Families Matter in Long-Term Care: Results of a Group-Randomized Trial

Abstract

This group-randomized trial implemented and evaluated an intervention to reduce staff burden and improve family and resident outcomes by helping families create meaningful roles for themselves in residential care/assisted living and nursing homes. Across 24 sites, families (N = 490) and staff (N = 397) provided data over six months about family involvement, family and staff well-being and attitudes, and resident quality of life. Intervention subjects participated in workshops and created service plans to identify family roles. For families, the intervention decreased burden and improved resident quality of life but also increased guilt and conflict. Staff reported less burnout and greater partnership with families, and felt families were more empathic. Consequently, there are benefits to increasing family involvement, but attention must be paid to potential barriers and negative outcomes.

INTRODUCTION

The long-term care (LTC) workforce is in short supply of crisis proportion, a problem that promises to worsen as the population ages. By 2020, the demand for direct care workers (referred to as nursing assistants in nursing homes [NHs] and personal care aides in residential care [RC]/assisted living [AL] settings) will increase by more than 70% as the baby boomers reach age 85 (U.S. Bureau of Labor Statistics, 2012). Supply will not meet demand for numerous reasons, including low wages, few benefits, and the lack of respect and opportunities for advancement that are attendant with this position (Pillemer, 1997). A less obvious reason, and one that is harder to combat through some of the culture change initiatives that are addressing workforce issues, speaks to the very core of the matter: this is a workforce that was developed on an assumption of endless supply—an assumption that is no longer valid, given the decreasing numbers of and alternate opportunities available to women ages 25 to 44 who comprise the bulk of the long-term care workforce (Dawson & Surpin, 2001; Koren, 2010).

Fortunately, there is at least a partial remedy to staff burden, already in place and waiting to be mobilized: residents’ families. The majority of LTC residents were cared for by family before their admission, and these same family members continue to visit on a regular basis and want to remain involved in care (Port et al., 2001; Port et al., 2005). At the same time, despite assertions that family involvement is important to care, LTC practices have tended to reflect a “visitor” philosophy that treats families as outsiders, or a “servant” philosophy that dictates the nature of their involvement, as opposed to a “client” philosophy that coordinates with family members to meet the needs of the resident, family, and staff (Caron, 1997; Montgomery, 1983). Consequently, family members are often at a loss as to their role after the resident is admitted to a LTC setting.

Tthus, it is of major significance that the LTC system has not appreciated the resource that families may constitute in supporting the work done by staff and in improving the quality of life of their family member. That said, this topic has received some research attention, evidenced by five prior efforts to address family involvement in care to improve family, staff, and/or resident outcomes. These prior efforts were promising, but due to their focus on select populations, settings, or outcomes, were limited. Most notably, all included only NHs, thereby excluding RC/AL settings, which provide supportive care to almost one million older adults (Park-Lee et al., 2011).

The earliest project, which examined outcomes of a dementia-specific Family Visit Education Program, focused on educational needs of families. It resulted in fewer behavioral symptoms and less resident depression, and also improved family and staff communication with residents, but there were few significant benefits for family members, none of which were sustained over six months (McCallion, Toseland, & Freeman, 1999). The second project, the Family Partnership Program (FPP), also focused only on families of residents with dementia and was grounded in the observation that families tend to advise staff regarding how to care for residents, rather than help provide care themselves. Results indicated that families did become more involved in care but were less satisfied than families in control sites because they felt obligated to substitute for staff as caregivers; in essence, families did not want to be responsible for routine care such as dressing, bathing, and toileting. Staff attitudes and resident outcomes were not assessed, however (Murphy et al., 2000).

The third program, Partners in Caregiving (PIC), addressed relationship and communication challenges between family and staff, based on the premise that families believe they must monitor care because they tend to distrust staff, and that staff believe families hold unrealistic expectations regarding the care that can actually be provided. A key limitation of PIC was that it addressed family-staff relationships and communication without considering the intent of the interactions; that is, it focused on the “how” without simultaneously addressing the “what” of collaborative care. Although the program resulted in reduced staff depression and resident behavioral symptoms, and improved family-staff perceptions of each other, it was unclear whether improved perceptions reduced conflict, family burden, or staff burnout (Pillemer et al., 2003; Robison et al., 2007).

Another dementia-specific program, Family Involvement in Care (FIC), tested a negotiated written family-staff partnership agreement and found significant beneficial intervention effects, primarily on family outcomes—emotional reactions, perceptions of staff relationships, and perceptions of care (Maas et al., 2004); In another application of FIC, resident deterioration slowed (Jablonski, Reed, & Maas, 2005). However, the intervention did not reduce perceived family-staff conflict, staff stress, or staff perceptions of a genuine partnership. Of note, the PIC and FIC interventions found positive staff outcomes limited to licensed nurses in leadership roles, the staff less likely to be available to families in RC/AL communities (Stearns et al., 2007).

One final small-scale randomized trial in Canadian NHs tested the efficacy of a psychoeducational group empowerment approach with daughters of residents with dementia (Ducharme et al., 2005). Daughters in the experimental group increased their perception of competence in dealing with staff. However, the primary outcomes were limited to improved mental health outcomes for the daughter caregivers.

Given the potentially significant resource that families could constitute in LTC, their desire for involvement and need for role clarification, promising results from prior work, and models upon which to expand, this article reports on a new program, called Families Matter in Long-Term Care. Families Matter was designed to help families work with staff and residents to create a role for themselves that would benefit the residents’ quality of life (or, in the case of residents who were severely incapacitated, benefit the overall setting), and also improve family-staff relations. Specifically, this article addresses whether a program designed to engage family members in a meaningful way with or on behalf of their relative changes the amount and type of family involvement, family and staff well-being, attitudes, and perceptions of resident quality of life. Families Matter was evaluated in both NHs and RC/AL communities, and was not specific to family members of residents with dementia.

METHODOLOGY

Families Matter was a group-randomized trial conducted in six NH and 18 RC/AL settings in North Carolina, participating in the Collaborative Studies of Long-Term Care (CS-LTC). The CS-LTC is a multistate consortium of LTC settings involved in research to inform the quality of care and quality of life in LTC. The Institutional Review Boards of the University of North Carolina at Chapel Hill and Duke University approved all procedures.

Identification and Recruitment of Settings

Power calculations and projections related to participation rates and the effects of clustering indicated a need to recruit 24 LTC settings. Due to the smaller average size of RC/AL settings, compared to NHs, more of the former were recruited. All settings had to be within an approximately one hour drive of UNC; they were matched in pairs by type (RC/AL or NH), region (urban or rural), chain affiliation (when possible), and size (within 25%), and one of each pair was randomly selected to be an intervention or a wait list control site. Administrators were sent a letter of invitation and then visited to explain the study procedures.

Identification and Recruitment of Subjects

To reduce the effects of clustering, a random sample of approximately 20 eligible residents was identified per setting, with eligibility limited to having a family member 21 years of age or older who visited at least once a month and was considered to be the individual who knew the most about the resident. For each eligible resident, a direct care worker was identified who knew the most about the resident, was 21 years of age or older, worked at least 20 hours a week during the morning or afternoon shift, and had worked in the setting for at least one month prior to study initiation. Administrators provided a list of eligible family-staff pair subjects, with the intent to maximize the number of different direct care staff. Families were sent a letter of invitation and then contacted by telephone to explain the study procedures; staff were informed about the study in person. All individuals provided consent before they were interviewed.

Families Matter Intervention

Families Matter encouraged residents’ families to work with the resident and LTC staff to identify an activity in which they could participate that would promote the residents’ quality of life in one or more of four areas: doing things, getting around, looking good, and eating well. These four areas were selected based on findings from the earlier FPP that identified them as areas in which families would be willing to be involved. A fifth area was later identified—helping the community—for families who could not conceive of a way to improve their resident’s quality of life. A meeting was convened with representatives from 10 of the intervention sites to discuss the suitability of these areas as well as the overall Families Matter program; modifications were made accordingly.

Families Matter was introduced into each setting through a workshop presentation made separately to families (and residents, if desired) and staff; the workshop focused on family involvement, suggestions for involvement in the five areas noted previously, and the importance of effective communication between families and staff (adapting content from the PIC Program, which focused on effective communication strategies). In the days following the presentation, individualized in-person service plan meetings were held between a trained member of the research team, the resident (if able), a staff member familiar with the resident (if available), and one or more of the resident’s family members. The purpose of these meetings was to identify one or more roles in which the family could be involved, in collaboration with the resident and staff, to improve the resident’s quality of life. The meeting concluded with a clearly articulated Families Matter Service Plan, which was written in triplicate so that the family, resident, and staff had a copy. If families could not attend the workshop and/or individualized meeting, the same material was covered through a telephone meeting. Materials were provided to support family involvement, such as jigsaw puzzles (doing things), pedometers (getting around), nail polish (looking good), and simple recipes (eating well).

Data Collection

Telephone interviews were conducted at baseline and six months later with family and staff to determine the outcomes of Families Matter. If the same family or staff member was not available at the time of the six-month interview, he/she was replaced with who was then considered to be the individual most knowledgeable about the resident. Fidelity interviews were conducted with family in intervention sites between one and three times for up to six months after the service plan was created.

Measures

Data were collected related to outcomes (family involvement, and family and staff well-being, attitudes, and perceptions of resident quality of life), descriptive information, and fidelity. To achieve comparability in scoring and ease respondent burden, some measures were changed to a five-point Likert scale for response consistency.

Family Outcomes

Several family outcome domains were assessed:

• Involvement in 16 areas (reported by both family and staff), using a modified and expanded version of Murphy et al.‘s (2000) Involvement scale, which assesses involvement in both direct and indirect (supportive) care. Family members reported the actual frequency of each activity for the past month. Because the distribution of frequencies was highly skewed and values grouped near round figures (e.g., 10, 20), the frequencies were grouped into ordered categories.

• Depressive symptoms, using the 10-item Center for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977)

• Perceptions of care, using three Family Perceptions of Care subscales (management effectiveness, physical care, and activities for residents) (Maas et al., 2004)

• Interpersonal conflict, using the Interpersonal Conflict Scale (Pillemer & Moore, 1989)

• Perception of caregiving role, using three subscales (conflict with staff, guilt, and loss) of the Family Perception of Caregiving Role instrument (Maas et al., 2004)

• Caregiving and burden, using the Lawton Caregiving Appraisal measure, which has three subscales (burden, satisfaction, and impact) (Lawton, Kleban, Moss, Rovine, & Glicksman, 1989), and the Zarit Burden Interview, which also has three subscales (social, psychological, and guilt) (Zarit, Reever, & Bach-Peterson, 1980)

Of note, when reporting on staff care and conflicts, families did so in reference to the staff members overall, not in reference to the individual staff respondent.

Staff Outcomes

Several domains also were assessed to evaluate staff outcomes:

• Burnout and stress, using the Maslach Burnout Inventory, which has three subscales (emotional exhaustion, depersonalization, and lack of personal accomplishment) (Maslach, Jackson, & Leiter, 1996); four of nine items from the Work Stressors Inventory caring for residents subscale (Schaefer & Moos, 1993); and the Hassles and Uplifts Scales (Elder, Wollin, Hartel, Spencer, & Sanderson, 2003)

• Depressive symptoms, using the Center for Epidemiological Studies Depression Scale (CES-D) (Radloff, 1977)

• Interpersonal conflict, using the Interpersonal Conflict Scale (Pillemer & Moore, 1989)

• Perceptions of caregiving role, using the Staff Perception of Caregiving Role instrument, which has four subscales (burden, frustration, dominion, and exclusion of families) (Maas et al., 2004)

• Attitudes toward families, using the Attitudes Toward Families Checklist, which has three subscales (disruption, family as partners, and family relevance) (Maas et al., 2004), and the Family Behaviors and Family Empathy Scales (Pillemer et al., 2003)

When reporting on conflict and attitudes toward family, staff did so in reference to families in general, not in reference to a particular family respondent.

Resident Outcomes

Family and staff both rated resident outcomes in three areas:

• Depressive symptoms, using the Patient Health Questionnaire (PHQ-9) (Kroenke & Spitzer, 2001)

• Activity involvement and enjoyment using the Pleasant Events Schedule - AD (asked only of staff) (Logsdon & Teri, 1997)

• Quality of life, using the Quality of Life - Alzheimer’s Disease (QOL-AD) measure (Logsdon, Gibbons, McCurry, & Teri, 1999) and a global question “Overall, would you rate (resident’s) quality of life as excellent, very good, good, fair, or poor?”

When reporting on residents, family and staff did so in reference to individual residents, namely the family member’s relation.

Descriptive Information

Descriptive information was obtained regarding the setting (e.g., age, size, case-mix), family (e.g., marital status, relationship to resident, health), staff (e.g., race, experience, role), and resident (e.g., dementia status, function, length of stay).

Fidelity and Dose of the Intervention

Fidelity interviews asked families to what extent the service plan was being followed as planned (scored as not at all [0], somewhat [0.5], or completely [1]). For families, the dose of the intervention was calculated using the mean score of three components: workshop attendance (scored 0 for no, 1 for yes), creation of a service plan (scored 0 or 1), and the fidelity interview score. The mean produced a dose score between 0 and 1, with 0 indicating no participation of any kind and 1 representing the fullest possible participation. For staff, the dose of the intervention was calculated as attending the workshop (1) or not (0).

RESULTS

Linear and nonlinear mixed models were used to adjust for the clustering of staff, family members, and residents within NHs and RC/AL settings. Models testing the intervention effect included as predictors intervention status, time period (baseline or follow-up), and the interaction of the two, as well as setting-level and individual-level characteristics that differed between control and intervention groups at baseline and were relevant to the type of participant whose outcomes were being considered. Models were constructed for both intention-to-treat (ITT) analyses and dose of intervention analyses, because dose was found to vary substantially within the intervention group. The dose analyses are highlighted here, with ITT analysis results footnoted. For testing of intervention effects on the frequency of family member activities, a modified version of a published SAS macro (Williamson, Lipsitz, & Kim, 1998) was used to adjust for clustering by carrying out a generalized estimating equations analysis of ordered categorical data. Analyses were completed using SPSS version 16 and SAS version 9.1.

Recruitment and Data Collection

Eight NHs and 21 RC/AL settings were invited to participate in the project; due to the time it took to obtain their agreement, it was not possible to engage two NHs and three RC/AL settings (75% and 86% participation rate, respectively). Baseline and follow-up data were obtained from 490 families and 397 staff (78% and 99% participation rate, respectively). If a family or staff member was no longer available at the six-month follow-up, he/she was replaced with the current best respondent; 1% of family and 24% of staff respondents at follow-up differed from those at baseline.

All settings were for-profit and on average, one-half of residents had a diagnosis of dementia (see Exhibit 1). In control sites, 5% of residents did not have a diagnosis of dementia but did have a memory problem or appeared disoriented; the comparable number for intervention sites was almost 28% (p = .01). Family members were on average 58 to 59 years of age, 73% of whom were female. In control sites, fewer family members were married (68% versus 78%; p = .03) and rated their health as good to excellent (84% versus 92%; p = .04); also, fewer were spouses and more were distant families and friends (p < .001). Staff averaged 39 to 40 years of age, and 95% were female; there were no differences in staff characteristics between control and intervention sites. Residents averaged 80 to 82 years of age, with 75% being female; those in control sites were somewhat less dependent in bed mobility (25% versus 33%; p = .02) but more often chairfast (37% versus 26%; p = .02).

Exhibit 1. Sample and Baseline Characteristics: Settings, Families, Staff, and Residents.

	Control Mean(SD) or n (%)	Intervention Mean (SD) or n (%)	P a
Settings	n = 12	n = 12
For-profit	12 (100)	12 (100)	--
Years in operation	13.0 (13.2)	11.7 (8.7)	.65
Number of beds	97.6 (33.1)	85.8 (24.8)	.47
Occupancy rate	0.84 (0.16)	0.88 (0.10)	.48
Percent of residents with dementia diagnosis	50.6 (31.5)	51.7 (30.2)	.97
Percent of residents without dementia diagnosis but who have memory problems/appear disoriented	5.1 (7.9)	27.5 (26.2)	.010
Percent of residents receiving Medicaid	49.7 (29.3)	44.6 (33.3)	.71
Families	n = 258	n = 230
Age	57.7 (11.1)	59.2 (11.3)	.26
Female	188 (73)	169 (73)	.86
Married	175 (68)	179 (78)	.029
Minority race	71 (27)	47 (20)	.34
Relationship to resident
Spouse	10 (4)	24 (11)
Sibling (includes by marriage)	23 (9)	22 (10)	<.001
Child	162 (63)	151 (66)
Other relative/friend	63 (24)	32 (14)
Education: more than high school	194 (75)	173 (75)	.98
Employed	157 (61)	125 (54)	.30
General health: good to excellent	217 (84)	211 (92)	.041
Staff	n = 202	n = 195
Age	40.1 (13.2)	39.1 (12.8)	.70
Female	192 (95)	185 (95)	.98
Minority race	133 (66)	137 (71)	.76
Months working in this setting	48.5 (66.8)	35.6 (48.8)	.29
Hours worked in typical week	37.3 (6.4)	38.2 (6.3)	.49
Education: more than high school	120 (59)	107 (55)	.59
Staff role			.64
Nurse (RN/LPN)	18 (9)	9 (5)
Direct care worker (certified and not certified)	150 (74)	152 (78)
Other	34 (17)	34 (17)
Residents	n = 258	n = 230
Age	80.2 (12.8)	81.9 (10.0)	.24
Female	196 (76)	171 (74)	.79
Education: more than high school	80 (31)	69 (30)	.95
Dementia	151 (60)	170 (74)	.21
Bed mobility, some dependence	65 (25)	77 (33)	.022
Chairfast in last seven days	97 (37)	60 (26)	.017
Length of stay (months)	32.9 (37.5)	25.4 (28.5)	.09

^ap values for family, staff, and resident characteristics are adjusted for clustering using mixed models with a random effect for setting.

At baseline, as shown in Exhibit 2, families visited on average 10 (control) to 12 (intervention) times per month (p = .13) and were involved in grooming/appearance 3 (control) to 5 (intervention) times per month (p = .05), and resident mobility 7 times per month (p = .34). Families in control sites had more guilt at baseline as measured on the Family Perception of Caregiving Role instrument (2.39 versus 2.16 on a range of 1 to 5; p = .01); the five items in this scale include feeling guilty about interactions with the resident, feeling that he/she isn’t doing as much now and did not do as much in the past (two items) for the resident as he/she could or should, and feeling nervous/depressed or comfortable (two items) about interactions with the resident. There were no significant differences between control and intervention sites on any other outcome measures, including resident quality of life, which was rated by family members to average 3.5 (range of 1 to 5). Given the extensive amount of data that were collected for this project, distributions are shown only for those measures that evidenced change; distributions of measures not shown are available upon request.

Exhibit 2. Baseline Means for Outcome Measures: Families, Staff, and Residents.

	Control Mean (SD)	Intervention Mean (SD)	P a
Family involvement, number of times/month	(n = 258)	(n = 230)
Visit resident	9.9 (14.8)	12.3 (10.7)	.13
Call or write resident	11.5 (17.1)	11.7(18.5)	.93
Take resident out of NH or RC/AL setting	1.9 (2.6)	2.8 (4.0)	.12
Involved in resident eating	4.8 (6.6)	6.3 (7.7)	.30
Involved in resident grooming/appearance	3.3 (6.2)	5.3 (9.3)	.047
Involved in resident mobility	6.6 (8.7)	7.1 (10.7)	.34
Involved in resident activities/conversation	14.1 (13.9)	14.9 (13.7)	.59
Involved in outside activities (e.g., shopping)	4.0 (3.9)	4.7 (4.8)	.15
Attend group or sponsored activity with resident	0.43 (1.1)	0.56 (1.1)	.21
Do resident laundry	1.2 (2.5)	2.3 (4.1)	.10
Monitor resident finances	5.6 (9.2)	7.0 (9.6)	.57
Discuss resident with staff	4.6 (6.6)	6.2 (7.7)	.09
Participate in service/care plan meetings with staff	0.23 (0.9)	0.3 (0.7)	.77
Attend training, seminar	0.03 (0.2)	0.05 (0.2)	.42
Participate in groups specifically for families	0.13(0.4)	0.2 (0.7)	.25
Do things on behalf of setting (e.g., lead activities)	0.7 (3.1)	0.7 (2.4)	.90
Family measuresb (lower scores are favorable)	(n = 258)	(n = 230)
Family Perception of Caregiving Role (FPCR) - guilt subscale (range: 1 to 5)	2.39 (0.68)	2.16 (0.68)	.013
FPCR - conflict subscale (range: 1 to 5)	2.49 (0.62)	2.44 (0.63)	.46
Lawton Caregiving Appraisal - burden subscale (range: 10 to 50)	17.82 (6.54)	19.01 (6.73)	.12
Zarit Burden Interview (ZBI) (range: 0 to 88)	21.7 (12.2)	23.6 (12.2)	.22
ZBI - social subscale (range: 0 to 24)	2.87 (3.71)	3.60 (3.73)	.07
ZBI - psychological burden subscale (range: 0 to 24)	4.73 (4.09)	5.34 (4.17)	.31
Staff measuresb (higher scores are favorable)	(n = 202)	(n = 195)
Maslach Burnout Inventory - personal accomplishment subscale (range: 0 to 48)	40.62 (7.36)	39.44 (7.40)	.80
Attitudes Toward Families Checklist - families are partners subscale (range: 1 to 5)	3.96 (0.47)	3.91 (0.48)	.26
Family Empathy (range: 3 to 15)	8.63 (2.43)	8.45 (2.28)	.50
Resident measuresb (higher score is favorable)	(n = 258)	(n = 230)
Quality of life (single rating, by family) (range: 1 to 5)	3.50 (0.87)	3.52 (0.88)	.98

^ap values for family, staff, and resident outcomes at baseline are adjusted for clustering using mixed models with a random effect for setting.

^bFamily, staff, and resident measures are those for which significant effects were found. Other measures, for which significant effects were not found, were: for families, the FPCR - loss subscale; the Lawton Caregiving Appraisal - satisfaction and impact subscales; the ZBI - guilt subscale; the Family Perception of Care Tool and all subscales; the Interpersonal Conflict Scale; and the Center for Epidemiological Studies Depression Scale (CES-D); for staff, the Maslach Burnout Inventory - emotional exhaustion and depersonalization subscales; the Attitudes Toward Families Checklist - disruption and relevance subscales; the Staff Perceptions of Caregiving Role and all subscales; Family Behaviors; the Work Stressors Inventory; the Interpersonal Conflict Scale; the Hassles and Uplifts Scales; and the Center for Epidemiological Studies Depression Scale (CES-D); and for residents, Quality of Life-AD and the Patient Health Questionnaire (PHQ)-9 (rated by both family and staff); and the Pleasant Events Schedule (rated by family). Means and SDs for all measures are available upon request.

Families Matter Implementation

All families from the 12 intervention sites were invited to attend the workshop and participate in service planning; 166 (72%) and 163 (71%) participated in each, respectively. In reference to staff, 88 of the 195 invited staff participated in workshops (45% participation rate). The individualized service planning meeting was rarely attended by residents (only two attended) and not consistently attended by staff (97 attended). All told, families developed 306 service plans, which included a wide range of activities such as facilitating the establishment of a Red Hat Social Club (doing things), taking the resident out for car rides (getting around), giving the resident a manicure (looking good), and having a picnic (eating well). Fidelity interview data were obtained from 132 of the 163 families who had created a service plan (81%); results indicated an average dose of 0.6 (SD = 0.4) for family members, with 53 (23%) having a dose of 0 and 32 (14%) having a dose of more than 0.9.

Intervention Effects

Family involvement was assessed in 16 areas by both family and staff. Of these, adjusting for clustering and significant baseline differences including family marital status, relationship, and health, and resident mobility and chairfastness, and in both ITT and the full-dose conditions (i.e., families attended the workshop and their service plan was completely followed as planned), compared to those in control sites, staff in intervention sites reported a significant increase in family involvement with resident mobility. Specifically, the ITT and dose odds ratios and 95% confidence intervals (95% CI) were 1.86 (95% CI = 1.06-3.26) and 2.11 (95% CI = 1.21-3.72), indicating that the odds of a family member in the intervention group having higher involvement at follow-up were roughly twice the odds for a family member in the control group. Of note, this effect was evidenced only for staff report of family involvement in resident mobility and not based on family report. Staff reports of family involvement also indicated that the odds of a family member in the intervention group having higher involvement monitoring the resident’s finances at follow-up were lower (O.R. = 0.42; 95% CI = 0.19-0.88) than those of a family member in the control group.

Exhibit 3 presents the impact of a full dose for scales that evidenced a statistically significant intervention effect; bold indicates effects in the desired direction, and footnotes indicate the significance of the ITT analyses. Six family outcomes, three staff outcomes, and one resident outcome showed an intervention effect; all except two family outcomes were in the desired direction. The first family outcome—Family Perceptions of Caregiving Role guilt subscale (which has a range of 1 to 5)—evidenced a change of −0.06 points with no dose, whether it was the control group or the intervention group and a change for the intervention group under a condition of full dose of 0.14 (p = .027); given that lower scores indicate less guilt, families who received the full intervention reported more guilt at follow-up. A virtually identical change was noted for the Family Perceptions of Caregiving Role conflict scale (which also has a range of 1-5), with the control and no dose groups showing a change of −0.04 and the full dose intervention group showing a change of 0.13 (p = .024), indicating more conflict. The 10 items on this measure elicited information about having to be careful about making suggestions or requests about the resident’s care so that staff would not think the family is interfering; feeling like an outsider/not having control in the resident’s care/being allowed to approve care (three items); thinking staff ignore family directions/do not reach consensus on resident care (two items); thinking staff have the say about care/inability to control care (two items); and agreeing on what is important or trivial and rules/routines/efficiency versus individualized care (two items).

Exhibit 3. Outcome Measures with an Intervention Effect for Dose, Adjusted.^a.

	Test of effects (mixed modelb)				Model estimated means
	Effect	SE	p		Baseline	Follow-up	Changec
Family Outcomes – effect of family dose
Family Perceptions of Caregiving Role - guilt subscale (range 1-5; lower scores favorable)
Change for control group over time	−0.06	0.05	.190	Control Dose = 0	2.38	2.32	−0.06
Difference between groups at baseline	−0.27	0.09	.007	Intervention Dose = 0	2.10	2.04	−0.06
Intervention effectd	0.20	0.09	.027	Dose = 1	2.10	2.24	0.14
Family Perceptions of Caregiving Role - conflict subscale (range 1-5; lower scores favorable)
Change for control group over time	−0.04	0.04	.27	Control Dose = 0	2.46	2.42	−0.04
Difference between groups at baseline	−0.06	0.11	.59	Intervention Dose = 0	2.40	2.36	−0.04
Intervention effecte	0.17	0.07	.024	Dose = 1	2.40	2.53	0.13
Lawton Caregiving Appraisal - burden subscale (range 10-50; lower scores favorable)
Change for control group over time	−0.90	0.33	.007	Control Dose = 0	17.5	16.6	−0.9
Difference between groups at baseline	0.86	0.82	.31	Intervention Dose = 0	18.3	17.4	−0.9
Intervention effectf	−1.64	0.66	.014	Dose = 1	18.3	15.8	−2.5
Zarit Burden Interview - overall scale (range 0-88; lower scores favorable)
Change for control group over time	−3.36	0.63	<.001	Control Dose = 0	20.1	16.8	−3.3
Difference between groups at baseline	1.77	1.76	.32	Intervention Dose = 0	21.9	18.6	−3.3
Intervention effectg	−2.54	1.24	.042	Dose = 1	21.9	16.0	−5.9
Zarit Burden Interview - social subscale (range 0-24; lower scores favorable)
Change for control group over time	−0.81	0.20	<.001	Control Dose = 0	2.61	1.80	−0.81
Difference between groups at baseline	0.54	0.44	.23	Intervention Dose = 0	3.15	2.34	−0.81
Intervention effecth	−0.87	0.39	.028	Dose = 1	3.15	1.47	−1.68
Zarit Burden Interview - psychological burden subscale (range 0-24; lower scores favorable)
Change for control group over time	−0.51	0.22	.019	Control Dose = 0	4.42	3.91	−0.51
Difference between groups at baseline	0.53	0.70	.45	Intervention Dose = 0	4.96	4.45	−0.51
Intervention effecti	−1.03	0.43	.017	Dose = 1	4.96	3.42	−1.54
Staff Outcomes – effect of workshop attendance
Maslach Burnout - personal accomplishment subscale (range 0-48; higher scores favorable)
Change for control group over time	−0.21	0.53	.70	Control Dose = 0	39.9	39.7	−0.2
Difference between groups at baseline	−0.15	1.91	.94	Workshop No	40.0	39.8	−0.2
Intervention effectJ	2.40	1.10	.029	Yes	40.0	42.2	2.2
Attitudes Toward Families Checklist - partnership subscale (range 1-5; higher scores favorable)
Change for control group over time	−0.13	0.03	<.001	Control Dose = 0	3.95	3.82	−0.13
Difference between groups at baseline	−0.03	0.09	.75	Workshop No	3.92	3.79	−0.13
Intervention effectk	0.14	0.06	.028	Yes	3.92	3.94	0.02
Staff Outcomes – effect of family dosel
Family Empathy (range 3-15; higher scores favorable)
Change for control group over time	0.26	.16	.11	Control Dose = 0	8.45	8.70	0.25
Difference between groups at baseline	0.22	0.48	.65	Intervention Dose = 0	8.67	8.93	0.25
Intervention effectm	0.76	0.38	.048	Dose = 1	8.67	9.68	1.01
Resident Outcomes – effect of family dose
Overall Quality of Life (range 1-5; higher scores favorable)
Change for control group over time	−0.11	0.06	.08	Control Dose = 0	2.40	2.29	−0.11
Difference between groups at baseline	0.03	0.13	.82	Intervention Dose = 0	2.37	2.26	−0.11
Intervention effectn	0.33	0.14	.015	Dose = 1	2.37	2.59	0.22

^aResults adjust for differences that were significant at baseline. For families, these include health, relationship to resident, and marital status, as well as resident bed mobility and chairfastness, plus the setting-level proportion of residents with memory/orientation problems; for staff, these include the setting-level proportion of residents with memory/orientation problems; for residents, these include the same variables as used for families. Intervention effects indicate the effect of the intervention given the full dose (i.e., dose = 1). Effects shown in bold are favorable. Measures with non-significant intervention effects are listed in the footnote of Exhibit 2.

^bWith a random effect for setting and individual participant, either family member, staff member, or resident, as appropriate.

^cFollow-up mean - baseline mean.

^dIntention-to-treat intervention effect (0.17) is statistically significant (p = .026).

^eIntention-to-treat intervention effect (0.11) is not statistically significant (p = .08).

^fIntention-to-treat intervention effect (−0.80) is not statistically significant (p = .14).

^gIntention-to-treat intervention effect (−1.51) is not statistically significant (p = .14).

^hIntention-to-treat intervention effect (−0.65) is not statistically significant (p = .18).

ⁱIntention-to-treat intervention effect (−0.58) is not statistically significant (p = .10).

^JIntention-to-treat intervention effect (3.39) is statistically significant (p < .001).

^kIntention-to-treat intervention effect (0.20) is statistically significant (p = .001).

^lIntention-to-treat intervention effect for Staff Perceptions of Caregiving Role - dominance subscale (−0.12) is statistically significant (p = .016). This is the only instance when intention-to-treat analyses were significant when the dose effect was not significant.

^mIntention-to-treat intervention effect (0.38) is not statistically significant (p = .16).

ⁿIntention-to-treat intervention effect (−0.06) is not statistically significant (p = .56).

All other results were in the desired direction. Families receiving the full dose of Families Matter reported less burden on the Lawton Caregiving Appraisal burden scale (−2.5 on a scale of 10 to 50), the Zarit Burden Interview overall scale (−5.9 on a scale of 0 to 88) and two of its subscales: social (−1.68) and psychological burden (−1.54), both on a scale of 0 to 24. Staff who attended the workshop reported less burnout (a higher score) on the Maslach Burnout personal accomplishment subscale and better attitudes on the Attitudes Toward Families Checklist partnership scale (2.2 on a scale of 0 to 48 and 0.01 on a scale of 1 to 5, respectively). Also, staff reported higher family empathy (1.02 on a scale of 3 to 15) when paired with residents whose family members experienced the full dose of the intervention. Finally, families in the intervention group reported improved resident quality of life as rated on the single item measure (0.22 on a scale of 1 to 5 with a full dose).

DISCUSSION

In an effort to help families identify and implement roles for themselves in the LTC setting that may benefit resident quality of life, this study found that doing so increased family guilt and sense of conflict but also decreased their burden; further, it improved their perceptions of resident quality of life. Staff benefits were uniformly positive, related to less burnout and more of a sense of working in partnership with families and that families were more empathic. Staff also perceived a change in the nature of family involvement but families themselves did not.

In putting these findings in context, it must first be noted that all effects were modest, in the small-medium effect size range (Cohen, 1988), with the largest being a standardized effect size of 0.38 for resident quality of life, as rated by the family. For other statistically significant outcomes, the average effect size was 0.25 for family and 0.31 for staff. Of course, it is not possible to know whether resident quality of life actually increased more than did family and staff outcomes because these results may reflect the family’s perception that their actions benefitted the resident. Also, intervention effects were not found in many areas, including depressive symptoms among family, staff, and residents, family perceptions of care, staff perceptions of caregiving role, resident activity involvement, and some subscales under study.

Effect size notwithstanding, a few interesting implications emerge. When increasing family involvement in LTC settings, staff members see most evidence of it in relation to resident mobility and monitoring finances (i.e., such was not true of the other 15 areas of family involvement listed in Exhibit 2). Of note, service plans related to “getting around” (resident mobility and physical functioning) were the second most frequent type of plan created and accounted for only 34 (11%) of the 306 service plan activities. The most frequent type of activity represented in service plans was related to “doing things” (recreation) and accounted for 202 (66%) of the service plan activities. Thus, it appears that staff members are more likely to notice a change in family involvement in resident mobility than in more recreational-based activities, perhaps because family involvement in this area is less common than it is in recreational activities, or perhaps because it, as opposed to recreational involvement, lessens staff burden. To the extent that resident mobility is important for function, social engagement, and quality of life (Bourret, Bernick, Cott, & Kontos, 2002) and is a contribution to care visible to the staff, it may be an area especially fruitful for family involvement. Similarly, staff (but not families) perceived a decrease in family involvement in monitoring finances. Other work has suggested that monitoring finances creates burden (Port et al., 2005), and it is plausible that the change in which families spent their time was in part responsible for their decreased burden.

The fact that family involvement in Families Matter decreased caregiving burden suggests that the tasks they identified were not taxing. The price, however, was that families felt more guilt and conflict. Indeed, identifying actionable roles for family may foster a greater sense of responsibility and perhaps a realization that they have not been doing as much as they could or should. Despite the fact that guilt compels people to act (Carlsmith & Gross, 1969; Taylor, 1991), there is need to temper this increased sense of obligation with support from staff. Over time, it is possible that guilt may lessen, but this has yet to be determined.

Increasing family involvement also increases the potential for conflict with staff. Our measure of family conflict included items about receptivity of staff to family instructions for care. It is possible that perceptions of conflict increased primarily because Families Matter necessitated more family-staff coordination. If so, this finding suggests that any efforts to increase family involvement in LTC must concurrently attend to potential conflict.

The need for family to coordinate their role with the staff speaks to the partnership that must be effected if family involvement is to be achieved, and in the context of staff perceptions, it indeed seems beneficial: staff felt less burnout, and more partnered with, and empathy from, families. It is possible that the task-natured focus of Families Matter was responsible for these benefits, as families became care partners in ways the staff appreciated (e.g., preparing special foods) (Natan, 2009).

Considering the results and their implications, the workforce shortages in LTC, and the presence of family members who want to be involved, it is important to consider next steps for Families Matter and related efforts. One step relates to how to implement such a program. In Families Matter, a train-the-trainer approach was initially envisioned, wherein the research staff would train a LTC staff member to coordinate and oversee service plan meetings with families, staff, and residents. Despite strong support from administrative leadership for Families Matter, this plan did not materialize; instead, the research team coordinated and oversaw these meetings and few staff ever attended them. Further, fidelity data indicate that 55% of staff did not attend workshops, one-quarter of families did not participate in workshops or service plan meetings, and overall, their fidelity was slightly more than half of what was optimal (i.e., the average dose was 0.6 on a scale of 0 to 1). Consequently, if a LTC setting is to promote family involvement, doing so must either become a component of an individual’s work responsibilities for which he/she is actually supported or tasked to a contract worker or consultant.

Indeed, implementing any new care practice in a LTC setting is a challenging process. The extent to which change is adopted relates to a host of factors, including the nature of the innovation itself (e.g., the extent to which it is seen as an advantage); antecedents for change (e.g., leadership support); readiness for change (e.g., dedicated resources to implement the new practice); linkages to promote change (e.g., the extent to which the agent for change is part of the system); the ease of assimilation (e.g., the extent to which the process of change is complex); communication and influence (e.g., the presence of champions to implement the change); the outer context (e.g., mandates for change from ownership); characteristics of those implementing the change (e.g., the skills of the staff); and the implementation process itself (e.g., the training needs of the staff) (Greenhalgh, Robert, Macfarlane, Bate, & Oliva, 2004). Thus, the modest fidelity achieved in this project and the uncertainty regarding its sustainability is not unexpected.

CONCLUSION

It is advantageous to streamline new care practices to the extent possible. Families Matter is a multicomponent intervention, and the question can be raised of whether it may be simplified by omitting either the workshop or service plan component. An in-depth examination of this matter found that both components related to decreased family burden: service plan creation related to reduced burden as measured by both the Zarit and Lawton measures, and workshop attendance related to reduced burden as measured by the Lawton measure (Washington et al., in press). Consequently, if one were to implement a more parsimonious intervention, it may be advisable to omit the workshop component.

A change addressing family relations with staff must be sensitive to the complex nature of interpersonal dynamics, which extends far beyond the focus on family involvement addressed in this project. A more thorough understanding of this topic would consider issues that pre-date admission to the LTC setting, such as caregiver stress and the quality of family members’ relationship with the LTC resident, the policies of the LTC settings such as the extent to which they are “permeable” to family involvement, the residents’ need for support, the proximity of the family to the LTC setting, and many other considerations (Gaugler, 2005). That said, the results from this project indicate that families do matter in long-term care, that there are important roles they can play to potentially improve resident quality of life, and that their involvement may actually decrease their sense of burden and improve their relations with the staff.