Volunteer Engagement and Systemic Inflammation: Does Helping Others Benefit Oneself?

Abstract

Background and Objectives

Although research on the health benefits of volunteering has proliferated, most studies are cross-sectional and rely on self-reported measures of health. Drawing from role theory, the objectives of this study are to examine if (a) volunteering engagement is related to systemic inflammation in later life, as measured by C-reactive protein (CRP); (b) the effect of volunteering varies by age; and (c) volunteering is related to change in CRP over time.

Research Design and Methods

This study uses 4 waves of data from the Health and Retirement Study, a nationally representative survey of adults 50 years or older. Nested linear regression models were used to examine the relationship between volunteer engagement and CRP concentration in later life. Residualized regression models were used to examine the effects of volunteer engagement on change in CRP.

Results

Results revealed that volunteering is modestly associated with lower CRP concentration, but only for respondents 65+. Highly engaged volunteers had lower CRP than both mid-level and nonvolunteers. Longitudinal analyses revealed a leveling of the beneficial effect of volunteering by age, indicative of reduced returns among the oldest respondents in our sample.

Discussion and Implications

These results support previous studies that volunteering, and doing so at a high engagement level, is associated with slightly lower levels of CRP. Leaders in medicine, public health, and social services should consider implementing volunteering programs for older adults.

Millions of Americans engage in volunteering each year, and the reported prevalence of this humanitarian behavior among older adults has risen during the past half-century. For example, in 1965, 10% of those aged 65 and older reported volunteering; by 2015, this number had more than doubled (Chambré, 2020). Not only are more older adults volunteering, but there is evidence that they have higher levels of volunteer engagement than their younger counterparts (Chambré, 2020). Because of the prevalence of volunteer engagement among older adults, studies have proliferated over the last two decades that examine the influence of volunteering on measures of well-being.

The bulk of this research focuses on formal volunteering, which refers to unpaid prosocial activities that occur within an organizational context (Ajrouch & Tesch-Roemer, 2017). The literature reveals that formal volunteering has a beneficial effect on an array of health outcomes in later life including life satisfaction (Matthews & Nazroo, 2021), mental health (Huo et al., 2021), functional health (Carr et al., 2018), cognitive functioning (Proulx et al., 2018), and longevity (Harris & Thoresen, 2005). Given that heart disease is the leading cause of death worldwide, many scholars have examined the relationship between volunteering and cardiovascular health. Results show that volunteering generally reduces the risk of hypertension (Sneed & Cohen, 2013) and incident cardiovascular disease (Burr et al., 2018).

Despite the preponderance of evidence that volunteering is beneficial, there are two limitations of the literature that we address directly in this investigation. First, the majority of studies on volunteering examine self-reported health outcomes only, which may be influenced by access to and use of medical care (Syed et al., 2013). In response to this limitation, some studies have turned to biomarkers to tap biological risk for disease (Burr et al., 2016; Kim & Ferraro, 2014; Kim & Yoon, 2020).

One such biomarker examined in the volunteering and health literature is C-reactive protein (CRP). CRP is an “upstream marker” of immunological functioning that is highly predictive of cardiovascular health events when it is systemically elevated (Willerson & Ridker, 2004). Previous studies report that elevated CRP is more common among adults with type 2 diabetes (Soinio et al., 2006), and cardiovascular diseases (Cozlea et al., 2013). Other studies report that not smoking, high physical activity, and having a normal body mass index (BMI) are associated with lower CRP (Colbert et al., 2004; McDade et al., 2006). We therefore examine volunteering’s association with CRP to investigate how volunteering “gets under the skin.”

Second, most studies of volunteering and health rely on measures of volunteering at one point in time as a binary variable (yes/no) or as annual time, in hours, of volunteering. Therefore, we created a construct of volunteer engagement that incorporates consecutive waves of volunteering with the number of hours spent volunteering in each wave. Compared to older people who do not volunteer or do so only occasionally, we ask whether being a highly engaged volunteer is associated with lower CRP and whether volunteer engagement is related to change in CRP over time. This study enhances our understanding of the relationship between volunteer engagement and health by using longitudinal data to examine an upstream biological marker of risk for cardiovascular disease.

Literature Review

A frequent approach in the literature on volunteering and health is to use cross-sectional data to compare volunteers to nonvolunteers. This approach has yielded notable findings, most often revealing that older adults (50 years or older) who volunteer have better health than nonvolunteers (Estrella et al., 2020; Kim et al., 2020). Despite the documented importance of the volunteer role, a growing number of studies have moved to investigating whether the number of hours spent volunteering is related to health benefits among older adults (Sneed & Cohen, 2013). The logic is to identify the “dose” of hours needed to achieve some protection against health threats. These studies generally reveal that individuals who volunteered more hours reported better health. For instance, Sneed and Cohen (2013) identified notably lower risk of hypertension among older adults who volunteered at least 200 hr during the past year. In their analysis, volunteering less than 200 hr per year did not confer any protection against incident hypertension. Further, results from a longitudinal study revealed that more frequent volunteer hours were associated with greater well-being (Morrow-Howell et al., 2003).

Examining the number of hours spent volunteering has been helpful for demonstrating that the dose of volunteering matters. However, one unanswered question in the current body of literature is how the number of volunteer hours is associated with health benefits when prolonged. Musick and Wilson (2003) created a measure of sustained volunteering over three waves but did not measure volunteer hours. Results indicated that volunteering two or three waves was associated with fewer depressive symptoms for respondents 65 years and older; volunteering one wave had no effect on depressive symptoms. We conceptualize volunteer engagement as involving (a) hours spent each year and (b) recurrent volunteering over years. Is volunteering at least 200 hr for 1 year sufficient to reap health benefits? Or must one be a “super-volunteer”—many hours over multiple years—to gain a benefit (Einolf & Yung, 2018)?

A review of the literature focused on systemic inflammation as the outcome of interest provides somewhat inconsistent evidence from two national surveys on the purported benefits of volunteering. Using the National Social Life Health and Aging Project, Kim and Ferraro (2014) reported that volunteering is associated with lower levels of CRP and that volunteers 70 years and older had lower CRP concentrations compared to volunteers under 70. By contrast, Burr and colleagues (2016) used the Health and Retirement Study (HRS) and showed that volunteering is associated with normal CRP concentration (i.e., <3 mg/L) among respondents in unadjusted models but not in the fully adjusted model. The inconsistency could be due to multiple reasons, but CRP measurement is a critical factor; the former team log transformed CRP, while the latter team dichotomized it. (In supplementary analyses, Burr and colleagues [2016] log transformed CRP and obtained parallel findings for the main effect reported by Kim and Ferraro [2014].)

Compared to the current body of literature examining the relationship between volunteering and CRP, our study is distinctive in several ways. First, Kim and Ferraro (2014) used cross-sectional data only, while Burr and colleagues (2016) examined CRP 2 years after the assessment of volunteering. Neither examined volunteer engagement over multiple waves.

Second, Burr and colleagues measured volunteer hours with three categories: 0, 1–99, and ≥100. The HRS, however, offers volunteers seven response categories, which we preserve in our analysis, because prior studies report beneficial effects only among older adults who volunteered 200 or more hours (Sneed & Cohen, 2013). Kim and Ferraro (2014) used an ordinal variable ranging from never to several times a week. As such, it is difficult in each of those approaches to detect exceptionally high volunteer activity.

Third, neither study examined change in CRP. We are aware of only one study that examines the relationship between volunteering and change in CRP, but it focuses on the role of sleep quality as a potential mediator or moderator of the relationship (Kim & Yoon, 2020).

Theoretical Framework

Scholars often use role theory to explain why volunteering promotes health. Indeed, many argue that the salutary effect of volunteering is due to the accumulation of roles, especially when the role involves “moral voluntarism” (Siebel, 1974, p. 572). Adding a role to one’s repertoire is advantageous because “role requirements give purpose, meaning, direction, and guidance to one’s life,” each of which is crucial to well-being in later life (Thoits, 1983, p. 175). Indeed, a recent study by Yang and Matz (2022) reported that purpose in life partially mediated the relationship between volunteering and depressive symptoms.

Researchers also argue that occupying roles in later life is beneficial because they increase social integration (Morrow-Howell et al., 2014), which is protective against the deleterious health effects of social isolation (Thoits, 1983). Greater social integration also provides access to an array of resources that promotes health, such as social capital (Onyx & Warburton, 2003), and social support (Parkinson et al., 2010), both of which can buffer the effects of stressors (Cohen, 2004). Sustaining high volunteer engagement may produce health benefits through role salience: the more one spends time volunteering, the greater the volunteer identity becomes salient, which contributes to well-being (Thoits, 2013).

The beneficial effect of volunteer engagement may vary such that the effect of volunteering is more pronounced at higher ages when role loss is more prevalent (Moen, 1995). Indeed, Greenfield and Marks (2004) reported that volunteering moderated the deleterious effect of role loss on psychological well-being among adults aged 65–74. Other studies report that, compared to their younger counterparts, higher volunteer engagement is especially beneficial for the health of older adults (Kim & Ferraro, 2014; Van Willigen, 2000). These studies suggest that occupying the volunteer role during a time in the life course when role loss is more common “might assume special importance” and therefore have a differential impact on health (Musick & Wilson, 2003, p. 261).

The present inquiry builds on studies of volunteering and health in several ways: (a) measuring volunteer engagement in a longitudinal study; (b) using all available categories to examine the number of hours spent volunteering without collapsing response categories; and (c) assessing the influence of volunteer engagement on both initial and repeated measures of CRP. We specify three hypotheses.

1. Greater volunteering engagement is associated with lower CRP, such that highly engaged volunteers have lower CRP compared to mid-level volunteers and nonvolunteers.

2. The effect of volunteering on CRP varies by age, such that the benefit of volunteering is greater among older respondents (65+) than middle-aged respondents (50–64).

3. Volunteering is related to a decrease in CRP over 4 years.

Design and Methods

Data

We analyzed data from the Health and Retirement Study (HRS). The HRS is a longitudinal panel study that surveys individuals aged 50 and older in the United States. We used data from the 2004 (W1), 2006 (W2), 2008 (W3), and 2010 (W4) waves. Biomarkers were collected every 4 years, starting in 2006, from a random one-half sample of HRS respondents. For cross-sectional models, we analyzed CRP from the initial 2006 half sample (W2). For longitudinal analyses examining volunteering’s relationship with change in CRP, we studied change in systemic inflammation over 4 years (W2–W4).

We limited our analysis sample to respondents who met the following criteria: (a) participated and had nonzero weights in the 2006 biomarker half sample (N = 6,213); (b) provided information for W1 and W2 volunteering (N = 6,059); (c) were at least 50 years old in 2004 (N = 5,780); (d) identified as non-Hispanic White, non-Hispanic Black, or Hispanic (N = 5,675); and (e) had no missing values on any of the included covariates (N = 5,540). These selection criteria yielded an analysis sample of 5,540 older adults at baseline.

Longitudinal analyses that examined change in CRP (i.e., Table 3) have a smaller sample size because respondents had to participate in W2 and W4 biomarker collection and have data for covariates in W1. Specifically, 1,692 respondents had CRP measured in W2 but not W4. Of the 1,692 respondents who did not have CRP measured in W4, 938 were missing because they died sometime between W2 and W4. Those who did not participate in the CRP collection in W2 or W4 were more likely to report high BMI and high physical activity.

Table 3.

Residualized Regression Models of Change in C-Reactive Protein (CRP), Health and Retirement Study

	Model 1	Model 2	Model 3
Variable	b (SE)	b (SE)	b (SE)
CRP (W2)	0.548 (0.013)***	0.546 (0.014)***	0.549 (0.013)***
Volunteer engagement, W1–W2	0.006 (0.004)		−0.055 (0.027)*
Volunteer engagement, W3–W4		0.003 (0.004)
Age	0.001 (0.002)	0.001 (0.002)	−0.002 (0.002)
Female	−0.050 (0.031)	0.047 (0.032)	−0.051 (0.031)
Black	0.001 (0.051)	0.004 (0.051)	0.006 (0.051)
Hispanic	0.049 (0.062)	0.039 (0.062)	0.047 (0.062)
Education	−0.004 (0.006)	0.006 (0.006)	−0.004 (0.006)
Wealth	−0.008 (0.005)	−0.004 (0.005)	−0.008 (0.005)
Current smoker	0.100 (0.048)*	0.097 (0.048)*	0.091 (0.048)
Physical activity	−0.008 (0.004)	−0.008 (0.004)	−0.008 (0.004)
Overweight	0.137 (0.037)***	0.137 (0.038)***	0.142 (0.037)***
Obese	0.188 (0.044)***	0.192 (0.044)***	0.188 (0.044)***
Hypertension	0.038 (0.032)	0.038 (0.032)	0.036 (0.032)
Self-rated health	−0.040 (0.017)*	−0.040 (0.017)*	−0.040 (0.017)*
Volunteer engagement W1–W2 × Age			0.001 (0.000)*
Constant	0.331	0.315	0.503
R 2	0.371	0.369	0.372
N	3,848	3,812	3,848

Notes: Dependent variable is the natural logarithm of CRP (W4). SE = standard error.

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

Measures

Systemic inflammation

During enhanced face-to-face interviews, HRS biomarkers were collected through a blood sample. Eighty-three percent of the W2 respondents consented to bloodspot data collection, and 97% completed the collection (Crimmins et al., 2013). Because CRP values are right-skewed, CRP was transformed into the natural logarithm, a common practice in the literature (Kim & Ferraro, 2014; Kim & Yoon, 2020). After conducting the natural log transformation, CRP approximated a normal distribution.

Volunteer engagement

HRS asked respondents at each biennial survey if they have “spent any time in the past 12 months doing volunteer work for religious, educational, health-related, or other charitable organizations?” If respondents answered affirmatively, HRS then asked how many hours respondents spent volunteering during the previous 12 months. Volunteers could select one of seven response categories for the number of hours they spent volunteering: 1–49 hr, about 50 hr, 50–99 hr, about 100 hr, 100–199 hr, about 200 hr, or 200 or more hours (coded 1–7, respectively, and nonvolunteers were coded 0). We used the information on whether a person volunteered at specific survey waves, as well as the number of hours spent in the role, to create a composite score for volunteer engagement. Volunteer engagement ranges from 0 to 14, where 0 represents respondents who did not volunteer and 14 represents respondents who reported volunteering more than 200 hr for both W1 and W2.

Based on the volunteer engagement construct, we conceptualize three profiles: highly engaged volunteers, mid-level volunteers, and nonvolunteers. Respondents who scored 14 are highly engaged volunteers (i.e., super-volunteers) because they report volunteering 200+ hr over two consecutive waves. Those who did not report volunteering at either wave are nonvolunteers (scored 0), while those who scored a 7 are referred to as mid-level volunteers. It is possible that mid-level volunteers reported volunteering many hours for one wave (i.e., 200+ hr) and very little the next, or reported a moderate number of volunteer hours for both waves. Analysis of mid-level volunteers for W1 and W2 reveals that nearly half (45.16%) reported volunteering 200 or more hours in one wave and none for the other. Alternatively, 30.5% of mid-level volunteers reported a moderate number of hours for both waves.

Covariates

Covariates were assessed at baseline (W1). Age was coded in years, and sex was dichotomized with 1 indicating female. Race/ethnicity was examined by using three binary variables representing non-Hispanic White (reference group), non-Hispanic Black, and Hispanic respondents (hereafter, White, Black, and Hispanic). Education was measured in years (0–17+). Wealth refers to participants’ net worth minus debt. Because the distribution was skewed, it was transformed by taking the cube root.

We also adjusted for five health indicators in our study. First, smoking status was measured with a dichotomous indicator, where 1 indicates currently smoking. Second, a self-reported physical activity scale gathered both frequency and intensity of physical activity. On a scale of 1–4, participants responded how often they participated in mild, moderate, and vigorous physical activity. The physical activity scale was creating by weighting the type of activity by the intensity based on metabolic equivalent recommendations (mild = 1.2, moderate = 1.4, and vigorous = 1.8). Possible scores range from 0 (no physical activity) to 17.6 (mild, moderate, and vigorous physical activity daily; Lathan & Williams, 2015). Third, BMI was separated into three categories: underweight or normal (BMI < 25; reference group), overweight (25 $\le$ BMI < 30), or obese (BMI $\ge$ 30). Fourth, a self-reported measure of hypertension is included, with the reference group being “does not have hypertension.” Finally, to capture overall health status, we adjust for self-rated health. Responses range from poor (1) to excellent (5).

Descriptive statistics for all variables are presented in Table 1.

Table 1.

Descriptive Statistics for the Analytic Sample, Health and Retirement Study

Variable	Range	Percent	Mean (SD)
CRP (W2)a	−3.50 to 5.63	—	0.74 (1.25)
CRP (W4)b	−3.00 to 5.22	—	0.60 (1.17)
Volunteer engagement, W1–W2a	0–14	—	2.59 (3.96)
Volunteer engagement, W3–W4b	0–14	—	2.65 (3.94)
Age	50–95		66.10 (9.81)
Female	—	41.72	—
Race/ethnicity
Whitec	—	79.62	—
Black	—	12.06	—
Hispanic	—	8.32	—
Education	0–17	—	12.56 (3.15)
Wealth ($\sqrt[3]{}$)	−12.60 to 31.00	—	5.95 (3.62)
Current smoker	—	13.69	—
Physical activity	0–17.6	—	7.86 (3.91)
Body mass index
Normal/underweightc	—	31.63	—
Overweight	—	38.88	—
Obese	—	29.49	—
Hypertension	—	49.11	—
Self-rated health	1–5	—	3.26 (1.09)

Notes: CRP is the natural logarithm of C-reactive protein; SD = standard deviation.

^a N = 5,540.

^b N = 3,812.

^cReference group.

Analytic Strategy

We adjusted for sample weights and clustering based on the 2006 biomarker subsample to correct for the multistage cluster design. We present two sets of analyses to examine (1) the relationship between volunteer engagement and CRP at W2 and (2) the relationship between volunteer engagement and change in CRP from W2 to W4. Both sets of analyses used ordinary least-squares regressions because our dependent variable is a continuous measure of systemic inflammation.

Results

As shown in Table 1, the mean CRP value at W2 was 0.74. The original, untransformed mean CRP concentration at W2 was 4.64; about 39% of the sample reported elevated CRP (i.e., ≥3 mg/L). At W4, the logged CRP was 0.60 and the untransformed CRP was 3.81; about 33% of the sample in W4 reported an elevated CRP. The prevalence of elevated CRP in older adults is similar to that of another study that examined volunteering and systemic inflammation in later life (Burr et al., 2016). The decline in CRP over 4 years was due largely to selective mortality. Respondents deceased by W4 were more likely than survivors to have high CRP (≥3.0) at the initial measurement. The mean volunteer engagement score was 2.59 and 2.65 for W1–W2 and W3–W4, respectively. Approximately 53% of the sample did not volunteer at W1 or W2.

Results from Model 1 of Table 2 reveal that volunteering is associated with lower CRP (b = −0.023, SE = 0.004, p < .001). Model 2 adjusts for age, sex, race/ethnicity, education, and wealth, revealing that women and Black respondents have higher CRP, while increases in age, higher education, and greater wealth are associated with lower CRP. Volunteer engagement during W1–W2 remains significant, although attenuated by the adjustment (b = −0.012, SE = 0.004, p < .05). After adjustment for health and lifestyle factors in Model 3, however, volunteer engagement during W1–W2 is no longer significantly associated with lower CRP. Being a current smoker, overweight, obese, and hypertensive were associated with higher CRP, but greater physical activity and higher self-rated health were associated with lower CRP.

Table 2.

Ordinary Least-Squares Models Predicting C-Reactive Protein (W2), Health and Retirement Study (N = 5,541)

	Model 1	Model 2	Model 3	Model 4
Variable	b (SE)	b (SE)	b (SE)	b (SE)
Volunteer engagement, W1–W2	−0.023 (0.004)***	−0.012 (0.004)*	−0.007 (0.004)	0.054 (0.028)
Age		−0.003 (0.002)*	0.001 (0.002)	0.003 (0.002)
Female		0.222 (0.033)***	0.263 (0.032)***	0.264 (0.032)***
Black		0.205 (0.053)***	0.081 (0.052)	0.077 (0.052)
Hispanic		−0.087 (0.065)	−0.095 (0.062)	−0.092 (0.062)
Education		−0.035 (0.006)***	−0.018 (0.006)**	−0.018 (0.006)***
Wealth		−0.033 (0.005)***	−0.012 (0.005)*	−0.012 (0.005)*
Current smoker			0.284 (0.049)***	0.292 (0.049)***
Physical activity			−0.017 (0.004)***	−0.016 (0.004)***
Overweight			0.328 (0.038)***	0.328 (0.038)***
Obese			0.778 (0.044)***	0.779 (0.044)***
Hypertension			0.106 (0.033)***	0.107 (0.033)***
Self-rated health			−0.083 (0.005)***	−0.082 (0.017)***
Volunteer engagement × Age				−0.001 (0.001)*
Constant	0.804	1.514	0.459	0.012
R 2	0.005	0.045	0.120	0.121

Notes: Dependent variable is the natural logarithm of C-reactive protein. SE = standard error.

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

To test if the effect of volunteer engagement varies by age, Model 4 of Table 2 includes a product term, which is significant. This indicates although volunteer engagement has no effect on CRP for respondents aged 50–64, volunteering is associated with lower CRP for older adults 65+. Indeed, additional analyses of the average marginal effects of volunteering revealed that volunteer engagement is associated with lower CRP among respondents 65 and older only.

We present Figure 1, derived from Table 2 analyses, to illustrate the cross-sectional relationship between three volunteer engagement profiles, age, and W2 CRP. Figure 1 demonstrates that highly engaged volunteers 65 years or older have lower CRP concentrations than nonvolunteers and mid-level volunteers of the same age. Figure 1 also illustrates that, for example, highly engaged volunteers who are 80 years old have lower CRP than respondents who are 65 and volunteer the same amount. Figure 1 and Model 4 of Table 2 lend support for Hypothesis 2.

Figure 1.

Predicted values of W2 C-reactive protein (CRP) by age and volunteer engagement level.

Table 3 presents results from analyses regressing CRP W4 on CRP W2, volunteer engagement, and covariates. In terms of change in CRP, Model 1 reveals that a one-unit increase in CRP at W2 was associated with a 0.548 increase in CRP at W4. Model 1 also shows that W1–W2 volunteer engagement was not associated with a change in CRP from W2 to W4. In Model 2, we tested whether more recent volunteering (W3–W4) is associated with a decrease in CRP from W2 to W4, but this relationship was nonsignificant. Hence, Models 1 and 2 provide no evidence to support Hypothesis 3.

Whereas analyses performed in Table 2 the effect of volunteering varies by age, we tested whether volunteering and age interact when predicting change in CRP (Model 3). The residualized regression predicting change in CRP from W2 to W4 reveals that volunteering and age interact in their effect on CRP, but not in the hypothesized direction. Note first that volunteering W1–W2 is significant and protective against high CRP once the volunteering-and-age product term is included in the estimation (b = −0.055, SE = 0.027, p < .05). The product term in Table 3 (Model 3), however, is positive, indicating that CRP rose among older volunteers compared to their younger counterparts (b = 0.001, SE = 0.000, p < .05). Although in Table 2 volunteer engagement was associated with slightly lower CRP for adults 65 and older, Model 3 of Table 3 reveals a leveling of that advantage. For the oldest respondents 4 years later, the salubrious effect of volunteering on CRP diminished.

Sensitivity Analyses

We also completed four sensitivity analyses. First, we completed a series of supplementary analyses to test interactions between volunteer engagement and (a) race/ethnicity, (b) BMI, and (c) gender. However, none of these interactions were significant and thus are not presented. Second, we estimated parallel models with separate variables for whether a person volunteered over two waves (coded as 0, 1, or 2) and hours volunteered (average over two waves), but the conclusions were identical, even for the interactions. Therefore, we presented the more parsimonious approach of a composite score, ranging from 0 to 14, which integrates both elements. Third, because previous studies have examined whether volunteering reduces the risk of having clinically high CRP (i.e., ≥3 mg/L; Burr et al., 2016), we also completed logistic regressions using the same clinical threshold for elevated CRP (0 = <3, 1 = ≥3). Results indicated the volunteer engagement had no effect on CRP. This result was somewhat expected given the results presented in the current study indicate that volunteering is modestly associated with lower CRP. Last, additional variables considered in preliminary analyses include marital status, occupation status, and diabetes. Because these covariates were not significantly related to CRP, they were omitted from the models presented.

Discussion

Curious if volunteering is good for the person doing prosocial work, we used longitudinal data to examine the relationship between volunteering and CRP. Several theoretical mechanisms could help explain how volunteering gets under the skin for older adults, but most emphasize the power of social connections. For instance, volunteer engagement generally increases social integration (Morrow-Howell et al., 2014), which then provides opportunities for social support (Parkinson et al., 2010) and social capital (Onyx & Warburton, 2003), all of which protect against stress—a contributor to systemic inflammation (Cohen, 2004; Gouin et al., 2012). Using nationally representative data from older adults, results from this study support prior findings that volunteering is associated with lower CRP but reveal contingencies for who might benefit the most and whether the benefit endures over time.

Our first hypothesis predicted that highly engaged volunteers will have lower levels of systemic inflammation compared to mid-level volunteers or nonvolunteers. We found partial support for this hypothesis: volunteering is associated with slightly lower CRP, but only for individuals 65 years and older. More specifically, our results indicate that highly engaged volunteers (i.e., volunteering many hours over multiple surveys) had lower CRP than both mid-level and nonvolunteers. Other studies support the result that highly engaged volunteers receive the greatest health return. For example, Sneed and Cohen (2013), reported that lower hypertension risk occurred only at the highest levels of volunteering (i.e., at least 200 hr per year). Our study contributes to the growing body of volunteering and health literature by assessing volunteer engagement over multiple waves. Doing so differentiates the super-volunteers from those who either give less of their time or do so for a limited period. Our measure of volunteer engagement, therefore, provides a more complete profile of volunteering than one can assess with a single measure of volunteering.

Our second hypothesis specified that the effect of volunteer engagement varies by age such that older adults benefit more from volunteering. Our results show that volunteer engagement was associated with lower CRP only among respondents 65 and older. Therefore, the reported interaction between volunteering and age in the current study is consistent with results from Musick and Wilson (2003). Results from the study revealed the effect of volunteering was significant only among respondents 65 and older. Many prior studies report that CRP rises with age, but there also is evidence that specific activities are associated with lower CRP. For instance, we found that physical activity by older adults was associated with lower initial CRP. Our findings about volunteer engagement and lower CRP provide evidence for another plausible avenue of intervention. To some degree CRP is modifiable in later life, and our results show that adults 65 years or older who volunteer reap a health benefit for their prosocial activity, albeit a modest one. Previous research suggests that altruistic acts may trigger hormones that help regulate the stress response, perhaps lowering the likelihood of systemic inflammation (Poulin & Holman, 2013). Future research that examines the effect of volunteering on both stress-related hormones and CRP is needed to verify this thesis.

Finally, our third hypothesis predicted that volunteering is related to a decrease in CRP over a 4-year period (W2–W4). We were somewhat surprised by the results. Without testing interactions, we found no relationship between volunteer engagement and change in CRP. Only by testing a multiplicative model did we uncover that the effect of volunteering on change in CRP is moderated by age. The main effect of volunteering W1–W2 was negative, meaning that volunteering was associated with a decline in CRP, but the product term of volunteering W1–W2 and age was positive, indicating that CRP increases for the oldest volunteers from W2 to W4.

Drawing from both the cross-sectional and the longitudinal analyses, we conclude that there is a leveling of the beneficial effect of volunteering for the oldest respondents, indicative of diminishing returns at advanced ages. Whereas CRP concentration cannot decrease unremittingly, the salutary effect of volunteering levels off at advanced ages, even among highly engaged volunteers. This may be due, in part, to the system dysregulation common at advanced ages resulting in organismal senescence (Bernard et al., 2020). Consistent with the concept of “inflammaging” (Franceschi et al., 2018), long-lived people face a cascade of biological challenges that eventually give way to inflammation. Volunteering has multiple mental and physical health benefits, but it is not a panacea for biological risks. Volunteering may slow the process of inflammaging, but inflammation is hard to contain at advanced ages, especially in modern societies beset by chronic disease. Given the dearth of research on change in systemic inflammation among older adults, we encourage other researchers to replicate—or refute—the result of diminishing returns reported herein.

Despite its contributions, this study is limited in four ways. The first is related to how the HRS asks about volunteer activity. Because the core survey is filled out every 2 years, we were able to tap volunteering data only in the year prior to survey completion. Thus, we were unable to account for volunteering that occurred prior to the year referenced in the survey question. Second, the coarseness of the response categories for volunteering means that it is impossible to differentiate volunteer hours within categories (e.g., 50–99). Third, we are unable to differentiate variability in the organizational setting for the volunteer activity. The HRS taps volunteer work from in a variety of settings, including religious, educational, health-related, or charitable organizations, but it is unclear if the findings reported herein vary across these settings. Last, we used prior research to suggest that volunteering increases purpose in life, social integration, and number of roles, which then may promote health, but we did not test these relationships in the current study.

Policy Implications

This study adds evidence to the accumulated body of research on the health benefits of volunteer engagement among older adults (Kim et al., 2020; Tan et al., 2009). Beyond respondent reports of better health, we uncovered a salubrious effect on systemic inflammation for persons at least 65 years old, albeit a modest one with diminishing returns over time. Although volunteer engagement should not be viewed as a remedy for health problems, we encourage leaders in medicine, public health, and social services to promote volunteering as a viable activity to enhance both social integration and physical health in later life.

Funding

This work was supported by a grant from the National Institute on Aging (AG043544 to K. F. Ferraro).

Conflict of Interest

None declared.