Background and Purpose
Ecological models of behavior change posit that individual, interpersonal, environment, & policy factors affect physical activity (PA). MIPARC (Multilevel Intervention for Physical Activity in Retirement Communities) is a group randomized controlled trial assessing the effectiveness of a 12 month multilevel intervention for improving PA in Continuing Care Retirement Community (CCRC) residents. CCRC settings provide an excellent microcosm in which to examine the effects of multi-level interventions. The intervention includes individual, interpersonal, environmental, and policy components. Previous interventions have not targeted all levels of the model. This paper describes the quantitative and qualitative methods employed to evaluate this multilevel PA intervention in older adults.
Residents at 7 intervention CCRCs were given pedometers and logs to track their daily walking. They received progress charts every 2 weeks to track their progress and individual telephone counseling to help them meet their step goals, adapt goals to overcome barriers and on-going health issues, and help participants develop goal setting skills. On the interpersonal level, participants engaged in bi-weekly 90 minute group education sessions led by a qualified health educator for 6 months. In the group sessions, participants were encouraged to share their progress, challenges and successes which provides the opportunity for social support and enhances group cohesion. Participants also attended active events led by a group of “peer leaders” who are active resident who complete training in organizing events. The peer leaders also advocated for environmental projects to improve walkabililty of their site or the surrounding neighborhood. Barriers to walking were identified following an introductory session by the pedestrian community advocacy organization Walk San Diego and action plans were developed to work toward solutions. Example projects included cleaning up a bridge that allowed participants to access a local farmer’s market more easily, redesigning a car park to prevent cars blocking a sidewalk allowing safer walking to a beach, trimming trees that were blocking a sidewalk, increasing pedestrian crossing times at an intersection as well as adding count down timers and auditory prompts. Process measures were used to assess implementation and exposure to all levels of the intervention. Surveys were developed to evaluate the mechanisms of change using validated scales from theoretical constructs, such as group cohesion, as well as scales adapted specifically for our target population. Individual level assessments include: self-efficacy, use of activity locations, satisfaction with facilities, barriers, walking location (self-report and GPS), satisfaction with each intervention component, and satisfaction with peer leaders. Interpersonal measures included: community cohesion, engagement, study group cohesion, social support, peer leader group cohesion, and session/event evaluations. The built environment was assessed through a quantitative audit tool that assesses the PA, educational and social opportunities on site (APARS) and site activity calendars. The policy and environment changes were tracked through progress reports and photo diaries.
We will present the evaluation results showing how different data sources can be summarized and patterns across sites investigated. We will demonstrate how qualitative data can be gathered together and given quantitative metrics. Data already demonstrated that attendance at group sessions has been high, 82%, and evaluations extremely positive. The high level of satisfaction however may reflect response bias, making other types of data helpful. The presentation will demonstrate how evaluation using mixed methods is important and what the challenges and solutions are. Case studies of successes and barriers to implementing the community projects and peer led activities will be highlighted. We will use configurational frequency analyses to detect variables occurring more or less frequently than by chance alone. Thresholds for each variable (high vs low) are created based on the numbers that present and the high and low group are compared. This method detects types and anti-types which is a way of indentifying clusters of communities that may share the same underlying systems.
Conclusions and Implications
A combination of quantitative and qualitative methods is necessary to assess community based interventions that use a multi level approach. Multi level interventions require evaluation tools at each level of the intervention, which can be challenging. There are more established quantitative techniques for the individual and interpersonal levels. Evaluating the impact of environmental changes remains difficult due to their diversity. In contrast to individual and interpersonal components that can be prescriptive and thus predictably assessed, environmental projects are context dependent, so measures need to be flexible and case specific. Comparing different projects and summarizing projects across multiple locations involves case study techniques and visual inspection of changes.
Future interventions will be informed by the current research results and we will present a model of an ecological intervention and its evaluation for others to use. We continue to learn the factors, such as local champions, that can result in more effective environmental projects. We also learn from the projects what features are common and may be more predictably measured in future.
Support / Funding Source
This study was funded by the National Heart Lung and Blood Institute, NIH. R01 HL098425.