Demiris et al., 2008; Tsai, Tsai, Wang, Chang, & Chu, 2010). Internet-based technologies, such as email, social media, online chat rooms, offer additional opportunities for improving social engagement (Crosby, Burbidge, & Mcpherson, 2013; McComish, Peura, & Richeson, 2010). It has been suggested that such technologies may also make differences in social engagement between rural and urban locales less pronounced (Lichter & Brown, 2014). For the NH population, research is needed to identify technological strategies that are useable and effective across a range of functional, communication, and cognitive capabilities.

Interviews of NH residents revealed that regular seating in the dining room fostered opportunities for making friends, and being offered a choice of leisure activities was also important to them (Thomas, O’Connell, & Gaskin, 2013). Others have observed that NH residents and staff can develop positive relational interactions that are meaningful to both (Cook & Brown-Wilson, 2010). Cook and Clarke (2010) included many of these ideas in a framework they proposed to support social interactions in NHs. The eight themes of their framework encourage individualizing the type and quality of opportunities for social engagement, creating an environment and community conducive to friendly interactions, and using supportive communication technologies.

The lack of racial and ethnic disparities in our study is in contrast to the one other study that has investigated this issue (Li & Cai, 2014). Li and Cai (2014) reported average social engagement values of racial and ethnic groups of older NH residents very similar to those in our study, but also found that these group differences, particularly between Hispanics compared with Whites, were statistically significant. There are a number of methodological distinctions between the studies that may explain the difference in outcome. Li and Cai (2014) used a simple, dichotomized social engagement score and different covariates. Similar to Won et al. (2006), our study used the original 6-point social engagement scale, perhaps granting greater precision of social engagement measurement. The study by Li and Cai (2014) had a very large sample of NH residents, making the likelihood of observing statistically significant differences among groups more probable. Won et al. (2006) identified a change of at least one in the social engagement score as clinically meaningful; using their assessment, a clinical interpretation of the small difference in social engagement scores among minority groups in our study supports the lack of a statistical significance. Importantly, the analyses in our study were restricted to Whites who were in the same NHs as the minority group being modeled to control for unmeasured NH confounders that might influence results if NHs with only White residents were included. Our analysis also controlled for the resident’s baseline social engagement at NH admission. Lastly, though the cross-sectional MDS observational time point was unspecified in the Li and Cai (2014) study, our approach measured social engagement at two explicit times—at admission and at 1-year follow-up—enabling a view of the lack of improvement in social engagement over time. This longitudinal perspective, the first of its kind, is another strength of our study.

There are several study limitations. Data were from a national chain of for-profit NHs; therefore, results may not be representative of all residents in all NHs, in particular, non-profit NHs. The results have import, however, as for-profit NHs comprise a majority, 69%, of all U.S. NHs (Centers for Medicare & Medicaid Services, 2013). Furthermore, in a previous study, we showed that the characteristics of older adults admitted to NHs analyzed in this study are comparable with those admitted to all Medicare/Medicaid certified NHs in the United States in the same time period (Bliss et al., 2013). Generalization of results is limited by sizes of minority groups; the observed mean social engagement, for example, will be much less precise (have a larger variance) for much smaller groups. However, one of the advantages of the Peters–Belson method is that the initial regression model used for modeling social engagement is conducted on White residents, which is the largest group. Because of the large size, the coefficients from the White models (which are then applied to the minority groups residing in the same NHs as those of Whites) are very precisely estimated. This precision is then carried over into the prediction of mean social engagement for the smaller sized minority groups. Another limitation may be that not all relevant risk factors for low social engagement predictors may be known or possible to include in the models. Although we focused on social engagement at 1 year after NH admission, changes in social engagement could have occurred at any time within that period. The years of our data, whose three levels are aligned at the time of the Census data that were available at the start of our grant, may be a limitation; however, the similarity in the levels of social engagement in our study and the later study of Li and Cai (2014) suggest a consistency in social engagement over time so that our findings have relevance and can serve to guide interventions for improving the well-being of older NH residents.

Social engagement has been shown to be essential part of the health and well-being of older adults in NHs. Older adults admitted to NHs tend to have a pre-existing low level of social engagement and enter the new social milieu with multiple types of limitations that lessen their ability for social interaction. Our findings suggest that these limitations place those admitted to NHs at risk for low social engagement up to a year later. In the national sample of NH residents that we studied, we found no racial and ethnic disparities in social engagement. A variety of creative ideas for strategies to improve social engagement of NH residents have been proposed. The findings of this study indicate that greater efforts by NHs are needed to increase social engagement of all groups of residents and encourage them to do so from the time of admission.

Footnotes

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 study was funded by National Institute of Nursing Research, NIH, (grant number) and the Minnesota Supercomputing Institute, University of Minnesota, Minneapolis, MN.

Author Biographies

Donna Bliss, PhD, RN, FAAN, FGSA, is a professor and holds the professorship in nursing research from the School of Nursing Foundation at the University of Minnesota. She is a Morse-Alumni Association Outstanding teacher. She is a senior investigator whose research expertise focuses on incontinence and associated skin problems and racial/ethnic disparities in these conditions in older adults.

Susan Harms, PhD, RPh, is a research associate at the University of Minnesota School of Nursing and adjunct associate professor at College of Pharmacy, University of Minnesota. Her expertise is pharmacological management of older adults and health disparities.

Lynn E. Eberly, PhD, is an associate professor of biostatistics in the School of Public Health at the University of Minnesota. She is a nationally recognized statistician who has been working with clinical research investigations for more than 15 years. Her expertise is in longitudinal and cluster correlated studies.

Kay Savik, MS, was a lead statistician, senior research fellow, Office of Nursing Research and Scholarship, University of Minnesota School of Nursing, at the time of the study.

Olga Gurvich, MA, is a lead statistician and senior research fellow at Office of Nursing Research and Scholarship, University of Minnesota School of Nursing.

Christine Mueller, PhD, RN, FAAN, FGSA, is a professor and associate dean for academic programs, University of Minnesota School of Nursing. Her research expertise focuses on quality and models of nursing practice in long-term care.

Jean F. Wyman, PhD, CNP, FGSA, FAAN, is a professor and the Cora Meidl Siehl Chair in nursing research at the University of Minnesota School of Nursing where she directs the John A. Hartford Foundation Center of Gerontological Nursing Excellence, and serves as the co-director of the Deborah E. Powell Center of Women’s Health.

Beth Virnig, PhD, is a professor and the senior associate dean of academic affairs and research, University of Minnesota School of Public Health Division of Health Policy and Management. She is involved in research that examines the use and quality of health care by the elderly in the Medicare program and managed care.

References

Achterberg

Pot

A. M.

Kerkstra

Ooms

Muller

Ribbe

(2003). The effect of depression on social engagement in newly admitted Dutch nursing home residents. The Gerontologist, 43, 213-218.

Bassuk

S. S.

Glass

T. A.

Berkman

L. F.

(1999). Social disengagement and incident cognitive decline in community-dwelling elderly persons. Annals of Internal Medicine, 131, 165-173.

Bliss

D. Z.

Harms

Garrad

J. M.

Savik

Gurvich

Wyman

J. F.

. . . Cunanan

(2013). Prevalence of incontinence by race and ethnicity of older people admitted to nursing homes. Journal of the American Medical Directors Association, 14, 451.e1-451.e7.

Brownie

Horstmanshof

(2011). The management of loneliness in aged care residents: An important therapeutic target for gerontological nursing. Geriatric Nursing, 32, 318-325.

Burrows

A. B.

Morris

J. N.

Simon

S. E.

Hirdes

J. P.

Phillips

(2000). Development of a minimum data set-based depression rating scale for use in nursing homes. Age and Ageing, 29, 165-172.

Centers for Medicare & Medicaid Services. (2013). Nursing home data compendium 2013 edition. Retrieved from http://www.cms.gov/Medicare/Provider-Enrollment-and-Certification/CertificationandComplianc/downloads/nursinghomedatacompendium_508.pdf

Charlson

M. E.

Pompei

Ales

K. L.

MacKenzie

C. R.

(1987). A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases, 40, 373-383.

Chiang

Chu

Chang

Chung

Chen

Chiou

Chou

(2010). The effects of reminiscence therapy on psychological well-being, depression, and loneliness among the institutionalized aged. International Journal of Geriatric Psychiatry, 25, 380-388.

Cook

Brown-Wilson

(2010). Care home residents’ experiences of social relationships with staff. Nursing Older People, 22(1), 24-29.

10.

Cook

Clarke

(2010). A framework to support social interaction in care home. Nursing Older People, 22(3), 16-21.

11.

Cordner

Blass

D. M.

Rabins

P. V.

Black

B. S.

(2010). Quality of life in nursing home residents with advanced dementia. Journal of the American Geriatrics Society, 58, 2394-2400.

12.

Crosby

J. T.

Burbidge

Mcpherson

(2013). Email, skype, pandora and YouTube: Pilot project with residents and Families in a nursing home [Abstract]. Journal of the American Medical Directors Association, 14, B15-B16.

13.

Degenholtz

H. B.

Kane

R. A.

Kane

R. L.

Bershadsky

Kling

K. C.

(2006). Predicting nursing facility residents’ quality of life using external indicators. Health Services Research, 41, 335-356.

14.

de Guzman

A. B.

Maravilla

K. N.

Maravilla

V. A. M.

Marfil

J. D.

Mariñas

J. A. R.

Marquez

J. M. B.

(2012). Correlates of geriatric loneliness in Philippine nursing homes: A multiple regression model. Educational Gerontology, 38, 563-575.

15.

Demiris

Oliver

D. R.

Hensel

Dickey

Rantz

Skubic

(2008). Use of videophones for distant caregiving: An enriching experience for families and residents in long-term care. Journal of Gerontological Nursing, 34(7), 50-55.

16.

Diez-Roux

A. V.

(1998). Bringing context back into epidemiology: Variables and fallacies in multilevel analysis. American Journal of Public Health, 88, 216-222.

17.

Dosa

Intrator

McNicoll

Cang

Teno

(2007). Preliminary derivation of a nursing home confusion assessment method based on data from the minimum data set. Journal of the American Geriatrics Society, 55, 1099-1105.

18.

DuBeau

C. E.

Simon

S. E.

Morris

J. N.

(2006). The effect of urinary incontinence on quality of life in older nursing home residents. Journal of the American Geriatrics Society, 54, 1325-1333.

19.

Fraher

Coffey

(2011). Older people’s experiences of relocation to long-term care. Nursing Older People, 23(10), 23-27.

20.

Fraser

R. J.

(2009). A comparison of rural and urban older adults in Iowa on specific markers of successful aging. Journal of Gerontological Social Work, 52, 423-438.

21.

Golden

Conroy

R. M.

Lawlor

B. A.

(2009). Social support network structure in older people: Underlying dimensions and association with psychological and physical health. Psychology, Health & Medicine, 14, 280-290.

22.

Graubard

B. I.

Sowmya Rao

Gastwirth

J. L.

(2005). Using the Peters–Belson method to measure health care disparities from complex survey data. Statistics in Medicine, 24, 2659-2668.

23.

Hartmaier

S. L.

Sloane

P. D.

Guess

H. A.

Koch

G. G.

(1994). The MDS cognition scale: A valid instrument for identifying and staging nursing home residents with dementia using the minimum data set. Journal of the American Geriatrics Society, 42, 1173-1179.

24.

Havens

Hall

Sylvestre

Jivan

(2004). Social isolation and loneliness: Differences between older rural and urban Manitobans. Canadian Journal on Aging, 23, 129-140.

25.

Health Services Advisory. (2006). Percent of residents who are more depressed or anxious. Phoenix, AZ.

26.

Hersch

Hutchinson

Davidson

Wilson

Maharaj

Watson

K. B.

(2012). Effect of an occupation-based cultural heritage intervention in long-term geriatric care: A two-group control study. American Journal of Occupational Therapy, 66, 224-232.

27.

Hirdes

J. P.

Frijters

D. H.

Teare

G. F.

(2003). The MDS-CHESS scale: A new measure to predict mortality in institutionalized older people. Journal of the American Geriatrics Society, 51, 96-100.

28.

Jang

Mortimer

J. A.

Haley

W. E.

Graves

A. R. B.

(2004). The role of social engagement in life satisfaction: Its significance among older individuals with disease and disability. Journal of Applied Gerontology, 23, 266-278.

29.

Kang

(2012). Correlates of social engagement in nursing home residents with dementia. Asian Nursing Research, 6, 75-81.

30.

Kiely

D. K.

Flacker

J. M.

(2003). The protective effect of social engagement on 1-year mortality in a long-stay nursing home population. Journal of Clinical Epidemiology, 56, 472-478.

31.

Kiely

D. K.

Simon

S. E.

Jones

R. N.

Morris

J. N.

(2000). The protective effect of social engagement on mortality in long-term care. Journal of the American Geriatrics Society, 48, 1367-1372.

32.

Krieger

Chen

J. T.

Waterman

P. D.

Rehkopf

D. H.

Subramanian

S. V.

(2005). Painting a truer picture of US socioeconomic and racial/ethnic health inequalities: The public health disparities geocoding project. American Journal of Public Health, 95, 312-323.

33.

Cai

(2014). Racial and ethnic disparities in social engagement among US nursing home residents. Medical Care, 52, 314-321.

34.

Lichter

D. T.

Brown

D. L.

(2014). The new rural-urban interface: Lessons for higher education. Choices, 29(1), 1-6.

35.

Lou

V. W.

Chi

Kwan

C. W.

Leung

A. Y.

(2013). Trajectories of social engagement and depressive symptoms among long-term care facility residents in Hong Kong. Age and Ageing, 42, 215-222.

36.

Lucas

G. M.

Knowles

M. L.

Gardner

W. L.

Molden

D. C.

Jefferis

V. E.

(2010). Increasing social engagement among lonely individuals: The role of acceptance cues and promotion motivations. Personality and Social Psychology Bulletin, 36, 1346-1359.

37.

McComish

Peura

Richeson

N. E.

(2010). Computer and internet engagement for older adults. Activities Directors’ Quarterly for Alzheimer’s & Other Dementia Patients, 11(3), 43-46.

38.

Mor

Branco

Fleishman

Hawes

Phillips

Morris

Fries

(1995). The structure of social engagement among nursing home residents. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 50(1), P1-P8.

39.

Mor

Zinn

Angelelli

Teno

J. M.

Miller

S. C.

(2004). Driven to tiers: Socioeconomic and racial disparities in the quality of nursing home care. The Milbank Quarterly, 82, 227-256.

40.

Morris

J. N.

Fries

B. E.

Morris

S. A.

(1999). Scaling ADLs within the MDS. The Journals of Gerontology, Series A: Biological Sciences & Medical Sciences, 54, M546-M553.

41.

Park

N. S.

(2009). The relationship of social engagement to psychological well-being of older adults in assisted living facilities. Journal of Applied Gerontology, 28, 461-481.

42.

Prosser

Townsend

Staiger

(2008). Older people’s relationships with companion animals: A pilot study. Nursing Older People, 20(3), 29-32.

43.

Rao

R. S.

Graubard

B. I.

Breen

Gastwirth

J. L.

(2004). Understanding the factors underlying disparities in cancer screening rates using the Peters-Belson approach: Results from the 1998 National Health Interview Survey. Medical Care, 42, 789-800.

44.

Resnick

H. E.

Fries

B. E.

Verbrugge

L. M.

(1997). Windows to their world: The effect of sensory impairments on social engagement and activity time in nursing home residents. The Journals of Gerontology, Series B: Psychological Sciences & Social Sciences, 52(3), S135-S144.

45.

Savik

Fan

Bliss

D. Z.

Harms

(2005). Preparing the MDS dataset for analysis in the study of risk factors of perineal dermatitis. Journal of Advanced Nursing, 52(4), 399-409.

46.

Schreiner

A. S.

Yamamoto

Shiotani

(2005). Positive affect among nursing home residents with Alzheimer’s dementia: The effect of recreational activity. Aging & Mental Health, 9, 129-134.

47.

Schroll

Jonsson

P. V.

Mor

Berg

Sherwood

(1997). An international study of social engagement among nursing home residents. Age and Ageing, 26(Suppl. 2), 55-59.

48.

Stevenson

K. M.

Brown

R. L.

Dahl

J. L.

Ward

S. E.

Brown

M. S.

(2006). The Discomfort Behavior Scale: A measure of discomfort in the cognitively impaired based on the Minimum Data Set 2.0. Research in Nursing & Health, 29, 576-587.

49.

Thomas

J. E.

O’Connell

Gaskin

C. J.

(2013). Residents’ perceptions and experiences of social interaction and participation in leisure activities in residential aged care. Contemporary Nurse, 45, 244-254.

50.

Tsai

Wang

Chang

Chu

H. H.

(2010). Videoconference program enhances social support, loneliness, and depressive status of elderly nursing home residents. Aging & Mental Health, 14, 947-954.

51.

Tse

M. M. Y.

(2010). Therapeutic effects of an indoor gardening programme for older people living in nursing homes. Journal of Clinical Nursing, 19, 949-958.

52.

van Beek

A. P.

Frijters

D. H.

Wagner

Groenewegen

P. P.

Ribbe

M. W.

(2011). Social engagement and depressive symptoms of elderly residents with dementia: A cross-sectional study of 37 long-term care units. International Psychogeriatrics, 23, 625-633.

53.

Verbeek

van Rossum

Zwakhalen

S. M.

Ambergen

Kempen

G. I.

Hamers

J. P.

(2009). The effects of small-scale, homelike facilities for older people with dementia on residents, family caregivers and staff: Design of a longitudinal, quasi-experimental study. BMC Geriatrics, 9, Article 3. doi:10.1186/1471-2318-9-3

54.

Won

Lapane

K. L.

Vallow

Schein

Morris

J. N.

Lipsitz

L. A.

(2006). Long-term effects of analgesics in a population of elderly nursing home residents with persistent nonmalignant pain. The Journals of Gerontology, Series A: Biological Sciences & Medical Sciences, 61, 165-169.