Risk of Social Isolation in Older Adults
Where are seniors that are more likely to experience social isolation?
Socially-isolated [1] seniors are at heightened risk for poor health outcomes [2] during the coronavirus pandemic if they lack access to help when needed, from transportation for medical care, to regular basic needs like food.
In this joint project between the Polis Center (Polis) and the Central Indiana Senior Fund (CISF), we have developed a population measure of the risk of social isolation in older adults (age 65 and older), based on an index previously developed by America’s Health Rankings (AHR) for their national Senior Report.
While little research has been done using aggregate data for measuring social isolation (as opposed to direct survey responses), the American Community Survey (ACS) has variables similar to what is often found to contribute to social isolation [3]. Specifically, the AHR social isolation index uses five individual-level risk factors that describe either socio-economic characteristics or physical limitations: poverty, living alone, marital status (never married and divorced, separated, & widowed), disability, and independent living difficulty. [4, 5]
Since ACS data are publicly available, Polis started with this same data source to recreate a social isolation risk index for older adults similar to that created by AHR, but with finer geographic detail than that available from AHR and with an interactive user experience. We have used this approach to create the map below, depicting older adult populations in Indiana potentially at higher risk of social isolation.
What We Found
Urban areas tend to have much higher scores than rural areas on the social isolation index, because factors like poverty, unmarried households, people living alone, and people with disabilities tend to be concentrated in cities and towns.
Because of this, we represent the social isolation risk scores using two different color gradients, one for urban areas and one for rural areas.
The chart below shows why two legend gradients make it easier to see hotspots of risk for social isolation in rural areas versus if we used a single legend gradient. A high score (80th percentile) for a rural tract is -0.3, while a high score for an urban tract is 1.6. For someone working in rural Indiana, the urban scores will dwarf the data of interest. Census tracts with the highest index values have the highest combination values of the four input variables (poverty, being unmarried, living alone, and disability).
Social Isolation Index in Rural and Urban Areas
Rural areas tend to have much lower social isolation risk scores than urban areas.
This map shows the senior social isolation index for Indiana's towns and cities (census tracts where at least half the population lives in places with populations of at least 25,000).
To use this map, click on a neighborhood to see which of the individual variables have the strongest impact on the component measure of social isolation risk.
We will take a closer look at three of the state's metro areas: Indianapolis, Fort Wayne, and Bloomington. If interested in looking at other parts of the state, simply pan to the area of the state of interest.
Indianapolis
In Central Indiana, several cities and towns are shown in this urban-only map. Indianapolis suburbs, like Carmel and Greenwood, have some of the lowest index scores (shown in the lightest shades).
In contrast, high index scores are concentrated in the downtown area of Indianapolis, with the notable exception of the IUPUI campus.
The town centers of Anderson, Plainfield, Crawfordsville, and Shelbyville also have higher scores. This suggests, perhaps not surprisingly, that our urban core and town centers are more likely to have seniors at risk for social isolation.
Fort Wayne
In Fort Wayne, the suburbs to the north (such as Huntertown) and west have low isolation risk scores compared to higher scores in the center of Fort Wayne. This is similar to what was found in Indianapolis.
In contrast to the downtown IUPUI campus in Indianapolis, the Purdue University Fort Wayne campus is in a tract with one of the highest index scores (5.7) in the Fort Wayne area. It is about two miles north of a tract with a very low isolation score (-3.0), a mostly residential area just west of Parkview Hospital Randallia and the VA Northern Indiana Health Care System.
Bloomington
Bloomington appears largely split between west and east, with higher isolation scores to the west (scores largely from 2-3.7), and lower scores to the east (averaging about -1.5).
Social Isolation in Rural Areas
The rural mapping shows rural areas of predicted social isolation relative to other rural areas. Southwest Indiana has a significant number of high vulnerability areas (darker-shaded tracts), with scores close to average for the state (about zero, when compared to all tracts in Indiana, including urban areas), whereas areas in northeast Indiana tend to have lower vulnerability, with many tracts ranging from -1 to almost -3.
Social Isolation Rural Areas
Credits
Ongoing analysis is being conducted by the Polis Center at IUPUI in partnership with the Central Indiana Senior Fund (CISF) as part of CISF’s initiative to develop an online, State of Aging in Central Indiana Report (SoAR). The goal of the CISF SoAR is to provide funders, policy makers, and decision makers with access to actionable community intelligence to inform decision making about programs and policy.
A beta-version of the online CISF SoAR will be available for user testing in the fall of 2020. If you are interested in user testing of the beta version of the online report, please contact polis@iupui.edu , with “SoAR user testing” in the Subject line.
Data Notes
Geographies shown are census tracts. Data come from the 2018 American Community Survey 5-year estimates (via SAVI). The rural versus urban categories used for these maps are based on Census assessments. If more than 50% of residents were designated as ‘rural dwellers’, we counted those tracts as ‘rural’, and vice-versa for urban.
Methodology
Whereas the AHR Senior Report used pre-determined weights to generate their Social Isolation Risk Index, Polis Center used Principal Component Analysis to generate a score for each census tract, using the software package R. [7] This approach standardizes each of the measures that go into this index, calculating a best-fit factor for summarizing how each of these variables relate to each other.
Since there is reason to believe each of these variables can play a role in social isolation, the resulting calculation will show census tracts most at risk from the combination of these variables. These factors are not mutually exclusive and any single factor may not lead to isolation. Similar to the AHR report, this exercise did not include potential societal-level risk factors such as crime, accessibility of public space, or presence of meaningful events.
Notes
[1] The simplest definition of social isolation is a “paucity of social contacts,” (Kobayashi & Steptoe, 2018) while a more complicated approach incorporates several aspects of social connectedness and its impacts, such as “social capital, social exclusion and social cohesion.” (Zavaleta & Mills, 2017
[2] Social isolation is linked to poor health outcomes,3 and by some estimates, is “a predictor of mortality on par with smoking, obesity, elevated blood pressure, and high cholesterol.” Klinenberg, Eric. 2016. Social isolation, loneliness, and living alone. AJPH 106: 786-87.
[3] America’s Health Rankings. Senior Report Spotlight. United Health Foundation. https://www.americashealthrankings.org/learn/reports/2018-senior-report/senior-report-spotlight
[4] Wigfield, Andrea. Alden, Sarah. 2018. Assessing the effectiveness of social indices to measure the prevalence of social isolation in neighborhoods. Social Indicators Research 140:1017-1034. Scholarly attempts to generate a social isolation index using aggregate data has been sparse, so validation efforts even more sparse. In this study, Wigfield and Alden perform a qualitative ‘sense-check’ of social isolation in communities tagged as having populations vulnerable to social isolation based on community-developed indices. Their results showed mixed predictive capacity.
[5] Most of the research on social isolation comes from household-level survey data, such as the European Social Survey, or individual measures, such as the Berkman-Syme Social Networks Index. Questions specific to social isolation are often included in surveys on aging populations. While social isolation and loneliness are often conflated, these are separate concepts and have differential mechanisms related to health outcomes, with isolation apparently being the more important factor. Most questions relating to both loneliness and social isolation involve describing the number of one’s weekly social contacts, participation in community activities, satisfaction with relationships, and household composition.
Kobayahsi and Steptoe. 2018. Social isolation, loneliness and health behaviors at older ages. Ann Behav Med 52: 582-93.
Nicholson, Nicholas. 2012. A Review of Social Isolation. J Primary Prevention 33: 137-152.
Steptoe, Shankar, Demakakos & Wardle. 2013. Social isolation, loneliness and all-cause mortality in older men and women. PNAS 110: 5797-5801.
Swader, C.S. 2019. Loneliness in Europe. Social Forces 97: 1307-1335
Tanskanen, Jussi, Antilla and Timo. 2016. A prospective study of social isolation, loneliness and mortality in Finland. AJPH 106: 2042-2048.
Wenger and Burholt. 2004. Changes in levels of social isolation and loneliness among older people in a rural area. Can J Aging 23: 115-27.
[6] America’s Health Rankings. Measures, Weights and Direction. United Health Foundation. https://www.americashealthrankings.org/about/methodology/measures-weights-and-direction
