Sage Journals: Discover world-class research

Abstract

We aimed to develop and test the psychometric properties of a new instrument to assess long-term care (LTC) home residents’ social connection. We developed a conceptual model and then used literature review and qualitative interviews to identify candidate items for the Social Connection in Long-Term Care Residents (SONNET) scale, which we refined following pilot and field testing and with patient and public involvement. We evaluated the final instrument’s feasibility, acceptability, reliability, and validity in 52 LTC resident-staff dyads in the UK and Canada. The 12-item SONNET scale comprises subdomains assessing social engagement and social connectedness, with versions to be rated by a resident or proxy (staff) informant. The SONNET scale is feasible and acceptable, and preliminary testing indicates its reliability and validity. It is available for use to examine social connection in LTC residents. Further testing in diverse samples and confirmation of psychometric properties is needed.

Keywords

measurement nursing homes loneliness dementia social engagement social connectedness long-term care

What This Paper Adds

• We developed a new instrument to measure social connection for LTC residents, guided by gold-standard psychometric methods and with methodological considerations for people with dementia.

• Through two-stage testing, we established its feasibility and acceptability, content and construct validity (structural validity and hypothesis testing), and reliability (internal consistency, test–retest, and inter-rater reliability).

Applications of Study Findings

• Our new measure is freely available for researchers and LTC settings to test the effects of interventions at the individual, home, and system levels.

• Addressing social connection has the potential to advance care and improve the health and quality of life of LTC residents.

Introduction

Social connection is an umbrella term encompassing the structure, functions, and quality of social relationships (Holt-Lunstad, 2021). Social connection is a core element of human experience, a basic need, and a key determinant of health (Thoits, 2011). Poor social connection has been linked to adverse health outcomes, including increased risk of mortality (Holt-Lunstad et al., 2015) and dementia (Sommerlad et al., 2023). Social connection may be particularly important for long-term care (LTC) home residents, most of whom have dementia (Stewart et al., 2014) and live apart from their previous social networks. In this setting, social connection is an important determinant of quality of life (O’rourke et al., 2015) and a marker of quality of care (Sion et al., 2020). Supporting social connection is a dementia care practice recommendation (Calkins, 2018) and important concept for person-centered care (Corazzini et al., 2019).

Evidence on effective interventions for improving social connection in LTC residents is sparse (Brimelow & Wollin, 2017), partly due to conceptual ambiguities (Liougas et al., 2024) and measurement difficulties (Chapman et al., 2024). LTC residents and homes have distinct considerations that must be reflected in a measure of social connection. Most residents are older adults with complex health needs, including cognitive, sensory, or mobility impairments. Within LTC homes, residents share space, take part in congregate activities, and receive daily care from staff. LTC homes are often separated from community activities, although families, volunteers, and other service providers may have important roles. Although over 30 measures of social connection have been tested with LTC residents, when evaluated against established criteria for evidence of psychometric properties, none are currently recommended for use because of lack of evidence of content validity, due to failing to demonstrate the relevance, comprehensiveness, and comprehensibility of measure items (Dewan et al., 2024).

We, therefore, aimed to develop a psychometrically robust, person-centered measure of social connection for residents in LTC homes, applicable to residents across the spectrum of cognitive impairment and assessing key domains of social connection. A measure of social connection in LTC residents, with acceptable validity and reliability, would facilitate rigorous study of determinants of social connection as well as the evaluation of interventions to improve social connection.

Methods

We used gold-standard psychometric methods (Streiner et al., 2024) and methodological considerations for people with dementia (Smith et al., 2005) to design a measure that could be used in settings that align with the international definition of nursing home (Sanford et al., 2015) and with all residents, irrespective of cognitive impairment (Hughes et al., 2021). We used the COnsensus-based Standards for the selection of health Measurement Instruments (COSMIN) guidelines (Mokkink et al., 2010) and the Person-Centered Measure Evaluation Tool (PC-MET) (Zimmerman et al., 2023) to guide our work. We followed a six-step process outlined by De Vet et al. (de Vet, 2011), adapted for concept elicitation, and which we summarize in Figure 1.

Figure 1.

Overview of six-step study process

Conceptual Model

Social connection is acknowledged to comprise multiple concepts (Holt-Lunstad, 2021) and thus presents implications for a multidimensional measurement. Yet, the inter-relationships between the aspects of social connection have not, to our knowledge, been articulated in a unifying conceptual model. Hence, we developed a conceptual model to summarize key aspects of social connection, and how they are related to each other, including along gradients as observable and experienced characteristics (Supplemental Material 1.1). We did this to also guide the development of our measure to focus on the most important aspects of social connection for LTC homes as identified by LTC residents and their relatives. We adapted a model (Berkman et al., 2000) by incorporating aspects of social connection highlighted for LTC home research (Leedahl et al., 2018) and iteratively refining it based on our literature review, qualitative interviews, patient and public involvement (PPI) input, and discussions with our multidisciplinary team.

Qualitative Interviews

Participants and Setting

We completed interviews, in Canada and the UK, with LTC residents, staff and clinicians, and family and friends of residents. We purposively sampled using the following criteria:

- LTC residents: aged 65+, had no active severe mental illness, and had mental capacity to consent to participation.

- Staff and clinicians: aged 18+ and who worked at an LTC home or had during the previous 2 years.

- Family and friends: aged 18+ and who visited an LTC resident at least monthly.

In the UK, we recruited participants from National Health Service (NHS) care home in-reach mental health teams from three NHS trusts, via newsletters from ENRICH, the national care home research network, and from LTC homes directly. In Canada, we recruited participants through organizations from across Ontario representing LTC residents, families, and staff and through LTC homes.

Procedures

We obtained informed consent from prospective participants and then conducted individual or dyadic interviews (for residents, family, and friends) and individual or dyadic interviews or focus groups (for staff and clinicians), in person or remotely, between May 2022 and June 2023. We followed pre-piloted interview guides (Supplemental Material 1.2a & b) derived from our conceptual model, previous study findings, and existing measures of social connection to explore views on the most important elements of social connection and how it should be measured. We audio-recorded and transcribed interviews/focus groups using MS Teams for both in-person and remote interviews and then anonymized and checked them for accuracy. We stopped recruitment when we judged that we had reached theoretical sufficiency (Saunders et al., 2018). We compensated participants with a gift voucher (£20 in the UK and $35 in Canada).

Analysis

Our analytic method, based on thematic analysis (Braun & Clarke, 2006), is described in detail elsewhere (Chapman et al., 2024). Using Dedoose 9.0.90 software (SocioCultural Research Consultants, 2023), five team members developed a codebook from a subsample of five interview transcripts. We then applied the codebook to four previously unread interviews whereby two members of the research team independently coded each transcript. We discussed the code book throughout and once inter-rater reliability was moderate or higher (Cohen’s kappa >0.60), one researcher coded all subsequent transcripts with regular research team meetings to discuss uncertainties in interview content or meaning thereby ensuring consistency throughout the coding process.

Item Generation

We used two approaches to generate candidate items. First, we extracted quotes from our qualitative interviews indicating positive and negative aspects of social connection, aligned with our conceptual model, and developed them into items. Second, we selected candidate items from a subset of scales with high or sufficient evidence of content validity and at least two measurement properties categorized as having adequately low risk of bias (Dewan et al., 2024): Laurens Well-Being Inventory for Gerontopsychiatry (van der Wolf et al., 2018), Lubben Social Network Scale Revised (Munn et al., 2018), QUALIDEM (Ettema et al., 2007), and the Revised Index of Social Engagement (Gerritsen et al., 2008). We further considered the UCLA three-item Loneliness Scale’s items (Russell, 1996), as the most widely used loneliness scale.

We shared the draft items as a survey with the study team (comprising nursing, psychiatry, epidemiology, rehabilitation, and psychology expertise), asking for them to indicate whether each item should be retained, rejected, or modified based on its relevance, comprehensibility, and contribution to a comprehensive measure and to nominate other items not covered by the long list. The survey also allowed for feedback, either on the content or form of the items, including any suggestions for modified wording. We then shared our preliminary version with our PPI groups in Canada and the UK for their feedback.

Pilot Testing

We purposively recruited residents and staff from the same settings as stage 2 for cognitive interviews. We followed recommendations for conducting cognitive interviews to ensure comprehension and comprehensiveness (Patrick et al., 2011), and we developed an interview guide (Supplemental Material 1.3a & b) to explore how respondents understood instructions, recall period, candidate items, and response options using the “verbal probing” method. We audio-recorded and transcribed interviews using MS Teams and then anonymized and checked them for accuracy. The two lead researchers and two research assistants iteratively amended the instructions, items, and response options as interviews were conducted, enabling us to pilot test the refined SONNET scale in subsequent interviews.

Initial Field Testing and Scale Refinement

Participants and Setting

We tested the scale with LTC resident–staff proxy dyads. We recruited dyads to test both the resident- and proxy-report versions and because staff provided some additional data about residents. Eligibility criteria for LTC were being aged 65+ and having adequate English language proficiency. Staff members were required to be aged 18+, have adequate English language proficiency, and have cared for the dyadic resident at least once weekly over the past month.

In Canada, we recruited participants from five LTC homes in Ontario, each with over 100 residents and located within 150 km of Toronto. In the UK, we recruited residents from the same sources as stage 2 and from four additional NHS research networks.

Procedures

We posted the research opportunity through an advert whereby potential participants could contact the study team directly. We also approached LTC home managers to identify potentially eligible resident participants. In the UK, we also used electronic health records to identify eligible resident participants who had previously consented to being contacted about research. For each resident, LTC homes identified potential staff participants. A trained member of the research team assessed mental capacity to consent against UK Mental Capacity Legislation or, in Canada, with an evaluation to sign consent (Resnick et al., 2007). Resident participants either gave informed consent for themselves, or, for those who lacked capacity, we sought to discuss the research study with the nearest relative or a nominated consultee. Once the consultee provided consent, the study was discussed with the LTC resident and assent obtained before administering questionnaires.

We interviewed LTC residents and staff separately, in-person. They were first asked for personal and demographic information (e.g., age, sex, and length of time living or working in an LTC, and whether the resident had a recorded diagnosis of dementia). Questions from the SONNET scale were then administered. A prompt sheet was provided to facilitate ordinal response options. Instructions to record binary (yes/no) responses or lack of response were also included. For residents with severe cognitive impairment, we corroborated personal and demographic information with staff.

Analysis

We aimed to recruit 100 dyads, estimated as five dyads per scale item (Rouquette & Falissard, 2011). Using Stata 16.1 (StataCorp LLC, 2020), we described the participants and their SONNET item responses using descriptive statistics (e.g., frequencies, proportions, ranges, and missingness)—as there was little missing data, we used complete case analysis in these descriptive analyses and full information maximum likelihood in the factor analyses. We examined dimensionality by conducting an exploratory factor analysis which used the principal factors method without rotation applied to the polychoric correlation matrix of ordinal item responses from the data. Items with low loadings below 0.3 on the general social connection factor or cross-loadings outside of domains were noted. We examined internal consistency by calculating inter-item and item-total correlations as well as Cronbach’s alpha. We conducted Horn’s Parallel Analysis (Dinno, 2009), using Stata’s paran command which implements Horn’s technique to contrast eigenvalues produced through a factor analysis on a random dataset with the same number of variables and observations as the observational dataset. We then reduced the items based on Horn’s Parallel Analysis results, missingness, response range, factor loading, and in consultation with the study team and PPI collaborators.

Final Psychometric Evaluation

Participants and Setting

We recruited participants from the same settings and with the same eligibility criteria as Stage 5. We purposively sampled participants aiming for two-thirds of residents with dementia and a range of demographic characteristics for both residents and staff.

Procedures

We followed the same procedures as stage 5, but using the final SONNET scale version and recording the time taken to complete the scale. At the same time as we collected the SONNET scale ratings, we also collected additional data from staff about the resident, namely:

(1) Revised Index of Social Engagement (RISE) (Gerritsen et al., 2008);

(2) Modified Clinical Dementia Rating (CDR) scale; this version has been tested and validated in people without dementia, those with mild cognitive impairment, and those with dementia (Duara et al., 2010). We used the total CDR score in our analyses and the “community affairs” subscale;

(3) Katz Index of Independence in activities of Daily Living (Katz et al., 1963);

(4) Neuropsychiatric Inventory (NPI) (Kaufer et al., 2000);

(5) Quality of Life in Alzheimer’s Disease scale, Nursing Home Version (QOL-AD-NH) (Dichter et al., 2018). A social subscale has previously been used, comprising four questions (i.e., living situation and relationships with each of family members, staff, and friends).

When available, a second researcher administered the SONNET scale (i.e., evaluating inter-rater reliability). When feasible, the researcher returned to participants 2 weeks later (+/− 1 week) to repeat the SONNET scale (i.e., evaluating test–retest reliability). All participants were provided with a £20/$35 gift voucher.

Analysis

We aimed to recruit 50 dyads (de Vet, 2011), providing 90% power at a significance level of 5% to find correlation of 0.42 between the SONNET and other measures, and adhering to the commonly used rule of thumb for factor analysis sample size calculation (Rouquette & Falissard, 2011). We used Stata 16.1 for statistical analysis. We first described the participant characteristics and their responses to the SONNET scale using descriptive statistics (e.g., frequencies, proportions, ranges, and missingness). We described feasibility (i.e., whether or not respondents are able to complete the scale) and acceptability (i.e., whether or not respondents are willing to complete the scale) with completion rates, item missingness, and mean time for completion. We conducted subsequent analyses using complete cases.

We assessed measurement properties for the SONNET resident- and proxy-report versions, namely:

• Construct validity (structural validity): We performed a confirmatory factor analysis, using full information maximum likelihood. To test the scale against our hypothesized bifactor structure, we used the comparative fit index, χ² statistic (with non-significant p-values indicating exact fit), coefficient of determination (CD, with values 0.08 or smaller indicating close fit), and Root Mean Square Error of Approximation (RMSEA, with values 0.06 or smaller indicating close fit) (Hu & Bentler, 1999). All subsequent analyses were based on the emergent subscales whereby we calculated subscale scores (as the sum of the items within the subscale).

• Construct validity (hypothesis testing): We calculated the Pearson’s correlation coefficient to estimate and characterize (Evans, 1996) the relationship between the SONNET subscales and SONNET summary question 12 and other measures (i.e., RISE; CDR total and community affairs domain; Katz index; NPI; and QOL-AD NH total and social subscale). We hypothesized that the SONNET subscales would be positively correlated with the SONNET question 12, RISE, QOL-AD NH, and Katz scores and inversely correlated with CDR and NPI.

• Reliability (internal consistency): We calculated and described (Morera & Stokes, 2016) Cronbach’s alpha for the subscales.

• Reliability (test–retest and inter-rater): We calculated Pearson’s correlation coefficients to estimate the association between the SONNET subscale totals from the two time points and from the two researchers (Cicchetti & Sparrow, 1981), respectively.

In sensitivity analyses, for the resident-report version, we categorized residents by level of cognitive impairment (based on the median CDR score) and examined measurement properties in stratified analyses.

Patient and Public Involvement/Engagement

We worked with PPI groups in Canada (Engagement of People with Lived Experience of Dementia (EPLED)) (Snowball et al., 2022) and the UK (Alzheimer’s Society Research Network), which included people with lived experience of dementia and LTC. The PPI group members provided feedback on and piloted the qualitative interview guide (step 2), assisted with interpreting findings to focus scale development on social engagement and social connectedness, refined and reduced the items (step 3), and offered insights on the measure (e.g., wording and format) (steps 4–6). We consulted PPI collaborators via online meetings, whereby members from Canada and the UK joined together, and email for ongoing input.

Results

Conceptual Model

Our conceptual model (Supplemental Material 1.1) includes components described along a continuum from more directly observable (e.g., social networks, social engagement and their converse, and social isolation) to constructs which reflect individuals’ experiences (e.g., social connectedness and its converse loneliness).

Qualitative Interviews

We interviewed 18 LTC residents, 32 friends and family, and 17 staff and clinicians (Table 1), and they described key elements of social connection (Table 2). We thus focused on social engagement and social connectedness domains, from our conceptual model, in developing the multidimensional SONNET scale. Interviews also identified key pragmatic considerations, including brevity, recency of recall period (i.e., 1 week), limited range of response options, and versions that might be completed by either residents or proxies and administered by an interviewer to accommodate residents with varying cognitive ability. Our PPI groups also interpreted the interviews in the same way.

Table 1.

Characteristics of Qualitative Interview Participants

	Residents (n = 18)	Family and friends (n = 32)	Staff and clinicians (n = 17)
Gender—n (%)
Female	13 (72)	25 (78)	16 (94)
Age (years)
Mean	82	64	40
Range	69–99	40–85	24–60
Marital status—n (%)
Single	2 (11)	5 (16)
Married	3 (17)	21 (66)
Common-law	0 (0)	2 (6)
Separated	1 (5)	1 (3)
Divorced	3 (17)	1 (3)
Widowed	9 (50)	1 (3)
Other	0 (0)	1 (3)
Employment status—n (%)
Employed	0 (0)	10 (30)
Retired	17 (94)	13 (41)
Unemployed	1 (6)	3 (9)
Other	0 (0)	6 (19)
Race—n (%) Canadian participants n = 30
East/Southeast Asian	0 (0)	0 (0)	1 (25)
White	5 (100)	21 (100)	2 (50)
Other	0 (0)	0 (0)	1 (25)
Race—n (%) UK participants n = 37
Asian	0 (0)	0 (0)	2 (15)
Mixed ethnicity	0 (0)	1 (10)	0 (0)
White	10 (77)	10 (90)	11 (85)
Other ethnic group	2 (15)	0 (0)	0 (0)
Prefer not to disclose	1 (8)	0 (0)	0 (0)
Education—n (%)
Secondary/high school or less	7 (38)	1 (3)	1 (6)
Degree	4 (21)	11 (34)	8 (47)
Postgraduate	1 (7)	9 (28)	6 (35)
Other	1 (7)	8 (25)	2 (12)
Unknown	5 (27)	3 (10)	0 (0)
Diagnosis of dementia of resident—n (%)
Yes	9 (50)	26 (81)
No	9 (50)	6 (19)
Length of time residing in care home of resident—n (%)
Less than 1 year	5 (28)	9 (28)
1–5 years	9 (50)	15 (47)
More than 5 years	4 (22)	7 (22)
Not disclosed	0 (0)	1 (3)
Relationship to resident—n (%)
Spouse		7 (22)
Child		22 (69)
Sibling		1 (3)
Friend		1 (3)
Other		1 (3)
Role—n (%)
Doctor/physician			3 (18)
Recreation therapist/activity worker			3 (18)
Personal support worker/care worker			2 (12)
Clinical psychologist			1 (6)
Care home manager/administrator			5 (28)
Nurse			2 (12)
Social worker			1 (6)
Years of experience—n (%)
Less than 5 years			5 (29)
5–10 years			5 (29)
More than 10 years			7 (42)
Working pattern—n (%)* (n = 13)
Full time			11 (85)
Part time			2 (15)
Shift pattern—n (%)* (n = 13)
Days only			10 (80)
Days and nights			3 (20)

Table 2.

Themes and Quotes From Qualitative Interviews With People With Dementia, Carers, and Expert Healthcare Professionals For Development and Testing of SONNET Instruments

Category	Theme	Exemplar quote
Social engagement	Relationships with other residents	“We eat together every day and we got friendly, that’s all. And we have a lot in common. We have been to the same places and we had all these funny little coincidences. No, it just developed here. Don’t forget we’re stuck here, aren’t we?” (83-year-old female resident without dementia, UK)
	Relationships with other residents	“Yeah, relationships with residents were... are important. Developed amongst people for whom it’s critically important” (65-year-old daughter of a resident with dementia, Canada)
	Relationships with staff members	“Certain members of the staff obviously are easier to get on with than others. But I’ve got on very well, I think with them all. I’ve sized them up. I know how to behave with whom” (87-year-old female resident without dementia, UK)
	Relationships with staff members	“There are certain staff that you know she enjoys seeing, and she’s had some support workers over the years that she’s been there, that she’s really quite liked. There’s a recreation person in there who she described to me as ‘the young man who comes and talks to me about subjects I’m not interested in’. But ... I think she was fine to chat with him. And, you know, it’s a little bit of variety in her day” (63-year-old daughter of a resident without dementia, Canada)
	Socializing at mealtimes with other residents	“Well I’ve got three people that I have my meals with, and one of them is an author and I found her fascinating” (76-year-old male resident without dementia, UK)
	Socializing at mealtimes with other residents	“It can be quite daunting because if you’re new and you’ve been in your home on your own for how many years, and then you’d walk into a lounge and there’s 12, 13 people just sat there staring, thinking ‘who are you’. And then they end up making friendships, and then they sit around the dining room table together and eat their lunch, and go into each other’s rooms and stuff like that” (35-year-old female care home manager, UK)
	Activities with other residents, staff, or volunteers in the care home	“I guess a lot of care homes have communal areas and I guess they often like to bring their residents to communal areas. But I guess those people can still be quite socially disconnected while in those communal areas. And it really depends then on I guess those activity coordinators, and the staff to maybe make them a bit more socially connected while they are there” (35-year-old female nurse, UK)
		“Oh, we do everything. Like painting and everything. We do a lot with [staff member]. Yeah, oh she’s great. Then we do our exercises, forget what they’re called, they’re on the television. [Staff member]’s good with activities, she definitely is” (91-year-old female resident with dementia, UK)
	Activities outside the care home	“We do lots of outings. So, we go out into the community, and there they can interact with other people; they interact with their group of people when they go out to like restaurants and cafes and ice cream parlors and stuff” (28-year-old female care home staff member, Canada)
	Activities outside the care home	“They would have outings that you could sign the resident up for, so I would sign my mom up for whatever was going on. Like a trip to (city) to see the Christmas lights. Once a month or twice a month, they would go to a nearby restaurant, and they had like a little bus and they could take up to eight residents. So, I would sign her up for all of those things” (63-year-old daughter of a resident with dementia, Canada)
	Visited by family or friends from outside the care home	“She wants me there all the time. Which... I can’t, I’m getting... I was getting burned out going everyday, so I tried to switch it to every other day, but then because of what I’m seeing happen more recently, I’m now feeling I need to go every day” (52-year-old daughter of a resident with dementia, Canada)
	Visited by family or friends from outside the care home	“I have three grown up children, two lovely grandchildren and they live within an environment. But whereas I was in the flat of my own the last year, you know, and they always thought one of them should come, I feel now they got their own lives. I like to see that too, and if I have any problem, they’ll be there and they always come one of them regularly” (84-year-old female resident without dementia, UK)
	Visiting family or friends outside the care home	“Sometimes they take me for dinner, I love it. Go back to their homes because then I’m in a normal ordinary environment and that’s what I love. I should think everybody here would say the same though some people, they don’t have anybody to come” (84-year-old female resident without dementia, UK)
	Visiting family or friends outside the care home	“She used to come and stay at my place and we would go out. We would go out for meals and we had quite a nice routine and we would stay at my place a couple of nights a week. And we would go out Friday night for a meal and we used to go Friday, Saturday morning and we’d go out for coffees and then she’d come back here” (74-year-old male partner of a resident without dementia, UK)
	Speaking to contacts from outside the care home on phone or other technology	“She communicates much better in writing than on the telephone. Sometimes on the phone she loses some words and uh, but as I said since moving to long-term care... whether it’s the setup of her computer or just her gradual decline from aging and frailty. She’s just not keen on walking across the room and sitting at the computer. It’s not that she never does it, but... she almost hardly ever would, would send an email. So, I now phone her every evening” (63-year-old daughter of a resident without dementia, Canada)
		“Every single floor has a tablet. So, and all the staff are trained to call family members if they need to and video call as well. So I think that’s been really helpful for those families that live far away” (28-year-old female care home staff member, Canada)
Social connectedness	Feeling accepted by other residents	“I think that is most important and that they feel that they belong, and they’re loved and cared for here” (28-year-old female care home staff member, Canada)
	Feeling accepted by other residents	“People get frailer, they’ll probably can’t communicate as well. But that seems to be accepted. That seems to be just part of it. We’re all in the same sort of boat type of a feeling” (59-year-old female care home manager, UK)
	Having things in common with other residents	“OK well this woman really enjoyed gardening her entire life and this gentleman really enjoys gardening and so if we kind of invite them both to a gardening group or, you know, maybe perhaps that would kind of begin a friendship” (31-year-old female activity worker, UK)
	Having things in common with other residents	“We pair people up with the volunteers as well. We have a coordinator. So if there’s people who just want letters writing or a book reading or newspapers or things like that, then we pair people up. If there’s somebody that wants to do gardening, then again, they’ll come and help with the group activities” (59-year-old female care home manager, UK)
	Feeling left out	“With residents, yeah, I mean, I’m… I wouldn’t say I’ve made very close friends” (87-year-old female resident without dementia, UK)
		“You can’t just build relationships because when you’re young and you build relationships, you give and take more. When you get to a certain age, people are very set in their ways and not maybe willing to give. Therefore … you’ll be friendly, but you’ll never have the same” (84-year-old female resident without dementia, UK)
		“Well, some people wouldn’t accept what I say, put it like that. I think that’s correct. I can tell him a spade is a spade and there is no come back, whereas other people there might be come back, they might object to what I said” (77-year-old male resident without dementia, UK)
	Loneliness	“We had one resident who was a bit, he was clearly a lonely but he also liked being on his own” (31-year-old female activity worker, UK)
		“She just doesn’t really know if there’s somebody there or not [due to vision impairment]. So, it’s kind of lonely unless somebody actually comes up and talks to her” (79-year-old daughter of a resident without dementia, Canada)
		“One of the biggest challenges in terms of long-term care or congregate living is... that people are surrounded... so residents are constantly surrounded with others and there is an assumption that they aren’t sort of deeply lonely or deeply alone. And I think that if we can assess that well and address that in a more effective way that I think that we would do very well to really get a clearer understanding of the resident experience” (50-year-old female recreation therapist, Canada)
	Being at ease with others	“It’s really about being present with that person in the most truest form of it, you know” (31-year-old female activity worker, UK)
	Being at ease with others	“And I seem to remember that actually soon after she got there, she was already referring to people as being new friends. So there was clearly… You know, something was clicking and the staff seemed to have a quite a good sense of which people are going to get along, which is good as well because they don’t always” (55-year-old daughter of a resident with dementia, UK)
	Staff and other residents knowing the individual	“Between the turnover and the staffing challenges, I don’t know that there’s anybody who I could say knows her well. I mean, certainly she notices and tells me about what happens when there’s nursing staff who don’t know her at all and they make mistakes, and they make assumptions and sometimes those include medication and it’s very frustrating” (63-year-old daughter of a resident without dementia, Canada)
	Staff and other residents knowing the individual	“When they come in, we try our best to get a life history. So we either work with the resident themselves or we work with relatives to build up that life history so that when they join us, we kind of at least have a base of what kind of things they enjoy doing and what kind of like, you know, what person, what person are they, who are they?” (31-year-old female activity worker, UK)

Item Generation

We developed 58 items (44 from our qualitative interviews and 14 from existing measures). We edited and reduced this to 30 items based on study team feedback and then to 20 items based on PPI group feedback.

Pilot Testing

We pilot tested the 20-item SONNET scale with five LTC residents and four staff (Supplemental Material 1.4). We refined the instructions, items, and response options based on input into the relevance and comprehensibility of items, and the comprehensiveness of the scale.

Initial Field Testing and Scale Refinement

We field tested the 20-item SONNET scale in 111 resident–staff dyads (Table 3 and Supplemental Material 1.5). Exploratory factor analysis provided evidence that both the resident-reported and proxy-reported scales had adequate goodness-of-fit. Horn’s Parallel Analysis justified no more than two factors. Based on missingness, range of responses, factor structure, and interviewer feedback, we further refined and reduced the SONNET scale to a 12-item version. Both the self- and proxy-reported versions presented 11 equivalent (but rephrased) items and one unscored summary question, all with ordinal response options. Questions 1–4 and 7–9 covered social engagement and questions 5, 6, 10, and 11 covered social connectedness. Higher scores represented greater social connection (questions 10 and 11 are reverse scored).

Table 3.

Characteristics of Participants for SONNET Scale Psychometric Testing

	Initial field testing		Final psychometric evaluation
	Residents (n = 111)	Staff informants (n = 77)	Residents (n = 52)	Staff informants (n = 38)
Gender—n (%)
Female	77 (69)	68 (88)	29 (55)	34 (89)
Country
Canada	30 (27)	23 (30)	27 (47)	17 (45)
UK	81 (73)	54 (70)	25 (53)	21 (55)
Age (years)
Mean (SD)	85	43	85 (8)	43 (13)
Range	65–106	20–77	65–100	21–71
Marital status—n (%)
Single	18 (16)		12 (23)
Married	20 (18)		12 (23)
Common-law	1 (1)		0 (0)
Separated	2 (2)		3 (6)
Divorced	10 (9)		4 (8)
Widowed	59 (53)		21 (40)
Other	1 (1)		0 (0)
First language—n (%)
English	100 (90)	57 (74)	47 (90)	29 (76)
Race—n (%) Canadian participants
East/Southeast Asian	0 (0)	0 (0)	1 (4)	1 (6)
South Asian	0 (0)	3 (13)	0 (0)	0 (0)
Black	2 (7)	3 (13)	0 (0)	2 (13)
Latino	0 (0)	1 (4)	0 (0)	0 (0)
White	28 (93)	13 (57)	26 (96)	13 (81)
Other	0 (0)	3 (13)	0 (0)	0 (0)
Race—n (%) UK participants
Asian	3 (4)	9 (17)	1 (4)	1 (5)
Mixed ethnicity	2 (2)	0 (0)	1 (4)	0 (0)
White	73 (90)	37 (69)	23 (92)	16 (76)
Black	1 (1)	7 (13)	0 (0)	4 (19)
Other	1 (1)	1 (2)	0 (0)	0 (0)
Missing	1 (1)	0 (0)	0 (0)	0 (0)
Education—n (%)
Secondary/high school or less	66 (55)	24 (31)	28 (54)	1 (6)
Degree	27 (24)	23 (30)	7 (13)	8 (47)
Postgraduate	7 (6)	8 (10)	1 (2)	6 (35)
Other	9 (8)	21 (27)	8 (15)	2 (12)
Missing	2 (2)	1 (1)	7 (13)	0 (0)
Known diagnosis of dementia—n (%)
Yes	64 (37)		33 (63)
No	41 (58)		18 (35)
Not known	6 (5)		1 (2)
Length of time residing in care home of resident—n (%)
Less than 1 year	39 (35)		18 (35)
1–5 years	51 (46)		25 (48)
More than 5 years	19 (17)		7 (13)
Missing	2 (2)		2 (4)
Years of experience - n (%)
Less than 5 years		24 (31)		13 (34)
5–10 years		18 (23)		7 (18)
More than 10 years		34 (44)		18 (47)
Working pattern—n (%)
Full time		70 (91)		29 (76)
Part time		7 (9)		9 (24)
Role—n (%)
Recreation therapist/activity worker		15 (19)		10 (26)
Care home manager/administrator		11 (14)		4 (11)
Personal support worker/care worker		42 (55)		12 (32)
Other		9 (12)		12 (32)

Final Psychometric Evaluation

We evaluated the 12-item SONNET scale with 52 resident–staff dyads, whereby 14 staff acted as proxy for more than one resident (Table 3).

Feasibility and Acceptability

Residents took a mean of 8.1 minutes (SD 3.6, range 3–16) and staff took a mean of 4.7 minutes (SD 3.0, range 2–15) to complete the resident- and proxy-reported versions. There were 53 (8.5%) and 10 (1.6%) missing responses for the resident and proxy-reported versions, respectively, but most residents (n = 44; 84.6%) and proxies (n = 45; 86.5%) had no missing data.

Construct Validity (Structural Validity)

Confirmatory factor analysis supported the hypothesized two-factor structure whereby seven questions measured social engagement and four questions measured social connectedness (Table 4). Comparative fit index was 1.00 for the resident-reported scale and 0.91 for the proxy-reported scale. Close model fit was demonstrated by χ² = 34.9 (p = 0.78), CD = 0.88, and RMSEA <0.001 (95% CI < 0.001, 0.07) for the resident-reported scale and χ² = 58.6 (p = 0.05), CD = 0.97, and RMSEA = 0.09 (95% CI: 0.01, 0.14) for the proxy-reported scale. The social engagement subscale mean score was 9.8 (SD 4.1, range: 0–17) for the resident-reported scale and 10.5 (SD 4.2 range: 3–20) for the staff-reported scale. The social connectedness subscale mean score was 7.6 (SD 3.0, range: 0–12) for the resident-reported scale and 6.5 (SD 2.8, range: 1–12) for the staff-reported scale (Supplemental Material 1.6).

Table 4.

Final Psychometric Evaluation Descriptive Results

	Median response		Factor loadings
	Resident-rated scale	Proxy-rated scale	Resident-rated scale	Proxy-rated scale
Social engagement subscale
Q1 How often do you/they enjoy the company of other residents?	Often	Often	0.78	0.79
Q2 How often do you/they enjoy the company of staff or volunteers?	Very often	Often	0.43	0.48
Q3 How often do you/they socialize with other residents at mealtimes?	Often	Often	0.54	0.78
Q4 How often do you/they take part in activities with other residents, staff, or volunteers?	Sometimes	Often	0.44	0.74
Q7 How often do people from outside the care home visit you/them?	Sometimes	Often	0.16	0.22
Q8 How often do you/they keep in touch with people from outside the care home using the phone or other technology?	Sometimes	Often	0.39	0.14
Q9 How often do you/they leave the care home to spend time with people from outside the care home?	Not in the past week	Not in the past week	0.34	0.41
Social connectedness subscale
Q5 How often do you/they think of other residents as your/their friends?	Often	Often	0.76	0.76
Q6 How often do you/they feel like you/they have things in common with other residents?	Sometimes	Sometimes	0.69	0.94
Q10 How often do you/they feel left out?	Not in the past week	Not in the past week	0.51	0.18
Q11 How often do you/they feel lonely?	Sometimes	Sometimes	0.27	0.42
Summary question
Q12 How is your/their social life overall?	Good	Good	N/A	N/A

Subscale scores	Resident-reported scale				Proxy-reported scale
Subscale scores	Mean	SD	Min	Max	Mean	SD	Min	Max
Social engagement	9.8	4.1	0	17	10.5	4.2	3	20
Social connectedness	7.6	3.0	0	12	6.5	2.8	1	12

Construct Validity (Hypothesis Testing)

Correlations of the SONNET subscales with hypothesized related constructs are summarized in Table 5. The resident-reported social engagement subscale had moderate correlations with the CDR community affairs domain and the SONNET summary question 12, and weak correlation with the RISE, QOL-AD NH social subscale, CDR total, and NPI. The social connectedness subscale had moderate correlation with the NPI and weak correlation with the SONNET summary question 12, RISE, QOL-AD NH social subscale, CDR total, and CDR community affairs domain.

Table 5.

Validity of Social Connection in Long-term Care Home Residents (SONNET) scale

	Resident-rated		Proxy-rated
	Social engagement	Social connectedness	Social engagement	Social connectedness
	r (95% CI), p-value	r (95% CI), p-value	r (95% CI), p-value	r (95% CI), p-value
SONNET summary question 12: “How would you describe your social life overall?”	0.42 (0.19, 0.65), <0.01	0.36 (0.07, 0.65), 0.02	0.62 (0.45, 0.79), <0.01	0.48 (0.30, 0.65), <0.01
Revised index of Social Engagement (RISE)	0.26 (−0.01, 0.52), 0.06	0.26 (0.04, 0.47), 0.02	0.54 (0.37, 0.72), <0.01	0.44 (0.23, 0.66), <0.01
QOL-AD NH social subscale	0.34 (0.07, 0.61), 0.02	0.35 (0.15, 0.55), <0.01	0.57 (0.36, 0.78), <0.01	0.34 (0.11, 0.57), 0.01
QOL-AD NH total	0.13 (−0.15, 0.40), 0.39	0.15 (−0.20, 0.49), 0.32	0.45 (0.25, 0.65), <0.01	0.33 (0.15, 0.51), 0.02
Clinical Dementia Rating Scale community affairs domain	−0.44 (−0.63, −0.26), <0.01	−0.36 (−0.59, −0.13), <0.01	−0.36 (−0.59, −0.13), 0.01	−0.34 (−0.58, −0.09), 0.01
Clinical Dementia Rating Scale total	−0.29 (−0.54, −0.04), <0.01	−0.24 (−0.46, −0.02), 0.03	−0.20 (−0.46, 0.05), 0.12	−0.13 (−0.41, 0.14), 0.33
Neuropsychiatric inventory (NPI)	−0.30 (−0.54, −0.06), 0.01	−0.46 (−0.72, −0.19) <0.01	−0.38 (−0.56, −0.21), <0.01	−0.29 (−.53, −0.05), 0.04
Katz Index of Independence in ADLs	0.05 (−0.22, 0.32), 0.71	0.01 (00.31, 0.33), 0.95	0.07 (−0.19, 0.34), 0.58	−0.24 (−0.49, 0.01), 0.06

Key: ADLs = activities of daily living; CI = confidence interval; QOL-AD-NH = Quality of Life in Alzheimer’s disease nursing home scale.

The proxy-reported social engagement subscale had strong correlation with SONNET summary question 12, moderate correlation with the RISE, QOL-AD NH total, and QOL-AD NH social subscale, and weak correlation with the NPI, CDR total, and CDR community affairs domain. The social connectedness subscale had moderate correlation with SONNET summary question 12 and the RISE, and weak correlation with the QOL-AD NH social subscale and QOL-AD NH total, CDR community affairs domain, NPI, and Katz index. There was moderate correlation between the resident-reported and proxy-reported social engagement scale (r = 0.35, p = 0.01) and weak correlation between the resident-reported and proxy-reported social connectedness subscale (r = 0.26, p = 0.08).

Reliability (Internal Consistency)

The internal consistency was acceptable; Cronbach’s alpha for the resident-reported scale was 0.64 for social engagement, 0.72 for the social connectedness, and 0.71 for both proxy-reported subscales.

Reliability (Test–Retest and Inter-rater)

Reliability (Table 6) was assessed in subsets of n = 37 residents and n = 29 proxies (inter-rater) and n = 34 residents and n = 38 proxies (test–retest). The resident-reported scale had good inter-rater reliability for the social engagement and connectedness subscales (r = 0.70 (p < 0.01) and 0.66 (p < 0.01), respectively) and fair to good test–retest reliability over mean interval 17.2 (SD 3.6) days (r = 0.62 (p < 0.01) and 0.51 (p < 0.01)). The proxy-report scale had excellent inter-rater reliability (r = 0.91 (p < 0.01) and 0.82 (p < 0.01) for the social engagement and connectedness subscales, respectively), and its test–retest reliability was good to excellent over mean 16.7 (SD 3.5) days (r = 0.79 (p < 0.01) and 0.67 (p < 0.01)).

Table 6.

Reliability of Social Connection in Long-term Care Home Residents (SONNET) scale

	Inter-rater reliability		Test–retest reliability
	Resident-rated scale (n = 37)	Proxy-rated scale (n = 29)	Resident-rated scale (n = 34)	Proxy-rated scale (n = 38)
	Tau (95% CI)	tau (95% CI)	tau (95% CI)	tau (95% CI)
Q1	0.41 (0.14, 0.68)	0.63 (0.38, 0.88)	0.54 (0.32, 0.75)	0.60 (0.40, 0.81)
Q2	0.41 (0.13, 0.68)	0.73 (0.59, 0.88)	0.32 (0.02, 0.62)	0.65 (0.48, 0.81)
Q3	0.63 (0.41, 0.84)	0.87 (0.76, 0.97)	0.69 (0.48, 0.89)	0.80 (0.68, 0.92)
Q4	0.63 (0.44, 0.82)	0.84 (0.73, 0.96)	0.37 (0.06, 0.68)	0.54 (0.31, 0.77)
Q5	0.40 (0.12, 0.67)	0.78 (0.65, 0.91)	0.45 (0.15, 0.74)	0.60 (0.40, 0.80)
Q6	0.34 (0.02, 0.66)	0.85 (0.71, 0.99)	0.28 (−0.02, 0.57)	0.74 (0.60, 0.87)
Q7	0.68 (0.47, 0.90)	0.76 (0.58, 0.94)	0.44 (0.19, 0.68)	0.71 (0.56, 0.85)
Q8	0.49 (0.24, 0.73)	0.72 (0.52, 0.93)	0.64 (0.45, 0.83)	0.74 (0.63, 0.85)
Q9	0.45 (0.16, 0.74)	0.92 (0.83, 1.00)	0.62 (0.36, 0.87)	0.64 (0.38, 0.90)
Q10	0.47 (0.17, 0.77)	0.35 (0.01, 0.69)	0.27 (−0.06, 0.60)	0.35 (0.05, 0.64)
Q11	0.60 (0.38, 0.83)	0.60 (0.37, 0.84)	0.27 (−0.06, 0.60)	0.75 (0.55, 0.94)
Q12	0.50 (0.19, 0.80)	0.70 (0.56, 0.85)	0.15 (−0.13, 0.44)	0.55 (0.30, 0.79)

	r (95% CI), p-value	r (95% CI), p-value	r (95% CI), p-value	r (95% CI), p-value
Social engagement	0.70 (0.53, 0.86), <0.01	0.91 (0.85, 0.97), <0.01	0.62 (0.38, 0.86), <0.01	0.79 (0.66, 0.92), <0.01
Social connectedness	0.66 (0.45, 0.88), <0.01	0.82 (0.67, 0.97), <0.01	0.51 (0.29, 0.74), <0.01	0.67 (0.41, 0.92), <0.01

Key: CI = confidence interval.

Sensitivity Analyses

In sensitivity analyses (Supplemental Material 1.7), inter-rater reliability for the social engagement subscale was higher for residents with less cognitive impairment (r = 0.86 (p < 0.01) and r = 0.55 (p = 0.02), respectively), which was similarly observed for the subscale’s test–retest reliability (r = 0.74 (p < 0.01) and r = 0.50 (p = 0.08), respectively). For the social connectedness subscale, inter-rater reliability was similar and, unlike social engagement, the test–retest reliability was higher among those who were more cognitively impaired (r = 0.63 (p = 04) and r = 0.33 (p = 0.14), respectively). There was no consistent difference in correlation of the SONNET subscales with hypothesized related constructs (Supplemental Material 1.8).

Discussion

We developed a new measure for assessing social connection with LTC residents. We followed key methodological approaches and used a conceptual model of social connection for LTC residents to align the development and content with principles of person-centeredness and ensure the content validity of the measure. We demonstrated feasibility and acceptability for both self-reported and staff proxy-reported versions, and provided evidence of their reliability and validity in UK and Canadian settings.

The structural validity of the scale was evidenced by our hypothesized scale structure, with two domains of social connection (i.e., social engagement and social connectedness) supported in confirmatory factor analysis. However, test–retest and inter-rater reliability was lower for both of these subscales in the resident-reported version when compared to the proxy-reported subscales; sensitivity analyses suggest a differential for both the subscale (observed social engagement vs. experienced social connectedness) as well as the resident’s level of cognitive impairment. No previously tested measures have reported high test–retest reliability indicating the challenge of establishing this domain (Dewan et al., 2024), and this finding extends to measures of deficit in social connectedness (loneliness) used across populations and settings (Maes et al., 2022).

In our analysis of construct validity, findings supported our hypotheses that the SONNET scale would be positively correlated with other measures of social connection and inversely correlated with neuropsychiatric symptoms. However, we did not find any association with independence in activities of daily living and only weak associations with dementia severity. This finding may be explained by the support provided from care staff to all residents, particularly to enable those who are most impaired, as well as preservation of some aspects of social functioning despite increasing dementia severity (Budgett et al., 2019). We also did not find any association between the resident-reported SONNET subscales and staff-reported overall quality of life, suggesting that social connection may differ from quality of life (Budgett et al., 2019) or reflecting a mismatch between residents and staff in quality of life ratings (Spector & Orrell, 2006). Similarly, independence in activities of daily living was not correlated with either resident-reported SONNET subscale. Despite the potential negative impact of disability on social connection, it is possible that it can also ensure social contact as staff or family help for ADLs is provided in LTC settings. Others have also recently reported no association between independence in activities of daily living and loneliness in this setting (Pavlovski et al., 2025).

We found stronger correlation between resident and staff ratings for the social engagement (vs. social connectedness) subscale. This suggests that social engagement items, which can be directly observed (e.g., participation in activities or visits from family and friends), are more readily assessed by staff or residents. The lower correlation for resident and staff ratings of social connectedness may reflect residents’ impairment of memory about social connection (Sommerlad et al., 2024) or the difficulty faced by a proxy-rater in accurately assessing other’s experiences (Albrecht & Devlieger, 1999). Taken together, this highlights the benefits of self- and proxy-reported versions of the SONNET scale. Although our study was not designed to specifically assess psychometric properties according to cognitive impairment, our sensitivity analyses suggest further testing is warranted for best approaches to assess social connection in people with severe cognitive impairment. However, we reiterate previous researchers’ caution (Hughes et al., 2021) against developing one measure for people with dementia and another for people without dementia.

Strengths of our study include the methodical and detailed development and psychometric evaluation processes, all underpinned by principles of person-centeredness (Zimmerman et al., 2023). The SONNET scale offers a measure of social connection that can be used by researchers and LTC settings to test the effects of interventions at the individual, home, and system levels. Given the broader movement to address social connection as a public health priority (Holt-Lunstad et al., 2017) and the distinct importance for LTC homes (Estabrooks et al., 2020), integrating robust measures of social connection with mechanisms for ongoing data collection could offer new opportunities for research and health surveillance. A formal analysis of the current study methods and results, as a mechanism to compare SONNET with other existing measures (Dewan et al., 2024), is beyond the scope of the current study. However, we highlight some key limitations and areas for building the evidence for the SONNET measure. First, generalizability is limited by our study design and sample. Participants were required to speak English, and there was a lack of racial diversity in the resident participants. Further, our approach to recruitment likely selected higher-quality homes and residents who were more socially active. Second, to minimize burden on residents, we did not administer other resident-rated scales and this limited our analysis of construct validity. Our findings also highlight areas for development, including longitudinal data to test responsiveness and predictive validity, translation and testing in other diverse settings as well as testing of a self-completed version. Third, testing of a proxy-reported version that could be completed by family may be useful in some settings, given staff and family can have differing views on quality of life in LTC residents with severe dementia (Clare et al., 2014; Robertson et al., 2017). Fourth, our confirmatory factor analysis sample size was small which may lead to unstable factor loadings, so further evaluation in larger samples is warranted. Finally, we did not define a threshold or cut-off point as a binary (e.g., high or low social engagement), despite the potential utility for both research and practice, but highlight this as an area for future investigation to generate normative data in a representative sample of LTC residents.

Conclusion

The SONNET scale is feasible and acceptable for use in measuring LTC resident’s social connection. It is freely available through our website: https://www.sonnetstudy.com/measure. The self- and proxy-reported versions each yield two scores, social engagement and social connectedness, intended to be treated separately. Our study provides initial data; however, further study is needed to adapt and establish psychometric properties for diverse uses and settings. Research using high-quality measurement instruments would improve our understanding of LTC system, home, and resident-level factors which facilitate social connection and, in turn, develop evidence-based approaches to improve social connection.

Supplemental Material

Supplemental Material - Development and Testing of the Social Connection in Long-Term Care Home Residents (SONNET) Scale

Supplemental Material for Development and Testing of the Social Connection in Long-Term Care Home Residents (SONNET) Scale by Andrew Sommerlad, Steven Stewart, Hannah Chapman, Neha Dewan, Gill Livingston, Katherine S. McGilton, Sube Banerjee, Kirsten Corazzini, David Edvardsson, Madalena P. Liougas, Hannah M. O’Rourke, and Jennifer Bethell in Journal of Applied Gerontology

Footnotes

Acknowledgments

We are grateful for the participation of the long-term care home residents, families, and staff who took part in our study. Our thanks to the Camden and Islington National Health Service (NHS) Foundation Trust; Oxford Health NHS Foundation Trust; Northumbria Healthcare NHS Foundation Trust; Devon Partnership NHS Trust; South London & Maudsley NHS Foundation Trust; Solent NHS Trust; Clinical Research Network (CRN) West Midlands; and CRN East of England. And special thanks to Mary Connell at Jarlette Health Services; Stacey Januszewski at Alexander Place (Waterdown, ON); Teresa Figliuzzi at Roberta Place (Barrie, ON); Denise Simpson at Avalon Care Centre (Orangeville, ON); Monica Bolger at Royal Rose Place (Welland, ON); Cristina de Santis at Responsive Group Inc.; Nicole Fisher at Cooksville Care Centre (Mississauga, ON); and Laura Petta at the Alzheimer Society of Toronto. We would like to thank Michal Misiak from University College London; Deborah Moll and Anouska Pearson from Oxford Health NHS Foundation Trust; Debera Drew and Louise Jones from Northumbria NHS Trust; Emma Searle and Amie Zolanver from Solent NHS Trust; Humberto Matos, Anna Percival, and Humberto from Devon Partnership NHS Trust; Roselyn Enesi, Yin Fan, Lisa Farrell, Susan Frost, Kathryn Howlett, and Adam Peel from CRN East of England for support in data collection.

ORCID iDs

Andrew Sommerlad

Steven Stewart

Hannah Chapman

Neha Dewan

Gill Livingston

Katherine S. McGilton

Kirsten Corazzini

Madalena P. Liougas

Hannah M. O’Rourke

Jennifer Bethell

Ethical Considerations

We pre-registered our study protocol on (ID NCT05315960) and obtained ethical approval in the UK from the NHS Health Research Authority (22/LO/0145) and in Canada from the University Health Network (21-5976).

Consent to Participate

Consent was obtained from participants, or, for those residents lacking capacity, consent was obtained from a consultee decision maker.

Author Contributions

All authors contributed to the conception or design of the study. All authors drafted the manuscript, or critically revised the manuscript, and gave final approval of the version that was submitted for publication. All authors agree to be accountable for all aspects of the work, ensuring integrity and accuracy.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Canada Brain Research Fund (CBRF), an innovative arrangement between the Government of Canada (through Health Canada) and Brain Canada Foundation, and the Alzheimer’s Association (grant number: ARCOM-22-875327). To date, Health Canada has invested over $130 million through the CBRF which has been matched by Brain Canada Foundation and its donors and partners. Ce projet a été rendu possible grâce au soutien financier de Santé Canada, par le biais du Fonds canadien de recherche sur le cerveau, un partenariat novateur entre le gouvernement du Canada (via Santé Canada) et Brain Canada, et d’Alzheimer’s Association (Re: ARCOM-22-875327). Andrew Sommerlad is funded by the Wellcome Trust (grant number 222932/Z/21/Z). Andrew Sommerlad, Hannah Chapman, and Gill Livingston are supported by the University College London Hospitals’ National Institute for Health Research Biomedical Research Centre. Jennifer Bethell and Katherine McGilton are supported by the Walter & Maria Schroeder Institute for Brain Innovation and Recovery. Sube Banerjee is supported by a National Institute of Health and Care Research (NIHR) Senior Investigator Award.

Declaration of Conflicting Interests

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

Data Availability Statement

Data cannot be made publicly available as a condition of our ethical approvals.*

Supplemental Material

Supplemental material for this article is available online.

References

Albrecht

G. L.

Devlieger

P. J.

(1999). The disability paradox: High quality of life against all odds. Social Science & Medicine, 48(8), 977–988. https://doi.org/10.1016/s0277-9536(98)00411-0

Berkman

L. F.

Glass

Brissette

Seeman

T. E.

(2000). From social integration to health: Durkheim in the new millennium. Social Science & Medicine, 51(6), 843–857. https://doi.org/10.1016/s0277-9536(00)00065-4

Braun

Clarke

(2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10.1191/1478088706qp063oa

Brimelow

R. E.

Wollin

J. A.

(2017). Loneliness in old age: Interventions to curb loneliness in long-term care facilities. Activities, Adaptation & Aging, 41(4), 301–315. https://doi.org/10.1080/01924788.2017.1326766

Budgett

Brown

Daley

Page

T. E.

Banerjee

Livingston

Sommerlad

(2019). The social functioning in dementia scale (SF-DEM): Exploratory factor analysis and psychometric properties in mild, moderate, and severe dementia. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 11, 45–52. https://doi.org/10.1016/j.dadm.2018.11.001

Calkins

M. P.

(2018). From research to application: Supportive and therapeutic environments for people living with dementia. The Gerontologist, 58(suppl_1), S114–S128. https://doi.org/10.1093/geront/gnx146

Chapman

Bethell

Dewan

Liougas

M. P.

Livingston

McGilton

K. S.

Sommerlad

(2024). Social connection in long-term care homes: A qualitative study of barriers and facilitators. BMC Geriatrics, 24(1), 857. https://doi.org/10.1186/s12877-024-05454-8

Cicchetti

D. V.

Sparrow

S. A.

(1981). Developing criteria for establishing interrater reliability of specific items: Applications to assessment of adaptive behavior. American Journal of Mental Deficiency, 86(2), 127–137. https://psycnet.apa.org/record/1982-00095-001.

Clare

Quinn

Hoare

Whitaker

Woods

R. T.

(2014). Care staff and family member perspectives on quality of life in people with very severe dementia in long-term care: A cross-sectional study. Health and Quality of Life Outcomes, 12, 175. https://doi.org/10.1186/s12955-014-0175-3

10.

Corazzini

K. N.

Anderson

R. A.

Bowers

B. J.

Chu

C. H.

Edvardsson

Fagertun

Gordon

A. L.

Leung

A. Y.

McGilton

K. S.

Meyer

J. E.

Siegel

E. O.

Thompson

Wang

Wei

Lepore

M. J.

WE-THRIVE . (2019). Toward common data elements for international research in long-term care homes: Advancing person-centered care. Journal of the American Medical Directors Association, 20(5), 598–603. https://doi.org/10.1016/j.jamda.2019.01.123

11.

de Vet

Terwee

C. B.

Mokkink

L. B.

Knol

D. L.

(2011). Measurement in medicine: A practical guide (Vol. 124). Cambridge University Press. https://doi.org/10.1017/cbo9780511996214

12.

Dewan

Sommerlad

Chapman

Banerjee

Corazzini

Edvardsson

Liougas

M. P.

Livingston

McGilton

K. S.

O'Rourke

H. M.

Bethell

(2024). Assessing social connection for long‐term care home residents: Systematic review using consensus‐based standards for the selection of health measurement instruments guidelines. Alzheimer’s & Dementia: Translational Research & Clinical Interventions, 10(3), Article e12492. https://doi.org/10.1002/trc2.12492

13.

Dichter

M. N.

Wolschon

E.-M.

Schwab

C. G.

Meyer

Köpke

(2018). Item distribution and inter-rater reliability of the German version of the quality of life in Alzheimer’s disease scale (QoL-AD) proxy for people with dementia living in nursing homes. BMC Geriatrics, 18(1), 145. https://doi.org/10.1186/s12877-018-0834-z

14.

Dinno

(2009). Exploring the sensitivity of Horn’s parallel analysis to the distributional form of random data. Multivariate Behavioral Research, 44(3), 362–388. https://doi.org/10.1080/00273170902938969

15.

Duara

Loewenstein

D. A.

Greig‐Custo

M. T.

Raj

Barker

Potter

Schofield

Small

Schinka

Potter

(2010). Diagnosis and staging of mild cognitive impairment, using a modification of the clinical dementia rating scale: The mCDR. International Journal of Geriatric Psychiatry: A Journal of the Psychiatry of Late Life and Allied Sciences, 25(3), 282–289. https://doi.org/10.1002/gps.2334

16.

Estabrooks

C. A.

Straus

S. E.

Flood

C. M.

Keefe

Armstrong

Donner

G. J.

Boscart

Ducharme

Silvius

J. L.

Wolfson

M. C.

(2020). Restoring trust: COVID-19 and the future of long-term care in Canada (Vol. 5(No. 1), pp. 651–691). Canadian Science Publishing 1840 Woodward Drive. Suite 1, Ottawa, ON K2C 0P7. https://doi.org/10.1139/facets-2020-0056

17.

Ettema

T. P.

Dröes

R. M.

De Lange

Mellenbergh

G. J.

Ribbe

M. W.

(2007). QUALIDEM: Development and evaluation of a dementia specific quality of life instrument––validation. International Journal of Geriatric Psychiatry: A Journal of the Psychiatry of Late Life and Allied Sciences, 22(5), 424–430. https://doi.org/10.1002/gps.1692

18.

Evans

J. D.

(1996). Straightforward statistics for the behavioral sciences. Thomson Brooks/Cole Publishing Co.

19.

Gerritsen

D. L.

Steverink

Frijters

D. H.

Hirdes

J. P.

Ooms

M. E.

Ribbe

M. W.

(2008). A revised index for social engagement for long-term care. Journal of Gerontological Nursing, 34(4), 40–48. https://doi.org/10.3928/00989134-20080401-04

20.

Holt-Lunstad

(2021). The major health implications of social connection. Current Directions in Psychological Science, 30(3), 251–259. https://doi.org/10.1177/0963721421999630

21.

Holt-Lunstad

Robles

T. F.

Sbarra

D. A.

(2017). Advancing social connection as a public health priority in the United States. American Psychologist, 72(6), 517–530. https://doi.org/10.1037/amp0000103

22.

Holt-Lunstad

Smith

T. B.

Baker

Harris

Stephenson

(2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237. https://doi.org/10.1177/1745691614568352

23.

L. t.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

24.

Hughes

L. J.

Farina

Page

T. E.

Tabet

Banerjee

(2021). Psychometric properties and feasibility of use of dementia specific quality of life instruments for use in care settings: A systematic review. International Psychogeriatrics, 33(9), 917–931. https://doi.org/10.1017/S1041610218002259

25.

Katz

Ford

A. B.

Moskowitz

R. W.

Jackson

B. A.

Jaffe

M. W.

(1963). Studies of illness in the aged: The index of ADL: A standardized measure of biological and psychosocial function. JAMA, 185(12), 914–919. https://doi.org/10.1001/jama.1963.03060120024016

26.

Kaufer

D. I.

Cummings

J. L.

Ketchel

Smith

MacMillan

Shelley

Lopez

O. L.

DeKosky

S. T.

(2000). Validation of the NPI-Q, a brief clinical form of the neuropsychiatric inventory. The Journal of Neuropsychiatry and Clinical Neurosciences, 12(2), 233–239. https://doi.org/10.1176/jnp.12.2.233

27.

Leedahl

S. N.

Sellon

Chapin

R. K.

(2018). Assessment of multiple constructs of social integration for older adults living in nursing homes. Journal of Gerontological Social Work, 61(5), 526–548. https://doi.org/10.1080/01634372.2018.1451938

28.

Liougas

M. P.

Sommerlad

O'Rourke

Chapman

Dewan

McGilton

K. S.

Bethell

(2024). Assessing social connection for long-term care home residents: A scoping review of measure content. Alzheimer’s & Dementia: Translational Research & Clinical Interventions, 23(10), Article e12488. https://doi.org/10.1002/trc2.12488

29.

Maes

Qualter

Lodder

G. M.

Mund

(2022). How (not) to measure loneliness: A review of the eight most commonly used scales. International Journal of Environmental Research and Public Health, 19(17), Article 10816. https://doi.org/10.3390/ijerph191710816

30.

Mokkink

L. B.

Terwee

C. B.

Knol

D. L.

Stratford

P. W.

Alonso

Patrick

D. L.

Bouter

L. M.

De Vet

H. C.

(2010). The COSMIN checklist for evaluating the methodological quality of studies on measurement properties: A clarification of its content. BMC Medical Research Methodology, 10(1), 22. https://doi.org/10.1186/1471-2288-10-22

31.

Morera

O. F.

Stokes

S. M.

(2016). Coefficient α as a measure of test score reliability: Review of 3 popular misconceptions. American Journal of Public Health, 106(3), 458–461. https://doi.org/10.2105/AJPH.2015.302993

32.

Munn

Radey

Brown

Kim

(2018). Revising the lubben social network scale for use in residential long-term care settings. Journal of Evidence-Informed Social Work, 15(4), 385–402. https://doi.org/10.1080/23761407.2018.1460734

33.

O’rourke

H. M.

Duggleby

Fraser

K. D.

Jerke

(2015). Factors that affect quality of life from the perspective of people with dementia: A metasynthesis. Journal of the American Geriatrics Society, 63(1), 24–38. https://doi.org/10.1111/jgs.13178

34.

Patrick

D. L.

Burke

L. B.

Gwaltney

C. J.

Leidy

N. K.

Martin

M. L.

Molsen

Ring

(2011). Content validity—establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments for medical product evaluation: ISPOR PRO good research practices task force report: Part 2—assessing respondent understanding. Value in Health, 14(8), 978–988. https://doi.org/10.1016/j.jval.2011.06.013

35.

Pavlovski

Stanojkovska

Dikken

van Hoof

Marston

H. R.

(2025). The experience of loneliness among nursing homes residents and its relationship with independence level and activities of daily living: Contemporary insights from North Macedonia. Cogent Gerontology, 4(1), Article 2574327. https://doi.org/10.1080/28324897.2025.2574327

36.

Resnick

Gruber-Baldini

A. L.

Pretzer-Aboff

Galik

Buie

V. C.

Russ

Zimmerman

(2007). Reliability and validity of the evaluation to sign consent measure. The Gerontologist, 47(1), 69–77. https://doi.org/10.1093/geront/47.1.69

37.

Robertson

Cooper

Hoe

Hamilton

Stringer

Livingston

(2017). Proxy rated quality of life of care home residents with dementia: A systematic review. International Psychogeriatrics, 29(4), 569–581. https://doi.org/10.1017/S1041610216002167

38.

Rouquette

Falissard

(2011). Sample size requirements for the internal validation of psychiatric scales. International Journal of Methods in Psychiatric Research, 20(4), 235–249. https://doi.org/10.1002/mpr.352

39.

Russell

D. W.

(1996). UCLA loneliness scale (version 3): Reliability, validity, and factor structure. Journal of Personality Assessment, 66(1), 20–40. https://doi.org/10.1207/s15327752jpa6601_2

40.

Sanford

A. M.

Orrell

Tolson

Abbatecola

A. M.

Arai

Bauer

J. M.

Cruz-Jentoft

A. J.

Dong

Goel

Hajjar

Holmerova

Katz

P. R.

Koopmans

R. T. C. M.

Rolland

Visvanathan

Woo

Morley

J. E.

Vellas

(2015). An international definition for “nursing home”. Journal of the American Medical Directors Association, 16(3), 181–184. https://doi.org/10.1016/j.jamda.2014.12.013

41.

Saunders

Sim

Kingstone

Baker

Waterfield

Bartlam

Burroughs

Jinks

(2018). Saturation in qualitative research: Exploring its conceptualization and operationalization. Quality & Quantity, 52(4), 1893–1907. https://doi.org/10.1007/s11135-017-0574-8

42.

Sion

K. Y.

Verbeek

Zwakhalen

S. M.

Odekerken-Schröder

Schols

J. M.

Hamers

J. P.

(2020). Themes related to experienced quality of care in nursing homes from the resident’s perspective: A systematic literature review and thematic synthesis. Gerontology and Geriatric Medicine, 6, Article 2333721420931964. https://doi.org/10.1177/2333721420931964

43.

Smith

Lamping

Banerjee

Harwood

Foley

Smith

Cook

Murray

Prince

Levin

Mann

Knapp

(2005). Measurement of health-related quality of life for people with dementia: Development of a new instrument (DEMQOL) and an evaluation of current methodology. Health Technology Assessment (Winchester, England), 9(10), 1–4. https://doi.org/10.3310/hta9100

44.

Snowball

Fernandez Loughlin

Eagleson

Barnett

K. M.

McLellan

O’Connor

Kelly

Thelker

McGilton

K. S.

Bethell

(2022). Engagement of people with lived experience of dementia advisory group and cross-cutting program: Reflections on the first year. Research Involvement and Engagement, 8(1), 28. https://doi.org/10.1186/s40900-022-00359-5

45.

SocioCultural Research Consultants, L . (2023). Dedoose. (Version Version 9.0.107).

46.

Sommerlad

Grothe

Umeda

Ikeda

Kanemoto

Livingston

Luppa

Rankin

K. P.

Riedel-Heller

S. G.

Röhr

Suzuki

Huntley

(2024). Awareness of social functioning in people with dementia and its association with dementia severity: Multi-center cross-sectional study. Journal of Alzheimer’s Disease, 100(4), 1183–1193. https://doi.org/10.3233/JAD-240311

47.

Sommerlad

Kivimäki

Larson

E. B.

Röhr

Shirai

Singh-Manoux

Livingston

(2023). Social participation and risk of developing dementia. Nature Aging, 3(5), 532–545. https://doi.org/10.1038/s43587-023-00387-0

48.

Spector

Orrell

(2006). Quality of life (QoL) in dementia: A comparison of the perceptions of people with dementia and care staff in residential homes. Alzheimer Disease & Associated Disorders, 20(3), 160–165. https://doi.org/10.1097/00002093-200607000-00007

49.

StataCorp LLC . (2020). StataCorp.

50.

Stewart

Hotopf

Dewey

Ballard

Bisla

Calem

Fahmy

Hockley

Kinley

Pearce

Saraf

Begum

(2014). Current prevalence of dementia, depression and behavioural problems in the older adult care home sector: The South East London care home survey. Age and Ageing, 43(4), 562–567. https://doi.org/10.1093/ageing/afu062

51.

Streiner

D. L.

Norman

G. R.

Cairney

(2024). Health measurement scales: A practical guide to their development and use (6th ed.). Oxford University Press.

52.

Thoits

P. A.

(2011). Mechanisms linking social ties and support to physical and mental health. Journal of Health and Social Behavior, 52(2), 145–161. https://doi.org/10.1177/0022146510395592

53.

van der Wolf

van Hooren

S. A.

Waterink

Lechner

(2018). Measurement of well-being in gerontopsychiatric nursing home residents: Development of the Laurens well-being inventory for gerontopsychiatry. Journal of Geriatric Psychiatry and Neurology, 31(3), 136–148. https://doi.org/10.1177/0891988718781031

54.

Zimmerman

Mast

Port

VanHaitsma

Fazio

(2023). Development and testing of the PC-MET (Person-Centered Measure Evaluation Tool). Innovation in Aging, 7(Supplement_1), 296–297. https://doi.org/10.1093/geroni/igad104.0989