Research in Gerontological Nursing

Instrument Development 

Measuring Social Isolation in the National Health and Aging Trends Study

Janet S. Pohl, PhD, RN; Barbara B. Cochrane, PhD, RN, FAAN; Karen G. Schepp, PhD, RN, FAAN; Nancy F. Woods, PhD, RN, FAAN

Abstract

The objectives of the current study were to describe the development of a social isolation measure based on Berkman and Syme's Social Network Index domains with data from the National Health and Aging Trends Study. A descriptive correlational design was used, establishing convergent and divergent validity of the measure with depression risk and well-being. Depression risk was measured with the Patient Health Questionnaire-2 and an ordinal well-being measure was developed based on measures in MIDUS–Midlife in the U.S.–A Study of National Health and Well-Being. Participants who scored ≥4 (cutoff point) on the social isolation measure represented 21.9% (N = 7,609) of the sample (95% confidence interval [20.6, 23.3]). Spearman's correlation with depression was 0.23 (p < 0.001) and −0.24 (p ≤ 0.001) with well-being. The weighted data complex samples general linear model with depression (R = 0.22, p ≤ 0.001) and well-being (R = −0.26, p ≤ 0.001) confirm the relationships. This measure offers conceptual clarity and measurement consistency for developing the components and targets for future social isolation intervention research.

[Res Gerontol Nurs. 2017; 10(6):277–287.]

Abstract

The objectives of the current study were to describe the development of a social isolation measure based on Berkman and Syme's Social Network Index domains with data from the National Health and Aging Trends Study. A descriptive correlational design was used, establishing convergent and divergent validity of the measure with depression risk and well-being. Depression risk was measured with the Patient Health Questionnaire-2 and an ordinal well-being measure was developed based on measures in MIDUS–Midlife in the U.S.–A Study of National Health and Well-Being. Participants who scored ≥4 (cutoff point) on the social isolation measure represented 21.9% (N = 7,609) of the sample (95% confidence interval [20.6, 23.3]). Spearman's correlation with depression was 0.23 (p < 0.001) and −0.24 (p ≤ 0.001) with well-being. The weighted data complex samples general linear model with depression (R = 0.22, p ≤ 0.001) and well-being (R = −0.26, p ≤ 0.001) confirm the relationships. This measure offers conceptual clarity and measurement consistency for developing the components and targets for future social isolation intervention research.

[Res Gerontol Nurs. 2017; 10(6):277–287.]

Social isolation in community-dwelling older adults can have life-altering outcomes that may profoundly influence individual health. Expanding knowledge regarding how social isolation affects older adult health is important because socially isolated older adults are at increased risk for multiple morbidities and mortality (Berkman & Syme, 1979; Saito, Kondo, Kondo, Ojima, & Hirai, 2012). However, research on this geriatric condition has been limited by a lack of conceptual clarity, resulting in numerous approaches to its measurement, difficulty in making comparisons across studies, and challenges in identifying appropriate components and targets of intervention development. Finding measurement approaches that reflect a consistent conceptual foundation to the study of social isolation is central to future research on factors that affect the health of older adults.

Social isolation is the absence of contact with other individuals and a lack of relationships and social integration (Coyle & Dugan, 2012; Hawton et al., 2011). The definition used in this current measurement study draws on the aforementioned definition and focuses on participation as a function of social networks and social integration, rather than focusing on perceptions. Social isolation is a social circumstance in which older adults have a deficiency of network contacts as well as a deficiency of integrating relationships with those contacts. Social isolation, as measured in the current study, addresses deficiency of network contacts and integrating relationships.

Social isolation can also have consequences for society, particularly when current trends in aging are considered. The population of older Americans has increased 10-fold in the past 100 years, and by 2030 an estimated 70 million, or 20%, of Americans will have passed their 65th birthday (Johnson, Bulot, & Johnson, 2008). Previous research indicates the prevalence of social isolation ranges from 15% to 43% (Iliffe et al., 2007; Nicholson, 2012; Taylor, Herbers, Talisman, & Morrow-Howell, 2016). This range in prevalence may in part be due to differences in social isolation definitions, measurement, and/or reporting. Research on social isolation is needed to enhance understanding, guide health care, and inform policy, but clarity and consistency in its definition and measurement are needed to ensure estimates and trends are as precise as possible.

The consequences of social isolation on individual health are seen in psychological and physiological health and health behavior. Psychologically, social interaction is important for preserving mental functioning in older adults, and a small social network is a risk factor for cognitive impairment (Bassuk, Glass, & Berkman, 1999; DiNapoli, Wu, & Scogin, 2014). Being socially isolated is positively associated with depression (Dorfman et al., 1995; Ebner, Maura, Macdonald, Westberg, & Fischer, 2013; Lee, Hasche, Choi, Proctor, & Morrow-Howell, 2013; McCrae et al., 2005) and negatively associated with high well-being scores (Bradley, 2015; Momtaz, Hamid, Ibrahim, Yahaya, & Chai, 2011). In addition, social isolation affects physical health. Coyle and Dugan (2012) found that individuals who reported social isolation had a 43% higher odds of also reporting fair or poor health compared to those who reported no or minimal social isolation. There is a wide range of health outcomes associated with social isolation, including general health, cardiovascular disease, and cancer (Courtin & Knapp, 2017; Uchino, 2006). There is also considerable evidence indicating that older adults who have few network contacts and integrating relationships are at increased risk for all-cause mortality (Berkman & Syme, 1979; Holt-Lunstad, Smith, Baker, Harris, & Stephenson, 2015). Social isolation is also associated with certain health-related behaviors, such as smoking and low physical activity (Pantell et al., 2013; Shankar, McMunn, Banks, & Steptoe, 2011).

Social Isolation Measured as Social Network

According to Victor, Scambler, and Bond (2008), there are two approaches to the measurement of social isolation: (a) counting the number of social contacts within a defined period of time or space; and (b) performing a calculation using indicators of isolation that combine measures related to network and integration, such as the availability of family and friends and varying aspects and types of social participation. A count of contacts was used as a single-item indicator of social isolation in the Rand Social Health Battery (Hawton et al., 2011), and a count of contacts measures social isolation in the Life Space Index (Baker, Bodner, & Allman, 2003; Havens, Hall, Sylvestre, & Jivan, 2004). Single-item indicators provide hints of network contacts, but do not examine integration of relationships. Although not technically a counting indicator, living alone status is also a single-item indicator that does not adequately measure isolation because older adults are more likely to live alone and may continue to be very active within their social networks (Lubben, 1988).

Measures that calculate indicators of social networks and integration domains include the Social Network Index (SNI), Lubben Social Network Scale (LSNS), and Lubben Social Network Scale-6 (LSNS-6). The SNI has been used in research to assess four social domains of network and integration in the general adult population: (a) marriage or partnership, (b) family and friends, (c) church participation, and (d) club participation (Berkman & Syme, 1979). The marriage or partnership domain encompasses an integrating relationship that is specific to this type of dyadic connection (Berkman, 1977). The original SNI was constructed by Berkman with secondary data from the Human Population Laboratory Study of Adults in Alameda County, California (Berkman, 1977). Berkman and Syme (1979) suggest that the SNI is a strong predictor of health and mortality outcomes. The SNI domains provide a foundation for the measure developed in the current article. Berkman (1977) has indicated that although the SNI has not been evaluated for reliability, the four domains provide strong face validity for the index. The SNI is not a validated index. However, the domains have been used within research to construct measures of social network and integration with data items that are not consistent across studies. Further, Nicholson (2010) reported that continued use of the SNI domains over the decades supports its credibility for identifying those who have small social networks, as borne out in published analyses of the index domains (Ford, Loucks, & Berkman, 2006; Heffner, Waring, Roberts, Eaton, & Gramling, 2011; Pantell et al., 2013; Shankar et al., 2011). The SNI domains are used in these studies with indicators that are specific to the data items in each study. The original LSNS, a refinement of the Berkman-Syme SNI (Lubben, 1988), was not developed for use in research, but instead for clinical assessment of older adult populations, with the LSNS-6 developed as a more concise social network assessment tool (Lubben et al., 2006).

The domains from the SNI are used in the current study, but the concept measured is a deficiency of network contacts and integrating relationships. The SNI and LSNS examine network contacts and integrating relationships and do not measure this deficiency. To date, a multiple-indicator measure of social isolation has not been found in the research literature. The measurement of isolation, and not network, with multiple-indicator measures in the study of social isolation will add conceptual clarity to research. The trends in aging and life-altering outcomes of social isolation highlight the need for a multiple-indicator measure of participation that will lead to expanding knowledge regarding how social isolation influences older adult health.

The source of the data for the current study was the National Health and Aging Trends Study (NHATS). Measuring social isolation was limited in the original NHATS protocol. Previous research using NHATS data used one item that identified whether participants had someone with whom to talk about important things as a measure of social isolation (Choi & DiNitto, 2015). A reporting of no one with whom to talk about important things was used to identify a participant as isolated. The current study focused on the identification of a new social isolation measure using NHATS items based on the SNI domains, expanding the family and friends domain, and measuring the concept of social isolation as a deficiency of network contacts and integrating relationships.

Precision in measuring social isolation is needed for appropriate comparisons of findings across studies and future development of social isolation interventions. The NHATS is a resource for the scientific study of functioning in later life and was used in the current study because it provides items that are consistent with the domains of the SNI and a focus on participation. The measure developed will provide a framework for future social isolation investigations with the NHATS data and other data sets. The measure also offers the conceptual clarity and measurement consistency for developing the components and targets for future intervention research.

The purpose of the current article is to describe the development of a new measure that expands the SNI domains of network contacts and integrating relationships to measure social isolation and establish the usefulness of the measure by examining the relationships between social isolation and depression risk (to establish convergent validity) and well-being (to establish divergent validity) in the NHATS. A secondary goal of this work is to establish a cutoff point for the measure of social isolation as a way to analyze social isolation status in future research.

Method

This measurement study involved a secondary analysis of cross-sectional (baseline) data from the NHATS using a descriptive correlational design to examine the association of a new social isolation measure with depression risk and well-being in a sample of older adults. The NHATS is an ongoing study that was designed originally to support the study of disability trends and dynamics in a nationally representative sample of non-nursing home Medicare beneficiaries age 65 or older (Kasper & Freedman, 2015).

Sample

Random subsamples from the Centers for Medicare & Medicaid Services Medicare enrollment database served as the sampling frame for NHATS (Montaquila, Freedman, Edwards, & Kasper, 2012). The NHATS study design includes baseline stratification of 5-year age groups (from 65 to 89, and 90+), an oversampling of those older than 90, and an oversampling of Black individuals. Participant interviews served as the main source of data for the NHATS. The inclusion criterion for the current study was completion of the baseline participant interview (N = 7,609). The participant interview was administered at baseline to all participants living in the community or residential care settings other than nursing homes. There were no exclusion criteria.

NHATS protocols were approved by the Johns Hopkins University review board, and all participants provided written informed consent. The University of Washington Human Subjects Division confirmed that the current study did not meet the definition of research involving human subjects because the data were de-identified. Access to the data required registering with the NHATS website ( https://www.nhats.org) before downloading the files. More sensitive demographic data, such as age, were available through a simple application process.

NHATS data collection involved annual in-person interviews that started in 2011; the baseline data were used for the current project. Specifically, the NHATS data in three different baseline data sets were used: (a) the participant or sample person data set contained data on NHATS respondents collected in the participant interviews; (b) the other person data set included data about a person identified based on one or more roles relative to the participant interviewed; and (c) the sensitive data set provided demographic data regarding the participant.

Measures

The development of the social isolation measure for the current study was based on research by Berkman (1977) and Berkman and Syme (1979). The measure was calculated using the following NHATS data items: marriage or partnership; family and friends; church participation; and club participation (Table 1). Contact with family or friends was determined by recoding responses to the item, “Looking back over the last year, who are the people you talked with most often about important things?” into separate variables identifying family and then friends. Based on previous research, increased weight was given to the family and friends domain by adding an item that indicates if the participant visited with family or friends in person (Faulkner, Cauley, Zmuda, Griffin, & Nevitt, 2003). All items were recoded to ensure that a higher score on the measure indicated greater isolation. One point was recorded for each item that had a negative (isolated) response, with the score calculated as the sum of those negative responses. Unlike the SNI, the relevant NHATS data items were dichotomous. Therefore, the frequency of network contact and integration were not available for this new measure. Figure 1 (adapted from Nicholson [2010]) represents the relationships among social isolation (theoretical construct), the SNI network contact and integrating relationship domains (participation), and the indicators (items) in the NHATS interview. The indicator scores were summed to create an ordinal social isolation score with a range of 0 to 6. Predicting a cutoff score to indicate who was isolated and who was not will be important for identifying the prevalence of social isolation.

Indicators of the Social Isolation Measure (National Health and Aging Trends Study Interview Items)

Table 1:

Indicators of the Social Isolation Measure (National Health and Aging Trends Study Interview Items)

Conceptualization of the social isolation (theoretical construct), the Social Network Index network contact and integrating relationship domains (participation), and the indicators (items) in the National Health and Aging Trends Study interview.Adapted from Nicholson (2010).

Figure 1.

Conceptualization of the social isolation (theoretical construct), the Social Network Index network contact and integrating relationship domains (participation), and the indicators (items) in the National Health and Aging Trends Study interview.

Adapted from Nicholson (2010).

Social isolation and depression are positively related, in that depression in old age contributes to social isolation and social isolation contributes to depression (McCrae et al., 2005). Depression risk was measured with the Patient Health Questionnaire-2 (PHQ-2), a brief ordinal response screening instrument for depression that asks about the frequency of depressed mood and anhedonia over the past 2 weeks (Löwe et al., 2010). The PHQ-2 has demonstrated criterion and construct validity (Kroenke, Spitzer, & Williams, 2003). Possible scores for the PHQ-2 range from 0 to 6 (Table 2), with higher scores indicating possible depression risk and the need for further depression assessment (Löwe et al., 2010). The measure has documented reliability and validity in the general population (Kroenke et al., 2003; Löwe et al., 2010).

Patient Health Questionnaire-2 Measure of Depression Risk (National Health and Aging Trends Study Interview Items)

Table 2:

Patient Health Questionnaire-2 Measure of Depression Risk (National Health and Aging Trends Study Interview Items)

The relationship between social isolation and well-being is negatively correlated (Bradley, 2015; Momtaz et al., 2011). Well-being items in the NHATS data are similar to items that have appeared in the study, MIDUS–Midlife in the U.S.–A Study of National Health and Well-Being (Kasper & Freedman, 2015). The ordinal response items contain information regarding affect, self-realization, and sense of control and were recoded similar to methods used in MIDUS (2004), with higher scores reflecting more positive feelings, a greater level of self-realization, and a higher sense of control. The score represents the summed items, with a range of 3 to 41 (Table 3). With a demonstrated Cronbach's alpha of 0.75 and strong face validity of the items, it was determined that the items in this measure tap into the underlying concept of well-being.

Items Used to Develop the Measure of Well-Being (National Health and Aging Trends Study Interview Items)

Table 3:

Items Used to Develop the Measure of Well-Being (National Health and Aging Trends Study Interview Items)

Data Analysis

The current analysis used NHATS raw unweighted and weighted data. NHATS used analytic sample weights to account for unequal probabilities of selection in the NHATS sample (Montaquila, Freedman, Spillman, & Kasper, 2012). The weights accounted for differential selection of Black individuals and those older than 90 and were adjusted for nonresponse. Variance estimates were calculated using a Taylor series linearization (Montaquila, Freedman, Spillman, et al., 2012). Data management and statistical analysis were performed with SPSS version 23.

Identifying a cutoff score to indicate those who were isolated and those who were not (i.e., social isolation status) was achieved by theoretically placing participants into tertiles. Placing participants into tertiles or quartiles to identify cutoff points for a measure is often done in social research studies (Nicholson, 2010; Reginold et al., 2013; Rutledge, Matthews, Lui, Stone, & Cauley, 2003).

The demographic, household, and health characteristics of the sample were compared by social isolation status in round one using chi-square and independent t test analysis. The usefulness of the social isolation measure was determined by establishing convergent and divergent validity. The measures used in this analysis include ordinal level data; therefore, Spearman's rho correlations were examined. Examining the relationship between the social isolation measure and depression risk measure was performed to establish convergent validity. Examining the relationship between the social isolation measure and well-being measure was performed to establish divergent validity. Spearman's rho and a complex samples general linear model (CSGLM) procedure was performed to examine each relationship. Spearman's rho correlation was used to evaluate the unweighted data to determine how variation in one variable was associated with variation in another. The CSGLM procedure enabled a correlation analysis with the weighted data to determine the proportion of variance in social isolation that was accounted for by depression risk and well-being, as well as the proportion of the variance in depression risk and well-being that was accounted for by social isolation. This procedure included regressing one variable on the other and taking the square root of the resulting R2 value. The regression result was used to confirm the unweighted correlation. The correlation with depression risk was expected to be low-to-moderate in strength, positive, and statistically significant; and the correlation with well-being was expected to be low-to-moderate in strength, negative, and statistically significant.

Results

The distribution of sample characteristics at baseline are reported by social isolation status (isolated, not isolated) in Table 4. Socially isolated participants were more likely to be older, White non-Hispanic, less educated, and live alone than those who were not isolated. Participants who reported excellent health were five times more likely to not be isolated than isolated. Participants who were isolated had higher mean depression risk scores than those who were not isolated, whereas those who were isolated had lower mean well-being scores than those who were not isolated.

Distribution of National Health and Aging Trends Study Demographic, Household, and Health Characteristics According to Social Isolation Status (N = 7,609)

Table 4:

Distribution of National Health and Aging Trends Study Demographic, Household, and Health Characteristics According to Social Isolation Status (N = 7,609)

The characteristics of the sample were further examined according to the indicators of social isolation across each of its four domains. More than one half (57%) of the sample was married or living with a significant other. Four times as many participants reported talking with family compared to talking with friends (78% versus 18%, respectively). Approximately 90% of participants reported visiting others in either their own homes or the homes of family and friends. More than one half (57%) attended church services, and approximately 40% reported participating in club or group activities.

Cutoff Point for Social Isolation Measure Established

The range of 0 to 6 for possible scores was divided into tertiles. Scores of 0 or 1 represented those who were least isolated. Scores of 2 or 3 represented those who were somewhat isolated, and scores of 4, 5, or 6 represented those who were socially isolated.

Distribution of Social Isolation Scores

The distribution of social isolation scores is identified in Figure 2. Social isolation was reported by 1,983 participants (weighted 21.9%; 95% confidence interval [20.6, 23.3]).

Distribution of social isolation scores among participants in the National Health and Aging Trends Study (N = 7,609) (weighted %, mean = 2.56; standard error = 0.02).

Figure 2.

Distribution of social isolation scores among participants in the National Health and Aging Trends Study (N = 7,609) (weighted %, mean = 2.56; standard error = 0.02).

Relationship of Social Isolation With Depression Risk and Well-Being

Spearman's rho correlations were used to examine convergent and divergent validity of the social isolation measure. The convergent relationship between social isolation and depression risk was examined using unweighted data (r = 0.22, p ≤ 0.001). The CSGLM results confirmed Spearman's rho using weighted data. Both regression results were identical (R = 0.22, p ≤ 0.001). The divergent relationship between social isolation and well-being was examined using unweighted data (r = −0.24, p ≤ 0.001). The CSGLM results confirmed Spearman's rho using weighted data. Both regression results were identical (R = −0.26, p ≤ 0.001).

Discussion

The current study is the first to examine a well-conceptualized, comprehensive, and domain-inclusive measure of social isolation with data from the NHATS. This work provides a foundation for the development of social isolation measures in future studies to examine the strength of this geriatric condition as a predictor of health outcomes in NHATS and using other study data. The current study contributes to science by (a) constructing a measure of social isolation that measures network and integration in a direction that truly reflects isolation; (b) recognizing aspects of social isolation in other data sets available for secondary analysis may offer conceptually sound ways to construct a measure of the concept; (c) developing a simple measure with dichotomous items that may be used in research to sufficiently capture social isolation; and (d) demonstrating that weighting of the SNI domains, reflective of previous research, is a sound method for social isolation measure construction. The usefulness of the measure was determined with the establishment of its convergent relationship with depression risk and divergent relationship with well-being. Measuring social isolation reflective of participation with a multiple-indicator and domain-inclusive measure may be particularly useful for future social isolation intervention studies.

The SNI has been used in research since the reporting of the Alameda County study, which brought the predictive usefulness of the index to the forefront of the research literature (Berkman & Syme, 1979), and variations of this index are described in many other study reports. With its strong face validity, the SNI has been used in psychological and behavioral studies (Bassuk et al., 1999; Pantell et al., 2013; Shankar et al., 2011), but it is found most often in physiological research studies (Ford et al., 2006; Heffner et al., 2011). All but one study using the SNI described it as a measure of social networks and integration that incorporated positive responses to SNI domain questions. Shankar et al. (2011) used the SNI domains to measure social isolation specifically, in a similar fashion to the current study by focusing on the negative responses to the domain questions. The construction of a social isolation measure that examines a deficiency of network contacts and integrating relationships is conceptually consistent with the definition of social isolation.

The performance of the measure developed in the current study is recognized with the reported 21.9% prevalence of social isolation in the current sample. This rate falls within the low-middle range of prevalence reported previously in research. Further, the performance of this measure acknowledges the estimated cutoff point for future research. Based on this performance, the identification of social isolation aspects in other data sets available for secondary analysis may be a conceptually sound method for the construction of a social isolation measure.

A strength of the current study is that it builds on the well-established predictive history of the SNI and its continued use in psychological, physiological, and behavioral research. A further strength is the nationally representative sample of Medicare beneficiaries ages 65 and older in the NHATS. Although the original SNI examined the frequency of contact with family and friends, church participation, and club participation (Berkman & Syme, 1979), that information was not available in the current data set. However, using the SNI domains with the NHATS items provides a more robust measure than was previously available with the data set. Historically, the use of the SNI varies widely in the extent to which indicator frequency has been measured, with each study also measuring the domains but with moderately different indicators (Bassuk et al., 1999; Ford et al., 2006; Heffner et al., 2011; Pantell et al., 2013; Shankar et al., 2011). With other study data in the future, it may be possible to develop ordinal response items to measure social isolation and compare measurement properties between the two levels of measurement (ordinal and dichotomous). Furthermore, older participants may prefer simplified, dichotomous response options, and such a measure may demonstrate more consistent test–retest reliability.

The social isolation measure developed in the current study reflects previous research and measures multiple items in a direction that reflects the concept of social isolation. The study reflects current research findings on social isolation with an increased weight given to the family and friends domain by scoring family and friends separately and by adding the visits in-person indicator to the domain. Considering three items as part of the family and friends domain highlights the importance of social ties. “Family” has shown predictive strength in falls research using the LSNS (Faulkner et al., 2003). Unlike the SNI or LSNS, this measure of social isolation examines isolation by focusing on a deficiency of network contacts and integrating relationships. Precisely measuring social isolation and not networks in this manner adds conceptual clarity and will facilitate the comparison of results across future studies .

The relationships of social isolation with depression risk and well-being were evaluated to determine the usefulness of the social isolation measure. The establishment of convergent and divergent validity has been used to demonstrate the utility of a measure and determine what it does and does not measure (Matson, Fodstad, & Rivet, 2008). Generally, when testing the relationship between two theoretically similar constructs, a higher correlation may be expected. However, Koenig (2012) discussed psychosocial variables that are known clinically to be strongly related, noting that they seldom exceed a correlation of 0.25–0.30. It was expected that the correlations would be low to moderate in strength, but the direction of the relationships helped establish convergent and divergent validity of the measure. The positive correlation with depression risk was consistent with previous research literature that examined the relationship between depression and social isolation (Dorfman et al., 1995; Ebner et al., 2013; Lee et al., 2013; Nicholson, 2010). The negative correlation with well-being was consistent with previous research literature that examined the relationship between well-being and social isolation (Momtaz et al., 2011). The results of these analyses, the historical and ubiquitous use of the SNI, and the predictive validity of the SNI domains validate the use of this measure to examine social isolation.

Implications and Conclusion

A well-conceptualized, comprehensive, and domain-inclusive social isolation measure was developed that provides a foundation for the precise measurement of social isolation and key foundations for the next steps in research. Using multiple-indicator measures offers an opportunity to capture the important integrating aspects of social isolation. Those aspects may be missed with single-item measures, which do not support research on indicators of isolation that may inform the components and appropriate targets for social isolation interventions.

Examining a deficiency of social network and social integration and using multiple-indicator measures advances precision in social isolation measurement and research. Precision in future studies may be advanced by constructing measures of social isolation that measure network and integration in a direction that truly reflects isolation and identifying aspects of social isolation in other data sets to construct a measure of the concept. Simple measures with dichotomous items may capture social isolation sufficiently to evaluate the concept in research, and the weighting of the SNI domains reflective of previous research may also be a sound method for social isolation measure construction. Knowledge regarding the current social isolation measure therefore provides a foundation for future intervention research that can positively impact health care and policy for older adults.

References

  • Baker, P.S., Bodner, E.V. & Allman, R.M. (2003). Measuring life-space mobility in community-dwelling older adults. Journal of the American Geriatrics Society, 51, 1610–1614. doi:10.1046/j.1532-5415.2003.51512.x [CrossRef]
  • Bassuk, S.S., Glass, T.A. & Berkman, L.F. (1999). Social disengagement and incident cognitive decline in community-dwelling elderly persons. Annals of Internal Medicine, 131, 165–173. doi:10.7326/0003-4819-131-3-199908030-00002 [CrossRef]
  • Berkman, L.F. (1977). Social networks, host resistance, and mortality: A follow-up study of Alameda County residents (Doctoral dissertation). Retrieved from ProQuest Dissertations & Theses Global. (288073491)
  • Berkman, L.F. & Syme, S.L. (1979). Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology, 109, 186–204. doi:10.1093/oxfordjournals.aje.a112674 [CrossRef]
  • Bradley, P. (2015). I don't want to be alone. British Journal of Health-care Assistants, 9, 408–411. doi:10.12968/bjha.2015.9.8.408 [CrossRef]
  • Choi, N.G. & DiNitto, D.M. (2015). Role of new diagnosis, social isolation, and depression in older adults' smoking cessation. The Gerontologist, 55, 793–801. doi:10.1093/geront/gnu049 [CrossRef]
  • Courtin, E. & Knapp, M. (2017). Social isolation, loneliness and health in old age: A scoping review. Health & Social Care in the Community, 25, 799–812. doi:10.1111/hsc.12311 [CrossRef]
  • Coyle, C.E. & Dugan, E. (2012). Social isolation, loneliness and health among older adults. Journal of Aging and Health, 24, 1346–1363. doi:10.1177/0898264312460275 [CrossRef]
  • DiNapoli, E.A., Wu, B. & Scogin, F. (2014). Social isolation and cognitive function in Appalachian older adults. Research on Aging, 36, 161–179. doi:10.1177/0164027512470704 [CrossRef]
  • Dorfman, R.A., Lubben, J.E., Mayer-Oakes, A., Atchison, K., Schweitzer, S.O., De Jong, F.J. & Matthias, R.E. (1995). Screening for depression among a well elderly population. Social Work, 40, 295–304.
  • Ebner, N.C., Maura, G.M., Macdonald, K., Westberg, L. & Fischer, H. (2013). Oxytocin and socioemotional aging: Current knowledge and future trends. Frontiers in Human Neuroscience, 7, 487. doi:10.3389/fnhum.2013.00487 [CrossRef]
  • Faulkner, K.A., Cauley, J.A., Zmuda, J.M., Griffin, J.M. & Nevitt, M.C. (2003). Is social integration associated with the risk of falling in older community-dwelling women?Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 58, M954–M959. doi:10.1093/gerona/58.10.M954 [CrossRef]
  • Ford, E.S., Loucks, E.B. & Berkman, L.F. (2006). Social integration and concentrations of C-reactive protein among US adults. Annals of Epidemiology, 16, 78–84. doi:10.1016/j.annepidem.2005.08.005 [CrossRef]
  • Havens, B., Hall, M., Sylvestre, G. & Jivan, T. (2004). Social isolation and loneliness: Differences between older rural and urban Manitobans. Canadian Journal on Aging, 23, 129–140. doi:10.1353/cja.2004.0022 [CrossRef]
  • Hawton, A., Green, C., Dickens, A.P., Richards, S.H., Taylor, R.S., Edwards, R. & Campbell, J.L. (2011). The impact of social isolation on the health status and health-related quality of life of older people. Quality of Life Research, 20, 57–67. doi:10.1007/s11136-010-9717-2 [CrossRef]
  • Heffner, K.L., Waring, M.E., Roberts, M.B., Eaton, C.B. & Gramling, R. (2011). Social isolation, C-reactive protein, and coronary heart disease mortality among community-dwelling adults. Social Science & Medicine, 72, 1482–1488. doi:10.1016/j.socscimed.2011.03.016 [CrossRef]
  • Holt-Lunstad, J., Smith, T.B., Baker, M., Harris, T. & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10, 227–237. doi:10.1177/1745691614568352 [CrossRef]
  • Iliffe, S., Kharicha, K., Harari, D., Swift, C., Gillmann, G. & Stuck, A.E. (2007). Health risk appraisal in older people 2: The implications for clinicians and commissioners of social isolation risk in older people. British Journal of General Practice, 57, 277–282.
  • Johnson, C.J., Bulot, J. & Johnson, R.H. (2008). Outcome assessment of mentorship program. Educational Gerontology, 34, 555–569. doi:10.1080/03601270701869303 [CrossRef]
  • Kasper, J.D. & Freedman, V.A. (2015). National Health and Aging Trends Study user guide: Rounds 1, 2, 3 & 4, final release. Baltimore, MD: Johns Hopkins University, Bloomberg School of Public Health.
  • Koenig, H.G. (2012). Spirituality and health research: Methods, measurements, statistics, and resources. Conshohocken, PA: Templeton Foundation Press.
  • Kroenke, K., Spitzer, R.L. & Williams, J.B. (2003). The patient health questionnaire-2: Validity of a two-item depression screener. Medical Care, 41, 1284–1292. doi:10.1097/01.MLR.0000093487.78664.3C [CrossRef]
  • Lee, M.J., Hasche, L.K., Choi, S., Proctor, E.K. & Morrow-Howell, N. (2013). Comparison of major depressive disorder and subthreshold depression among older adults in community long-term care. Aging & Mental Health, 17, 461–469. doi:10.1080/13607863.2012.747079 [CrossRef]
  • Löwe, B., Wahl, I., Rose, M., Spitzer, C., Glaesmer, H., Wingenfeld, K. & Brähler, E. (2010). A 4-item measure of depression and anxiety: Validation and standardization of the Patient Health Questionnaire-4 (PHQ-4) in the general population. Journal of Affective Disorders, 122, 86–95. doi:10.1016/j.jad.2009.06.019 [CrossRef]
  • Lubben, J., Blozik, E., Gillmann, G., Iliffe, S., von Renteln Kruse, W., Beck, J.C. & Stuck, A.E. (2006). Performance of an abbreviated version of the Lubben social network scale among three European community-dwelling older adult populations. The Gerontologist, 46, 503–513. doi:10.1093/geront/46.4.503 [CrossRef]
  • Lubben, J.E. (1988). Assessing social networks among elderly populations. Family and Community Health, 11, 42–52. doi:10.1097/00003727-198811000-00008 [CrossRef]
  • Matson, J., Fodstad, J. & Rivet, T. (2008). The convergent and divergent validity of the Matson Evaluation of Drug Side-effects (MEDS) and the Dyskinesia Identification System: Condensed User Scale (DISCUS). Journal of Intellectual and Developmental Disability, 33, 337–344. doi:10.1080/13668250802478799 [CrossRef]
  • McCrae, N., Murray, J., Banerjee, S., Huxley, P., Bhugra, D., Tylee, A. & Macdonald, A. (2005). ‘They're all depressed, aren't they?’ A qualitative study of social care workers and depression in older adults. Aging & Mental Health, 9, 508–516. doi:10.1080/13607860500193765 [CrossRef]
  • MIDUS. (2004). Documentation of scales in MIDUS I. Retrieved from http://www.midus.wisc.edu/midus1/documentationofscales.pdf
  • Momtaz, Y.A., Hamid, T.-A., Ibrahim, R., Yahaya, N. & Chai, S.T.. (2011). Moderating effect of religiosity on the relationship between social isolation and psychological well-being. Mental Health, Religion & Culture, 14, 141–156. doi:10.1080/13674676.2010.497963 [CrossRef]
  • Montaquila, J.D., Freedman, V.A., Edwards, B. & Kasper, J.D. (2012). National Health and Aging Trends Study round 1 sample design and selection (NHATS Technical Paper #1). Baltimore, MD: Johns Hopkins University, Bloomberg School of Public Health.
  • Montaquila, J.D., Freedman, V.A., Spillman, B. & Kasper, J.D. (2012). National Health and Aging Trends Study round 1 development of round 1 survey weights (NHATS Technical Paper #2). Baltimore, MD: Johns Hopkins University, Bloomberg School of Public Health.
  • Nicholson, N.R. (2010). Predictors of social isolation in community-dwelling older adults (Doctoral dissertation). Retrieved from Pro-Quest Dissertations & Theses Global. (109853772)
  • Nicholson, N.R. (2012). A review of social isolation: An important but underassessed condition in older adults. The Journal of Primary Prevention, 33, 137–152. doi:10.1007/s10935-012-0271-2 [CrossRef]
  • Pantell, M., Rehkopf, D., Jutte, D., Syme, S.L., Balmes, J. & Adler, N. (2013). Social isolation: A predictor of mortality comparable to traditional clinical risk factors. American Journal of Public Health, 103, 2056–2062. doi:10.2105/AJPH.2013.301261 [CrossRef]
  • Reginold, W., Duff-Canning, S., Meaney, C., Armstrong, M.J., Fox, S., Rothberg, B. & Marras, C. (2013). Impact of mild cognitive impairment on health-related quality of life in Parkinson's disease. Dementia and Geriatric Cognitive Disorders, 36, 67–75. doi:10.1159/000350032 [CrossRef]
  • Rutledge, T., Matthews, K., Lui, L.Y., Stone, K.L. & Cauley, J.A. (2003). Social networks and marital status predict mortality in older women: Prospective evidence from the Study of Osteoporotic Fractures (SOF). Psychosomatic Medicine, 65, 688–694. doi:10.1097/01.PSY.0000041470.25130.6C [CrossRef]
  • Saito, M., Kondo, N., Kondo, K., Ojima, T. & Hirai, H. (2012). Gender differences on the impacts of social exclusion on mortality among older Japanese: AGES cohort study. Social Science & Medicine, 75, 940–945. doi:10.1016/j.socscimed.2012.04.006 [CrossRef]
  • Shankar, A., McMunn, A., Banks, J. & Steptoe, A. (2011). Loneliness, social isolation, and behavioral and biological health indicators in older adults. Health Psychology, 30, 377–385. doi:10.1037/a0022826 [CrossRef]
  • Taylor, H.O., Herbers, S., Talisman, S. & Morrow-Howell, N. (2016). Assessing social isolation: Pilot testing different methods. Journal of Gerontological Social Work, 59, 228–233. doi:10.1080/01634372.2016.1197354 [CrossRef]
  • Uchino, B.N. (2006). Social support and health: A review of physiological processes potentially underlying links to disease outcomes. Journal of Behavioral Medicine, 29, 377–387. doi:10.1007/s10865-006-9056-5 [CrossRef]
  • Victor, C., Scambler, S. & Bond, J. (2008). The social world of older people: Understanding loneliness and social isolation in later life. Maidenhead, UK: Open University Press.

Indicators of the Social Isolation Measure (National Health and Aging Trends Study Interview Items)

DomainIndicator QuestionsIndicator Responses
Marriage/partnerAre you currently married, living with a partner, separated, divorced, widowed, or never married?No marriage or partner
Family and friendsLooking back over the past year, who are the people you talked with most often about important things?No family identified No friend identified
In the last month, did you ever visit in-person with friends or family not living with you either at home or theirs?No
Church participationIn the past month, did you ever attend religious services?No
Club participationIn the past month, did you ever participate in clubs, classes, or other organized activities?No

Patient Health Questionnaire-2 Measure of Depression Risk (National Health and Aging Trends Study Interview Items)

Question AnchorQuestionsResponse Options
Over the last month, how often have you:

had little interest or pleasure in doing things?

felt down, depressed, or hopeless?

0 = not at all

1 = several days

2 = more than half the days

3 = nearly every day

Items Used to Develop the Measure of Well-Being (National Health and Aging Trends Study Interview Items)

Question/Statement AnchorQuestion/StatementResponse Options
Affect/feelings
  During the last month, how often did you feel….

cheerful?

full of life?

bored?

upset?

Never

Rarely

Some days

Most days

Every day

Self-realization/well-being
  Tell me whether you agree with the following statements about your life.

My life has meaning and purpose.

I feel confident and good about myself.

I like my living situation very much.

I gave up trying to improve my life a long time ago.

Not at all

A little

A lot

Sense of control
  For the last few statements, please tell me if you agree a lot, a little, or not at all.

I have an easy time adjusting to change.

When I really want to do something, I usually find a way to do it.

Other people determine most of what I can and cannot do.

Agree a lot

Agree a little

Agree not at all

Distribution of National Health and Aging Trends Study Demographic, Household, and Health Characteristics According to Social Isolation Status (N = 7,609)

CharacteristicSocially Isolatedap Value
Yes (Weighted %)No (Weighted %)
Age (years)<0.001b
  65 to 69237 (19.2)1,172 (30.4)
  70 to 74318 (21.4)1,261 (25.9)
  75 to 79320 (17.1)1,193 (19.6)
  80 to 84419 (17.4)1,086 (13.9)
  85 to 89366 (15.7)587 (7.3)
  ≥90323 (9.2)327 (2.9)
Gender0.178b
  Female1,217 (58.2)3,221 (56.2)
  Male766 (41.8)2,405 (43.8)
Race/ethnicityc<0.001b
  White, non-Hispanic1,181 (73.9)4,005 (83.6)
  Black, non-Hispanic546 (11.6)1,116 (7.3)
  Hispanic152 (9.4)302 (6.1)
  Other76 (5.1)143 (3.1)
Educationc<0.001b
  No schooling to 12th grade/no diploma832 (37.3)1,215 (17.4)
  High school graduate, some college863 (48.3)2,697 (49.2)
  Associate degree to graduate degree251 (14.4)1,655 (33.4)
Household<0.001b
  Lives alone908 (48.4)1,566 (24.6)
Overall health (self-report)c<0.001b
  Excellent151 (8.7)787 (16.5)
  Very good357 (19.4)1,672 (32.3)
  Good573 (28.4)1,853 (31.3)
  Fair581 (28.6)1,026 (15.5)
  Poor318 (14.9)285 (4.4)
Mean (SD)
Depression (PHQ-2)1.4 (1.7)0.88 (1.3)<0.001d
Well-being measure32.95 (5.16)35.03 (4.22)<0.001d
Authors

Dr. Pohl is PhD Graduate, School of Nursing, Dr. Cochrane is de Tornyay Endowed Professor for Healthy Aging, Family and Child Nursing, Dr. Schepp is Professor and Chair, Psychosocial & Community Health, and Dr. Woods is Professor Dean Emerita, Biobehavioral Nursing and Health Systems, University of Washington, Seattle, Washington.

The authors have disclosed no potential conflicts of interest, financial or otherwise. This work was supported by the Myrene C. McAninch Doctoral Scholarship, de Tornyay Center for Healthy Aging, University of Washington School of Nursing.

Address correspondence to Janet S. Pohl, PhD, RN, PhD Graduate, School of Nursing, University of Washington, Box 357262, Seattle, WA 98195; e-mail: janpohl@uw.edu.

Received: April 17, 2017
Accepted: August 02, 2017
Posted Online: October 06, 2017

10.3928/19404921-20171002-01

Sign up to receive

Journal E-contents