Voluntary work for the physical and mental health of older volunteers: A systematic review

There are increasing numbers of older adults who no longer work. Volunteering has the double benefit of the activity for these adults and the value of the services they provide as volunteers. The evidence suggests that volunteering improves the physical and mental health of volunteers, notably a reduction in mortality. The evidence is inconclusive for other outcomes because of the small number of studies.

The increasing imbalance between the number of older adults not working and the number of adults in the age range of labour force participation is a fundamental public policy challenge in high‐income countries. Retiring may carry the risk of a fast decline in health.

Volunteering can play a significant role in people's lives as they transition from work to retirement. Volunteering may contribute to individuals ageing well and society ageing well because volunteering by older adults relieves a societal burden while helping these volunteers to maintain health and functionality.

This review examines the evidence of impact of volunteering on the physical and mental health of older people who volunteer.

Included studies had to examine the impact of formal volunteering by people aged 65 or over on their physical and mental health. Studies had to have a comparison group.

Ninety studies analysing 47 different populations were identified. Of these, only 26 studies, analysing 19 different populations, could be used in the data synthesis. The studies were from Australia, Ireland, Israel, Japan, Korea and United States. Three were randomised controlled trials (RCTs) and 21 were nonrandomised studies. The studies contained data for over 47,000 volunteers.

What is the effect of volunteering on the physical and mental health of older people?

Volunteering improves the physical and mental health of volunteers. The effect is best documented for mortality, with too few studies to draw conclusions for other outcomes.

For physical health, the evidence shows that there is a positive effect on reducing mortality (10 studies) and possibly incident functional disability (three studies). Volunteering may support the maintenance of functional competence (three studies) and slow the decline in instrumental activities of daily living (IADL; two studies).

For mental health, volunteering may reduce depression (three studies).

There was no evidence of adverse effects.

In all cases, there is little variation in the estimated effects from the different studies. However, the effects are small, and except in the case of mortality, based on a small number of studies.

Voluntary work reduces the mortality hazard of volunteers aged 65 and above. The effect corresponds to a 43% chance of the volunteers dying first which should be compared to a fifty‐fifty chance (50%) of dying first if the intervention had no effect. The evidence seems robust as there is little variation in findings between the studies.

As the intervention—unlike most other interventions in the social welfare area—is not costly, it could be prescribed to more older adults. In fact, contrary to carrying a cost, volunteering is a productive activity contributing directly to community well‐being.

As volunteering has a positive effect on the volunteers, it probably should be prescribed universally. Given the very nature of the intervention, however, it is voluntary and cannot be prescribed. Nevertheless, measures can be taken to encourage people to engage in voluntary work by making opportunities more immediately available and visible.

The review authors searched for studies published up to October 2019.

The increasing imbalance between the number of older adults not working and the number of adults in the age range of labour force participation (age range 20–64) has long been a fundamental public policy challenge in the Organization for Economic Co‐operation and Development (OECD) member countries1 (OECD, 2015). The large cohorts born after World War II (WWII), born appr. from 1946 to 1955, grow older and as a consequence, social scientists and policy makers have taken an intense interest in how their aging and eventual retirement from the full‐time labour force will affect society.

In only 15 years, the share of the population aged 65 and over in the OECD countries has increased by more than 4% points; from 13% in 2000 to more than 17% in 2018 (OECD. Stat, data extracted April 3, 2020). The effect of an aging population on a country's societal support burden is often measured by the older dependency ratio, which is the ratio of the older population to the working‐age population. The OECD average older dependency ratio (ratio of individuals aged 65 and above to those aged 15–64) has increased considerably over the last half century, from 15.2 in 1970 to 26.4 in 2018 (OECD. Stat, data extracted April 3, 2020). The problem is more pronounced in Europe than in the United States; the older dependency ratio was 24.5 in the United States in 2018 and as high as 30.7 in Europein 2018 (OECD. Stat, data extracted April 3, 2020). 2

In addition to the large post‐WWII cohorts growing older, the average duration of expected years in retirement has increased. In 1970, men in the OECD countries spent on average 11 years in retirement and by 2014, this average had increased to almost 18 years (OECD, 2014). The increase for women has been from 15 years in 1970 to 22.3 years in 2014.

The increase in average duration of years in retirement is partly due to increased longevity and partly due to earlier retirement. Although the effective age of retirement (the average effective age at which workers withdraw from the labour force) has slowly started to increase since 2004, it decreased steadily for 30 years between 1970 and 2001 (OECD, 2014). In 2018, the effective retirement age was on average 64.2 for men (63.5 for women) in the OECD countries; somewhat higher in the United States (66.0 for both men and women) than in Europe3 (64.2 for men and 63.3 for women) (OECD. Stat, data extracted April 3, 2020).

By 2050, the population aged 65 and over in the United States is expected to grow to almost 21% and the older dependency ratio is estimated to increase to 38 (OECD. Stat, data extracted August 22). In Europe, the percentage of the population aged 65 and over is expected to grow to almost 29% by 2050, and the older dependency ratio is estimated to increase to 55 (OECD. Stat, data extracted August 22). At a societal level, this growing imbalance raises serious concerns about the viability and funding of social security, pensions, and health programmes.

At an individual level, the concern is probably more that of aging well with the prospect of many years in retirement. Some research suggests that retiring for some carries the risk of a fast decline in health (Dave et al., 2008; Szinovacz & Davey, 2004). Evidence on whether, and if so, how retirement influences health and wellbeing is however inconclusive (Biggs et al., 2017). The effects of retirement have been found to depend, for example, on the type of work (e.g., Vickerstaff, 2010) and whether the retirement is forced or voluntary (Hershey & Henkens, 2013). Nevertheless, for many people, work can provide meaningful roles, social contacts and structured everyday life experience (Jahoda, 1981, 1982, 1984; Seeman, 1996; Warr, 1994). Thus, loosing these aspects in one's life when transitioning to retirement can have negative implications for wellbeing and health of individuals. Several studies have demonstrated that subjective usefulness is strongly related to both physical and psychological health (Ranzijn et al., 1998; Ryan & Frederick, 1997; Ryff, 1989). The performance of other meaningful (for the individual) activities than working for pay may thus help maintain health and functional ability for older people.

Using U.S. data from 1995 and 2005, Einolf (2009) predicted that post‐WWII cohorts' rate of volunteering at the time of their retirement will be higher than that of earlier cohorts. Given the large size of these post‐WWII cohorts, Einolf concluded that the total number of older volunteers would increase. The prediction seems to hold true, at least for those countries where it has been possible to locate relevant numbers for the baby boom generation's rate of volunteering. In Canada, the rate of volunteering for those aged 65 and over increased from 32% in 2004 to 36% in 2010 (Vézina & Crompton, 2012), and in Denmark the rate increased from 23% in 2004 to 34% in 2012 (Fridberg & Henriksen, 2014). A recent study in Denmark showed that while the rates of volunteering remained stable for younger seniors, there was a large increase over time among those aged 67–77 years (Amilon & Larsen, 2020).

Researchers have partly explained the increasing in the rates of older volunteers by increasing heath and active ageing lifestyles among older adults (e.g., Amilon & Larsen, 2020; Schippers & Conen, 2014). Partly the development has been explained by volunteer work becoming increasingly professionalized and the recruitment strategies to integrate older people in the volunteer organisations being increasingly efficient (Amilon & Larsen, 2020). Many countries in Europe have explicit strategies to increase older adults' volunteering (Ehlers, Naegele, & Reichert, 2011). In the United States, specific programmes aiming to integrate the aging population into voluntary work exist. Some programmes are organised in local nonprofit organisations, referred to as Senior Corps Programs, a network of national service programmes that provides the opportunity for people aged 55 years or above to apply their life experience to meeting community needs (see www.seniorcorps.org/rsvp/senior-corps-programs-2/↗).

Volunteering is a complex phenomenon and spans a wide variety of types of activities, organisations and sectors. The intervention of interest in this review is formal volunteering. Formal volunteering can be described as voluntary, on‐going, planned, helping behaviour that intend to increase the well‐being of strangers, offers no monetary compensation and typically occurs within an organisational context (Clary et al., 1998; Penner, 2002). We will define formal volunteering centred on four axes (as defined in Hustinx et al., 2010). These are:

(a) Free will: volunteering is a free choice; it is a voluntary action as opposed to a compulsory action. (b) Remuneration: the voluntary work offers no monetary compensation. There may be reimbursement for expenses incurred but otherwise the work is unpaid. (c) Intended beneficiaries: volunteer work can be described as “unpaid work provided to parties to whom the worker owes no contractual, familial or friendship obligations” (Tilly & Tilly, 1994, p. 291). Thus, formal volunteer work typically benefits strangers and is often referred to as nonobligatory helping (Omoto & Snyder, 1995). (d) Structure: volunteering as defined here should involve planned and ongoing activities (as opposed to a spontaneous one‐time activity). Such planned and ongoing activities often occur in some type of organisational context (Penner, 2002). An organisation defines the content of the volunteer work and formulates some expectations to the volunteer, including the tasks of the volunteer worker. The organisation produces plans, recruits the volunteers, educates them if necessary, and leads them. Thus, the relations that occur in the voluntary work are formal and different from the informal relations that are found between friends and family members to whom the volunteer may feel obliged (La Cour, 2014).

Volunteering can play a significant role in people's lives as they transition from work to retirement. According to Smith and Gay (2005), retirement is a trigger for volunteering for some older people, as it offers a “structured” means of making a meaningful contribution in society once the opportunity to do so through work has been cut off. Some older people consider voluntary work as a way to replicate aspects of paid work lost upon retirement, such as organisational structure and time discipline (Smith & Gay, 2005). The same line of arguments for volunteering can be found in several other studies (see, Chappell & Prince, 1997; Fischer et al., 1991; Greenfield & Marks, 2004; Newman et al., 1985; Widjaja, 2010). Volunteering thus seems to provide a way of compensating for the losses due to retirement as identified by Jahoda (1981, 1982, 1984), such as the need for time structure, social contact, collective effort or purpose, social identity or status, and regular activity. Several studies argue that there is a potential health benefit to older volunteers and in particular retirees (Moen & Fields, 2002; Musick & Wilson, 2003; Young & Glasgow, 1998).

The exact mechanisms and processes linking volunteering and health for older people has however not been sufficiently explored and may be very complex (Warburton, 2006). Using a qualitative approach Warburton (2006) explored this relationship and identified six potential themes and the impacts on health is discussed. The six themes and their impacts on health are illustrated in Figure 1.

In many countries, volunteering of the older adults is increasing and programmes designed specifically for this subpopulation are emerging. Volunteering may contribute to both individuals aging well and society aging well, as volunteering by the older adults at the same time relieves the societal burden if it helps maintain health and functionality for those who volunteer. It thus remains to be established to what extent volunteering impacts on the physical and mental health of those who volunteer.

Health status is often found to be an important predictor of volunteering among those aged 65 years or more, see, for example, Brown (2000) and Young and Glasgow (1998). The question that is important to answer is: does good health predict volunteering or does volunteering improve health (or maybe both)? Studies that simply assess the association between voluntary work and health outcomes cannot answer this question. Research using appropriate controls and outcome measures can, however, provide some relevant evidence on whether engaging in voluntary work might cause good health outcomes on older people. It is vital that an appropriate comparison group is used to establish the direction of cause. Does volunteering make people healthier, or are healthier people more likely to volunteer? Likewise, it is vital that the health measures are objective. As stated in Wilson and Musizk (1999, p. 153): “[C]ross‐sectional designs that use participants to self‐assess the impact of a volunteer program function as little more than market research for the agency concerned. Without a pre/posttest design and a control group, and without more objective and generalisable outcome measures, little can be learned of the benefits of volunteering from these studies”. The same worries concerning reliance on cross‐sectional designs and self‐assessment of health to establish causality can be found in Lum and Lightfood (2005). Hence, considering the fact that the population under investigation in this review by nature volunteer into the intervention, we believe it is vital that an appropriate comparison group and access to relevant pre health measures and objective health measures are used to establish causality.

We are very clear that firm causal conclusions probably cannot be drawn from the studies we included in the review, as we found only a few studies based on randomised trials. However, a distinction can be drawn between studies that simply assess the association between voluntary work and health outcomes, and studies that control for important confounding factors, in particular pre health measures, and use objective health measures. Studies that control for important confounding factors and use objective health measures provide some evidence for considering possible causal effects. While conclusions about causal effects must be very tentative, it is important to extract and summarise the best evidence available.

An obvious question arises: is there any value in conducting a systematic review when it is likely that there are no trial based studies available? We think it is worthwhile as a systematic review may uncover high quality studies that may not be found using less thorough searching methods. Furthermore, if a systematic review demonstrates that high quality studies are lacking, this could encourage a new generation of primary research. Therefore, even though we did not expect to find any trial based studies (and only found a few) and only a limited number of studies of voluntary work based on appropriate outcome measures and control group comparison, we still believe there is value in conducting a review in order to gather and highlight the best available knowledge.

The search was performed by one review author (B. C. N.) and two members of the review team (CAKI and ANDA). 4

Relevant studies were identified through searches in electronic databases, governmental and grey literature repositories, hand search in specific targeted journals, citation tracking, contact to international experts and internet search engines.

The searches on other resources and for unpublished literature was done between the September 14, 2019 and October 9, 2019. Terms used to search others resources were based on the general search strategy. Combinations of terms such as volunteering with terms for the population (i.e., old or elderly) or the outcome terms (i.e., health outcomes) were utilised. As example of these searches can be seen in the Supporting Information Appendix under the Google Scholar search strings.

Most of the resources searched for unpublished literature includes multiple types of references. As an example, the resources listed to identify reports from national bibliographical resources also include working papers and dissertations, as well as peer‐reviewed references. The same is true for Google Scholar. In the following documentation, we have split the resources for each type of unpublished literature. The resources are listed under the category of literature we expect to be most prevalent in the resource, even though multiple types of unpublished/published literature might be identified in the resource.

To account for possible statistical dependencies, we examined a number of issues: whether individuals had undergone multiple interventions, whether there were multiple treatment groups and whether several studies were based on the same data source.

Sensitivity analysis was carried out by restricting the meta‐analysis to a subset of all studies included in the original meta‐analysis and was used to evaluate whether the pooled effect sizes were robust across components of risk of bias.

For methodological quality, we performed sensitivity analysis for the incomplete outcome data, the other bias and the confounding item of the risk of bias checklists, respectively. For the remaining components of the risk of bias checklist, there were no variation in the scores given.

Reporting bias refers to both publication bias and selective reporting of outcome data and results. Here, we state how we assessed publication bias.

We used funnel plots for information about possible publication bias.

The risk of bias coding for each of the 90 studies is shown in a Supporting Information document.

Three studies reported on the same RCT but none of the studies reported the method of randomisation and was judged unclear on the sequence generation and allocation concealment items. A further three studies used a randomised design of which two reported using an appropriate method of randomisation and were rated low risk of bias on the sequence generation and allocation concealment items. The remaining RCT did not report the method of randomisation and was judged unclear on the sequence generation and allocation concealment items. One of the RCTs had a published a priori protocol and a priori analysis plan.

The remaining 84 studies used nonrandomised designs, they were all judged to have a high risk of bias on the sequence generation item and the allocation concealment item. None of the nonrandomised studies had an a priori protocol or an a priori analysis plan.

A summary of the risk of bias associated with blinding, incomplete data, selective reporting, other bias and confounding is shown in Table 3. Due to the nature of the intervention it was not possible to blind the participants and the majority of studies were given a score on 4 on the blinding item. If, however, the study reported on an objective outcome (as, e.g., mortality or long‐term care certificates in Japan) or outcome assessors were blinded, the study was given a score of 3 (20 studies) and one study applied further procedures to blind participants and members of the study team and was given a score of 2. Almost a third (28) of the studies did not report any information concerning attrition or missing data and could therefore not be rated on the incomplete data item. Three studies were given a score of 5 on the incomplete data item, corresponding to a risk of bias so high that the findings should not be considered in the data synthesis. A few studies had selective reporting issues, although two studies were given a score of 5 on the selective reporting item, corresponding to a risk of bias so high that the findings should not be considered in the data synthesis. Almost a third (29) of the studies were given a score of 5 on the other bias item, corresponding to a risk of bias so high that the findings should not be considered in the data synthesis. The confounding item was only judged for the 84 nonrandomised studies, of these 43 were given a score of 5, corresponding to a risk of bias so high that the findings should not be considered in the data synthesis and 29 studies did not either show or discuss imbalances on any of the confounders and were thus rated unclear on the confounding item.

Some of the studies were given a score of 5 on several items. The total number of studies with a risk of bias so high that the findings should not be considered in the data synthesis was 46, corresponding to a little more than half of the studies. Most of the studies (30) with a risk of bias so high that the findings should not be considered in the data synthesis were not able to identify a causal effect, as all variables used (treatment, confounders and outcome) were measured at the same time. Others (13) failed to establish a comparison group that was balanced on important confounders and also controlled for outcomes that occur posttreatment, that is, they were bad controls. The remaining three studies scored 5 on the Incomplete data item (2) and Selective reporting item (1).

Three studies analysed the effect of voluntary work on depression, and reported results that enabled the calculation of SMD and variance. The effect sizes are measured such that a positive effect size favours the volunteers, that is, when an effect size is positive the voluntary workers are better off than control groups of nonvolunteers, and when an effect size is negative the voluntary workers are worse off than control groups of nonvolunteers.

The random effects weighted SMD was 0.12 (95% CI, 0.00–0.23) and statistically significant. The Forest plot is displayed in Figure 7. There is a very small amount of heterogeneity between the studies; the estimated τ² is 0.00, Q = 2.28, df = 2 and I² is 12% as displayed in Figure 7.

In addition, a number of mental health outcomes were reported in a single study only. The outcomes were psychological well‐being, life satisfaction, self‐efficacy, quality of life, purpose in life, personal growth, self‐esteem and loneliness. The effect sizes and 95% CIs are reported in section Effect sizes.

A number of secondary outcomes were reported in a single study only. The outcomes were functional limitations, self‐rated health, decline in cognitive performance and life satisfaction. The effect sizes and 95% CIs are reported in section Effect sizes.

We carried out a sensitivity analysis in the analysis of mortality based on hazard ratios only; as the number of studies included in the other meta analyses were too low (three studies as maximum).

Sensitivity analyses were planned to evaluate whether the pooled effect sizes were robust across study design and components of methodological quality.

No RCTs were included in the meta analysis on mortality, so the impact of study design could not be evaluated. For methodological quality, we carried out sensitivity analyses for the incomplete outcome data and confounding risk of bias components of the risk of bias checklists. We examined the robustness of our conclusions when we excluded studies with risk of bias score Unclear and 3 on the incomplete outcome data item and studies with risk of bias score Unclear and 4 on the confounding item. 7

The results of the sensitivity are provided in Table 4 and displayed in Forest plots in section Funnel Plots.

There were no appreciable changes in the results following removal of any of the studies.

In summary, the conclusion of the main synthesis do not change.

This review focused on the effect of voluntary work on the physical and mental health of the volunteers.

In this review, we included total of 24 studies in the data synthesis. This number is relatively low compared to the large number of studies (90) meeting the inclusion criteria. The reduction was caused by three different factors. Forty‐six studies were judged to have a very high risk of bias (5 on the scale) and, in accordance with the protocol, we excluded these from the data synthesis on the basis that they would be more likely to mislead than inform on the size of the effect of the intervention. Eighteen of the 90 studies did not report effect estimates or provide data that would allow the calculation of an effect size. Two studies could not be used because of overlapping data samples.

If all the 90 studies had provided an effect estimate with lower risk of bias, the final list of useable studies in the data synthesis would have been largerwhich, in turn, would have provided a more robust literature on which to base conclusions. 8

In total, 47 different populations from 16 countries were represented by the 90 studies meeting the inclusion criteria. The 24 studies used in the data synthesis covered only six countries (Australia, Ireland, Israel, Japan, Korea and United States). The more narrow geographical coverage may limit the applicability of the evidence.

We analysed all available physical and mental health outcomes. At most three studies were eligible for analysis of each of these outcomes with the exception of mortality. The small number of studies reporting other outcomes than mortality makes us reluctant to draw conclusions except concerning mortality.

We found no strong indication of publication bias.

Overall the risk of bias in the majority of included studies was high. Forty six studies were judged to be at very high risk of bias. The risk of bias was examined using a tool for assessing risk of bias incorporating nonrandomised studies. We attempted to enhance the quality of the evidence in this review by excluding studies judged to be at very high risk of bias using this tool. We believe this process excluded those studies that are more likely to mislead than inform.

Furthermore, we performed a sensitivity analysis (where possible) to check whether the obtained result is robust across methodological quality. The overall conclusion did not change.

There was overall consistency in the direction of effects and there was no heterogeneity between studies.

We performed a comprehensive electronic database search, combined with grey literature searching, and hand searching of key journals. All citations were screened by two independent screeners from the review team (CAKI and ANDA), and one review author (TRF) assessed all included studies against inclusion criteria.

We believe that all the publicly available studies on the effect of voluntary work on the health of volunteers up to the censor date were identified during the review process. However, two references were not obtained in full text and 17 studies provided insufficient information to determine whether the study participants met the age eligibility criteria of the review. Despite attempts to contact the author(s) of all studies with insufficient information (we could not locate a valid e‐mail address for the authors of five of the studies) we are still awaiting answers with regard to 10 studies.

We were unable to comment on the possibility of publication bias as at most eight studies was included in the same meta‐analysis. Thus, we cannot rule out that there are still some missing studies.

We believe that there are no other potential biases in the review process as two members of the review team (CAKI and ANDA) independently coded the included studies. Any disagreements were resolved by discussion. Further, decisions about inclusion of studies were made by two members of the review team (CAKI and ANDA) and one review author (TRF). Assessment of study quality and numeric data extraction was made by one review author (TRF) and was checked by a second review author and two members of the review team (TF, ND, and AB).

We identified seven systematic reviews in the area of voluntary work and the health of older volunteers. Four of these offers only a narrative description of the studies, one an unweighted average of correlations, one reported effect sizes but did not pool them and one conducted a meta‐analysis that compare to our systematic review although it solely considered mortality.

Cattan et al. (2011) provides a comprehensive review of current knowledge (at that time, i.e., articles published between 2005 and 2011) regarding the role of volunteering in improving older (age not specified) people's quality of life. Studies that only analysed the impact on mortality or risk of disease were excluded. All study designs including descriptive and qualitative studies were eligible. Their search was systematic but they only offer a narrative description of the 22 included studies. The authors conclude that: “Although there are indications that volunteering may help to maintain and possibly improve some older adults' quality of life there are still major gaps in our knowledge regarding who actually benefits, the social and cultural context of volunteering and its role in reducing health and social inequalities “ (p. 331).

Bonsdorff von and Rantanen (2011) performed a systematic search (until November 30, 2009) of studies of volunteer work on personal well‐being: physical health, mental health and psychosocial resources. To be included, the study had to be published in a peer‐reviewed journal, written in English and conducted in western countries, participants were at least 60 years of age. Sixteen studies were included and a narrative review was provided. The conclusion offered by the authors was: “Volunteering in old age predicted better self‐rated health, functioning, physical activity and life satisfaction, as well as decreased depression and mortality. However, it did not decrease the risk of chronic diseases or nursing home admission in old age” (p. 167).

Milbourn et al. (2018) performed a systematic search of studies that assessed the relationship between time spent volunteering and quality of life in adults aged ≥50 years. They searched up to December 2017 and included only English language peer‐reviewed article published after January 2000. Eight studies were included (and one was excluded due to too low quality). The authors provided a narrative analysis of primary study results and concluded “Volunteering is identified as an important occupation for adults over the age of 50, although the relationship between time spent volunteering and the impact on quality of life outcomes is still not fully understood” (p. 613).

Anderson et al. (2014) performed a systematic search of studies that assessed the relationship between volunteering and benefits to the volunteers in adults aged ≥50 years. Searches were performed up to April 2014 and only English language peer‐reviewed articles were eligible. Seventy three studies were included of which 17 were descriptive studies without a comparison group. A narrative analysis of primary study results was provided. The conclusion is that “volunteering is associated with reduced symptoms of depression, better self‐reported health, fewer functional limitations and lower mortality” (p. 1).

Wheeler et al. (1998) searched systematic for studies of older volunteers (age not specified) doing all kinds of volunteer work (formal as well as informal) from 1965 to “the present”. Twenty nine studies were included of which 13 were descriptive studies without a comparison group or a one group before‐after study. An unweighted average of correlations (Pearson's linear correlation coefficient) between volunteering and life satisfaction was provided and the authors concluded that “engagement in meaningful volunteer activities probably positively affects nearly all older volunteers” (p. 76).

Guiney and Machado (2018) performed a systematic search of studies that assessed the relationship between volunteering (predictor) and cognitive functioning in adults aged ≥55 years. They searched up to June 2017 and included only English language peer‐reviewed article. They included 15 studies and reported effect sizes from primary studies (various measures, Cohen's d, Hedges g, unspecified betas and correlation) but did not provide a meta‐analysis. The overall conclusion was that there is evidence that “supports the idea that volunteering can protect against cognitive aging with respect to global functioning and at least some specific cognitive domain” (p. 399).

Contrary to the above mentioned six reviews, the conclusion in our review concerning mental health and physical outcomes (other than mortality) is that the evidence was inconclusive because too few studies contributed data.

Okun et al. (2013) performed a systematic search of studies that assessed the relationship between volunteering and mortality of the volunteers in adults aged ≥55 years up to November 3, 2011. They included only English language peer‐reviewed articles and book‐chapters. Fourteen studies were included. They performed a meta‐analysis using eleven studies, examining the relation between organisational volunteering and risk of mortality among adults 55 years old and older using hazard ratios. The weighted mean of these hazard ratios was 0.76, favouring the volunteers, with a 95% CI of 0.69–0.84. A moderate amount of heterogeneity between effect sizes were found (I² = 59%, τ² = 0.01). The weighted average effect size reported in Okun et al. (2013) is thus comparable to ours but where Okun et al. (2013) finds there is a moderate amount of heterogeneity between the studies we find no heterogeneity between studies. An explanation to this discrepancy may be that the effect sizes used in Okun et al. (2013) is a mix of estimated hazard ratios and standard errors as reported in the primary studies and “conversion” of estimated odds ratios and relative risk ratios (estimated using multinomial logistic models). Neither OR's nor RR's takes into account the differences in time in which each person is at risk of experiencing the event and there is no way to convert neither an OR nor a RR to a HR. The conversion formula reported in Okun et al. (2013) is not correct and may seriously bias the estimate of the HR. Another, or a supplementary, explanation may be that the outcomes from two studies performing multinomial logistic regression models are included. They are converted into a HR using first a conversion from OR into RR and then from RR into HR. However, the outcome from a multinomial logistic regression model is not an OR but a RR; and further the alternative to death (using a multinomial model) is not just staying alive but staying alive and being in good health. In one of the studies, Luoh and Herzog (2002), staying alive and being in good health is measured by self‐rated overall health and ADL limitations and in the other study, Gruenewald et al. (2007), measured by the lack of increases in self‐reported mobility disability and onset of difficulty in performing ADL). Okun et al. (2013) performs a meta regression in order to investigate if a number of factors (journal impact factor, year of publication, minimum age of sample, percentage of sample deceased and percentage of sample volunteering) can explain the observed heterogeneity but does not succeed. The reason may very well be that these factors are not the source of the observed heterogeneity but rather the inclusion of studies measuring mortality versus staying alive and being in good health.

The review aimed to examine effects on all types of physical and mental health outcomes. With the exception of mortality, there was insufficient evidence available. Nevertheless, we found evidence that voluntary work reduces the mortality hazard of the volunteers aged 65 and above and the evidence seems robust in the sense that we did not find any heterogeneity between the studies even though they represented participants from a number of different countries (Ireland, Israel, Japan and United States) and spanned time periods ranging from 1984 to 2013. We cannot comment on the variety of types of volunteer work; that is, the extent to which it was done in different organisational contexts such as religious organisations, educational organisations, health organisations political groups, sports clubs, cultural organisations, senior citizen groups or related organisations, as none of the studies reported the rates of participants engaging in work done for different kinds of organisations (one study reported that all participants engaged in educational work).

Although the intervention is not completely free; organisations need staff time dedicated to train, support and recognise volunteers which is not costless, it is probably less costly than most other interventions in the social welfare area, and could be prescribed to more older people. In fact as the intervention in contrary to only carrying a cost is a productive activity too contributing directly to community well‐being and has a positive effect on the volunteers it probably should be prescribed universally. However, due to the very nature of the intervention, it is voluntary and it cannot be prescribed, although it can be encouraged to take up. More people could probably be encouraged to take up voluntary work if the opportunity was immediately available.

Increasing the availability of opportunities through organisations that define the content of the volunteer work, produce plans, recruit the volunteers, educate them if necessary, and lead them is vital. Equally important is the visibility of the organisations through advertisement in newspapers, on the Internet and public appeals on television or radio.

In the United States, programmes such as Senior Corps Programs, are organised in local nonprofit organisations. Such programmes aims to integrate the aging population into voluntary work by providing the opportunity for older people to apply their life experience to meeting community needs. Such programmes could be more widespread also outside of the United States.

In this review we found evidence that voluntary work reduces the mortality hazard of the volunteers aged 65 and above. Concerning other physical and mental health outcomes, the evidence was inconclusive.

By excluding from the data synthesis studies judged to be at very high risk of bias this review aimed at enhancing the quality of the evidence on the effects of doing voluntary work on the volunteers. We believe this process excluded those studies that are more likely to mislead than inform on the true effect sizes. Overall the risk of bias in the studies included in the review was high. Forty six studies were judged to be at very high risk of bias, corresponding to a risk of bias so high that the findings should not be considered in the data synthesis, leaving only 24 studies to be meta analysed. In almost all of the studies excluded from the data synthesis (43) the concern of too high risk of bias was on the confounding item. Most of these studies (30) were not able to identify a causal effect, as all variables used (treatment, confounders and outcome) were measured at the same time. Others failed to establish a comparison group that was balanced on important confounders and also controlled for outcomes that occur posttreatment.

These considerations point to the need for future studies that more thoroughly make use of the existing longitudinal data in order to identify a causal effect on health outcomes other than mortality.

Further, many of the available studies (18) did not provide data that permitted the calculation of an effect size and standard error or did not provide results in a form enabling us to use it in the data synthesis. If effect sizes and standard errors of these studies had been available, valuable information about the effect of voluntary work on outcomes other than mortality may possibly have been provided.

These considerations point to the need for studies reporting detailed results that permit their inclusion in systematic reviews.