Loneliness is defined as subjective distress resulting from a less than desired level of social connectedness (Hawkley and Cacioppo, Reference Hawkley and Cacioppo2010). The deleterious effects of loneliness on a range of health outcomes such as stroke or coronary artery disease (Donovan and Blazer, Reference Donovan and Blazer2020), as well as premature mortality (Donovan and Blazer, Reference Donovan and Blazer2020; Holt-Lunstad et al., Reference Holt-Lunstad, Smith, Baker, Harris and Stephenson2015), have been documented. To date, however, literature on the relation between loneliness and dementia risk has been mixed, with some (Sundström et al., Reference Sundström, Adolfsson, Nordin and Adolfsson2020; Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020), but not all (Joyce et al., Reference Joyce2022) studies observing significant associations.
Using data from 492,322 individuals from the UK Biobank, Sutin and colleagues (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) examined the relation between loneliness and risk of incident all-cause dementia over 15 years. Loneliness was assessed using a single item and diagnoses of dementia were derived from health and death records. This study showed that endorsement of loneliness was associated with a 45% increased risk of all-cause dementia, which ranged from 30% increased risk of Alzheimer’s disease to 57% increased risk of vascular dementia. Of note, these associations were independent of a broad range of risk factors for dementia, including age, depression, physical exercise, and vascular risk factors. Furthermore, the association between loneliness and risk of all-cause dementia and Alzheimer’s disease was stronger in individuals who did not carry the apolipoprotein E (APOE) ε4 allele, suggesting that loneliness may be a key moderating risk factor for dementia in these individuals.
Results of Sutin et al.’s (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) study complement those of a recent longitudinal study that examined latent social network types and cognitive function in 50,071 European adults aged 50 years or older (Cohn-Schwartz et al., Reference Cohn-Schwartz, Levinsky and Litwin2021). In this study, older adults with more varied sources of network ties (e.g. friendship ties and/or several types of family relationships) performed better on measures of cognitive function (e.g. immediate recall, delayed recall, and fluency) at a 2-year follow-up relative to older adults who had network ties with mainly relatives other than their spouse and children. The authors concluded that older adults in the latter group may have increased risk for cognitive decline and could benefit from interventions to enrich their social-interpersonal environments. Similarly, in another longitudinal study of 897 older adults, having a larger number of close friends at a baseline assessment was associated with slower decline in executive function over six years (Ihle et al., Reference Ihle2021). Interestingly, this association was mediated by leisure activities at baseline, suggesting that social reserve may help promote greater engagement in leisure activities, which may in turn help preserve executive function in late-life.
The study by Sutin and colleagues has several notable strengths. First, the large, well-characterized UK Biobank cohort with linked hospital and death records and 15 years of follow-up data provides a unique opportunity to examine the relation between loneliness and incident dementia risk. Specific clinical diagnoses also allowed the investigators to identify unique associations with rarer causes of dementia, such as frontotemporal dementia. Second, the UK Biobank data set contains rich data on clinical and behavioral factors that may confound the association between loneliness and dementia risk, as well as information regarding APOE ε4 status. Third, in addition to self-report measures, the investigators included objectively assessed measures of body mass index, C-reactive protein, low-density lipoprotein, and hemoglobin A1c. Fourth, the analytic approach was systematic and considered key variables for adjustment in a staged manner, as well as sensitivity analyses to evaluate the robustness of model estimates.
There were, however, some notable limitations of this study. First, the measurement of loneliness was based on a single question: “Do you often feel lonely?” which had a Yes/No response. While this face-valid item has demonstrated similar association with dementia risk (Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020) and other cognitive outcomes (Sundström et al., Reference Sundström, Adolfsson, Nordin and Adolfsson2020) relative to multi-item scales, it does not allow the examination of whether particular aspects of loneliness (e.g. satisfaction with social relationships) or loneliness in general is more strongly related to dementia risk. Second, loneliness data were unavailable for a large number of participants (N = 9869). As the reasons for this missing data were unclear, it is possible that the observed associations may be underestimated or contain some bias, particularly in light of prior work suggesting an association between nonresponse to survey items and mental disorders (Kessler et al., Reference Kessler, Little and Groves1995). Third, while the authors considered a wide range of potential confounding variables that were available in the UK Biobank data set, there are several other dementia risk factors and indicators, such as mild cognitive impairment, substance use disorders, concussion/traumatic brain injury, posttraumatic stress disorder (PTSD), and engagement in cognitively stimulating activities, that were not examined. Further, loneliness has been linked to greater amyloid (Donovan et al., Reference Donovan2016) and tau (Uquillas et al., Reference Uquillas2018) pathology in cognitively normal older adults, so it is possible that the association between loneliness and dementia risk may, at least in part, be related to the early manifestation of these neuropathological changes. It is important to note that loneliness is a strong risk factor for Alzheimer’s disease (Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020) as well as depression (Wang et al., Reference Wang, Mann, Lloyd-Evans, Ma and Johnson2018). Therefore, further studies are needed to elucidate the shared and/or unique variance of loneliness and depression, as well as other psychiatric factors (e.g. PTSD) in contributing to dementia risk. Indeed, attenuation of observed associations between loneliness and incident dementia risk after accounting for clinical and behavioral factors such as depression suggest that these factors may potentially mediate this association. Further research is needed to identify factors that may play key roles and serve as candidate mechanisms. Finally, given that all predictors of incident dementia, including loneliness, were assessed at baseline in this study, it is unclear whether time-varying changes in loneliness or certain risk factors, such as increasing depressive symptoms, may mediate the link between baseline loneliness and dementia risk.
Despite these limitations, results of Sutin and colleagues’ study provide critical new insight into the importance of loneliness as a modifiable risk factor for dementia. Their findings underscore the importance of assessing loneliness as part of dementia risk assessments; monitoring loneliness in individuals at risk for dementia; and treating loneliness as part of dementia prevention efforts. Furthermore, the results suggest that a single-item measure of loneliness may be useful in predicting dementia risk. Given the robustness of the observed associations in this study, it may be useful to further validate this single-item measure, as well as other brief standardized measures of loneliness.
In a recent commentary (Na et al., Reference Na, Jeste and Pietrzak2022), we outlined five recommendations to help address the global behavioral epidemic of loneliness. First, we suggested educating the public and medical community about the negative health impact of loneliness, as well as the importance of social connectedness in medicine and health care more generally. The goal for such educational initiatives would be to help health professionals recognize the importance of loneliness and social connectedness and to implement strategies to assess, monitor, and treat loneliness and promote social connection in health care settings.
Second, we called for increased health policy efforts to build a public health infrastructure to mitigate loneliness. Several population-based interventions have been implemented to mitigate loneliness in different nations. In the UK, there are ongoing efforts to promote “social prescribing” in order to improve the health of individuals who are identified as lonely by their primary care physicians (Roland et al., Reference Roland, Everington and Marshall2020). In this model, those who are referred are connected to a volunteer (i.e. “link”) worker who helps connect individuals to community- and voluntary sector-based groups who can provide social and emotional support. In the USA, the Veterans Health Administration has a similar program called Compassionate Contact Corps (Taylor, Reference Taylor2021), in which clinicians refer veterans who are lonely and connect them with a volunteer worker who calls them weekly via phone or videoconferencing. Emerging research has revealed that social prescribing is associated with health care cost savings, particularly among older and non-White older adults, and those with higher levels of engagement with the intervention (Wildman and Wildman, Reference Wildman and Wildman2022). However, further research that employs rigorous methodologies is needed to evaluate the effectiveness of such interventions in mitigating loneliness and related outcomes.
Third, we underscored the need for research grants and resources to support the development of assessment instruments to assess loneliness and social connectedness and evaluation of health policy efforts and clinical interventions to help promote social connectedness. Given the pervasiveness of loneliness and its link to a broad range of mental, physical, and cognitive health outcomes (Holt-Lunstad et al., Reference Holt-Lunstad, Smith, Baker, Harris and Stephenson2015), it represents an intervention target that cuts across diagnostic boundaries and has broad relevance within diverse health care settings.
Fourth, we highlighted the importance of developing and validating measures to assess and monitor loneliness in health care settings. Currently, there are no validated brief standardized measures of loneliness. Therefore, more research is needed to develop and validate measures of loneliness that are easily administered. For example, validating the question used in Sutin and colleagues’ study (i.e. “Do you often feel lonely?”) against validated research measures of loneliness (e.g. the Revised UCLA Loneliness Scale (Russell et al., Reference Russell, Peplau and Cutrona1980) may yield an easily deployable question—similar to those used to screen for alcohol and nicotine use—that can be rapidly administered in health care settings. This would enable primary care providers to use such a measure to screen and monitor older adults who may be at increased risk for dementia.
Finally, we suggested rigorous evaluation of clinical interventions that target loneliness. There have been increasing efforts to develop and implement clinical interventions to help mitigate loneliness and help promote health outcomes. For example, a recent pilot controlled clinical trial of a tele-health resilience- and wisdom-focused intervention in 20 older adults showed a small-to-medium effect size reduction in loneliness, as well as perceived stress (Jeste et al., Reference Jeste2023). To date, however, few large-scale randomized clinical trials to mitigate loneliness have been conducted (Kahlon et al., Reference Kahlon2021). Furthermore, large-scale RCTs are critically needed to evaluate the efficacy and effectiveness of interventions to mitigate loneliness and bolster social connectedness in different populations.
Social restrictions of the COVID-19 pandemic have led to much of the world experiencing first-hand the profound negative impact of loneliness. The results of Sutin et al.’s (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) study, as well as a large and growing body of literature, support the need for the health care community to develop comprehensive strategies to address this public health crisis.
You have
Access
Loneliness is defined as subjective distress resulting from a less than desired level of social connectedness (Hawkley and Cacioppo, Reference Hawkley and Cacioppo2010). The deleterious effects of loneliness on a range of health outcomes such as stroke or coronary artery disease (Donovan and Blazer, Reference Donovan and Blazer2020), as well as premature mortality (Donovan and Blazer, Reference Donovan and Blazer2020; Holt-Lunstad et al., Reference Holt-Lunstad, Smith, Baker, Harris and Stephenson2015), have been documented. To date, however, literature on the relation between loneliness and dementia risk has been mixed, with some (Sundström et al., Reference Sundström, Adolfsson, Nordin and Adolfsson2020; Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020), but not all (Joyce et al., Reference Joyce2022) studies observing significant associations.
Using data from 492,322 individuals from the UK Biobank, Sutin and colleagues (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) examined the relation between loneliness and risk of incident all-cause dementia over 15 years. Loneliness was assessed using a single item and diagnoses of dementia were derived from health and death records. This study showed that endorsement of loneliness was associated with a 45% increased risk of all-cause dementia, which ranged from 30% increased risk of Alzheimer’s disease to 57% increased risk of vascular dementia. Of note, these associations were independent of a broad range of risk factors for dementia, including age, depression, physical exercise, and vascular risk factors. Furthermore, the association between loneliness and risk of all-cause dementia and Alzheimer’s disease was stronger in individuals who did not carry the apolipoprotein E (APOE) ε4 allele, suggesting that loneliness may be a key moderating risk factor for dementia in these individuals.
Results of Sutin et al.’s (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) study complement those of a recent longitudinal study that examined latent social network types and cognitive function in 50,071 European adults aged 50 years or older (Cohn-Schwartz et al., Reference Cohn-Schwartz, Levinsky and Litwin2021). In this study, older adults with more varied sources of network ties (e.g. friendship ties and/or several types of family relationships) performed better on measures of cognitive function (e.g. immediate recall, delayed recall, and fluency) at a 2-year follow-up relative to older adults who had network ties with mainly relatives other than their spouse and children. The authors concluded that older adults in the latter group may have increased risk for cognitive decline and could benefit from interventions to enrich their social-interpersonal environments. Similarly, in another longitudinal study of 897 older adults, having a larger number of close friends at a baseline assessment was associated with slower decline in executive function over six years (Ihle et al., Reference Ihle2021). Interestingly, this association was mediated by leisure activities at baseline, suggesting that social reserve may help promote greater engagement in leisure activities, which may in turn help preserve executive function in late-life.
The study by Sutin and colleagues has several notable strengths. First, the large, well-characterized UK Biobank cohort with linked hospital and death records and 15 years of follow-up data provides a unique opportunity to examine the relation between loneliness and incident dementia risk. Specific clinical diagnoses also allowed the investigators to identify unique associations with rarer causes of dementia, such as frontotemporal dementia. Second, the UK Biobank data set contains rich data on clinical and behavioral factors that may confound the association between loneliness and dementia risk, as well as information regarding APOE ε4 status. Third, in addition to self-report measures, the investigators included objectively assessed measures of body mass index, C-reactive protein, low-density lipoprotein, and hemoglobin A1c. Fourth, the analytic approach was systematic and considered key variables for adjustment in a staged manner, as well as sensitivity analyses to evaluate the robustness of model estimates.
There were, however, some notable limitations of this study. First, the measurement of loneliness was based on a single question: “Do you often feel lonely?” which had a Yes/No response. While this face-valid item has demonstrated similar association with dementia risk (Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020) and other cognitive outcomes (Sundström et al., Reference Sundström, Adolfsson, Nordin and Adolfsson2020) relative to multi-item scales, it does not allow the examination of whether particular aspects of loneliness (e.g. satisfaction with social relationships) or loneliness in general is more strongly related to dementia risk. Second, loneliness data were unavailable for a large number of participants (N = 9869). As the reasons for this missing data were unclear, it is possible that the observed associations may be underestimated or contain some bias, particularly in light of prior work suggesting an association between nonresponse to survey items and mental disorders (Kessler et al., Reference Kessler, Little and Groves1995). Third, while the authors considered a wide range of potential confounding variables that were available in the UK Biobank data set, there are several other dementia risk factors and indicators, such as mild cognitive impairment, substance use disorders, concussion/traumatic brain injury, posttraumatic stress disorder (PTSD), and engagement in cognitively stimulating activities, that were not examined. Further, loneliness has been linked to greater amyloid (Donovan et al., Reference Donovan2016) and tau (Uquillas et al., Reference Uquillas2018) pathology in cognitively normal older adults, so it is possible that the association between loneliness and dementia risk may, at least in part, be related to the early manifestation of these neuropathological changes. It is important to note that loneliness is a strong risk factor for Alzheimer’s disease (Sutin et al., Reference Sutin, Stephan, Luchetti and Terracciano2020) as well as depression (Wang et al., Reference Wang, Mann, Lloyd-Evans, Ma and Johnson2018). Therefore, further studies are needed to elucidate the shared and/or unique variance of loneliness and depression, as well as other psychiatric factors (e.g. PTSD) in contributing to dementia risk. Indeed, attenuation of observed associations between loneliness and incident dementia risk after accounting for clinical and behavioral factors such as depression suggest that these factors may potentially mediate this association. Further research is needed to identify factors that may play key roles and serve as candidate mechanisms. Finally, given that all predictors of incident dementia, including loneliness, were assessed at baseline in this study, it is unclear whether time-varying changes in loneliness or certain risk factors, such as increasing depressive symptoms, may mediate the link between baseline loneliness and dementia risk.
Despite these limitations, results of Sutin and colleagues’ study provide critical new insight into the importance of loneliness as a modifiable risk factor for dementia. Their findings underscore the importance of assessing loneliness as part of dementia risk assessments; monitoring loneliness in individuals at risk for dementia; and treating loneliness as part of dementia prevention efforts. Furthermore, the results suggest that a single-item measure of loneliness may be useful in predicting dementia risk. Given the robustness of the observed associations in this study, it may be useful to further validate this single-item measure, as well as other brief standardized measures of loneliness.
In a recent commentary (Na et al., Reference Na, Jeste and Pietrzak2022), we outlined five recommendations to help address the global behavioral epidemic of loneliness. First, we suggested educating the public and medical community about the negative health impact of loneliness, as well as the importance of social connectedness in medicine and health care more generally. The goal for such educational initiatives would be to help health professionals recognize the importance of loneliness and social connectedness and to implement strategies to assess, monitor, and treat loneliness and promote social connection in health care settings.
Second, we called for increased health policy efforts to build a public health infrastructure to mitigate loneliness. Several population-based interventions have been implemented to mitigate loneliness in different nations. In the UK, there are ongoing efforts to promote “social prescribing” in order to improve the health of individuals who are identified as lonely by their primary care physicians (Roland et al., Reference Roland, Everington and Marshall2020). In this model, those who are referred are connected to a volunteer (i.e. “link”) worker who helps connect individuals to community- and voluntary sector-based groups who can provide social and emotional support. In the USA, the Veterans Health Administration has a similar program called Compassionate Contact Corps (Taylor, Reference Taylor2021), in which clinicians refer veterans who are lonely and connect them with a volunteer worker who calls them weekly via phone or videoconferencing. Emerging research has revealed that social prescribing is associated with health care cost savings, particularly among older and non-White older adults, and those with higher levels of engagement with the intervention (Wildman and Wildman, Reference Wildman and Wildman2022). However, further research that employs rigorous methodologies is needed to evaluate the effectiveness of such interventions in mitigating loneliness and related outcomes.
Third, we underscored the need for research grants and resources to support the development of assessment instruments to assess loneliness and social connectedness and evaluation of health policy efforts and clinical interventions to help promote social connectedness. Given the pervasiveness of loneliness and its link to a broad range of mental, physical, and cognitive health outcomes (Holt-Lunstad et al., Reference Holt-Lunstad, Smith, Baker, Harris and Stephenson2015), it represents an intervention target that cuts across diagnostic boundaries and has broad relevance within diverse health care settings.
Fourth, we highlighted the importance of developing and validating measures to assess and monitor loneliness in health care settings. Currently, there are no validated brief standardized measures of loneliness. Therefore, more research is needed to develop and validate measures of loneliness that are easily administered. For example, validating the question used in Sutin and colleagues’ study (i.e. “Do you often feel lonely?”) against validated research measures of loneliness (e.g. the Revised UCLA Loneliness Scale (Russell et al., Reference Russell, Peplau and Cutrona1980) may yield an easily deployable question—similar to those used to screen for alcohol and nicotine use—that can be rapidly administered in health care settings. This would enable primary care providers to use such a measure to screen and monitor older adults who may be at increased risk for dementia.
Finally, we suggested rigorous evaluation of clinical interventions that target loneliness. There have been increasing efforts to develop and implement clinical interventions to help mitigate loneliness and help promote health outcomes. For example, a recent pilot controlled clinical trial of a tele-health resilience- and wisdom-focused intervention in 20 older adults showed a small-to-medium effect size reduction in loneliness, as well as perceived stress (Jeste et al., Reference Jeste2023). To date, however, few large-scale randomized clinical trials to mitigate loneliness have been conducted (Kahlon et al., Reference Kahlon2021). Furthermore, large-scale RCTs are critically needed to evaluate the efficacy and effectiveness of interventions to mitigate loneliness and bolster social connectedness in different populations.
Social restrictions of the COVID-19 pandemic have led to much of the world experiencing first-hand the profound negative impact of loneliness. The results of Sutin et al.’s (Reference Sutin, Luchetti, Aschwanden, Zhu, Stephan and Terracciano2022) study, as well as a large and growing body of literature, support the need for the health care community to develop comprehensive strategies to address this public health crisis.
Conflict of interest
Dr. Na has received royalty from Wolters Kluwer. Drs. Lim and Pietrzak report no competing interests.