Community empowerment and mental wellbeing: longitudinal findings from a survey of people actively involved in the big local place-based initiative in England

Abstract Background Community empowerment initiatives are receiving increased interest as ways of improving health and reducing health inequalities. Purpose Longitudinally examine associations between collective control, social-cohesion and mental wellbeing amongst participants in the Big Local community empowerment initiative across 150 disadvantaged areas of England. Methods As part of the independent Communities in Control study, we analysed nested cohort survey data on mental wellbeing (Short Warwick Edinburgh Mental Wellbeing Scale—SWEMWBS) and perceptions of collective control and social-cohesion. Data were obtained in 2016, 2018 and 2020 for 217 residents involved in the 150 Big Local areas in England. Adjusted linear mixed effect models were utilized to examine changes in SWEMWBS over the three waves. Subgroup analysis by gender and educational level was conducted. Results There was a significant 1.46 (0.14, 2.77) unit increase in mental wellbeing score at wave 2 (2018) but not in wave 3 (2020) (0.06 [−1.41, 1.53]). Across all waves, collective control was associated with a significantly higher mental wellbeing score (3.36 [1.51, 5.21]) as was social cohesion (1.09 [0.19, 2.00]). Higher educated participants (1.99 [0.14, 3.84]) and men (2.41 [0.55, 4.28]) experienced significant increases in mental wellbeing in 2018, but lower educated participants and women did not. Conclusion Collective control and social cohesion are associated with better mental wellbeing amongst residents engaged with the Big Local initiative. These health benefits were greater amongst men and participants from higher educational backgrounds. This suggests that additional care must be taken in future interventions to ensure that benefits are distributed equally.


Background
Community empowerment initiatives have gained greater attention in recent years as potentially effective approaches to improve health and reduce inequalities in disadvantaged areas. 1 These initiatives are built upon theoretical foundations 2 and empirical evidence that demonstrates how control over one's life is a significant determinant of health outcomes. [3][4][5] Indeed, the social gradient in health is, in part, produced by variations in how much control people have over their lives and where they live. 6 There is now increasing evidence demonstrating positive associations between collective control, social cohesion and improvements in health outcomes. 5 For example, a systematic review of collective empowerment group-based micro-financing interventions found they were associated with positive health and social outcomes. 7 However, much of this research is cross-sectional. 8 The research reported here adds a longitudinal dimension to this evidence base.
Furthermore, there is less empirical evidence on the mechanisms by which collective control and social cohesion result in improvements in health and social outcomes. Yet, it is precisely the identification of these pathways that will produce effective actions to redress inequalities in health. 9 Such processes and mechanisms can occur via both direct and indirect pathways. 10 Direct pathways to health improvement are hypothesized to arise from community empowerment initiatives where residents engage in neighbourhood activities that, for example, prevent toxic waste facilities being placed in their area 10 or maintain services threatened by state disinvestment. 11 Indirect routes can arise if these collective activities increase social cohesion 5 or increase a community's sense of control over where they live. 5,10 The latter is examined in the research reported here.
In our analysis of the 2016 baseline survey of participants in the Big Local (BL) community empowerment initiative, we demonstrated that people with higher levels of collective control and area-belonging had better mental wellbeing. 12 In related work, we identified that active involvement with the BL led to increased feelings of belonging to the places people live, which was associated with improved mental wellbeing. 13 Furthermore, although our findings highlighted that pathways to health improvement are complex, there were clear indications that when residents reported no improvements in collective control or social cohesion, there was a decline or no improvement in mental wellbeing. 13 This paper presents the findings from analysis of a longitudinal nested cohort survey of people actively involved in BL conducted over 4 years and three waves (2016, 2018, 2020), examining in further detail the mechanisms of collective control, social cohesion and mental wellbeing arising from the BL initiative. This research was part of a larger independent Communities in Control (CiC) study (https://communitiesi ncontrol.uk/) that uses a mixed methods approach to evaluate the health and social impacts of BL.

The BL initiative
BL is a place-based resident-led initiative funded by the National Lottery Community Fund (formerly Big Lottery Fund) and managed by a charitable organization, Local Trust (LT). The BL initiative, which runs from 2010 to 2026, provides 150 disadvantaged areas in England with at least £1 million to support residents in making their neighbourhood a better place to live. There is considerable flexibility in how residents decide to use the funds but a core element of the initiative is placing residents in control over these decisions. This is done through the development of a BL partnership in each area, which must be comprised of at least 51% resident membership but can also include other local stakeholders (such as professionals who work in the area, people who were previously residents but have moved, relatives of residents, people who live in a neighbouring area, or people with other connections to the community). Although some members of the partnership boards do not live in the BL area they either support residents in making decisions or they are directly involved with decision-making. Moreover, the funding is accompanied by the provision of support and guidance from LT, which aims to build capabilities among residents including sharing information and developing skills (http://www.localtrust.org.uk).

Data sources
This research used data from three waves (2016, 2018, 2020) of a biannual survey conducted by LT of all BL partnership members in all 150 BL areas across England. BL partnerships may have members from local stakeholders (e.g. the church, the NHS or third sector) but the majority must be BL residents. The Partnership Members survey is a repeat crosssectional survey but we were able to construct a small nested cohort of linked individual records of those who participated across all three waves (n = 217, of which > 70% were residents). LT manage this survey biannually between May and August each year via an online and/or postal questionnaire and from 2016 onwards we were allowed to include additional questions on mental wellbeing (Short Warwick Edinburgh Mental Wellbeing), collective control and social cohesion.
Obtaining and managing the data LT collated the survey responses and sent them to the research team in an anonymized SPSS spreadsheet in October 2016, 2018 and 2020. They uploaded the anonymized data to a shared Box folder only accessible to named collaborators. Individual records were linked over the three waves via unique numerical identifiers for the purpose of the nested cohort. The data were stored in electronic form on secure university servers and were accessed through password-protected networked PCs and laptops.

Variables
The survey collected data on the characteristics of BL partnership members (demographic data, socio-economic status [education]), perception of collective control and perception of the BL area, levels of participation (number of unpaid hours per week on BL activities) and self-perceived mental wellbeing using the validated measure Short Warwick Edinburgh Mental Wellbeing Scale (SWEMWBS).
The SWEMWBS (mental wellbeing) is designed to measure positive mental health states (as opposed to symptoms of mental ill-health such as anxiety and depression). Questions include the degree to which a participant 'feels useful', 'feels relaxed', 'feels with problems well', 'feels close to other people', 'feels able to make up my mind about things'. The scale has seven domains and scores range from 7 to 35 and higher scores indicate higher positive mental wellbeing. The scale has been validated for the general population. 12 At baseline, we examined any association with mental wellbeing with: (i) whether respondents felt able to influence decisions affecting their area, either collectively with others (collective control) or as individuals (individual control); (ii) perceptions of social-cohesion around involvement (feels got to know more people in the area, feels more connected, feels more positive about BL area, feels stronger sense of community) and (iii) area perception (feels people in the BL can be trusted; feels people in their BL are willing to help each other; feels belong to the area); and (iv) hours of involvement amongst participants. 12 Only those explanatory variables significantly associated with mental wellbeing at baseline were then included in the follow-up analysescollective control, social cohesion (people in area are willing to help) and hours of involvement (as time varying variables, specific for each wave). 12 We also included resident status (lives in BL area/not).

Population sample
In 2016, potential respondents were identified using a common sampling frame: all BL partnership members who provided contact details for the annual partnership review carried out by LT were approached via email (for an online questionnaire submission) or could be given one by the BL rep supporting their partnership. The aim was to reach as many partnership members as possible. This gave a total baseline potential sample of over 1600 partnership members across all 150 BL areas. About 862 participants submitted a completed questionnaire in 2016 (a baseline response rate of over 50%), 1011 in wave 2 (63%) and 1023 at wave 3 (64%). These repeat cross-sections provided the basis for a small nested cohort, whereby individual records were linked over the three waves; 217 participants (of which > 70% were residents) provided linked data over all three waves and comprise the nested cohort analysed in this paper.

Analysis
We examined whether mental wellbeing changed over time across the three waves (wave 2 compared with baseline and wave 3 compared with baseline) using the nested cohort design. Our baseline analysis has previously been published in this journal. 12 For the 2018 and 2020 data, we examined associations over time with those explanatory variables that were significant at baseline (collective control, social cohesion, hours involved, resident status). Linear mixed effect models (accounting for the clustering of participants within sites and the repeated measures per participants across the three waves) examined changes in mental wellbeing over the three waves. The different waves were treated as categorical variables. Baseline age categories, sex, highest educational qualification and ethnicity were held as constants, whereas the other variables (collective control, willingness to help each other, resident status and hours volunteered) were used as wave-specific timevarying factors. The most parsimonious model was selected using a likelihood ratio test.
We also investigated whether there were differences in effects for our primary outcome across a number of predefined sub-groups. Firstly, we investigated potential effects on health inequalities by analysing whether any mental wellbeing effects differed by education or gender. Secondly, we investigated any differences in terms of levels of participation in the BL (measured using hours involved) to see if there was a graded effect of participation. Thirdly, we also examined differences by resident versus non-resident status of BL partnership members. All analyses used baseline as a reference to compare the change over time for the cohort between (a) baseline and wave 2 [2018] and (b) baseline and wave 3 [2020]. The presented models examine the associations between mental wellbeing and the explanatory factors across all waves together (baseline, wave 2, wave 3) and only included wave-specific results when the associations varied between waves.
The cohort analysis performed utilized a likelihood-based mixed effects model, which assumes that data are missing at random. Missing data for SWEMWBS were 5% and 11% for the 2018 and 2020 cohort waves, respectively. Missing data rates were tabulated and examined. Only gender predicted missingness and so we adjusted for this in the main models. Table 1 summarizes the socio-demographic, health and explanatory characteristics of the 217 nested cohort survey respondents. There were a higher proportion of women than men, and those aged 45-64 made up the majority (50.5%). Most survey respondents were White (91.5%): only 8.5% were from Black and Minority Ethnic groups and a sizeable minority (37.8%) had one or more degrees. The mean SWEMWBS score was 24. 5 (±4.5) at baseline, 25.0 . Hours volunteered was also associated with a small positive improvement in mental wellbeing (0.08 [0.03, 0.12) across all waves. Across all waves, there was no significant association between mental wellbeing and resident status: both residents and non-residents reported similar impacts from participation in the BL.

Results
The subgroup analysis by education-which used similar models as the main analysis-found that there was no significant difference in 2018 or 2020 in mental wellbeing for those without a degree-level education (Table 3). However, for those with one or more degrees, a proxy for higher socio-economic status (

Main findings of this study
In this small uncontrolled longitudinal study, we found that there was a statistically significant increase in mental wellbeing for all respondents in 2018 but that this declined again by 2020. This suggests that being involved in the BL potentially had a small positive impact on wellbeing in the medium term (2018) but not over the longer term (2020). However, this might be explained by the fact that the 2020 survey was delivered in summer 2020, during the COVID-19 pandemic when the mental wellbeing of the whole country had declined. Respondents who perceived that people in the area were willing to help each other had a significantly higher wellbeing score as did those who agreed that collectively they could influence decisions in the area. Hours volunteered also had a small positive association with increased mental wellbeing for men. This suggests that increased involvement in the BL intervention both in terms of time and influence was associated with increases in mental wellbeing. This may have been a result of the direct (for example influencing what happens in their neighbourhood) or indirect (e.g. higher social cohesion) pathways potentially influenced by the BL. In terms of our sub-group analysis, among those with 'no degree', there was no significant change in mental wellbeing score in 2018 or 2020, but there was a significant improvement in 2018 among those who had one or more degrees. However, for both educational groups, those who felt they could collectively influence area decisions, had a positive increase in mental wellbeing. For both educational groups, there was also a small significant positive association between hours volunteered and mental wellbeing amongst men. In terms of gender, there was no significant difference in mental wellbeing for women, whereas men had a significant increase in 2018.

What is already known on this topic
Previous research has found, like ours, that collective control and social cohesion are important mechanisms through which area-based interventions impact on mental wellbeing. For example, a survey of communities undergoing regeneration in Glasgow found that residents' perceptions of their ability to influence decisions about where they lived were positively associated with mental health outcomes. 14 Similarly, a study of neighbourhood belonging found moderate associations with wellbeing stemming from greater social participation and increased feelings of belonging to the neighbourhood. 15 Our results also suggest that increased involvement in the BL intervention increased mental wellbeing. This is in keeping with other studies of volunteering which have found a relationship between the amount of time spent volunteering and health benefits. 16 We also found evidence of unequal impacts: only higher educated participants experienced an increase in mental wellbeing and men's wellbeing increased, whereas women's did not. This is in keeping with previous research which found that there may be negative or unintended consequences if initiatives succeed in building social trust within communities and engaging people in collective action, but gains are not equally distributed-with residents from higher socio-economic backgrounds having greater benefits. 17 Indeed, some studies have reported that living in areas with higher elements of social cohesion can actually be harmful for excluded residents. 18 More generally, our findings are also consistent with theory that suggests factors such as control and feeling part of a neighbourhood can help to facilitate collective action in pursuit of shared goals. 19

What this study adds
This research provides a longitudinal analysis of the role that community empowerment interventions can have on health. Our findings, showing an association between collective control, social cohesion and mental wellbeing, suggest that although participation and volunteering matters for health so too does feeling that it is possible to have influence over local decisions. This adds to the literature which suggests that increasing collective control over decisions affecting where communities live could be a potential pathway to better health. However, our results also add a note of caution that not all participants may benefit equally from involvement. 17,18 This has important implications for the future intervention design and implementation.

Limitations of this study
This study is subject to a number of important limitations. Firstly, the sample is small and restricts statistical power.
Secondly, the survey population is skewed towards the highly educated, White people and older age groups. This is unsurprising given the evidence that shows how people from more disadvantaged social backgrounds or those with poor health are less likely to participate in volunteering activities. 20 But, it does further limit generalizability. Thirdly, the study relies on self-reported outcomes which may limit the precision and reliability of our findings. However, there is evidence that shows a strong association between self-reported health and more objective outcomes such as mortality. 21 Fourthly, our nested cohort analysis assumed that data were missing at random which might not be the case (although in our examination only sex predicted missingness so we adjusted for this in the main models). Fifthly, the study had no control group and this restricts our ability to conclude that the changes observed were related to participation in BL. Our study can therefore only assess correlation between participation, collective control, etc. and mental wellbeing-it cannot establish causation. Finally, the impact of the COVID-19 pandemic is an extremely important-and unforeseeable-issue in our trend data as the bias it creates potentially affects both the outcome and all of the explanatory variables too (control, willingness, hours of volunteering). It is therefore very difficult to interpret the 2020 survey data. Indeed, other research has suggested that the pandemic and the associated restrictions negatively impacted on mental wellbeing. 22

Conclusion
This research significantly extends the empirical evidence base on the health effects of community empowerment interventions by determining-using a longitudinal design-that collective control and social cohesion are associated with better mental wellbeing amongst people actively involved in the BL in relatively disadvantaged areas. It has also found that the health benefits of involvement in BL were greater amongst men and those from higher educational backgrounds. This suggests that additional care must be taken in future interventions to ensure that benefits are equally distributed. Future design, implementation and evaluation work should explore these issues further.

Supplementary data
Supplementary data are available at the Journal of Public Health online.