Prevalence and determinants of loneliness in people living with dementia: Findings from the IDEAL programme

To establish the prevalence and determinants of loneliness among people living with dementia.


| INTRODUCTION
Worldwide around 50 million people are living with dementia and this is predicted to increase and reach 152 million by 2050. 1 Identifying factors that may help or hinder the ability to 'live well' with dementia is therefore valuable. The Institute of Medicine defines 'living well' as 'the best achievable state of health that encompasses all dimensions of physical, mental and social well-being' (p 32). 2,3 Our recent comprehensive model examining factors associated with living well for people with dementia identified loneliness as one of the psychological indicators that affected the ability to 'live well'. 3 Loneliness is potentially modifiable and, as such, the aims of the current study are to identify the prevalence and determinants of loneliness in people living with dementia.
Loneliness may be defined as the discrepancy between expectations of the quantity and/or quality of relationships and what is actually experienced. 4 It has been widely linked to depression and mental health 5 and is a potential risk factor for the development of poorer physical health outcomes 6 and mortality, 7 and compromised quality of life, life satisfaction and wellbeing. 3,8 In older populations, factors such as female gender, lower levels of education, living alone, marital status, lack of or low quality social relationships and poorer physical health have been identified as determinants of loneliness. 9,10 One area where there is potential for studies of loneliness in later life to be more inclusive is by encompassing people with dementia. 3 However our current evidence base has focused either on loneliness as a potential risk factor for dementia, where findings have been mixed, [11][12][13][14] or on its influence on cognitive function, where some but not all studies have reported that it is predictive of faster cognitive decline. 15 Few existing prevalence studies of loneliness have included people with dementia. [16][17][18][19] Where studies have looked at loneliness and dementia we can identify two approaches; (a) comparison of loneliness amongst those with or without dementia or (b) comparison of dementia status amongst those who are and are not lonely. 20,21 Our focus is with the former approach as we wish to establish the prevalence and predictors of loneliness among people with dementia. A recent British study of 93 people with mild-to-moderate Alzheimer's disease, using the 11-item De Jong Gierveld loneliness scale, reported a mean score of 2.0 (scores of between 0 and 2 on the 11-item scale indicates no loneliness) suggesting that on average participants did not feel lonely. 16 An earlier Swedish study examined social and emotional loneliness using two single-item measures in 154 people with dementia and 435 older people without a dementia diagnosis. 18 People living with dementia were significantly more likely to report often feeling socially lonely than respondents without dementia, but no differences were observed for emotional loneliness; 46% of the participants with dementia reported often feeling lonely (social loneliness) and 53% said they experienced loneliness (emotional loneliness). A smaller European study explored the relationship between loneliness, social isolation and hallucinations in 22 people with Alzheimer's disease in comparison with 24 healthy controls using the UCLA loneliness scale. 17 Participants with Alzheimer's disease reported greater levels of loneliness than the controls (mean scores of 61.3 and 50.1 on the 20-item UCLA loneliness scale, respectively). Studies of loneliness determinants have looked at specific symptoms of dementia such as neuropsychiatric symptoms or more specifically hallucinations rather than more established loneliness risk factors. No relationship was observed for neuropsychiatric symptoms in general, 10

| Procedure
Information was collected by direct interview from people with dementia who were visited at home by a researcher on three separate occasions. All measures used in the current study were completed by the person with dementia and reflect self-reported experience.

| Loneliness
Loneliness was assessed using the six-item version of the De Jong Gierveld Loneliness Scale. 25 Total scores range from 0 to 6, where a

Key Points
• This is one of the few studies to explore the effect of loneliness on people living with dementia.
• Approximately one-third of people with dementia reported experiencing loneliness • People with dementia experiencing social isolation and depression, and those living alone, were more likely to report feeling lonely.
• No relationship was observed between dementia diagnosis, cognitive function or marital status. score of 0 to 1 indicates no loneliness, scores of 2 to 4 moderate loneliness and 5 to 6 severe loneliness. 26 We selected this because it has been successfully used in two previous studies involving people with dementia 16,18 and mindful of participant burden, takes less time to complete than the 11-item version. None of the studies identified reported problems with the acceptability of the scale.

| Covariates
We included as co-variates established factors associated with loneliness in the general population including age, sex, marital status, education and living situation (whether the person lived alone or with a spouse or others). The number of additional chronic health conditions was counted using the Charlson comorbidity index 27 and was grouped into four severity levels. 28 Depressive symptoms were assessed using the 10-item Geriatric Depression Scale (GDS-10), 29 with higher total scores indicating greater depressive symptoms. The six-item Lubben Social Network Scale was used to gauge social isolation by measuring perceived social support received from family and friends 30 ; total scores ranged from 0 to 30, where a score of less than 12 is seen to indicate a higher risk of social isolation. Life satisfaction was measured the Satisfaction with Life Scale. 31 Possible scores ranged from 5 to 35, with higher scores indicating greater life satisfaction. Well-being was measured using the World Health Organization-Five Well-Being Index (WHO-5). 32 The raw score was transformed into a percentage score where 0 signifies the worst possible well-being and 100 represents the best possible well-being. The QoL in Alzheimer's disease (QoL-AD) scale 33 which was developed specifically for people with dementia, was used to gauge QoL. Possible scores range from 13 to 52, with higher scores indicating greater QoL.
Dementia-specific factors were cognitive function at baseline as measured using the MMSE 24 and the type of dementia as recorded by the diagnosing clinicians at recruitment sites and retrieved from medical records when participants were enrolled into the study. Neuropsychiatric symptoms were not included in our analysis because information about these was provided by care partners, thereby excluding the 17% in our study who did not have a participating care partner.

| Statistical analyses
Our analysis consists of three phases linked to our two research questions. To establish the prevalence of loneliness we classified participants into the three loneliness categories defined by the de Jong Gierveld scale (no loneliness, moderate loneliness and severe loneliness). 26 To establish loneliness determinants, we compared the baseline characteristics of respondents in the three loneliness groups using Chi-Squared tests for categorical variables and analysis of covariance for continuous variables. Finally, multinomial logistic regression was used to examine the determinants of loneliness in people with dementia and findings are presented as relative risk ratios (RRR) of reporting moderate or severe loneliness in comparison with no loneliness. To investigate the influence of missing data, we imputed missing values using multivariate imputation by chained equations. Missing data ranged from 0.4% to 10.9%, and 33% of participants had missing data on one or more variable of interest. We included all variables from the analysis in the imputation model. Estimates from 30 imputed data sets were combined using Rubin's rules. 34 All data were analysed using Stata 14.2 (TX: StataCorp LP).

| RESULTS
Overall 93% of participants, 1445 out of 1547, completed the loneliness measure.
There were no significant differences in non-response to the loneliness questions across our included analytical variables except for marital status where the measure was more likely to be completed by the married (see Table S1). Table 1 shows the characteristics of the 1445 people with dementia who completed the loneliness measure.
Most participants had a diagnosis of Alzheimer's disease (55.5%) and at least one additional chronic health condition (75%) and the mean age was 76 (SD 8.6). Just over half of respondents were male. The majority of respondents, 64.7%, were classified as not lonely, 30.1% as moderately lonely and 5.2% severely lonely (see Table 1). Loneliness was significantly associated with increased age, living alone, wid-

| Determinants of loneliness
We undertook multi-variable regression analyses to identify the key independent loneliness determinants, with no loneliness as our reference category. In our final fully adjusted model two variables, depression and isolation, were associated with both moderate and severe loneliness and hinted at a 'dose-response' relationship. For depressive symptoms the RRR for moderate dementia was 1.18 (95% CI 1.08, 1.28) and (95% CI 1.14, 1.58;

| Strengths and limitations
To the best of our knowledge, this study has used a larger cohort of people with dementia than previous studies investigating the prevalence and determinants of loneliness in this population. As well as numerical size, which enables us to undertake analyses that can look at a broader range of determinants than previous studies, [16][17][18][19] we included the full range of dementia sub-types and all participants had a clinical diagnosis. We included an established measure of loneliness that was completed by 93% of our total cohort, confirming that it is appropriate and acceptable for people with mild to moderate dementia. Additionally, we included in our analysis both dementia-specific and generally established loneliness determinants. However, we fully acknowledge the main limitation of our study which is the crosssectional design. This does not compromise our prevalence estimates, it does, however, mean that it is not possible to establish causal relationships or the direction of the observed associations. For some determinants, this is not problematic. For example, based upon previous studies, it is highly likely that loneliness is, in part, a consequence of living alone rather than vice versa. For some other observed relationships, such as depression or quality of life, these could be either a cause or consequence of loneliness. However, the IDEAL cohort study is longitudinal which will enable us to investigate the robustness of our observations and directionality of relationships when data from further time-points become available.

| CONCLUSION
This is one of only a few studies to explore the prevalence and predictors of loneliness in people living with dementia. We found that people with dementia had a loneliness prevalence which was broadly comparable with that of their peers who did not have dementia. We demonstrated that established loneliness predictors such as living alone were also important for people with dementia and that there was a relationship between loneliness and depression. Whilst we cannot establish directionality at this stage, it seems likely that addressing depressive symptoms will offer benefits in terms of reduced loneliness. It is important that interventions designed to help alleviate loneliness should be person-centred and focused on the specific needs of people living with dementia. 37 Our study produced the unique and counterfactual finding of no relationship between loneliness and marital status. This highlights an important area for further research in the field of loneliness and dementia, but also with regard to wellbeing more generally. It further suggests that we need to look not just at marital status but also at relationship quality 38 when trying to understand these associations, and focus more on the dyad rather than simply looking at individual wellbeing in isolation.