Social Relationships and Functional Impairment in Aging Cancer Survivors: A Longitudinal Social Network Study

Abstract

Background and Objectives

The intersection of cancer, treatment, and aging accelerates functional decline. Social networks, through the provision of social support and resources, may slow the progression of functional deterioration. Socioemotional selectivity theory posits that aging and major life events, like cancer, cause an intentional social network pruning to procure and maintain emotionally fulfilling bonds, while shedding weaker, less supportive relationships. However, it is relatively unknown if such network changes impact functional impairment in cancer survivors. This study examined the relationships between changes in the egocentric social network and functional impairment in older adult cancer survivors and a similarly aged group without cancer (older adults).

Research Design and Methods

Data were analyzed from 1,481 community dwelling older adults (n = 201 cancer survivors) aged 57–85 years, from Waves 1 and 2 (2005–2006 and 2010–2011) of the National Social Life, Health and Aging Project. Associations were analyzed with multiple logistic regression.

Results

Cancer survivors and older adults reported similar levels of functional impairment and social network change. Adding 2 new relationships exhibited protective effects against functional impairment, irrespective of cancer status (odds ratio [OR]: 0.64, 95% confidence interval [CI]: 0.41–0.99). Declines in frequent contact were associated with higher odds of functional impairment among cancer survivors (OR: 1.92, 95% CI: 1.15–3.20). Social network components were not significantly associated with functional impairment in older adults.

Discussion and Implications

Adding new relationships may reduce disability in older adults and increasing network contact may help cancer survivors remain independent. Social network interventions may improve quality of life for older adults.

Over the next 5 years, the number of cancer survivors living in the United States will approach 19 million individuals (DeSantis et al., 2014) and the majority of these survivors will be aged 65 and older (Parry, Kent, Mariotto, Alfano, & Rowland, 2011). The “hallmarks of aging” include a gradual decline in functional capacity, including deterioration of the cardiovascular and musculoskeletal systems (Jones, Eves, Haykowsky, Freedland, & Mackey, 2009; Sawhney, Sehl, & Naeim, 2005). The intersection of cancer, cancer treatment, and aging accelerate functional decline and mortality (Brown, Harhay, & Harhay, 2015; Henderson, Ness, & Cohen, 2014; Hurria, Jones, & Muss, 2016). Cancer survivors perceive fair or poor health (Bloom, Petersen, & Kang, 2007) and report more functional limitations in everyday life compared with healthy controls (Hewitt, Rowland, & Yancik, 2003; Keating, Nørredam, Landrum, Huskamp, & Meara, 2005). Older adult cancer survivors fatigue faster (Gresham et al., 2018), and are more likely to have comorbidities (Deimling, Arendt, Kypriotakis, & Bowman, 2009; Sogaard, Thomsen, Bossen, Sorensen, & Norgaard, 2013), including higher rates of heart conditions, lung disease, arthritis, incontinence, pain, and obesity (Keating et al., 2005); underscoring the need to design effective interventions to prevent or halt the progression of functional decline.

Adequate social network support may slow the progression of physical decline (Rottenberg et al., 2014; Unger, McAvay, Bruce, Berkman, & Seeman, 1999), while receiving less support may exacerbate disability (Stuck et al., 1999). Social networks are “the web of social relationships that surround an individual” (Kroenke, 2018). Close contacts and their attributes can have profound effects on health through social influence, norms, the perception of social support, and the flow of information and resources (Smith & Christakis, 2008). In the general population, the presence of quality social relationships has been shown to benefit physiological (Uchino, 2006; Yang, Schorpp, & Harris, 2014), physical, and mental health (Kawachi & Berkman, 2001) and increase longevity (Yang et al., 2016). Among breast cancer survivors, social isolation, or the absence of social ties, is associated with all-cause and cancer-specific mortality, and the reported effect size is of similar magnitude to established cancer risk factors, such as smoking and obesity (Hinzey, Gaudier-Diaz, Lustberg, & DeVries, 2016). Social networks may bolster physical functioning through access to information and resources, shared decision making, the provision of tangible and emotional support, and modeling positive health behaviors and coping strategies.

Socioemotional selectivity theory posits that social networks are constructed based on the perception of time. As individuals age, time is perceived to be limited, and as a result, the size and composition of the social network shifts to include a smaller number of meaningful relationships comprised primarily of family members. In this context, emotionally fulfilling bonds are procured and maintained, while weaker, less supportive relationships are pruned (English & Carstensen, 2014; Lansford, Sherman, & Antonucci, 1998). A life changing event, such as a cancer diagnosis, can also trigger the perception that end of life is near, causing cancer survivors to rearrange their social networks in similar ways (Fisher & Nussbaum, 2015). Creating a strong, supportive network may provide important advantages for long-term cancer recovery, shared decision making, survivorship care planning, psychosocial well-being, and functional independence (Thompson, Rodebaugh, Pérez, Schootman, & Jeffe, 2013). Recently, a call was put forth to understand the impact of social networks on cancer survivorship endpoints using more robust social network measures (Kroenke, 2018). To address this research gap, we conducted a secondary analysis of egocentric social network data over a 5-year follow-up period. Egocentric social networks focus on an individual, “ego” and their personal social network of close contacts (alters; Borgatti, Everett, & Johnson, 2018). Understanding how social networks evolve over time and their influence on physical decline is critical to devise strategies to address the needs of the aging cancer survivor. The objective of this study was to: (a) compare the patterns of social network change between older adult cancer survivors and a similarly aged group without cancer; and (b) investigate if changes to social networks are associated with perceived physical functioning over a 5-year period. We hypothesize that cancer survivors and older adults without cancer (older adults) will maintain frequently contacted, emotionally close, and supportive relationships over time, in accordance with socioemotional selectivity theory. Further, adding new relationships and preserving high levels of social support and network contact among enduring relationships will be protective of functional impairment in both groups, while losing specific relationships may not impact disability due to intentional network pruning.

Research Design and Methods

Study Population

A secondary analysis of the National Social Life Health and Aging Project (NSHAP) was conducted to assess whether changes in the social network was associated with functional limitations 5 years later. NSHAP is a cohort of community dwelling older adults aged 57–85 years old. Wave 1 (W1) and Wave 2 (W2) data were collected in 2005–2006 and 2010–2011, respectively. The study design, methods of recruitment, and sampling have been well described elsewhere (Hayward & Wallace, 2014; O’Muircheartaigh, English, Pedlow, & Kwok, 2014; Waite, Laumann, Levinson, Lindau, & O’Muircheartaigh, 2014). Briefly, NSHAP is a complex multistage probability sample of older adults. Oversampling by age, race/ethnicity, and gender was conducted to ensure adequate representation of subgroups. The W1 sample included 3,005 participants, of whom 2,261 participated again in W2. Participants were considered cancer survivors if they indicated on the NSHAP W1 questionnaire that they had been diagnosed with cancer (excluding nonmelanoma skin cancer). Participants who were diagnosed with cancer after W1 (n = 148) were excluded because their social network at baseline would not reflect their cancer experience. Individuals with missing functional impairment data (n = 1) or missing social network data (n = 631) were excluded from the analysis, resulting in a final sample size of 1,481 men and women (Figure 1). This study was approved by the Institutional Review Board at University of Maryland, College Park.

Figure 1.

Derivation of the analytic sample.

Measures

Functional Impairment

Disability is the inability to perform tasks essential for autonomous living (Cornwell & Laumann, 2015) and is predictive of future health complications and mortality (Stuck et al., 1999). The Activities of Daily Living (ADL) scale is a 7-item instrument that asks respondents to self-report difficulties performing activities lasting 3 months or more: walking across a room, dressing, bathing, eating, getting in or out of bed, using the toilet, and driving during the day. Responses ranged from 0 (no difficulty) to 3 (difficulty). Scores were summed with higher scores indicating more severe impairments. The Cronbach’s alpha for the scale was 0.84 in W1 and 0.81 in W2. We created a binary variable (1= any impairment vs 0= no impairment) because more than 70% of the sample indicated no functional impairments.

Social Network Change

The NSHAP egocentric social network module has been previously described (Cornwell, 2015; Cornwell, Schumm, Laumann, & Graber, 2009; Cornwell, Schumm, Laumann, Kim, & Kim, 2014). Strong ties were elicited from the name generator, “from time to time, most people discuss things that are important to them with others. For example, these may include good or bad things that happen to you, problems you are having, or important concerns you may have. Looking back over the last 12 months, who are the people with whom you most often discussed things that were important to you?” (Marin, 2004; Ruan, 1998). For the current study, social network change was measured in three ways: (a) network turnover, (b) changes to the characteristics of relationships that last over time, and (c) changes to the overall network structure.

Network Turnover

To distinguish between ties that were added, lost, or persisted between waves, NSHAP developed a computer-assisted personal interviewing (CAPI) exercise. During the W2 interview, NSHAP respondents were first asked the “important matters” question, as done in W1. Interviewers then presented the respondent’s W1 and W2 alters and asked participants to verify the linked matches. To capture potential differences in the direction and magnitude of the associations of interest, we created categorical variables for the numbers of lost and added ties. Lost ties were defined as alters who were named in W1, but not in W2. Categories were created based on the distribution: 0 = no lost alters (reference group), 1 = one lost alter, 2= two lost alters, and 3 = three or more lost alters. Added ties were alters that were named for the first time in W2 and were coded in the same fashion as lost ties.

Changes to the Characteristics of Persistent Ties

Understanding the attributes of ties that persist over time can help distinguish why some ties endure and why others do not. NSHAP measured emotional closeness by the question “How close do you feel is your relationship with [name]?” Response options ranged from 0 = not very close to 3 = extremely close. The frequency of contact indicates how often ego interacts with their personal network. NSHAP asked each respondent to rate on an ordinal scale how often they talked to each alter, including via telephone and email. Responses ranged from 1 = once a year to 8 = every day. Perceived social support was measured via six questions that asked participants how often they could open up to and rely on their spouse/partner, family, and friends. All responses were measured on a three-point scale (0 = hardly ever/never to 2 = often). The six social support questions were summed to calculate the total amount of support the ego received from each specific relation. Average difference scores for closeness and contact were calculated by computing the mean of each variable for each ego at both waves, and then taking the absolute difference between W1 and W2. Change in social support was computed by taking the absolute difference between waves.

Changes to the Overall Network Structure

Density is the proportion of alters within ego’s network who know each other. Each participant was asked how often each of the alters talked to each other. Alters who had never spoken to each other were considered to have no connection (no tie). Density was calculated by dividing the total number of actual ties by the number of potential ties. The number of potential ties is the proportion of [k(k − 1)/2] pairs, where k is the total number of alters in ego’s network (not including ego) and ego has at least two alters. Density ranges from 0 to 1, where higher network density signifies more connections among alters. Variations in network density over time were assessed with difference scores.

Covariates

Confounders were identified a priori from the literature. Age was categorized as 57–64 (reference), 65–74, 75–85 because older age groups have different levels of risk for functional impairments (Jette & Branch, 1981). Body mass index (BMI, kg/m²) was categorized based on established cut points (BMI: <18.5 = underweight, 18.5–24.9 = normal, 25–29.9 = overweight, >30 = obese; Williams, Pham-Kanter, & Leitsch, 2009). Underweight and normal categories were combined due to small sample size and were used as the reference group. Race (white [reference], black, and other), gender (male [reference] vs female), education (high school education or less vs some college or more [reference]), marital status (married/cohabitating partner [reference] vs unmarried), smoking (yes/no [reference]), and the number of comorbidities were also controlled for in the analysis. Comorbidities were assessed using a modified version of the Charlson Comorbidity Index (Vasilopoulos et al., 2014) including 12 chronic conditions (hypertension, heart attack/myocardial infarction, congestive heart failure, stroke, any procedure for coronary artery disease, depression, diabetes, COPD/asthma, arthritis, Alzheimer’s disease/dementia, urinary/stool incontinence, and other urinary problems). Scores were summed for a total of 12 possible points.

Statistical Analysis

Means and proportions were used to describe the distributions of variables by cancer status. Descriptive characteristics of the individual ADL items are presented in Supplementary Table 1. Functional impairment and social networks by cancer type is described in Supplementary Table 2. Simple linear regression and chi-squared tests were used to examine differences between cancer survivors and older adults for continuous and categorical variables, respectively. All p values are two-sided. Multiple logistic regression was used to assess whether the change in social network components were associated with functional impairment in W2, adjusting for existing impairments in W1 and all covariates. A priori, power analyses indicated that we would have sufficient power to detect an effect size of 0.15 with at least 175 participants in each of the two groups (power = 0.80, alpha = 0.05). We analyzed the reported reasons alters were “lost” between waves of data collection to provide context to the results (data not shown). Additionally, we separated tangible and emotional support in logistic regression models to assess the individual contribution of the different support types by relation (Supplementary Table 3). The complex survey design was taken into account using the survey procedures in SAS version 9.3 (SAS, Inc., Cary, NC) and by utilizing the W1 survey weights adjusted for nonresponse (Cornwell et al., 2014).

Results

The final analytic sample included 1,481 participants, of whom 13.8% (n = 201) indicated that they had a cancer diagnosis. About half the sample was aged 57–64 years (47.7%) and female (57.6%). The majority were white (83.8%), married (71.7%), overweight or obese (75.4%), nonsmokers (88.2%), and had received some college or more (62.4%) in W1. Overall, 26.7% of the sample reported impairments in their activities of daily living in W2, which was an increase from W1 (21.2%; Table 1). Cancer survivors were statistically significantly older (p < .01) and unmarried (p = .02) compared with older adults, but did not differ in terms of functional impairment (cancer survivors: W1 = 22.6%, W2 = 29.4%; older adults: W1 = 21.0%, W2 = 26.3%; p = .57). Cancer survivors reported statistically significantly higher percentages of difficulty bathing in W2 compared with older adults (11.9% vs 7.3%; Supplementary Table 1). Most cancer survivors either had breast (n = 48), prostate (n = 40), or gynecological cancer (n = 35). Leukemia/lymphoma survivors reported the highest levels of functional impairment (39.5%; Supplementary Table 2).

Table 1.

Characteristics of Cancer Survivors and Older Adults

Sociodemographic and social network characteristics	Overall (n = 1,481)		Cancer survivor (n = 201)		Older adults (n = 1,280)
	N	%	N	%	N	%	p Valuea
Age (years)
57–64	592	47.7	49	33.4	543	50.0	<.01
65–74	561	35.5	89	44.6	472	34.0
75–85	328	16.8	63	22.0	265	16.0
Gender
Male	628	42.4	82	41.7	546	42.5	.82
Female	853	57.6	119	58.3	734	57.5
Race
White	1,099	83.8	167	89.5	932	82.9	.07
Black	214	8.4	24	5.8	190	8.8
Hispanic/other	162	7.9	10	4.7	152	8.4
Education
Less than high school	235	12.1	23	10.6	212	12.4	.16
High school	373	25.5	42	23.5	331	25.8
Some college	477	33.6	82	41.4	395	32.3
Bachelor’s degree or higher	396	28.8	54	24.5	342	29.4
Married/cohabitating partner
Yes	970	71.7	120	65.1	850	72.8	.02
No	511	28.3	81	34.9	430	27.2
Smoking
Nonsmoker	1,303	88.2	183	90.4	1,120	87.8	.42
Smoker	177	11.8	18	9.6	159	12.2
BMI
Underweight	2	0.1	0	0.0	2	0.1	.72
Normal	344	24.4	44	25.0	300	24.4
Overweight	506	35.1	66	32.5	440	35.5
Obese	556	40.3	81	42.4	475	40.0
Comorbidity Index (mean, SE)	2.14	0.05	2.20	0.14	2.13	0.05	.65
Functional impairment W2
No impairment	1,058	73.3	139	70.6	919	73.7	.80
1–2 impairments	202	13.3	26	12.8	176	13.4
3–4 impairments	114	6.7	23	9.6	91	6.2
5–6 impairments	54	3.2	7	3.3	47	3.2
6–7 impairments	23	1.7	3	1.1	20	1.8
8 or more impairments	30	1.8	3	2.6	27	1.7
Functional impairment W1
No impairment	1,034	72.8	137	71.7	897	73.0	.93
1–2 impairments	270	16.6	43	18.2	227	16.3
3–4 impairments	100	5.7	15	5.7	85	5.6
5–6 impairments	39	3.0	3	2.4	36	3.1
6–7 impairments	18	1.0	1	0.3	17	1.1
8 or more impairments	19	1.0	2	1.7	17	0.9
Change in functional impairment score (mean, SE)	0.18	0.05	0.46	0.26	0.13	0.04	.22
% Declined	272	18.0	41	20.0	231	17.7	.62
% Stable	932	65.5	126	66.0	806	65.4
% Improved	277	16.5	34	14.1	243	16.9
Lost alters between W1 and W2
No lost alters	238	15.4	25	12.5	213	15.9	.74
Lost one alter	414	27.2	65	29.4	349	26.8
Lost two alters	388	27.2	47	26.7	341	27.2
Lost three or more alters	441	30.3	64	31.4	377	30.1
Added alters between W1 and W2
No added alters	319	21.5	38	21.5	281	21.5	.28
Added one alter	391	27.0	61	29.6	330	26.5
Added two alters	396	27.5	45	20.4	351	28.6
Added three or more alters	375	24.1	57	28.5	318	23.4

Note: For continuous variables, the mean is presented with the standard error of the mean. For categorical variables, the number and percentage are presented. Responses are weighted for nonresponse. BMI= body mass index; SE = standard error of the mean.

^a p Value is testing the difference between cancer survivors and older adults with chi-squared or simple linear regression.

Most egocentric networks experienced some network turnover, with both cancer survivors and older adults reporting similar percentages of lost (p = .74) and added (p = .28) alters. Among stable relationships over the 5-year period, the frequency of contact, emotional closeness, density, and social support from spouses and friends declined slightly over time and the patterns of change were similar between the two groups (Table 2). Results from the multiple logistic regression model are presented in Table 3. In the overall sample, adding two new alters to the network was associated with lower odds of impairment, adjusting for W1 functional impairments and other confounding factors (odds ratio [OR]: 0.64, 95% confidence interval [CI]: 0.41–0.99). Among cancer survivors, declines in frequent contact were associated with higher odds of impairment (OR: 1.92, 95% CI: 1.15–3.20). No social network components were associated with functional impairments among the older adults. The stratified results did not reach statistical significance after adjusting for multiple comparisons using the Bonferroni method. Additionally, tangible and emotional support were not associated with functional impairment when modeled separately (Supplementary Table 3).

Table 2.

Network Characteristics Among Persistent Ties Among Cancer Survivors and Older Adults

Social network characteristics of persistent ties	Cancer survivors (n = 201)						Older adults (n = 1,280)
	Wave 1		Wave 2		Change		Wave 1		Wave 2		Change		p Valuea
	Mean	SE	Mean	SE	Mean	SE	Mean	SE	Mean	SE	Mean	SE
Mean frequency of contact	7.08	0.07	6.95	0.07	−0.13	0.06	7.05	0.04	6.99	0.04	−0.06	0.03	0.22
Mean closeness	2.33	0.05	2.27	0.04	−0.05	0.04	2.33	0.02	2.29	0.02	−0.04	0.02	0.73
Density	0.80	0.05	0.74	0.04	−0.05	0.07	0.81	0.02	0.75	0.01	−0.06	0.02	0.96
Social support
Spouse	2.37	0.12	2.12	0.15	−0.25	0.11	2.64	0.06	2.38	0.06	−0.26	0.04	0.97
Family	3.22	0.07	3.12	0.08	−0.11	0.07	3.07	0.03	3.12	0.03	0.04	0.03	0.05
Friends	2.51	0.08	2.38	0.14	−0.13	0.16	2.47	0.04	2.38	0.04	−0.09	0.04	0.77

Note: Responses are weighted for nonresponse. SE= standard error of the mean.

^a p Value is testing the difference in network change between cancer survivors and older adults using simple linear regression to take the complex survey design into account.

Table 3.

Adjusted Associations for Social Network Characteristics and Functional Impairment in W2 by Cancer Status^a

Social network characteristics	Overall	Cancer survivors	Older adults
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Lost ties between W1 and W2
No lost alters (ref.)
Lost one alter	1.47 (0.85–2.54)	1.57 (0.36–6.76)	1.46 (0.82–2.60)
Lost two alters	1.10 (0.62–1.94)	1.26 (0.26–6.02)	1.05 (0.57–1.94)
Lost three or more alters	1.77 (0.90–3.48)	4.32 (0.68–27.33)	1.48 (0.73–3.02)
Added ties between W1 and W2
No added alters (ref.)
Added one alter	0.67 (0.41–1.09)	0.39 (0.11–1.39)	0.75 (0.42–1.33)
Added two alters	0.64 (0.41–1.00)*	0.27 (0.07–1.11)	0.74 (0.45–1.22)
Added three or more alters	0.63 (0.37–1.08)	0.31 (0.08–1.17)	0.70 (0.39–1.25)
Change in closeness	0.82 (0.60–1.12)	0.69 (0.25–1.91)	0.84 (0.60–1.17)
Change in frequency of contact	1.16 (0.98–1.37)	1.92 (1.15–3.20)*	1.09 (0.91–1.30)
Change in density	1.02 (0.73–1.41)	0.63 (0.26–1.51)	1.09 (0.75–1.58)
Change in social support
Spouse	1.05 (0.95–1.16)	1.15 (0.82–1.61)	1.05 (0.94–1.17)
Family	1.06 (0.94–1.20)	0.97 (0.65–1.45)	1.08 (0.96–1.22)
Friends	0.99 (0.87–1.13)	1.24 (0.90–1.72)	0.94 (0.83–1.07)

Note: Responses are weighted for nonresponse. OR = odds ratio; CI = confidence interval.

*p < .05. **p < .01. ***p < .001.

^aModel adjusted for age, gender, race, marital status, education, BMI, number of comorbidities, smoking, and functional impairment in Wave 1.

Discussion and Implications

The objective of this study was to describe the evolving social environment among cancer survivors and older adults and its association with functional impairment. The proportion of functional impairments was similar in the two groups, and is consistent with estimates from other studies of cancer survivors (Gewandter et al., 2013; Goodwin, Hunt, & Samet, 1991), but higher than reported national estimates (CDC, NCBDDD, 2017; Chavan, Kedia, & Yu, 2017). Our findings suggest that the patterns of social network change in older adults with and without cancer were quite similar, in line with the socioemotional selectivity theory. Adding new relationships was protective of disability in the overall sample. Declines in the frequency of contact were associated with having at least one functional impairment among cancer survivors, controlling for impairments at baseline and other confounding factors.

Although both groups reported high levels of contact with network members, and the level of decline was relatively small over the 5-year period, frequent contact may be especially important for cancer survivors for two reasons. First cancer survivors are more vulnerable to functional decline due to the genotoxic and cytotoxic effects of cancer therapies (Henderson et al., 2014). Second, frequent contact with network members is especially important to arrange tangible assistance to complete activities of daily living. Cancer survivors may require additional assistance due to ongoing or intermittent treatments and have follow-up care needs that those without cancer do not have to endure. Additionally, informational support may facilitate resources for individual cancer patient needs, such as rehabilitation services (Kroenke, 2018) that may help to stabilize or mitigate cancer- and treatment-related physical decline. Therefore, diminishing contact may suggest a weakening of ties over time and/or reduced network support available to the cancer survivor, which may impact physical functioning.

Our study indicates that both cancer survivors and individuals without cancer were both adding and losing at least one alter over time. Adding two alters to the network was associated with lower disability in the overall NSHAP sample. Network loss was not associated with functional impairment in both groups. Losing network members over time may impact functional impairment because it eliminates potential sources of support required to sustain an independent lifestyle (Avlund et al., 2004; de Leon, Gold, Glass, Kaplan, & George, 2001). Although different measures were used to evaluate social network connectedness, our results are consistent with Michael and colleagues’ (2002) study, indicating that changes in social network integration were not associated with disability in cancer survivors (Michael, Berkman, Colditz, Holmes, & Kawachi, 2002). However, the relatively high and unchanged levels of emotional closeness and social support from family members among lasting ties suggests that strong relationships were maintained for both groups, providing support for the socioemotional selectivity theory. Among participants who lost alters, the majority reported that the alter moved, died, or they lost touch for other reasons, suggesting that the specific relationships lost over time may have been weaker ties less likely to provide help.

To expand on this point, studies on aging suggest that older adults maintain a close-knit network consisting of mainly familial ties that provide social support (Antonucci & Akiyama, 1987; Carstensen, Isaacowitz, & Charles, 1999). Van Tilberg’s 4-year longitudinal study in older Dutch adults showed that seniors tend to preserve a strong circle of familial ties and lose more friend ties over time (Van Tilburg, 1998). Keeping a close network may be one strategy to ameliorate some of the challenges of aging and cope with the transitions to come (Southwick, Litz, Charney, & Friedman, 2011). While losing certain network members to death, conflict, or other reasons may be a stressful situation to endure, a wealth of literature suggests that the process of aging itself can help older adults become resilient to future stressors (Helgeson, Snyder, & Seltman, 2004; Southwick et al., 2011).

The majority of cancer survivors in our sample were long-term survivors (e.g., surviving >5 years) and therefore, their social networks may reflect resilience to life challenges as they transition to the permanent survivorship phase. First, cancer survivors are also aging, which may cause them to arrange their networks into similar patterns as their cancer-free counterparts. Second, network loss may have little impact functional impairment because new relationships are created and support and resources from existing network members remain relatively stable (Southwick et al., 2011). Moreover, the relationships lost are intentionally pruned in favor of deep, supportive relationships, irrespective of age at diagnosis (Pinquart & Silbereisen, 2006). Our findings indicate that this preference is maintained among older cancer survivors many years after a cancer diagnosis. Future studies should replicate these findings in large-scale epidemiologic studies.

Given the importance of frequent contact for cancer survivors, interventions focusing on strengthening communication may improve quality of life. Family-level public health interventions have also been shown to improve access to resources among individuals with disabilities (Heller, Gibbons, & Fisher, 2015). Web- and mobile-based technologies hold promise for functionally impaired older adults as a means to create new connections or enhance existing social relationships, although their efficacy has yet to be established (McAlpine, Joubert, Martin-Sanchez, Merolli, & Drummond, 2015). For example, a randomized trial of a technology application among seniors demonstrated enhanced social connectivity and reduced feelings of loneliness (Czaja, Boot, Charness, Rogers, & Sharit, 2018). Further, a multicomponent trial targeting cancer survivors with high levels of distress showed that an online social networking and coping-based program significantly reduced cancer-related fatigue (Owen, O’Carroll Bantum, Pagano, & Stanton, 2017), a common posttreatment symptom associated with poor functional outcomes (Giacalone et al., 2013; Hofman, Ryan, Figueroa-Moseley, Jean-Pierre, & Morrow, 2007).

The strengths of our study include utilizing a prospective egocentric social network framework over a 5-year period from a large cohort of older adults with the ability to evaluate differences amongst cancer survivors and older adults. The limitations of the study include the small number of cancer survivors in the NSHAP dataset, limiting the ability to assess differences by cancer type. Prognostic features such as cancer stage and treatment information were not collected by NSHAP and the sample consisted of community dwelling older adults. Therefore, it is possible that the cancer survivors in our study may be healthier and/or diagnosed at earlier stages with better prognosis than other cancer samples. Additionally, the provision of social support from ego to network members was not collected, but offers important context, since higher levels of impairment may hinder one’s ability to provide support. The measures of disability in our study were self-reported measures of ADL and individuals with low levels of social support and network contact may rate their impairments as more severe. Finally, regression analyses can only explore associations among factors and does not imply causation.

Our results suggest that the social networks of cancer survivors and older adults have similar impacts on functional impairments, albeit a few subtle differences. Maintaining frequent contact and adding new relationships may be strategies to halt the progression of functional decline. Recommendations for future research include the use of a sample with greater variability in functional impairment and a personal network approach to describe the social networks of older adult cancer survivors. The health promoting aspects of social relationships should be further studied to optimize independence in late life.

Funding

This work was supported by the University of Maryland, College Park.

Conflict of Interest

None reported.