Background
Exposure to public open spaces or parks has a positive effect on health and health related behaviors. Parks have been identified as important settings within the local neighborhood for leisure-time physical activity. Park exposure may also buffer against the adverse effects of socioeconomic disorder on health. For example, the increased risk of all-cause and circulatory disease mortality related to socio-economic disadvantage has been shown to be lower in those who live in the ‘greenest’ areas or have more parks compared to those who have less exposure to parks. There is a particular need for strategies to promote physical activity in disadvantaged neighborhoods where residents are at an increased risk of inactivity and associated poor health. Natural experiments have been identified as a top research priority in investigating casual associations between the built environment and physical activity. Due to the substantial financial costs and logistical challenges of conducting research requiring major modification of the built environment, research in this area is scarce and natural experiments involving parks have rarely been conducted. In 2009, an opportunity arose to conduct a natural experiment in Victoria, Australia through collaboration with a local government who were planning to improve a neighborhood park located in an area of socio-economic disadvantage. This intervention represented a rare opportunity to undertake a natural experiment to examine the impact of park improvement on park use and physical activity.

Objectives
This study aimed to examine whether improvements in park design led to changes in: 1) park use; 2) the active (or sedentary) nature of activities undertaken in the park; and 3) whether any observed changes from pre- to post-improvement were maintained over time. A further aim was to examine whether perceptions of the park changed following park improvement among residents living near the intervention park.

Methods
Park use was measured using direct observations with research assistants recording counts of park users and park-based activity at the intervention and control park at three time points; baseline (T1-August 2009), post-improvement (T2-March 2010) and 12 months after baseline (T3-August 2010). At each time point, observations were undertaken in the morning (7.30-9.00am), midday (11.30-1.00pm) and afternoon (3.30-5.00pm), on weekdays and weekend days. Surveys assessing park satisfaction were also completed by local residents pre- and post park improvement. Two-way ANOVAs examined the effects of park (intervention vs. control) and time point (T1 vs T2 vs T3) on the number of people walking, being vigorously active, and total number of people observed in the park. For the survey data, chi-square tests of independence were used to test for differences in the statements regarding park satisfaction at T1 and T2.

Results
In the intervention park there were significant increases post-improvement in the number of park users and the number of people observed walking and being vigorously active. In the intervention park, there was a significant interaction between park and time, (F(2,154) = 14.99, p < 0.0005) with more people observed at T2 and T3, compared to T1. The overall counts of youth (5-18 years) observed in the intervention park increased from 57 at T1 to 359 at T3. At the control park, counts of usage decreased over the same period. In regards to the number of park users walking in the intervention park, there was a significant interaction between park and time, (F(2,154) = 11.70, p < 0.0005), with more people walking at T3, compared to T1 and T2. There was also a significant interaction between park and time in regards to vigorous activity in the intervention park, (F(2,154) = 4.98, p =0.008) with more people being vigorously active at T3, compared to T1, and no differences between T1 and T2. The overall counts of youth observed being very active in the intervention park increased from 29 at T1 to 198 at T3. At the control park there were no significant differences between the three time points in the number of people walking or being vigorously active. A significantly greater proportion of respondents reported greater satisfaction with the quality of the intervention park and the facilities available, thought that the park was a good place for families to visit, and felt safe at the park, post park-improvement.

Conclusions
This study provides evidence for policy and decision-makers that park renewal has the potential to positively influence park use and park-based physical activity among youth in an area of socio-economic disadvantage. Our findings confirm the importance of parks as settings for physical activity and have practical implications which will assist urban planners and designers to develop neighborhood parks that attract users and facilitate greater levels of physical activity. Future research should explore the impact of structural modifications further in diverse neighborhoods and parks.

Support/Funding
This research was funded by an Australian National Health and Medical Research Council Strategic Award (NHMRC ID:374241).