Factors influencing trust and mistrust in health promotion partnerships

Trust has been described from a number of different perspectives, including interpersonal, inter-organizational, and societal (12). It has a number of different dimensions, including trusting (openness and sharing), trustworthiness (support and acceptance), vulnerability, and expectations (reciprocity) (13,14). The literature disagrees as to whether trust and mistrust lie at opposite ends of a continuous variable or whether they are distinct, but linked, dimensions. Studies by Jones and Barry and Benamati et al. found that they are distinct constructs and that low trust is not the same as mistrust (11,14). Whether distinct or not, mistrust is a primary barrier to collaboration (9). Health promotion partnerships often have a history of mistrust, which can exist between public sector organizations such as health and education, between statutory and voluntary organizations, and/or the private and public sectors (8,15). Trust is a prerequisite for effective partnerships (16).

Despite its importance, trust has seldom been measured. A recent literature review of health promotion partnerships found only one study that measured trust specifically, although trust emerged as a component of effective partnerships in a number of qualitative studies (5). A systematic literature review found that the measurement of trust is very underdeveloped (17). Costa et al. (18), in a study of 112 health and social care teams, showed that trust is positively related to performance. Bahraminejad et al. found that a lack of communication led to ‘inadequate or even lack of trust between partners’ and this lack of trust (15), especially at the start-up stage, made partnership ‘impossible’. Their study showed that communication was particularly difficult between academics and community members. Armistead et al. found that (19), although trust is a key element of multi-sector partnerships, it is an intangible phenomenon experienced more when absent than when present.

Many factors influence partnership functioning. A scoping review by Corbin et al. identified nine core elements that support and inhibit partnership functioning (5), including trust, leadership, and having a wide range of partners. Previous research shows that other factors include how the partnership is administered and managed, efficiency, finance, boundary spanners, community involvement, and shared power (3). Administration and management of a partnership involves inter alia communicating effectively, managing grants and funds, orientating new partners, and evaluating the impact of the partnership on health (4). Jones and Weiss et al. found that efficiency – how resources and time are used by the partnership – was a significant predictor of synergy (20,4), an indicator of successful partnerships. Baron-Epel et al. showed that effective management of a partnership is a key component of its success (8).

Boundary spanners have a set of partnership skills that enable partnerships to function more effectively, such as negotiating skills and being able to see new opportunities (3). Challis et al. and Alter and Hage have identified the need for boundary spanners who can connect partners up with common goals (21,22). Boundary spanners bring a range of skills to partnership functioning, serving as ‘collabronauts’ (23), and establishing a climate of trust, optimism, and perseverance (16). Mays et al. found that boundary spanners are necessary for partnership effectiveness (24).

A key element of health promotion partnerships is that community members are actively involved (25), and Green et al. argue that if they are not (26), the partnerships are not health promoting. Robertson and Minkler define community involvement as communities working with professionals to define and solve health problems (27). Lasker and Weiss stress the critical role of community stakeholders and of sufficient heterogeneity of partners to supply the range of perspectives required (28). Winer and Ray argue that successful partnerships ‘need to involve minority, grassroots, and end-user groups’ (29). Weiner and Alexander and Minkler et al. found limited evidence of community involvement in community health partnerships (30,31). Zahner showed that having a broad array of partners contributed to effectiveness (32).

Collaborative leadership can fix ‘public problems in a shared-power world’ (33). Collaborative leaders understand social and political contexts, can communicate and share a vision, and implement policy decisions. Integrative leadership is required in situations where there is no one person in charge and power is distributed across a number of organizations, as happens in health promotion partnerships. Leadership has been measured more often than any other partnership functioning factor (3), and has been found to have a positive effect on levels of partner participation (34). Lempa et al. found that leadership contributed to more than five times the variance of other factors and concluded that the ‘importance of leadership is the greatest implication for practice that emerges from the study’ (7).

Power is always present in partnerships and it can have a positive (shared power) or a negative (power abuse) effect (29). Power is an important factor in terms of facilitating cooperation and is the functional equivalent of trust (35). The 68^th World Health Assembly recognizes the importance of shared power and recommends supportive structures to deal with power dynamics in partnerships (36). Hemphill et al. argue that research studies have largely ignored the influence of power (37), although anecdotal evidence suggests that shared power is far from the norm. Pratt et al. note that ‘local power struggles … become a painful distraction that can last for years’ (38). A systematic literature review by Gillies found that durable structures which support power-sharing between partners – particularly between professionals and lay partners – are vital to successful partnerships (39). Gillies argues that participation by the lay public must not be ‘mere tokenism’ and ‘must connote a sharing of power and control between the public and key protagonists’. Several analyses of Healthy City projects ‘made visible the difficulties they had in shifting the balance of power and control in cities’ (39).

This is a new analysis of data collected for a cross-sectional study of partnership functioning. The original analysis explored the relationship between synergy and partnership functioning factors in health promotion partnerships. In the previous study, Jones and Barry found that leadership and efficiency were significant predictors of partnership synergy (3). Administration and management, community assets, and boundary-spanning skills also approached significance. Power was included in the new analysis as it featured in the literature as an important functioning factor although it has seldom been measured. This new analysis explores the relationship between trust and mistrust (the dependent variables) and six other partnership functioning factors (the predictor variables) in the original sample of health promotion partnerships.

Eligible partnerships were identified from a database of health promotion partnerships developed for a previous study (3). Partnerships had to have been in existence for a year or more, have a minimum of five partners and a health promotion purpose. Initial contact was made by telephone with the partnerships’ chairs/leads. Of the 73 partnerships deemed to be eligible 42 partnerships agreed to participate. Two partnerships participated in the pilot study leaving 40 for the main study. The partnerships had members from a range of sectors including the community, voluntary organizations, health professionals, and the education sector, and represented all four provinces of the Republic of Ireland. Partners were involved in a wide variety of projects and programs including: increasing involvement in physical activity, improving the health of children and families, and alleviating the effects of poverty. The original data were collected between January and March 2007 via questionnaires to partnerships’ chairs/leads and partners (total N for both groups = 469). Two reminders were sent after two and four weeks, respectively. The methodology is described in full in previously published papers (3,11). The present study is a new analysis of the original data set.

The questionnaire incorporated specifically designed and validated multi-dimensional scales designed to assess the relationship of six functioning factors to partnership trust and mistrust (3). Most questions were in the form of five-point Likert scales, where ‘always’ scored 5 and ‘never’ scored 1. A ‘don’t know’ option was included (40). New scales were developed and validated by Jones and Barry for trust and mistrust (11). The nine-item trust scale measured low to high trust and the five-item mistrust scale measured low to high mistrust. Sample trust items included ‘partners keep the promises they make to the partnership’ and ‘partners eagerly volunteer to take on tasks associated with the partnership’. Sample items for mistrust included ‘decisions are made by a minority of partners outside of the main partnership meetings’ and ‘partners’ time and energy is wasted due to mistrust’. Predictor measures included an ‘administration and management’ scale (8 items), an ‘efficiency’ scale (3 items), and a ‘leadership’ scale (11 items), all developed by Weiss et al. (4). New scales were developed for ‘boundary-spanning skills’ (14 items), ‘community assets’ (5 items), and ‘shared power’ (9 items), based on a literature review.

SPSS version 20 was used to carry out all statistical analyses. Variability within partnerships was significantly less (p < 0.0005) than variability between partnerships for all scales, allowing individual responses to be aggregated at the partnership level (4). Reliability and validity tests of all measures were carried out on individual partner-level data. Pearson’s correlations and regression analysis were carried out on aggregated partnership-level data. Partner- and partnership-level scores for all scales were obtained by calculating the scale scores for each respondent and then calculating an aggregate score for respondents within each partnership. A detailed description of the statistical analyses was given in a previous paper (3), including Pearson’s correlation coefficients for the predictors and dependent variables.

The study has practical implications for health promotion partnership working. HiAP calls on governments to strengthen partnership capacity across all sectors (2). Health promotion involves influencing the determinants of health including education, poverty, social inclusion, inequalities, employment, and the overall environment. Indeed it can be argued that population health cannot be promoted without partnerships because so many sectors create health. Climate change, for example, will not be tackled without a global partnership between nations. On a smaller scale, a community might want to develop a safer environment for children and will require a partnership between themselves, law enforcement officials, health professionals, and local authorities. Partnerships often begin enthusiastically and subsequently collapse when trust, mistrust, and shared power are not taken seriously. Jones and Barry found that a majority of partners in health promotion partnerships think trust is a ‘taken for granted’ phenomenon (3). Those involved in partnerships may need training on the concepts of trust and power to ensure that partners understand their importance, know how to build and maintain trust, how to share power, and how to deal with mistrust when it arises (1). Weiner et al. recommend procedural fairness as a way of building trusting relationships (45). Other methods that can be used to build trust include having a common purpose or mission all are agreed on, clear roles, and valuing the contributions of all partners equally (5). Partnerships characterized by mistrust and power abuses will not succeed and will waste resources that could be spent on effective health promotion interventions. Boundary-spanning skills did not emerge as a significant predictor for either trust or mistrust in the regression analysis. However, they were highly correlated with community assets (0.85). It can be inferred that boundary spanners are needed to fully and fairly integrate the public’s perspective into the partnership’s mission. Boundary spanners are particularly important in health promotion partnerships because of the well-established vertical hierarchies of professional groups (3). Further research is needed into the role of boundary spanners.

The study had a number of limitations. The study sample was a convenience and not a random sample. Although every effort was made to include partnerships that were performing less well, it was inevitable that those performing adequately to very well were more likely to participate. The resulting restricted variance of the dependent variable made statistical analysis more conservative. Loss of variability is an inevitable consequence of using aggregated scores, leading to inflated correlations between variables (42), and consequent multicollinearity which was a problem in this study and the previous study. The only way to avoid multicollinearity is to recruit a large enough sample, not always an easy task.