Introduction

Social prescribing is a person-centred approach involving referral to non-clinical services including those within the third sector (Public Health England, 2019), which describes groups or organisations operating independently to government, where social justice is the primary goal (Salamon & Sokolowski, 2016). It is an intervention that directs patients with non-medical health needs away from healthcare and towards social means of addressing their needs (Muhl et al., 2022), such as support with the social determinants of health including finance and housing, activities around art and creativity, and exercise (Thomson et al., 2015). Social prescribing can also involve referring clients to engage in volunteering (Thomson et al., 2015; Tierney et al., 2022), defined as unpaid work or activity to benefit others outside of the family or household, in which the individual freely chooses to participate (Salamon & Sokolowski, 2016). Volunteering, also known as community service in the USA, can be regular and sustained or ad hoc and short term (episodic) (Macduff, 2005) and encompasses activity directed towards helping others (civic) (Jenkinson et al., 2013), environmental conservation (environmental) (Husk et al., 2016), and as part of education (service learning), often accompanied by structured reflection of the voluntary activity (Conway et al., 2009).

Unique to other referrals within social prescribing, volunteering may provide a twofold benefit. Volunteering provides clear economic benefits to organisations (NCVO, 2021a) and acts as a ‘bridge’ of welfare services to deprived communities (South et al., 2011). There are also distinct benefits for recipients in comparison with professional help including increased sense of participation, self-esteem and self-efficacy, and reduced loneliness, due to a more neutral and reciprocal relationship (Grönlund & Falk, 2019). As utilised by social prescribing, volunteering as an intervention in itself is supported by clear health benefits to the volunteer, particularly improved mental health and reduced mortality (Jenkinson et al., 2013). There are many primary studies which find significant positive effects of volunteering on social, physical and mental health, including mortality and health behaviours (Casiday et al., 2008; Linning & Volunteering, 2018). Furthermore, there is evidence that these benefits occur from adolescence across the lifespan (Mateiu-Vescan et al., 2021; Piliavin, 2010), although they may increase with age (Piliavin, 2010). However, due to the poor quality of this evidence, it is unclear which of the benefits, particularly concerning mental health, predict rather than result from volunteering (Stuart et al., 2020; Thoits & Hewitt, 2001).

An investigation of the benefits of volunteering can therefore inform on the utility of this practice in improving the health and well-being of clients (Tierney et al., 2022) and support a twofold benefit (Mateiu-Vescan et al., 2021). Also, establishing the benefits may help retain volunteers within organisations (Mateiu-Vescan et al., 2021), as low volunteer retention (Chen et al., 2020) has been a key debated issue (Snyder & Omoto, 2008; Studer & Schnurbein 2023), with suggested solutions including maintaining motivation through opportunities for evaluation and self-development (Snyder & Omoto, 2008), improved management of volunteers (Studer & Schnurbein 2023), and recognising their value (Studer & Schnurbein 2023). However, outcomes of volunteering such as self-efficacy (Harp et al., 2017) and sense of connection (Dunn et al., 2021) have also been shown to predict retention.

An umbrella review methodology is appropriate to provide a systematic and comprehensive overview of the vast evidence on the benefits of volunteering and to determine which are most supported, making clear and accessible recommendations for research and policy (Pollock et al., 2020). An umbrella review can also help establish what works, where, and for whom, through comparison of different settings, volunteering roles, and populations from systematic reviews with different focuses (Smith et al., 2011). Thus, it is important that an exploration of the benefits of volunteering consider potential moderators. Umbrella reviews also assess the quality of the included systematic reviews and weight findings accordingly (Smith et al., 2011), which may help to establish a causal influence of volunteering. The emerging use of an umbrella review methodology in third sector research has enabled clear recommendations for practice, exploration of moderators and mediators, identification of gaps in the research, and recommendations for future reviews (Saeri et al., 2022; Woldie et al., 2018).

Aims

The aims of this umbrella review were to;

  1. 1)

    Assess the effects of volunteering on the social, mental and physical health and well-being of volunteers, and;

  2. 2)

    Investigate the interactions between outcomes and other factors as moderators or mediators of any identified effects.

Establishing clear conclusions to these aims helped identify gaps in the literature to direct future research and provided directions to support research and implementation of interventions involving volunteers. Specific outcomes explored within this review are displayed in Fig. 1.

Fig. 1
figure 1

Outcomes identified and analysed within the current umbrella review, grouped by coding of outcome

Full size image

Methods

This umbrella review was pre-registered on the International Prospective Register of Systematic Reviews (PROSPERO) (Nichol et al., 2022) following scoping searches but prior to the formal research (registration number: CRD42022349703). Reporting of the umbrella review methodology followed the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) (Page et al., 2020). Prior to formulating the research question, the International Prospective Register of Systematic Reviews (PROSPERO), the Joanna Briggs Institute (JBI) Systematic Review Register, and the Open Science Framework Registry were checked for pre-registrations of umbrella reviews of the same or a similar topic. No such umbrella review protocols were retrieved.

Inclusion Criteria

Intervention: Volunteering

Volunteering was defined as conducting work or activity without payment, for those outside of the family or household. Participants of all ages were included. There were no limits by country or organisation or group that the volunteering was for. Although part of the definition of volunteering is that it is sustained (Salamon & Sokolowski, 2016), all durations of volunteering were included in this review to ensure a comprehensive search. Additionally, only reviews of volunteering involving some interpersonal contact with other volunteers or recipients were included. Reviews of volunteering in disaster settings such as warzones and aid for natural disasters were excluded, as these represent volunteering in extreme circumstances that is unusual and highly stressful (Thormar et al., 2010).

Systematic reviews were required to investigate the effect of volunteering on the volunteer. Reviews were excluded if volunteering was a component of a wider intervention. Reviews only assessing the effect of volunteering on the recipient were also excluded. The distinction between volunteer and recipient was sometimes less clear for reviews assessing the effect of intergenerational programmes. In this case, outcomes were only extracted for the group(s) that were performing work or activity, and no data was extracted from primary studies where neither group were.

Outcomes

The outcome of interest was health and well-being. This was categorised into general, psychological, physical, and social. Of additional interest was the interaction between these effects and with other factors such as demographics or factors associated with volunteering such as duration and type. Outcomes could be self-reported, or objective for physical outcomes (e.g. body mass index (BMI)). Reviews that did not assess effect were excluded, such as those exploring implementation, feasibility, or acceptability of volunteering as an intervention.

Types of Studies

The focus of this umbrella review was on systematic reviews of quantitative studies with or without meta-analyses to assess effect, although reviews of mostly quantitative studies were also included. The adopted definition of a systematic review was a documented systematic search of more than one academic database. Primary studies, reviews of qualitative or mostly qualitative literature, opinion pieces and commentaries were excluded.

Search Strategy

The search was conducted on the 28th July 2022 via 11 databases including EPISTEMONIKOS, Cochrane Database, and PsychARTICLES, ASSIA and the Health Research Premium collection via ProQuest (Consumer Health Database, Health & Medical Collection, Healthcare Administration Database, MEDLINE®, Nursing & Allied Health Database, Psychology Database and Public Health Database). The search was applied to title and abstract and restricted to peer-reviewed systematic reviews published in English, as all reviewers were English language speakers with no translation services available. Initial scoping searches helped to build the search strategy (Supplementary Material 1). To maximise scope, forward and backward citation searching was applied, and the results of scoping searches and further sources such as colleagues and other academics were combined into the final umbrella review.

Study Selection

Search results were exported via a RIS file and uploaded onto Rayyan for screening. Reviewer BN screened all reviews by title and abstract against the inclusion criteria, before screening the remaining (not previously excluded) articles based on full text. Details on independent screening and inter-rater reliability are available in Supplementary material 2.

Quality Appraisal

Quality was assessed using the AMSTAR 2 checklist (Shea et al., 2021), which is designed to assess the quality of quantitative systematic reviews of healthcare interventions (Shea et al., 2021) and has the highest validity in comparison to other quality assessment tools (Gianfredi et al., 2022). Also, the accompanying guidance sheet ensures consistent use across reviewers. The 16 checklist items are presented under Table 1. Further details on quality appraisal for both the included reviews and primary included studies are available in Supplementary Material 3.

Table 1 Quality of the included reviews, as rated using the AMSTAR 2
Full size table

Data extraction and Synthesis

The data extraction form was created with guidance from Cochrane (Pollock et al., 2020). To increase transparency, data extraction was completed via SRDR plus, and made publicly available (https://srdrplus.ahrq.gov//projects/3274). Further information on data extraction, including on inter-rater agreement, is available in Supplementary Material 4.

Data Analysis

The strategy of summarising rather than re-analysing the data of the reviews was adopted (Pollock et al., 2020). Vote counting by direction of effect was applied (McKenzie & Brennan, 2019), relying on the reporting of included systematic reviews. Variables were formed to allow for votes to be counted across reviews (e.g. self-esteem, self-efficacy and pride and empowerment were collapsed due to them regularly being combined by reviews). To test for significance, a two-tailed binomial test was applied with the null assumption that positive effects were of a 50% proportion (McKenzie & Brennan, 2019). Given that vote counting does not indicate magnitude of effect, results of meta-analyses are also presented. To estimate the degree of overlap of primary studies between the included reviews, the equation for calculated covered area (CCA) (Pieper et al., 2014) was applied. To prevent underestimating overlap, only primary studies addressing the effect of volunteering on the health of the volunteer were included when calculating overlap. Although vote counting also accounts for overlap, the resulting CCA was used as an additional tool for assessing the credibility of conclusions made.

Results

Search Outcomes

Initially 8325 articles were retrieved, as shown in Fig. 2. After removal of duplicates, 7118 remained for screening based on title and abstract and 62 articles remained to screen based on full texts, of which 21 reviews were included in the final review. A further 10 articles were retrieved from google scholar and citation searching, of which 7 were included, providing a total of 28 reviews. Excluded articles and the reasons for exclusion are available in Supplementary Material 5. Details on the inter-rater agreement of article screening can be found in Supplementary Material 6.

Fig. 2
figure 2

PRISMA flow diagram of retrieved articles (Page et al., 2020)

Full size image

Overlap

Authors of three included reviews were contacted to gain sufficient information to accurately calculate overlap, for example to separate studies of volunteering from those on prosociality in general (Goethem et al., (2014); Howard & Serviss, 2022; Hui et al., 2020). For one review (Goethem et al., (2014)), sufficient information to calculate true overlap was not obtained and thus it was excluded from the calculation of CCA. The excluded review was the only one that focused on adolescents; thus the exclusion is more likely to result in a conservative estimate of overlap rather than an underestimation. Despite this, CCA was 1.3%, indicating slight overlap. The overlap table used to calculate CCA is available from the corresponding author on request.

Methodological Quality of Included Primary Studies

Only 12 of the included reviews assessed primary studies for quality or risk of bias (Chen et al., 2022; Filges et al., 2020; Gualano et al., 2018; Hui et al., 2020; Hyde et al., 2014; Jenkinson et al., 2013; Lovell et al., 2015; Manjunath & Manoj, 2021; Marco-Gardoqui et al., 2020; Milbourn et al., 2018; Owen et al., 2022; Willems et al., 2020). The tools most commonly used to assess study quality were the Effective Public Health Practice Project tool (Lovell et al., 2015; Owen et al., 2022) and JBI checklists (Manjunath & Manoj, 2021; Marco-Gardoqui et al., 2020). Those that assessed risk of bias mainly utilised Cochrane tools ROB-2 for randomised controlled trials (RCTs) (Gualano et al., 2018; Jenkinson et al., 2013), and ROBINS-I for non-RCTs (Chen et al., 2022; Filges et al., 2020; Gualano et al., 2018). Only two reviews removed studies from the narrative review (Milbourn et al., 2018) or meta-analysis (Filges et al., 2020) based on quality. Reported study quality varied, but most often was reported as mainly poor quality or high risk of bias.

Methodological Quality of Included Reviews

As shown in Table 1, the quality of included reviews varied hugely. Only seven reviews scored more than 50% (Chen et al., 2022; Filges et al., 2020; Gualano et al., 2018; Jenkinson et al., 2013; Marco-Gardoqui et al., 2020; Owen et al., 2022; Willems et al., 2020). One review was found to be significantly higher quality than the rest (Filges et al., 2020). None of the included reviews reported the funding source of the included studies, and most did not report a pre-registration or protocol, or reference to excluded studies.

Characteristics of Included Reviews

The main characteristics of included reviews are displayed in Table 2. Publication of reviews spanned from 1998 (Wheeler et al., 1998) to 2022 (Chen et al., 2022; Howard & Serviss, 2022; Owen et al., 2022), with search dates up to 2020 (Chen et al., 2022; Howard & Serviss, 2022; Owen et al., 2022). Most reviews focused on older people (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Cattan et al., 2011; Chen et al., 2022; Filges et al., 2020; Gualano et al., 2018; Manjunath & Manoj, 2021; Milbourn et al., 2018; Okun et al., 2013; Onyx & Warburton, 2003; Owen et al., 2022; Wheeler et al., 1998), with inclusion criteria ranging from aged over 50 years (Anderson et al., 2014; Cattan et al., 2011; Manjunath & Manoj, 2021; Milbourn et al., 2018) to a sample with a mean age of 80 years or above (Owen et al., 2022). Only one review focused specifically on adolescents (Goethem et al., (2014)). The number of included primary studies included in the reviews ranged from 5 (Blais et al., 2017) to 152 (Kragt & Holtrop, 2019), although not all related to the benefits of volunteering. For those that reported on location of included samples, most reviews included participants mostly from the USA (Anderson et al., 2014; Blais et al., 2017; Bonsdorff & Rantanen 2011; Cattan et al., 2011; Farrell & Bryant, 2009; Filges et al., 2020; Giraudeau & Bailly, 2019; Gualano et al., 2018; Jenkinson et al., 2013; Marco-Gardoqui et al., 2020; Milbourn et al., 2018; Okun et al., 2013; Onyx & Warburton, 2003; Owen et al., 2022; Wheeler et al., 1998), followed by North America (Anderson et al., 2014; Blais et al., 2017; Hyde et al., 2014; Jenkinson et al., 2013), the UK (Farrell & Bryant, 2009; Lovell et al., 2015), and Australia (Kragt & Holtrop, 2019; Onyx & Warburton, 2003). Four reviews focused on intergenerational programmes (Blais et al., 2017; Galbraith et al., 2015; Giraudeau & Bailly, 2019; Gualano et al., 2018), two on service learning (Conway et al., 2009; Marco-Gardoqui et al., 2020), and five on specific roles including crisis line (Willems et al., 2020), environmental conservation (Chen et al., 2022; Lovell et al., 2015), care home work (Blais et al., 2017), and water sports inclusion (O’Flynn et al., 2021). One review limited the search to volunteering at a frequency less than seasonally (Hyde et al., 2014).

Table 2 Characteristics of included reviews
Full size table

Several of the included meta-analyses, whilst employing a systematic search, did not perform any form of narrative synthesis alongside the results of the meta-analyses, meaning information about the characteristics of included studies was missing.

Publication Bias

Seven of the included reviews applied a meta-analysis (Conway et al., 2009; Filges et al., 2020; Goethem et al., (2014); Howard & Serviss, 2022; Hui et al., 2020; Okun et al., 2013; Wheeler et al., 1998). Of these, five reported testing for publication bias (Filges et al., 2020; Goethem et al., (2014); Howard & Serviss, 2022; Hui et al., 2020; Okun et al., 2013; Wheeler et al., 1998). Generally, there was no strong evidence to indicate publication bias, although one review found a likelihood of publication bias specifically for the analyses of moderators on the risk of mortality (Okun et al., 2013). Also, one review reported three approaches to assess publication bias which gave mixed findings (Hui et al., 2020), and as the remaining reviews assessed publication bias in a variety of ways such as funnel plots (Filges et al., 2020), publication as a moderator (Goethem et al., (2014)), trim and fill procedure (Okun et al., 2013), and Rosenthal’s failsafe (Wheeler et al., 1998), results may not be reliable.

Findings

Results of vote counting by direction of effect from the 18 included reviews are shown in Table 3. Five meta-analysis did not provide sufficient information to be included (Conway et al., 2009; Goethem et al., (2014); Howard & Serviss, 2022; Hui et al., 2020; Wheeler et al., 1998), and one only provided sufficient information to include one variable (Cattan et al., 2011).

Table 3 Summary table of direction and strength of evidence for each outcome, and strength of potential moderators and mediators
Full size table

General Effects on Health and Well-being

Fifteen of the included reviews reported general effects on health and well-being (Table 4). Reviews reporting on composite, general measures of health mainly assessed well-being, although others measured quality of life. Generally, most reviews reported that volunteering improved well-being (Anderson et al., 2014; Cattan et al., 2011; Gualano et al., 2018; Hui et al., 2020; Jenkinson et al., 2013; Kragt & Holtrop, 2019; O’Flynn et al., 2021; Onyx & Warburton, 2003; Owen et al., 2022) and quality of life (Anderson et al., 2014; Cattan et al., 2011; Höing et al., 2016). However, the relationship with well-being was often small and with exceptions (Conway et al., 2009), and one review found most studies reported no significant impact on well-being or quality of life (Lovell et al., 2015), possibly because the review assessed environmental volunteering specifically. The review that reported on quality of life with the highest quality reported only significant positive relationships between volunteering and well-being and quality of life (Jenkinson et al., 2013), although there was evidence to suggest an impact on quality of life only when volunteers felt their contribution was appreciated (Jenkinson et al., 2013). One review found only organisational level and not individual level participation in volunteering to significantly increase well-being (Howard & Serviss, 2022), another found increased well-being for older but not younger people (Farrell & Bryant, 2009), and another found a curvilinear relationship such that a moderate intensity of volunteering was most beneficial (Bonsdorff & Rantanen 2011).

Table 4 General benefits
Full size table

Psychological Effects on Health and Well-being

Psychological effects were the most commonly reported health and well-being outcome of volunteering, reported by 23 reviews (Table 5). The reviews that reported on general mental health reported mixed findings (Farrell & Bryant, 2009; Lovell et al., 2015; Milbourn et al., 2018), likely due to the large variation in how mental health was defined and measured. Whilst some considered mental health to be a distinct factor (Farrell & Bryant, 2009; Lovell et al., 2015), others combined factors such as life satisfaction into a composite measure of mental health (Milbourn et al., 2018).

Table 5 Psychological benefits. Displayed in brackets are the number of primary included studies to support the review findings. Where no brackets are provided, findings are the result of meta-analyses
Full size table

The main effects of volunteering on psychological well-being clustered around those affecting mood and affect, and self-evaluations and concepts. For affect outcomes, reviews mostly reported a significant positive improvement in depression scores (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Cattan et al., 2011; Filges et al., 2020; Giraudeau & Bailly, 2019; Höing et al., 2016; Onyx & Warburton, 2003). Only one review reported highly mixed findings (Jenkinson et al., 2013), possibly attributable to the higher quality of included primary studies (Jenkinson et al., 2013). Reviews reporting a smaller number of contributing studies found possible moderators; two reported a reduction in depression in women but not men (Anderson et al., 2014; Cattan et al., 2011), one found a reduction in older but not younger populations (Farrell & Bryant, 2009), and another found a reduction for general volunteering but increased depression for volunteering involving high empathetic arousal (Höing et al., 2016). In support of age as a moderator, the reviews finding a consistent positive effect on depression mainly focused on older adults (Bonsdorff & Rantanen 2011; Cattan et al., 2011; Filges et al., 2020), and the review with mixed findings included adults of all ages (Jenkinson et al., 2013).

There was more consistent evidence to support other mood and affect benefits, such as life satisfaction (Anderson et al., 2014; Cattan et al., 2011; Chen et al., 2022; Farrell & Bryant, 2009; Höing et al., 2016; Jenkinson et al., 2013; Manjunath & Manoj, 2021; Onyx & Warburton, 2003; Owen et al., 2022), positive affect (Anderson et al., 2014; Chen et al., 2022; Höing et al., 2016; Kragt & Holtrop, 2019; Manjunath & Manoj, 2021; Willems et al., 2020), and motivations (Goethem et al., (2014); Marco-Gardoqui et al., 2020), although a minority of evidence found non-significant effect of volunteering on life satisfaction (Anderson et al., 2014; Höing et al., 2016; Howard & Serviss, 2022; Jenkinson et al., 2013) and positive affect (Anderson et al., 2014; Jenkinson et al., 2013). The heterogeneity of findings is most likely attributable to all volunteering types being included (Anderson et al., 2014; Cattan et al., 2011; Farrell & Bryant, 2009; Höing et al., 2016; Jenkinson et al., 2013). Additionally, single reviews found a significant reduction in anxiety (Galbraith et al., 2015) and an increase in psychological well-being (Cattan et al., 2011). Although symptoms of burnout and emotional exhaustion was cited as a significant consequence of volunteering by one review (Höing et al., 2016), this included emotionally demanding volunteering roles including working with medium to high risk sex offenders.

Some reviews grouped prominent psychological benefits into self-evaluations or self-concepts (Conway et al., 2009; Goethem et al., (2014)). The most commonly reported effects on self-concepts were an increase in self-esteem (Anderson et al., 2014; Chen et al., 2022; Farrell & Bryant, 2009; Höing et al., 2016; Marco-Gardoqui et al., 2020; Onyx & Warburton, 2003), purposefulness, meaningfulness, satisfaction or accomplishment (Chen et al., 2022; Galbraith et al., 2015; Gualano et al., 2018; Höing et al., 2016; Willems et al., 2020), pride and empowerment (Farrell & Bryant, 2009; Giraudeau & Bailly, 2019; Höing et al., 2016; Marco-Gardoqui et al., 2020), and self-efficacy (Goethem et al., (2014); Marco-Gardoqui et al., 2020). However, there was some evidence of no significant effect on self-esteem (Anderson et al., 2014; Jenkinson et al., 2013) or purposefulness (Jenkinson et al., 2013).

Physical Effects on Health and Well-being

Outcomes relating to physical effects were the least commonly investigated, reported by only 13 reviews (Table 6). The most consistent positive effect on physical health was an increase in physical activity (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Cattan et al., 2011; Chen et al., 2022; Lovell et al., 2015; Onyx & Warburton, 2003). Increased self-reported health (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Cattan et al., 2011; Chen et al., 2022; Gualano et al., 2018; O’Flynn et al., 2021; Onyx & Warburton, 2003) and functional independence (Anderson et al., 2014; Cattan et al., 2011; Filges et al., 2020; Gualano et al., 2018; Höing et al., 2016) and reduced functional disability (Bonsdorff & Rantanen 2011; Höing et al., 2016; Milbourn et al., 2018) and mortality (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Filges et al., 2020; Höing et al., 2016; Jenkinson et al., 2013; Okun et al., 2013; Onyx & Warburton, 2003) were also commonly cited benefits, although the evidence for these effects was more inconsistent (Anderson et al., 2014; Jenkinson et al., 2013). For example, there was evidence to suggest that benefits associated with self-reported health find a curvilinear relationship with intensity of volunteering, such that benefits only increase up until a moderate amount of hours spent volunteering (Anderson et al., 2014). The evidence for a decrease in mortality was the most substantial and, although reduced by the inclusion of covariates including SES, age, religious attendance, social support and health habits, remained significant (Jenkinson et al., 2013; Okun et al., 2013; Onyx & Warburton, 2003).

Table 6 Physical benefits
Full size table

Evidence for improvements in blood pressure (Chen et al., 2022; Lovell et al., 2015) and grip strength (Anderson et al., 2014; Chen et al., 2022; Lovell et al., 2015) was sparse and inconsistent. There was no evidence for volunteering as a significant predictor of number of medical conditions (Anderson et al., 2014; Bonsdorff & Rantanen 2011; Milbourn et al., 2018), BMI (Chen et al., 2022; Lovell et al., 2015), frailty (Anderson et al., 2014; Jenkinson et al., 2013), or living in a nursing home (Anderson et al., 2014; Bonsdorff & Rantanen 2011). One review concluded that whilst volunteering helped to maintain good health, it did not improve bad health (Höing et al., 2016). Only one review reported decreased smoking (Onyx & Warburton, 2003).

Social Effects on Health and Well-being

A total of 15 reviews reported social outcomes from volunteering (Table 7). When social support, sense of community and social network were combined, the evidence mostly found volunteering to improve social outcomes (Anderson et al., 2014; Cattan et al., 2011). Individually, there was evidence in support of volunteering increasing social integration (Lovell et al., 2015; Marco-Gardoqui et al., 2020), but most commonly social network (Blais et al., 2017; Farrell & Bryant, 2009; Gualano et al., 2018; Höing et al., 2016), and social connectedness or a sense of community (Chen et al., 2022; Kragt & Holtrop, 2019; O’Flynn et al., 1971; Willems et al., 2020), with only a minority of evidence indicating no significant effect of volunteering in increasing one’s social network (Anderson et al., 2014). Volunteering was found to increase social support from both other volunteers (Höing et al., 2016) and friends and neighbours (Milbourn et al., 2018). There also appeared to be some knock-on effects, as an increased number of friendships in turn increased social integration (Farrell & Bryant, 2009) and increased social connectedness increased motivations (Willems et al., 2020). Only one review reported a negative effect, namely that whilst the number of positive social ties were increased, so were the number of negative social ties (Milbourn et al., 2018). Another caveat reported was that although social ties was beneficial, less than half of volunteers reported forming connection with volunteers (Hyde et al., 2014).

Table 7 Social Benefits
Full size table

Moderators and Mediators on the Effects on Health and Well-being

Several moderators were explored around the aspects of volunteering. Evidence for the most beneficial frequency of volunteering was mixed; whilst some reviews reported a positive linear relationship between volunteering frequency and benefits (Cattan et al., 2011; Goethem et al., (2014); Höing et al., 2016), others including the best quality evidence to report on optimal frequency (Jenkinson et al., 2013) reported inconsistent findings (Anderson et al., 2014; Cattan et al., 2011; Jenkinson et al., 2013; Okun et al., 2013). Some reviews reported a curvilinear relationship between frequency and benefits (Conway et al., 2009; Höing et al., 2016; Milbourn et al., 2018; Onyx & Warburton, 2003), such that a moderate intensity of volunteering maximised the benefits, although these reviews were poor quality. The suggested optimal intensity was suggested to be around 2 h per week or 100 h per year (Anderson et al., 2014; Höing et al., 2016; Milbourn et al., 2018). There was disagreement as to whether formal volunteering is more (Cattan et al., 2011; Conway et al., 2009; Wheeler et al., 1998) or less (Cattan et al., 2011; Hui et al., 2020) beneficial than informal volunteering. This was possibly due to the outcome measure, as direct formal volunteering significantly increased life satisfaction (Wheeler et al., 1998), whilst mixed or informal helping significantly increased well-being and psychological functioning compared to formal volunteering (Hui et al., 2020). One review focusing on adolescents found no moderation of type of volunteering (Goethem et al., (2014)), but another higher quality review reported only beneficial effects of environmental volunteering on physical health in comparison to civic volunteering (Jenkinson et al., 2013). In contrast, there was consistent evidence that structured reflection was an important positive predictor of health outcomes (Conway et al., 2009; Goethem et al., (2014)). Religious volunteering was also a consistently reported moderator for positive health benefits (Bonsdorff & Rantanen 2011; Höing et al., 2016; Manjunath & Manoj, 2021; Okun et al., 2013), with one review finding a partially mediating role of volunteering on the beneficial effects of religiosity on well-being (Kragt & Holtrop, 2019).

Several factors were explored in relation to the characteristics of the volunteer. Age was the most consistently reported demographic factor as a significant moderator of the effects of volunteering on well-being. Generally, older age predicted larger effects on positive health outcomes (Anderson et al., 2014; Goethem et al., (2014); Gualano et al., 2018; Höing et al., 2016; Jenkinson et al., 2013), and there was inconsistent evidence to suggest these increased effects were related to retirement (Höing et al., 2016; Hui et al., 2020). Whilst one review reported older adults volunteering to experience greater satisfaction than older adults in employment (Kragt & Holtrop, 2019), another higher quality review found older adults both working and in employment saw the most beneficial effects on health and well-being (Milbourn et al., 2018). On the other hand, younger age predicted higher emotional exhaustion and distress in emotionally demanding volunteering roles such as crisis line, with positive coping strategies and organisational support key to reducing this (Willems et al., 2020). There was minimal evidence of gender as a moderator of volunteering and well-being (Okun et al., 2013), with mostly no effect found (Goethem et al., 2014; Hui et al., 2020). The issue of self-selection was frequently discussed. Some reviews reported that those of higher SES were more likely to volunteer, creating a sampling bias in the results (Bonsdorff & Rantanen 2011; Cattan et al., 2011). However, the effect of volunteering on mortality was reduced but still significant when adjusting for covariates such as SES (Okun et al., 2013). Also, there was some evidence to suggest that those of lower SES felt more empowered by volunteering (Cattan et al., 2011) and reported more health benefits (Cattan et al., 2011; Höing et al., 2016). However, higher education was found to decrease stress when volunteering for crisis line (Willems et al., 2020).

Motivations for volunteering was found to be a significant moderator, such that those with altruistic or intrinsic motivations for volunteering saw increased benefits than those motivated for other reasons (Anderson et al., 2014; Höing et al., 2016; Okun et al., 2013). In support, one review found prosociality to be a far stronger predictor of health and well-being than volunteering alone (Hui et al., 2020). Feeling appreciated was found to be necessary to see improvements in quality of life (Jenkinson et al., 2013) or moderated the effects (Anderson et al., 2014). A moderating effect of feeling appreciated on health outcomes was also reported for depression, life satisfaction, and general well-being (Anderson et al., 2014). Although empathising with the recipient was important for spiritual development, it also increased the likelihood of burnout in emotionally demanding volunteering roles (Willems et al., 2020).

Some interactions were explored between the effects. The most frequently discussed was social factors including social connection, support, and interaction, which often moderated the relationship between volunteering and other health outcomes (Höing et al., 2016; Milbourn et al., 2018; Okun et al., 2013; Onyx & Warburton, 2003), with one review finding them to be a complete mediator of volunteering and life satisfaction (Anderson et al., 2014). For emotionally demanding volunteering such as crisis line, social support helped to increase well-being and buffer any negative effects (Wheeler et al., 1998). In keeping with this, one review hypothesised that volunteering generates social capital for both the recipient and the volunteer, with subsequent benefits on health and well-being (Onyx & Warburton, 2003).

Findings from Meta-Analyses

Results from reported meta-analyses (Table 8) varied on measures used to calculate both pooled estimates and heterogeneity, meaning comparison between reviews was difficult. There was also a lack of reporting heterogeneity at all, reflecting the general poor quality of included reviews. There were no available meta-analyses for social outcomes, aside from an aggregate measure of personal and social competence. Although many were significant, the pooled estimates for most outcomes were small, aside from mortality (Filges et al., 2020; Okun et al., 2013), and measures of physical functionality such as maintenance of functional competence (Filges et al., 2020). Mortality (Filges et al., 2020; Okun et al., 2013) and well-being (Conway et al., 2009; Howard & Serviss, 2022; Hui et al., 2020) were the only two outcomes reported by meta-analyses of more than one review. For both outcomes, pooled estimates were similar across reviews.

Table 8 Table of meta-analyses
Full size table

Discussion

The current umbrella review identified 28 eligible reviews, mostly focusing on older adults, based in the USA, and including a range of forms volunteering. An overview of the strength of the evidence for each variable is shown in Fig. 3. Reduced mortality and improved physical functioning showed the largest effect sizes with consistent supporting evidence. There was also consistent evidence to support effects on general health and well-being and quality of life, psychological well-being, pride and empowerment, motivation, self-efficacy, life satisfaction, positive affect, reduced depression, and purposefulness related to psychological constructs, improved self-reported health and physical activity relating to physical benefits, and improved social support, sense of connectedness and community, and network. The evidence suggests no effect of volunteering on medical conditions, BMI, frailty, or living in a nursing home. More research is required to establish whether there are effects of volunteering on blood pressure and grip strength. Organisational-level participation, older age, reflection, religious volunteering, altruistic motivations, and feeling appreciated all amplify the relationship between volunteering and health and well-being. Additionally, social factors have a knock-on effect for other health and well-being outcomes, with protective effects for any potential negative outcomes. There was no evidence of moderation of gender. More research is needed to explore the optimal intensity of volunteering, the role of SES, whether formal or informal volunteering is most beneficial, and whether the moderation of age is related to retirement, as current evidence is inconsistent.

Fig. 3
figure 3

Summary of strength of evidence for each variable outlined in Fig. 1. Labelled according to vote counting results; ‘very strong’, ‘strong’, ‘moderate’, ‘weak’, and ‘very weak’

Full size image

Age was the most supported moderator, namely that those of older age received greater health benefits from volunteering. One reason is that volunteering compensates for the loss of the health and well-being benefits of career success (Spurk et al., 2019), easing the adjustment to retirement. In support of this, work related satisfaction and perceived rewards significantly predicted life satisfaction in retired volunteers, even when controlling for demographic factors and self-efficacy (Wu et al., 2005). However, the current umbrella review found inconsistent evidence to support retirement as the explanation. Instead, the findings indicate that although many of the benefits associated with volunteering do relate to a sense of purpose, the benefits of volunteering are also distinct from usual work activity, through feelings of altruism and self-actualisation. This perhaps explains the complex relationship with age. Age has been established as a positive predictor of altruistic motivations (Sparrow et al., 2021), which was found to predict better health outcomes of volunteering. More research is needed to explore the role of retirement and alternate explanations in the relationship between age and the benefits of volunteering, including the interaction of age with other moderators.

On the contrary, there was no evidence to support gender as a moderator for the relationship between volunteering and health and well-being. Although women are more likely to volunteer than men (NCVO, 2021b), the results of this review indicated that once volunteering, there is no effect of gender on the subsequent health benefits. This provides a case for future volunteering initiatives to be targeted towards men, and for more research to explore the barriers to volunteering for men specifically, such as through qualitative methodology (Males, 2015).

The findings of this review suggest a complex relationship between SES and volunteering and its benefits. There is vast research to support the finding that those of higher SES are twice as likely to volunteer than those of the lowest SES (NCVO, 2021b). However, the current review also indicated that those of lower SES may benefit more from volunteering. If so, the use of volunteering must be maximised to help reduce health inequalities. It is key to note that those of lower SES are more likely to engage in informal volunteering, which is often overlooked by the volunteering literature (Dean, 2022). Thus, it is important that future research further explore the influence of the formality of volunteering on the health benefits, as the current umbrella review found inconsistent results. Dependent on this, particularly during retirement, the findings of this review indicate that public health campaigns to enable volunteering should be particularly focused on those of lower SES.

More research is needed to determine the relationship between frequency of volunteering and health and well-being, as the current review found it was not related to the age of volunteers or type of volunteering. The rationale behind a curvilinear relationship is that time spent volunteering positively predicts burnout (Moreno-Jiménez & Villodres, 2010). However, the only evidence linking volunteering to burnout in the current umbrella review related to volunteering that was emotionally demanding (Höing et al., 2016; Willems et al., 2020) rather than frequency, as suggested by Linning and Jackson (Linning & Volunteering, 2018). Indeed, emotional exhaustion is one of three subscales within the concept of burnout, which is explained as a result of prolonged and intense emotional involvement (Maslach & Jackson, 1981). The current umbrella review found that sufficient support from the organisation helped mitigate the effects of emotionally demanding volunteer roles on burnout and increased well-being (Höing et al., 2016; Kragt & Holtrop, 2019; Willems et al., 2020). Systematic reviews of healthcare providers have found a negative prediction of positive social support to burnout, leading the authors to recommend that interventions to reduce burnout should focus on social support (Guilaran et al., 2018; Velando-Soriano et al., 2020). Thus, it is at upmost importance that organisations recruiting for emotionally demanding volunteer roles must ensure a sufficient and positive support network to avoid negative health and well-being outcomes such as burnout. For example, sufficient support from supervisors and a stable and supportive organisational environment are essential.

A particularly useful finding of this review is that positive social outcomes of volunteering in turn encourage other positive health and well-being outcomes. Indeed, social capital has been established to reduce mortality and improve physical and mental health (Ehsan et al., 2019). Interestingly, the current review also found that volunteering predicted self-reported health, functioning, mortality, and mental health outcomes much better than for other objective indicators of health such as living with medical conditions, BMI, and frailty. This highlights the need for a holistic view of health to assess mortality risk rather than only focusing on physical indicators. For example, lack of flourishing mental health was shown to significantly predict mortality in a 10-year longitudinal analysis, even when controlling for a number of factors including physical disease (Keyes & Simoes, 2012). Another longitudinal study found that although the prediction of life satisfaction on mortality was partially shared with physical health and social orientation, it also exerted an independent effect on mortality (Hülür et al., 2017). Thus, it is essential to also focus on the mental and social outcomes of volunteering to capture all the potential benefits.

There was consistent evidence to suggest religious volunteering to be a moderator of the effects of volunteering on health and well-being. Whilst one suggested explanation for the moderating effect on well-being is that religiosity is an indication of benevolent and altruistic motives (Krause et al., 2017), the social science literature suggests that volunteering offers a chance to enact a group identity (Caricati et al., 2020; Gray & Stevenson, 2020), in this case a religious group (Wakefield et al., 2022). Indeed, for volunteers high in religiosity, identification with the religious organisation they were volunteering for predicted their sense of being enable to enact their religious group three months later, which in turn predicted mental health improvements (Wakefield et al., 2022). Subsequently, the relationship between religion, volunteering and well-being is not only explained through altruistic motives, but also because volunteering provides those high in religiosity a space to enact their religious norms, strengthening their group identity and consequently their well-being (Wakefield et al., 2022). However, more research is needed to determine whether this also applies when volunteering for secular organisations.

Strengths and Limitations

The current umbrella review provides a comprehensive overview of the literature on the benefits of all types of volunteering (Gianfredi et al., 2022). Furthermore, the very low overlap of primary studies provides credibility to the conclusions drawn. However, there are a number of limitations to consider. The relatively high proportion of articles retrieved from other sources, despite scoping searches being conducted prior to the search, indicates that the databases searched were not comprehensive. Forward and backward citation searching aimed to address this limitation. Secondly, the included reviews were mainly low quality, and for those reviews that assessed quality, the quality of primary studies was mixed. However, as higher quality reviews tended to use a more stringent measure of risk of bias (Chen et al., 2022; Filges et al., 2020; Gualano et al., 2018; Jenkinson et al., 2013), it is important that the quality of the review was also considered when weighting findings. Whilst the very low percentage of overlap between primary was a strength, it also may indicate that the included reviews were not thorough, reflected in the general poor quality ratings. Also, the vote counting method applied could not account for the curvilinear relationships identified, highlighting the importance of describing these within the text. More significantly, although efforts were made to conduct vote counting via direction of effect rather than significance, this was not always possible to attain due to insufficient reporting of reviews.

Another limitation is that although three reviews were published in 2022, none of the searches went beyond 2020, meaning no research conducted during or after the COVID-19 pandemic was included. There is evidence that the COVID-19 pandemic created lasting changes to volunteering, mainly that it encouraged digital volunteering which has sustained even after restrictions were lifted (Kanemura et al., 2022). This digitalisation has attracted a new group of volunteers who may experience volunteering differently (Kanemura et al., 2022). More importantly, digitalisation has impacted on the opportunity for social connection (Kanemura et al., 2022), which, as established by this review, has a knock-on effect on the mental and physical benefits of volunteering. A systematic review of research conducted after 2020 would be useful to compare to the findings of the current umbrella review to explore these differences further.

Conclusion

This review has established a multitude of benefits of volunteering on mental, physical, and social health and well-being, particularly reduced mortality, and increased functioning, quality of life, pride, empowerment, motivation, social support, and sense of community. To ensure the generalisability of these findings, more research is needed outside of the USA, and specifically focusing on adolescents. More quantitative research to aid meta-analyses on the social benefits of volunteering would be beneficial to quantify the effects and aid comparison with the mental and physical benefits. However, any future systematic review and meta-analysis on the topic should ensure to follow quality criteria from the AMSTAR-2 (Shea et al., 2021), specifically ensuring to pre-register methods and hypotheses, cite excluded studies, report their funding source, and account for their risk of bias. Concerning interacting factors, more research is needed to explore the likely complex relationship of volunteering with both SES and religiosity, and the optimum ‘dose’ of volunteering to gain the established benefits. Volunteering should be considered as an intervention in itself, particularly within the context of social prescribing, where referral to engage in volunteering should be encouraged. Where volunteering roles are emotionally demanding, an appropriate support system should be ensured by the organisation to prevent negative health outcomes such as burnout.