Emotional Loneliness Is Associated With a Risk of Dementia in a General Japanese Older Population: The Hisayama Study

Abstract

Objectives

To investigate the association of loneliness and its component subscales with the risk of dementia in a general Japanese older population.

Method

A total of 1,141 community-dwelling Japanese residents aged ≥65 years without dementia were prospectively followed up for a median 5.0 years. We evaluated any loneliness and its component subscales—namely, social and emotional loneliness—by using the 6-item de Jong Gierveld Loneliness Scale. Cox proportional hazards models were used to estimate hazard ratios (HRs) of each loneliness type on the risk of dementia controlling for demographic factors, lifestyle factors, physical factors, social isolation factors, and depression.

Results

During the follow-up, 114 participants developed dementia. The age- and sex-adjusted incidence rate of dementia was significantly greater in participants with any loneliness and emotional loneliness than those without. The multivariable-adjusted HRs (95% confidence intervals) of participants with any loneliness and emotional loneliness on incident dementia were 1.61 (1.08–2.40) and 1.65 (1.07–2.54), respectively, as compared to those without. However, there was no significant association between social loneliness and dementia risk. In subgroup analyses of social isolation factors, excess risks of dementia associated with emotional loneliness were observed in participants who had a partner, lived with someone, or rarely communicated with relatives or friends, but such association was not significant in participants who had no partner, lived alone, or frequently communicated with friends or relatives.

Discussion

The present study suggested that loneliness, especially emotional loneliness, was a significant risk factor for the development of dementia in the general older population in Japan.

Dementia is a major health problem due to its psychological, social, and economic impacts. A report from the World Health Organization (2020) has emphasized the importance of identifying risk factors for dementia to prevent or delay its incidence.

Loneliness, which is generally defined as “the unpleasant experience that occurs when a person’s network of social relations is deficient in some important way, either quantitatively or qualitatively” (Perlman & Peplau, 1981), has been a focus of attention as one of the modifiable risk factors for incident dementia. Several studies found a significant association between loneliness and incident dementia, but the findings across the studies are inconsistent (Chen et al., 2011; He et al., 2000; Lobo et al., 2008; Rawtaer et al., 2017; Sundström et al., 2020). Loneliness consists of two components—namely, social loneliness and emotional loneliness (Weiss, 1973). Social loneliness arises from the subjective perception of absence of a broader group of contacts or an engaging social network (e.g., friends and people in the neighborhood), whereas emotional loneliness stems from the subjective perception of the absence of an intimate person or a close emotional attachment (e.g., a partner and a best friend). Previous epidemiological studies have assessed the association between loneliness and dementia without distinguishing the two components of loneliness (Holwerda et al., 2014; Lobo et al., 2008; Rafnsson et al., 2017; Sundström et al., 2020). Examining these two components separately could help to clarify their discrete effects on the development of dementia. Several previous studies have reported that emotional loneliness is more likely to be involved in lower life satisfaction, risk of depression, and mortality than social loneliness (O’Súilleabháin et al., 2019; Peerenboom et al., 2015; Salimi, 2011). O’Súilleabháin and colleagues (2019) described “emotional loneliness, which is often associated with feelings of abandonment and anxiety, is the toxic component of loneliness,” and discussed that “increases in emotional loneliness resulted in an increase of the effect rate of functional status on mortality.” It has also been reported that emotional loneliness is more associated with lower life satisfaction rather than social loneliness, because the absence of close relationships (i.e., emotional loneliness) is more painful than the absence of social friendships (i.e., social loneliness) (Peerenboom et al., 2015). Considering these findings, social and emotional loneliness may be differentially associated with chronic diseases. Understanding the impact of different types of loneliness on dementia risk may help to clarify the appropriate targets for prevention and treatment of dementia.

The purpose of the present study is to elucidate the association of loneliness and its component subscales with the development of dementia in a community-dwelling older Japanese population using a prospective cohort analysis.

Method

Participants

The Hisayama Study is an ongoing population-based prospective study for cardiovascular disease and dementia in the town of Hisayama, a suburb of Fukuoka City in southern Japan. In this study, health examinations of physical and neurological conditions have taken place every 1–2 years since 1961 (Ninomiya, 2018). We invited all residents aged 40 years or over in the town of Hisayama to participate in the annual health examination by mail, telephone, or in-person interview with the cooperation of the Division of Health and Welfare of Hisayama. Further, follow-up surveillance for incident dementia has been performed since 1985, in combination with comprehensive screening surveys for dementia among older residents every 5–7 years (Ohara et al., 2017). A total of 1,906 residents aged 65 years or older underwent the health examination in 2012–2013 (participation rate: 93.6%). Of those, 1,519 attendees who were free of dementia and provided informed consent at baseline were included in this study. After excluding 363 participants who did not complete the loneliness questionnaire, and 15 participants who had missing data on covariates, the final analyses included 1,141 participants (507 men and 634 women), whose mean (±SD) age was 74 ± 6 years (range: 65–92 years). This study was approved by the Kyushu University Institutional Review Board for Clinical Research. We obtained written informed consent from all participants.

Follow-up Survey of Dementia

We followed up the participants for incident dementia for a median of 5.0 years (interquartile range: 4.9–5.1 years) from their baseline screening examination. As previously described, we established a daily follow-up system comprising the study team, local physicians, and members of the town’s Health and Welfare Office to collect information on new neurological events, including dementia and stroke (Ohara et al., 2017). Physicians of the study team regularly visited hospitals, clinics, and the town’s office to collect the information on dementia cases. Additionally, we repeated neurological and physical examinations at the annual health examination to identify incident dementia cases. For participants who did not participate in the health examination or moved out of the town, we performed postal and telephone surveys. Moreover, to precisely detect dementia cases to the greatest extent possible, we conducted comprehensive neuropsychological screening for dementia in 2017–2018 for most of the participants (88.9%). Once dementia was suspected, a psychiatrist and stroke physician from the study team carefully examined the individual for the absence or presence of dementia. When a participant died, we interviewed the family and the attending physician, and reviewed all the available clinical information, including neuroimaging and information on cognitive function and activities of daily living. The participants were followed up until the date of neuropsychological screening for dementia in 2017–2018 or March 31, 2018 (for participants who did not participate in the neuropsychological screening in 2017–2018). No participants were lost to follow-up during the follow-up period.

Diagnosis of Dementia

The diagnosis of dementia was made based on the guidelines of the Diagnostic and Statistical Manual of Mental Disorders, third edition, revised (American Psychiatric Association, 1987). First, the interview survey for the screening of cognitive function was conducted by trained doctors and clinical psychologists using the Mini-Mental State Examination (MMSE) (Folstein et al., 1975). Next, the presence of cognitive impairment was determined by expert psychiatrists based on physical and neurological examinations, including the delayed recall test of the logical memory IIA subscale of the Wechsler Memory Scale-Revised according to the Alzheimer’s Disease Neuroimaging Initiative (ADNI) procedures (Petersen, 2008), the information from the patient, interviews with family members and attending physicians, and medical records. The diagnosis of each dementia case was adjudicated by expert psychiatrists and stroke physicians together, who were blind to the participants’ loneliness scores.

Assessment of Loneliness

We assessed loneliness by using the six-item De Jong Gierveld Loneliness Scale (de Jong Gierveld & van Tilburg, 2006). The scale has two component subscales: social loneliness and emotional loneliness. The scale for social loneliness includes three negatively formulated items as follows: “There are plenty of people that I can lean on in case of trouble,” “There are many people that I can count on completely,” and “There are enough people that I feel close to.” The scale for emotional loneliness includes three positively formulated items as follows: “I experience a general sense of emptiness,” “I miss having people around,” and “Often, I feel rejected.” The original possible answers are “yes,” “more or less,” and “no,” but we modified possible answers simply using “yes” and “no” and dropped “more or less.” For the items of social loneliness, “no” scored 1 point and “yes” scored 0 points, whereas the scoring for the items of emotional loneliness was reversed. The score for each subscale thus ranges from 0 to 3. The total score was calculated by combining the scores for each subscale and ranged from 0 to 6. The presence of social loneliness and emotional loneliness were each defined by a subscale score of 1 point or more. Participants were classified with presence of any loneliness if they according to the presence of either social or emotional loneliness (i.e., total score of 1 point or more). The participants who had both social and emotional loneliness were classified into the “presence” category for both loneliness subtypes. The questionnaire has been well examined its reliability, validity and external consistency (de Jong Gierveld & van Tilburg, 2010).

Other Covariates

Information on the following factors was collected using a questionnaire administered by trained interviewers at baseline: marital status (with or without partner), living situation (alone or with someone), monthly frequency of communication (e.g., meet or telephone etc.) with friends or relatives (few, or several times or more), educational level, occupation, alcohol habits, smoking habits, regular exercise, daily life activity, antihypertensive agent use, the use of glucose-lowering agents, and past history of disease, including cardiovascular disease, cancer, respiratory disease, hepatic disease, brain injury, and psychiatric disease. Expert psychiatrists diagnosed depression based on the guidelines of the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (American Psychiatric Association, 1994) by using the Mini-International Neuropsychiatric Interview (Sheehan et al., 1998). We measured body height and weight and calculated body mass index (BMI). Electrocardiogram abnormalities were defined as ST depression (Minnesota Code, 4-1, 2, 3), left ventricular hypertrophy (3-1), or atrial fibrillation (8-3). Blood pressure was measured three times after more than 5 min of rest in the sitting position and the mean value of the three measurements was calculated. Hypertension was defined as a systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, and/or current use of antihypertensive agents (Mancia et al., 2007). Diabetes mellitus was defined as a fasting plasma glucose level of ≥7.0 mmol/L (126 mg/dL), a 2-hr post-load or causal glucose level of ≥11.1 mmol/L (200 mg/dL), and/or current use of glucose-lowering agents (Genuth et al., 2003).

Statistical Analysis

Comparisons of characteristics between the participants with and without loneliness subscales were performed by a Student’s t test for continuous variables or a chi-squared test for dichotomous variables. The incidence rates of dementia according to the status of any loneliness or its component subscales were calculated by the person-year method with adjustment for age and sex by a direct method. We used Cox proportional hazards regression models to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) for incident dementia according to the status of any loneliness or its component subscales. The proportional hazards assumption was visually checked using the log cumulative hazard plots. In the multivariable-adjusted analysis, we built four models including the status of any loneliness or its component subscales and the following covariates: Model 1: age [continuous] and sex [men or women]; Model 2: the covariates in Model 1+ lifestyle and physical factors (educational level [≤9 or >9 years], occupation [employed or not], current smoker [yes or no], current drinker [yes or no], regular exercise [not more than three times per week or three or more times per week], daily life activity [none to light level or middle to high level], BMI [continuous], hypertension [yes or no], diabetes [yes or no], electrocardiogram abnormalities [yes or no], past history of cardiovascular disease [yes or no], past history of cancer [yes or no], and past history of other chronic diseases [respiratory disease, hepatic disease, brain injury, and psychiatric disease] [yes or no]); Model 3: the covariates in Model 2 + social isolation factors (marital status [with or without partner], living situation [alone or with someone], and monthly frequency of communication with friends or relatives [few or several times or more]); and Model 4: the covariates in Model 3 + depression [yes or no]. The heterogeneity in the association of any loneliness or its component subscales with risk of dementia between subgroups was assessed by adding the multiplicative interaction term to the relevant Cox model. To assess the possibility of selection bias, we analyzed the difference in characteristics between the included and excluded participants in the same manner as for the aforementioned table of main characteristics. All statistical analyses were performed using the SAS software (version 9.4; SAS Institute, Cary, NC). Statistical significance was defined as a two-tailed p-value of <.05.

Results

The baseline characteristics of all participants according to loneliness subscales are summarized in Table 1. The participants with any loneliness were significantly more likely to be older and to be men than those without any loneliness. Also, they were significantly more likely to have a past history of cancer or other disease, to live alone, to be depressed, and to have fewer monthly communications with friends or relatives than the participants without loneliness. The participants with social loneliness had significantly higher frequencies of men, depression, and past history of other diseases, and fewer communications with friends or relatives than those without social loneliness. The participants with emotional loneliness were significantly more likely to be older, women, unemployed, without a partner, and living alone, and significantly more likely to have depression, a past history of other diseases, lower levels of daily activity, and fewer communications with friends or relatives than those without emotional loneliness.

During the follow-up, 114 participants (49 men and 65 women) developed dementia. As shown in Figure 1, the age- and sex-adjusted incidence of dementia was significantly higher in participants with any loneliness and emotional loneliness than in those without, whereas there was no significant difference in the incidence of dementia between participants with and without social loneliness. Table 2 shows the association of the status of loneliness and its component subscales with the development of dementia after adjusting for potential cofounding factors. The presence of any loneliness and the presence of emotional loneliness were significantly associated with an approximately 60% greater risk of developing dementia after adjustment for age, sex, and lifestyle and physical factors (Model 2). Such increased risks of dementia were still significant after adjusting for social isolation factors (Model 3) and depression (Model 4). Multivariable-adjusted HRs of dementia increased significantly for every 1 point increment in the scores for any loneliness and emotional loneliness. Significant associations of any loneliness and emotional loneliness with greater risk of dementia were still observed in the sensitivity analyses which excluded participants with MMSE of <24 points at baseline (Supplementary Table 1) or dementia cases occurring within the first 1 year of follow-up (Supplementary Table 2). On the other hand, there was no evidence of significant association between social loneliness and the incidence of dementia (Table 2 and Supplementary Tables 1 and 2).

Next, we estimated the combined risks of developing dementia according to the presence or absence of social and emotional loneliness (Figure 2). The participants with emotional loneliness and any loneliness had significantly higher risk of dementia than those without any loneliness. The multivariable-adjusted HR for dementia tended to be higher, but did not reach the statistically significant level, in participants with both social and emotional loneliness than in those with neither of them, and the HR for dementia was significantly greater in participants with emotional loneliness without social loneliness. The effect of the interaction between social loneliness and emotional loneliness for the risk of dementia was not statistically significant (p for interaction = .42 for categorical score, = .07 for continuous score). In the subgroup analyses of social isolation factors—namely, marital status, living situation, and monthly frequency of communication with friends or relatives (Figure 3)—the presence of emotional loneliness was significantly associated with a greater risk of dementia among participants who had a partner, lived with someone, or had few communications with friends or relatives (all p < .05), but such association was not observed among the participants who had no partner, lived alone, or had several or more monthly communications with friends or relatives. The p-value for the heterogeneity in the association of emotional loneliness with the risk of dementia was significant only in the marital status subgroup as follows: marital status (.02 for categorical score and .049 for continuous score); living situation (.81 for categorical score and .56 for continuous score); and monthly frequency of communication with friends or relatives (.11 for categorical score and .19 for continuous score).

Discussion

The present prospective cohort study clearly demonstrated that any loneliness, especially emotional loneliness, was significantly associated with an increased risk of dementia in a community-dwelling older population of Japanese. The positive association between any loneliness or emotional loneliness and risk of dementia persisted after adjusting for sociodemographic factors, lifestyle factors, physical factors, social isolation factors, and depression. However, there was no significant association between social loneliness and dementia risk. Intriguingly, emotional loneliness was associated with a greater risk of dementia in participants who had a partner or lived with someone, but not in those who did not, probably suggesting that older participants who felt emotional loneliness even in the situation of living with others were at an increased risk of dementia. Further, the excess risk of dementia associated with emotional loneliness was observed among participants who rarely communicated with friends or relatives, but it disappeared among the participants who frequently communicated with friends or relatives. These findings highlight the significance of closely communicating with others and easing emotional loneliness in the elderly for preventing the development of dementia.

Our results are consistent with those from previous prospective studies in which loneliness was associated with an elevated risk of all-cause dementia (Holwerda et al., 2014; Rafnsson et al., 2017; Sutin et al., 2018; Zhou et al., 2018) and Alzheimer disease (Sundström et al., 2020; Wilson et al., 2007). However, several previous cohort studies failed to confirm a significant association between loneliness and risk of dementia (Chen et al., 2011; He et al., 2000; Lobo et al., 2008; Rawtaer et al., 2017). This discrepancy in the association between loneliness and incident dementia among the studies may be due to differences in the methods used to measure loneliness (e.g., single-item measures vs multi-item scales, and nonvalidated vs validated) or an insufficient number of dementia events. In the present study, therefore, we improved these points by using validated multi-item scales on a sufficient number of the study population. Notably, lifestyle, physical health, and particularly depression have been considered to be confounders of or potential mechanisms underlying the excess dementia risk conferred by loneliness in the above-mentioned studies conducted in Western populations. In the present study, however, the magnitudes of the association between loneliness and dementia were not reduced after adjusting for these factors, possibly suggesting these factors are unlikely to be confounders or possible mechanisms. The precise reasons for the discrepant findings between the present and previous studies are unclear, but we speculate that the findings from Western populations may not necessarily be generalizable to Asian populations, probably due to the difference in social and cultural backgrounds. In addition, depression was diagnosed by psychiatrists in the present study, which is different from the questionnaire-based assessment method in the previous studies. That may also have contributed to the difference in results from previous studies.

As far as we know, there have been no previous prospective studies separately assessing the influence of social loneliness and emotional loneliness on dementia risk. Notably, the present study found that emotional loneliness was significantly associated with risk of dementia, but social loneliness was not. In particular, the presence of emotional loneliness even in the situation of having a partner or living with someone, which might derive from a feeling of subjective social exclusion or alienation, was significantly associated with greater risk of dementia in the elderly. Social exclusion includes the feeling of being rejected, ignored, and isolated by others (Vijayakumar et al., 2017). Elderly individuals who have a partner or live with someone may tend to have a much higher expectation of intimate emotional communication than those who live alone. When they fail in establishing an intimate relationship with a cohabiting individual, they may feel greater loneliness or even a sense of being excluded. In this way, the impact of emotional loneliness on dementia may be augmented. Previous population-based studies have reported that emotional loneliness or social exclusion are associated with an increased risk of all-cause mortality (O’Súilleabháin et al., 2019) and depression and poor self-rated health (Feng et al., 2019) in the elderly. Therefore, emotional loneliness may be a more adverse risk factor for dementia than social loneliness. On the other hand, we found that the significant positive association between emotional loneliness and risk of dementia was attenuated in participants who had communication with relatives or friends several or more times per month. Maintaining frequent communications with friends or relatives may satisfy their expectation for intimate relationships, and thereby reduce the enhancement of dementia risk by emotional loneliness in the elderly. This finding raises the possibility that frequent communications with others may be a protective factor against the development of dementia. On the other hand, this finding could simply reflect the play of chance due to the small sample size.

The exact mechanisms underlying the association between loneliness and the risk of dementia are not clear at present. However, we consider that there are several possible mechanisms. Some neuroimaging studies conducted in humans have reported that participants with loneliness showed decreased gray matter volume in the left posterior superior temporal sulcus (Kanai et al., 2012) and decreased regional white-matter density in the bilateral inferior parietal lobe, right anterior insula, posterior temporoparietal junction, and other regions (Nakagawa et al., 2015). Additionally, other neuroimaging studies found that participants with social exclusion, which is closely related to emotional loneliness, had decreased activities in the dorsomedial prefrontal cortex, posterior cingulate cortex, and posterior insular cortex (Olié & Courtet, 2018; Powers et al., 2013). Since these brain regions have been reported to be related to attention control, executive function, and social brain functions such as theory of mind and mentalizing, the dysfunction or atrophy of these regions may be related to the development of dementia. In support of this possibility, an animal study showed that social isolation was associated with Alzheimer’s disease through decreased brain-derived neurotrophic factor and increased oxidative stress (Hsiao et al., 2018). Since social isolation is likely to cause the feeling of loneliness, a similar mechanism may underlie the association between loneliness and increased risk of dementia.

The present study showed a significant association between emotional loneliness—rather than social loneliness—and dementia risk even after adjusting for lifestyle, physical health, and depression. Emotional loneliness is defined as a subjective perception of an absence of close or intimate relationships, whereas social loneliness is defined as a subjective perception of a lack of social networks. Since close relationships are more difficult to achieve than social networks, emotional loneliness is assumed to be more psychologically stressful than social loneliness (Blow et al., 2019). In addition, emotional loneliness has been reported to make a greater contribution to low life satisfaction, which is considered to be a risk factor for depressive symptoms and cognitive impairment, than social loneliness (Peitsch et al., 2016; Salimi, 2011). Moreover, it has been shown that emotional loneliness is associated with low self-esteem and attachment problems, both of which are mainly developed through negative experiences in early life (Dahlberg & McKee, 2014; DiTommaso, 1997). Previous studies revealed that adverse childhood experiences (e.g., maltreatment) and long-standing stress from early life were associated with lower hippocampal volume (Lupien et al., 2009; Riem et al., 2015). Further investigations of the differential influences of emotional and social loneliness on mental health are warranted.

The strengths of our study are as follows.

1. This is the first population-based prospective cohort study clarifying the influence of loneliness and its component subscales on dementia.

2. Loneliness was measured by using the multi-item loneliness scale, which has been well examined in terms of reliability, validity, and external consistency.

3. Dementia was accurately diagnosed by experts and the study had a high follow-up rate (100%).

However, this study also had several limitations. First, we could not rule out the possibility of bias due to residual confounding effects (e.g., genetic factors, borderline mental retardation), although we accounted for a wide range of confounders. Second, there is a possibility that more cases of prodromal dementia were included in the participants with loneliness at baseline. However, the sensitivity analysis after excluding participants with MMSE of <24 points at baseline (Supplementary Table 1) or dementia cases occurring within the first one year of follow-up (Supplementary Table 2) did not alter the present findings substantially. Third, in the subgroup analyses of social isolation factors, the sample sizes of some of the subgroups were small. Fourth, we cannot deny the possibility of selection bias, because 25% of subjects (=378/1,519) were excluded from the analysis. To examine the potential for selection bias, we analyzed the difference in characteristics between the included and excluded participants (Supplementary Table 3). We found that the excluded subjects were older, and showed higher risk of incident dementia, and higher frequencies of low education level, unemployment, hypertension, history of cardiovascular disease, and depression, and lower frequencies of current smoking and the presence of regular exercise than the included subjects. In addition, they were more likely to be socially isolated. Therefore, the findings of the present study may have underestimated the association. Finally, we urge caution in generalizing the findings to populations with different backgrounds, because participants were recruited from a single town in Japan.

In conclusion, the present study demonstrated that loneliness, especially emotional loneliness, was significantly associated with an elevated risk of dementia. Subgroup analysis revealed that this association existed in the participants who had a partner or lived with someone. The present findings suggest that social supports for easing emotional loneliness in the elderly by helping them to maintain social communications with others, participate frequently in social activities and avoid social exclusion may be beneficial for reducing the risk of dementia. The present findings should be validated in other populations of different races and with different cultures.

Funding

This work was supported in part by Grants-in-Aid for Scientific Research (A) (JP16H02692), (B) (JP16H05850, JP17H04126, JP18H02737, and JP19H03752), and (C) (JP17K09114, JP17K09113, JP17K01853, JP18K07565, JP18K09412, JP19K07890, JP26460911, and JP17K09304) and Early-Career Scientists (JP18K17925 and JP18K17382) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; by Health and Labour Sciences Research Grants of the Ministry of Health, Labour and Welfare of Japan (H29-Junkankitou-Ippan-003, H30-Shokuhin-[Sitei]-005, and 201811071A); and by the Japan Agency for Medical Research and Development (JP19dk0207025, JP19ek0210082, JP19ek0210083, JP19km0405202, JP19ek0210080, JP19fk0108075, JP18ek0610015h0002, and JP19ek0610015h0003).

Conflict of Interest

None declared.

Acknowledgments

We thank the staff members of the Division of Health and Welfare of Hisayama for their cooperation in this study. We thank Ryota Nakayama for technical support with the electronic processing of the data. We are very grateful to Professor Yoshinao Oda, Professor Toru Iwaki, and their colleagues at the Department of Anatomic Pathology and Department of Neuropathology, Graduate School of Medical Sciences, Kyushu University, who provided expertise and insight into the autopsy findings that greatly assisted our research. The statistical analyses were carried out using the computer resources offered under the category of General Projects by the Research Institute for Information Technology, Kyushu University. This study was not preregistered.

Data Availability

The datasets used in the current study are not publicly available because they contain confidential clinical data on the study participants. However, the data are available on reasonable request and with the permission of the Principal Investigator of the Hisayama Study, Toshiharu Ninomiya (Department of Epidemiology and Public Health, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan).

References