Social Support, Self-Efficacy for Decision Making, and Follow-up Care Use in Long-term Cancer Survivors

Abstract

Objective

Cancer survivors play an important role in coordinating their follow-up care and making treatment-related decisions. Little is known about how modifiable factors like social support are associated with active participation in follow-up care. This study tests associations between social support, cancer-related follow-up care use, and self-efficacy for participation in decision making related to follow-up care (SEDM). We also identified sociodemographic and clinical factors associated with social support among long-term survivors.

Methods

The FOllow-up Care Use among Survivors (FOCUS) study is a cross-sectional, population based survey of breast, prostate, colon, and gynecologic cancer survivors (n=1522) 4 to 14 years post-diagnosis. Multivariable regression models were used to test associations between perceived social support (tangible and emotional/informational support modeled separately), follow-up care use (past two years), and SEDM, as well as to identify factors associated with perceived support.

Results

Neither support type was associated with follow-up care use (all p>0.05), although marital status was uniquely, positively associated with follow-up care use (p<0.05). Both tangible support (B for a standard deviation increase (SE)=9.75(3.15), p<0.05) and emotional/informational support (B(SE)=12.61(3.05), p<0.001) were modestly associated with SEDM. Being married, having adequate financial resources, history of recurrence, and better perceived health status were associated with higher perceived tangible and emotional support (all p<0.05).

Conclusions

While perceived social support may facilitate survivor efficacy for participation in decision making during cancer follow-up care, other factors, including marital satisfaction, appear to influence follow-up care use. Marital status and social support may be important factors to consider in survivorship care planning.

Introduction

Growth in the number of survivors who are living longer after cancer [1, 2] has generated increased attention to optimizing cancer follow-up care [3]. Follow-up care encompasses prevention, detection, and treatment of recurrence, second cancers, and long-term and late physical and psychological effects of cancer or its treatment [3]. Patient-centered models of care recognize the key role that patients play in the receipt and coordination of cancer follow-up care, as well as in making decisions about the management of the adverse effects of cancer [4]. Yet, little is known about how modifiable factors like social support may enable survivors to actively participate in their cancer follow-up care. Perceived social support, a multidimensional construct including tangible aspects (i.e., perceived availability of practical resources such as transportation) and emotional or informational aspects (i.e., perceived acceptance, empathy, assistance with coping) [5, 6], is widely acknowledged as beneficial for health after cancer, including psychological adaptation [7–10], quality of life [11–13], and, in some cases, survival [14, 15]. However, associations between social support and participation in health care have received less attention in the literature.

Perceived social support may affect receipt of cancer follow-up care through the availability of practical assistance obtaining and attending appointments, or, as observed for other chronic health conditions, through creation of an emotionally supportive environment which facilitates healthy behavior and treatment adherence [16–19]. However, attendance at follow-up care appointments does not guarantee active participation in patient-centered care. Perceived social support may also increase active involvement in follow-up care; for example, through increasing self-efficacy for participating in decision-making [20], a core skill necessary for self-management of chronic illness [4]. Self-efficacy for decision making (SEDM) has been positively associated with survivors’ sense of control, asking questions of providers, actual participation in treatment-related decisions, and health-related quality of life [21–23]. Thus, perceived social support might affect health and adjustment in long-term cancer survivors via self-efficacy for, and ultimately participation in, treatment-related decision making.

Most research on social support in cancer survivors is based on middle-aged, Caucasian, breast cancer survivors within 5 years of diagnosis. The Follow-up Care Use among Survivors (FOCUS) study offers a unique opportunity to examine the role of social support in cancer-related follow-up care using a population-based sample of long-term survivors that is diverse in terms of ethnicity, age and cancer site. The current study tested associations between both tangible and emotional/informational support and 1) receipt of recent follow-up care and 2) SEDM, hypothesizing positive associations between both support types and both outcomes. Additionally, because identification of those survivors most likely to report low social support may facilitate referral of such individuals to support-enhancing interventions, we also identified sociodemographic and clinical factors associated with social support.

Methods

Sample and Data Collection

This inquiry utilized data from the cross-sectional FOllow-up Care Use among Survivors (FOCUS) study (http://cancercontrol.cancer.gov/ocs/focus.html). A detailed description of the methodology employed in FOCUS has been published elsewhere [24]. Briefly, FOCUS is a population-based, cross-sectional survey of adult survivors of breast, prostate, colorectal, endometrial and ovarian cancer who were 4 to 14 years post-diagnosis (n=1666). Survivors were recruited from two Surveillance, Epidemiology and End Results (SEER) cancer registries in California: the Los Angeles County Cancer Surveillance Program and the Greater Bay Area Cancer Registry at the Cancer Prevention Institute of California. The study was approved by the Institutional Review Boards at the Northern California Cancer Center (NCCC, now Cancer Prevention Institute of California, CPIC) and the University of Southern California (USC), in accord with assurances filed with and approved by the US Department of Health and Human Services. Mailed surveys were fielded between March 2005 and July 2006. Survivors eligible for the FOCUS study were diagnosed with cancer at age 21 or older, able to read English, and had completed active treatment. A total of 4,981 eligible cases were sampled by the registries. Of the eligible cases, 2,977 were located and 1,666 completed the survey (overall participation rate based on located cases=56%). Among located cases, non-respondents were more likely to be older, non-White, diagnosed with colorectal or gynecologic cancers, and diagnosed longer ago (all p<0.05). The current analytic sample (n=1522) excluded respondents who were not cancer free at the time of survey (n=84), who were missing >25% of social support items (n=32), or who reported American Indian/Alaska Native descent (n=28, small sample would preclude comparisons by race/ethnicity). Male survivors and survivors of prostate or ovarian cancer (compared to breast cancer) were more likely to be excluded (p<0.05).

Measures

Social support

Perceived social support was measured using a short-form of the MOS Social Support Scale, which measures perceived instrumental, informational, and emotional support [6]. The full measure contains 19 items and is scored from 0 to 100, with higher scores indicating better social support. The instrument was shortened in consultation with the instrument's author [25] to a 12-item version scored similarly to the longer version. The shortened version has been used in large epidemiological surveys [19, 25]. The measure does not specify a referent time frame over which support should be evaluated. Exploratory factor analysis using principle components analysis and promax (oblique) rotation provided a 2-factor solution accounting for 74.9% of the variance: tangible support (4 items, Cronbach’s α=0.92) and emotional/informational support (8 items, Cronbach’s α=0.94). The inter-factor correlation was 0.64. As part of the MOS scale, survivors also identified the one person who is most likely to help them with day-to-day activities (spouse/significant other, sibling, parent, child, friend) and the number of close friends or relatives with whom they feel at ease and can talk about what is on their mind. We measured perceived support rather than received support or other definitions. Perceived support, which is moderately correlated with received support, is the conceptualization of social support most consistently linked to health [26].

Follow-up care use

The FOCUS team developed questions to assess cancer follow-up care use. Specifically, survivors indicated whether or not they had seen any doctor for cancer-related follow-up care in the past two years, as well as when they last saw any doctor for cancer-related follow-up care (>1 year ago, 7–12 months ago, 4–6 months ago, 1–3 months ago, <4 weeks ago). The National Comprehensive Cancer Network® Clinical Practice Guidelines in Oncology (NCCN Guidelines®) currently recommend annual follow-up including appropriate surveillance examination for survivors ≥5 years post-treatment of breast [27], prostate [28], ovarian [29], and endometrial cancer [30]. However, because these guidelines were released between November 2003 and January 2004, only approximately one year before FOCUS was fielded, adequate time may not have elapsed for these guidelines to have become common practice. Additionally, survivors on an annual follow-up schedule may schedule appointments immediately after a full year has passed (i.e., in the 13 month), and guidelines for follow-up frequency do not extend beyond 5 years post-treatment for colon [31] and rectal cancer [32]. Given these limitations, we defined recent follow-up care as any cancer-related visit in the past two years (yes/no) and we replicated our analyses examining cancer-related follow-up care in the past 12 months as a sensitivity analysis.

Self-efficacy for decision making (SEDM)

The Decision-making Participation Self-efficacy scale [21] was used to measure survivors’ confidence about participating in multiple aspects of medical decisions related to their cancer follow-up care (e.g., letting the physician know he/she has questions, telling the physician which option he/she prefers). Survivors respond to 5 items on a 5-point Likert scale ranging from “not at all confident” to “completely confident.” A SEDM score was computed by summing item scores, and then linearly transformed on a 0–100 metric (higher scores reflect greater self-efficacy). This measure has demonstrated high internal consistency among cancer survivors (Cronbach’s α=0.89) and convergent validity such that SEDM is positively associated with perceived control and health-related quality of life [33].

Survivor characteristics

We collected self-report information on socio-demographic characteristics including age at survey, sex, marital status, race/ethnicity, and education. Additionally, participants indicated if they had adequate financial resources to meet their daily needs over the past 4 weeks (yes/no). Reports of having adequate financial resources were used as an economic indicator rather than annual household income because there was less missing data for adequate financial resources (2.5% vs. 10.2%, respectively); income is a limited predictor of health outcomes, particularly in an older adult population [34]; and perceived financial adequacy may be more predictive of follow-up care use than income [35].

Survivors rated their general health status on a 5-point Likert scale as assessed by the SF-12 [36]. We obtained data on comorbidity burden using an index of 14 possible conditions (congestive heart failure, cardiomyopathy, heart attack, angina, hypertension, peridcarditis, blood clots in the legs or lungs, stroke, chronic lung disease, lung fibrosis, liver disease, diabetes, arthritis, and depression/anxiety) adapted from previous studies of cancer survivors [19, 37] and symptom burden using an index of 26 possible problems experienced in the last 6 months [19]. We obtained data on cancer site, stage at diagnosis, and date of diagnosis from the registry. Survivors self-reported history of cancer recurrence and cancer treatment history.

Statistical Analysis

Analyses were conducted using the Statistical Analysis Software (SAS) callable version of SUDAAN 10.0 (RTI International, Research Triangle Park, NC). All analyses incorporated population weights, which is necessary in order to account for the complex survey design (sampling based on cancer site, age group, race/ethnicity, gender, time since diagnosis, and registry site). Weighted analyses also allow for inferences intended to reflect the population of cancer survivors, rather than a convenience sample. Prior to analysis, the data were examined using frequencies and descriptive statistics. Multiple logistic or linear regression models tested the association between social support (tangible and emotional/information support modeled separately) and receipt of cancer related follow-up care or SEDM. For these models, social support was rescaled by the standard deviation (regression coefficients reflect differences in the dependent variable per standard deviation increase in social support). Potential confounders (age, sex, marital status, race/ethnicity, education, insurance, adequate financial resources, cancer site, time since diagnosis, treatment history, history of recurrence, general health status, comorbidity burden, symptom burden) were modeled using backward elimination, with a >10% change in the coefficient for social support indicating confounding to appropriately control confounding and preserve statistical power [38, 39]. For models of recent follow-up care use, we developed the fully adjusted models for follow-up care in the past two years and then tested models of follow-up care in the past year including the same confounders. To identify the unique predictors of social support, multiple linear regression models simultaneously tested associations between socio-demographic and cancer characteristics selected a priori because of previous or hypothesized associations with support (age, race, gender, marital status, education, adequate financial resources, cancer site, time since diagnosis, history of cancer recurrence, perceived health status)[7, 40–43] and tangible and emotional/informational support (in separate models). Survivors in the analytic sample excluded from any multivariable regression models due to missing data on follow-up care use, decision-making self-efficacy, and/or a covariates were older than those with complete data (p<0.001) but were otherwise similar on race, gender, cancer site, and time since diagnosis (all p>0.05).

Results

Sample Characteristics

Approximately half of the sample was diagnosed with cancer >10 years ago and reported non-White race/ethnicity (Table 1). Most survivors were older than 65 years (67.0%), married (66.0%), reported at least some college education (75.0%) and reported adequate financial resources (88.3%).

Table 1.

Descriptive characteristics of long-term cancer survivors in FOCUS (N=1522)

		N (Weighted %)
Age at survey (years)	<50	3.9
	50–64	29.1
	65+	67.0
	mean (SD)	69.4 (43.7)
Race/ethnicitya	Non-Hispanic White	590 (51.6)
	Hispanic White	217 (16.5)
	African-American	367 (18.3)
	Asian/Pacific Islander	347 (13.6)
	Male	572 (49.8)
Marital Statusa	Married/living as married	926 (66.0)
	Single, widowed, separated, divorced	595 (34.0)
Educationa	High school or less	400 (23.9)
	Some college	550 (35.9)
	College graduate or more	558 (39.1)
Adequate financial resourcesa	No	165 (10.0)
Insurancea	None/Public Only	341 (23.0)
	Other	1125 (73.8)
Cancer site	Breast	379 (24.4)
	Prostate	376 (38.2)
	Colon/rectum	380 (21.5)
	Ovary	194 (5.4)
	Endometrial	193 (10.6)
Stage at diagnosisa	Localized	708 (39.3)
	Regional	334 (19.9)
	Distant	95 (2.7)
	Localized regional (prostate)	359 (36.5)
Years since cancer diagnosis	4 to 9	798 (52.0)
	10 to 15	724 (48.0)
Treatment history	Chemotherapy	576 (31.1)
	Radiation	515 (36.4)
	Chemoprevention	324 (19.5)
Recurrence/second cancer	Yes	218 (15.9)
Perceived Health Status (1=Excellent, 5=Poor), mean (SD)		2.7 (3.7)
# of comorbid conditions	0	275 (19.2)
	1–2	756 (49.9)
	3+	491 (31.0)
# of symptoms	0	128 (8.0)
	1–2	241 (16.0)
	3+	1153.0)

^aVariable also includes ≤3% missing data

Perceived Social Support

Cancer survivors reported relatively high perceived tangible and emotional/informational social support (weighted mean(sd)=75.2(105.9) and 75.1(90.7), respectively).¹ Most survivors indicated that the one person most likely to help with day to day activities was their spouse/significant other (59.5%), followed by child (19.4%), friend (6.5%), sibling (3.6%), or parent (1.0%). More than half (52.0%) of survivors reported at least 5 close relatives or friends, while more than a third (38.6%) indicated having 2 to 4 close relatives or friends and a small proportion of survivors reported 0 to 1 close friends or relatives (9.2%). Survivors who were married/living as married reported higher tangible and emotional/informational support than those who were not married (B(se)=21.97(2.17), p<0.001 and B(se)=14.55(1.84), p<0.001, respectively).

Social Support, Cancer Follow-up Care Use, and Self-Efficacy for Decision Making

Approximately three-quarters of survivors (75.7%, 95% CI=72.1–78.8) reported using cancer-related follow-up care in the past 2 years. The weighted mean SEDM score was 82.5 (sd=74.0). Follow-up care use in the past 2 years was not associated with SEDM (B(se)=3.22(1.82), p=0.08). Tangible support was positively associated with follow-up care use in the past two years in the unadjusted model (p=0.04, Table 2), but this effect was attenuated and non-significant after adjustment for confounders. Emotional/information support was not significantly associated with follow-up care use in the past two years either before or after adjustment for confounders (all p>0.05). It is notable that survivors who were married/living as married were more likely than unmarried survivors to report follow-up care use in the past two years, over and above tangible support and emotional/informational support (OR, 95% CI= 1.48, 1.01–2.15, p=0.04 and OR, 95% CI= 1.53, 1.04–2.25, p=0.03, respectively). Approximately two-thirds of survivors (67.3%, 95% CI=63.5–70.8%) reported follow-up care in the past year. Repeating our final models using follow-up care in the past year also revealed non-significant associations in adjusted models for both tangible and emotional/informational support (Table 2); marital status was again a significant predictor beyond tangible support (OR, 95% CI=1.70, 1.17–2.47, p=0.01) and emotional/informational support (OR, 95% CI=1.63, 1.11–2.39, p=0.01).

Table 2.

Associations between social support, follow-up care use, and self-efficacy for decision making among long-term cancer survivors

	Unadjusted Models		Multivariable Models
	OR (95% CI)†	p	OR (95% CI)†	P
Follow-up Care Use (past two years)
Tangible supportd, h	2.25 (1.05–4.84)	0.04	1.52 (0.69–3.32)	0.30
Emotional/informational supporta, d,e, g, h, l	1.81 (0.85–3.85)	0.12	1.20 (0.56–2.55)	0.64
Follow-up Care Use (past year)
Tangible supportd, h	1.73 (0.86–3.49)	0.13	1.07 (0.52–2.21)	0.86
Emotional/informational supporta, d,e, g, h, l	1.89, 0.94–3.81	0.07	1.27 (0.62–2.60)	0.51
	Β (se)†	p	Β (se)†	P
Self-efficacy for Decision Making
Tangible supportd, o	13.60 (2.77)	<0.001	9.75 (3.15)	0.002
Emotional supporto	15.61 (2.96)	<0.001	12.61 (3.05)	<0.001

^{a – q}indicate confounders included in multivariable models:

^arace/ethnicity,

^bsex,

^cage,

^dmarital status,

^eeducation,

^fadequate financial resources,

^ginsurance,

^htime since diagnosis,

ⁱcancer site,

^jstage at diagnosis,

^khistory of chemotherapy,

^lhistory of radiation,

^mhistory of chemoprevention,

ⁿrecurrence or second cancer (considered as confounder only for association between social support and self-efficacy for decision making),

^operceived health status,

^pcomorbidities,

^qsymptom burden. Confounders were modeled using backward elimination based on a 10–15% change in the coefficient for social support indicating significant confounding.

^†per standard deviation increase in social support.

Both tangible and emotional/informational social support were positively associated with SEDM before adjustment for confounders (all p<0.001, Table 2) and after adjustment (p=0.002 and p<0.001, respectively), although the overall variance in SEDM accounted for by these models was modest (R²=0.07 and 0.08, respectively).

Factors Associated with Perceived Social Support

Based on multivariable models, survivors who were married, had adequate financial resources, had a history of cancer recurrence or multiple cancers, and perceived better health status reported higher tangible and emotional/informational support (all p<0.05, Table 3). Furthermore, survivors with a college degree reported lower emotional/informational support than those with less education (p=0.02). Neither tangible nor emotional/informational support were associated with age, race/ethnicity, gender, cancer site, or time since cancer diagnosis (all p>0.05).

Table 3.

Multivariable models examining factors associated with tangible and emotional/informational support among long-term cancer survivors

	Tangible Support (N=1469)		Emotional Support (N=1470)
	B (SE)	p	B (SE)	p
Age at survey (years)	0.10 (0.09)	0.27	0.04 (0.08)	0.5767
Race/ethnicity	Wald F=1.78, p=0.15		Wald F=1.54, p=0.20
Non-Hispanic White	Ref		Ref
Hispanic White	0.36 (2.80)	0.90	−0.09 (2.72)	0.97
African-American	5.25 (2.42)	0.03	2.73 (2.43)	0.26
Asian/Pacific Islander	1.08 (2.05)	0.60	−2.37 (1.92)	0.22
Gender
Female	Ref		Ref
Male	4.40 (3.62)	0.22	3.89 (2.89)	0.18
Marital Status
Unmarried	Ref		Ref
Married	19.06 (2.53)	<0.0001	12.33 (2.18)	<0.0001
Education	Wald F=1.31, p=0.27		Wald F=4.09, p=0.02
College graduate or more	Ref		Ref
Some college	1.93 (2.12)	0.36	3.90 (1.94)	0.04
High school or less	4.14 (2.58)	0.11	6.36 (2.34)	0.007
Adequate financial resources
Yes	Ref		Ref
No	−12.14 (3.82)	0.002	−11.15 (3.24)	0.0006
Cancer site	Wald F=1.15, p=0.33		Wald F=0.71, p=0.59
Breast	Ref		Ref
Prostate	−2.12 (4.13)	0.61	−5.23 (3.54)	0.14
Colon/rectum	2.64 (3.48)	0.45	−1.34 (2.63)	0.61
Ovary	1.53 (2.73)	0.58	−0.49 (2.32)	0.83
Endometrial	−1.16 (3.45)	0.74	−0.30 (2.56)	0.91
Time since cancer diagnosis (years)
4 to 9	Ref		Ref
10 +	−2.09 (1.84)	0.26	−1.37 (1.62)	0.40
History of recurrence or second cancer
No	Ref		Ref
Yes	5.03 (2.43)	0.04	5.17 (1.91)	0.007
Perceived Health Status (1=Excellent, 5=Poor)	−4.28 (1.05)	<0.0001	−4.94 (1.06)	<0.0001

Discussion

This study examined associations between perceived social support, receipt of recent cancer follow-up care, and self-efficacy for participation in decision making related to follow-up care among long-term survivors. Contrary to expectations, neither tangible nor emotional/informational support was associated with use of recent cancer follow-up care. Although this finding is consistent with one previous study that failed to find an association between social support and surveillance for recurrent cancer among breast cancer survivors [44], this null association was surprising because social support has been positively associated with receipt of health care in other circumstances, including receipt of cancer screening in the general population [45, 46] and adherence to recommended care for other chronic conditions [17]. Ours and other studies that did not observe an association between social support and receipt of care were all conducted in cancer survivors. Particularly for this population, other factors (e.g., comorbid conditions or symptoms, survivorship care planning, provider recommendations) may play a stronger role in receipt of cancer care than social support. Although effects of marital status on health are often explained by perceived social support, in our study there was an effect of marital status on recent follow-up care, over and above perceived social support. This finding suggests there is something unique about the resources (either practical or emotional) associated with the marital/partnered relationship as opposed to other relationships (e.g., relatives, friends) [47]. For example, spousal relationships, particularly those perceived as high quality [47], may foster a sense of obligation to manage one’s health. Marriage may also facilitate access to medical care, including cancer-related follow-up care, through increased financial resources, access to health insurance, transportation assistance, or reminders to keep medical appointments. More research is needed to explore whether and how support from spouses differ from other relationships. However, because marital status may be an important factor in identifying survivors at greatest risk for inadequate follow-up care, marital status may be an important factor to consider in survivorship care planning with the patient and the patients’ other healthcare providers.

Both tangible and emotional/informational support were positively associated with self-efficacy for decision making among long-term survivors. Survivorship care can involve many complex medical decisions which affect long-term health [48], thus one way that family and friends of survivors may play an ongoing role in survivors’ health long after cancer treatment ends is through facilitation of self-efficacy for participating in decisions. However, associations between social support and decision making self-efficacy were modest. While other factors, such as survivorship care planning, patient-provider communication and support for disease self-management, may play a stronger role in facilitating decision making self-efficacy than social support, social support may be more strongly related to self-efficacy for decision making than receipt of follow-up care due to the more social nature of active participation in follow-up care.

Although the associations with social support tested in this study were modest, clinicians caring for survivors may foster numerous aspects of health and well-being in their patients, including mood, quality of life, management of physical symptoms [49, 50], and, potentially, engagement in follow-up care, by making appropriate referrals to support-enhancing interventions. Additionally, healthcare providers can be an important source of emotional and informational support [51, 52], for example, through active listening and empathic or empowering statements, that might potentially promote survivor self-efficacy for decision making. Clinicians can use the characteristics identified in this study to understand which survivors are at risk of experiencing low levels of social support to facilitate referral to support-enhancing interventions. We found that survivors who were not married/living as married, had inadequate financial resources, and reported poor health status perceived lower levels of tangible support and emotional/informational support. Additionally, in this study, education was inversely associated with emotional/informational support. In contrast, research on recently-diagnosed survivors found that higher education was associated with greater perceived emotional support [51]. Perhaps long-term survivors with higher education have different priorities or demands (e.g., prioritizing career over relationships), evaluate their available support more critically, or perceive fewer emotional needs over time and therefore identify less available support to meet such needs compared to less educated survivors. Surprisingly, history of recurrence or second cancers was associated with greater social support, suggesting that perceived support systems may be galvanized, rather than ‘burnt out,’ by multiple cancer treatment episodes. Further, although prior research has reported that support declines during the first year after cancer treatment [7, 40, 51], in our study, the level of support reported was similar among survivors diagnosed 4 to 9 versus 10 to 15 years ago suggesting that support may stabilize over time in the absence of new major cancer events.

To our knowledge, this study is the first to examine associations between social support and participation in cancer-related follow-up care. We utilized a large, population-based sample of long-term cancer survivors that was heterogeneous with respect to cancer site and race/ethnicity. However, the registries from which the FOCUS sample was selected cover California, including predominantly urban/suburban regions and our sample reported notable levels of higher education and adequate financial resources. One reason may be that eligible participants with lower socioeconomic status may be more difficult to locate using mail based surveys. Thus, Findings may not be applicable to survivors in other regions of the United States, those with fewer economic resources, or those living in rural areas. Also, these results may reflect respondent bias: survivors who were younger and diagnosed more recently were more likely to participate in FOCUS, responding survivors may be more involved in cancer care and/or more socially connected, and non-locatable survivors may differ from those who were located in unmeasured but perhaps important ways. In particular, non-respondents were more likely to be older, perhaps due to age-associated limitations in health status or changes in residence, in which case these results may not reflect the experiences of older adults with the poorest health. However, our survey sample reflects the age distribution of cancer survivors [53]. Given that older adults tend to have fewer social relationships from which greater support is obtained relative to younger adults [54], our findings may have limited generalizability for young adults.

Follow-up care was assessed using one-item indicators and survivors may misreport follow-up care, particularly the timing of care. Furthermore, this survey was conducted at little over one year after follow-up care guidelines were released. Findings may have differed if this study were fielded after more time had passed for these guidelines to be adopted. The lack of clear guidelines for long-term follow-up care across the cancers surveyed make it difficult to define which survivors are “current” on follow-up care. Consistent findings across analyses of multiple follow-up care intervals increase confidence in our results. Further, we measured perceived social support; results may have differed if we measured support differently (e.g., received support or social network size). Given the limitations of cross-sectional data, no causal inferences can be made regarding the association between social support and decision-making self-efficacy, or between SEDM and better health outcomes. Finally, while FOCUS data were collected several years ago (2005–2006); we see no indication that associations between social support and follow-up care have changed since that time, although financial resources and the ability of social networks to provide support may have decreased with recent economic declines.

Although factors other than social support determine recent follow-up care use, marital status was robustly associated with follow-up care use. Additionally, social support was modestly associated with self-efficacy to participate in follow-up care decision making among long-term survivors. Recent research has found associations between social support and other aspects of patient-centeredness; emotional/informational support was positively associated with satisfaction with providers and inversely associated with difficult interactions with providers [55]. Future research should examine the role of social support and marital status in additional aspects of patient-centered care (e.g., survivor adherence to treatment or behavior change recommendations, trust in physicians, and survivor attitudes towards survivorship care). Additionally, future research could investigate whether explicitly incorporating spouses/partners as key members of the survivorship care team might increase survivors’ receipt of, and active participation in, follow-up care. Other factors not measured in this study, such as survivorship care planning, communication with and among members of the healthcare team, and provider or survivor attitudes and preferences also merit future research attention as these factors may play a stronger role than social support in determining receipt of cancer follow-up care. Also, longitudinal research is needed to fully understand how social support changes over time in long-term survivors. Finally, future work should determine whether associations between social support and improved health among cancer survivors can be partially explained by active participation in cancer follow-up care. Ultimately, a better understanding of the pathways through which social support affects health in cancer survivors will lead to more targeted, maximally effective interventions for empowering survivors to manage their health and actively participate in their care.