BackgroundThis study was set in 39 neighbourhoods involved in a government-funded regeneration programme called New Deal for Communities (NDC) that began in 1998. We tested whether or not different approaches to engage residents in decision-making in these areas had different social and health impacts.MethodsFirst, NDC approaches to community engagement (CE) were grouped into four types. We then assessed the impact of these types and whether or not their cost-effectiveness could be calculated. We used existing data from surveys and from NHS and government sources. New data were collected from interviews with residents of NDC areas and former staff. We have also made these data publicly available so that other researchers can assess impacts over a longer time period.ResultsThe four CE types included an empowering resident-led approach (type A), in which residents had a lot of control over decisions, and an instrumental professional-led approach (type D), in which CE was more often used to promote the priorities of public sector organisations. Type B was initially empowering but over time became instrumental and type C balanced empowerment and instrumental approaches from the beginning. There were few statistically significant differences in health and social impacts by CE type. However, when there were statistically significant differences, the results suggest that type A, and to a lesser extent, types B and C approaches may have had better outcomes than the type D approach in relation to levels of participation and trust between residents, control or influence over decisions, social cohesion and mental health. NDC areas with a type D approach were the only ones where residents’ ‘sense of control’ deteriorated over time. Residents of these areas were less likely to feel that the NDC had improved their area and to experience improvements in mental health. However, some aspects of cohesion and trust improved in type D areas. The findings of our economic analyses are mixed. It was difficult to cost engagement activities, measures of effectiveness were not robust and relating costs that could be calculated to specific measures of effectiveness was difficult. There were almost as many negative as positive scores, making the calculation of cost-effectiveness an arbitrary exercise.ConclusionsOur results are consistent with a theory that the greater the levels of control that residents have over decisions affecting their lives the more likely there are to be positive impacts. It is plausible that an empowerment approach to CE would help build trust and community cohesion, and that having a greater influence over NDC decisions could lead to more people feeling that the NDC initiative had improved an area. Conversely, our results are also consistent with a theoretical position which suggests that instrumental approaches, which try to engage residents in agendas that are not theirs, will have relatively little positive impact and that community cohesion and well-being may be undermined. The study has not produced firm evidence on the effectiveness of different approaches to CE. However, the findings do suggest that programmes involving CE will be more likely to have positive impacts if the approaches to CE are experienced as more empowering and less instrumental (i.e. less focused on the agendas of external agencies). Future methodological research is needed to develop better measures of empowerment at the collective level and more robust approaches to empowerment on health and well-being at the population level.FundingThe National Institute for Health Research Public Health Research programme.