RAND Corporation Santa Monica, 1776 Main St, Santa Monica, CA, 90401, USA.
RAND Corporation Pittsburgh, 4570 Fifth Ave #600, Pittsburgh, PA, 15213, USA.
BMC Public Health. 2020 Oct 7;20(1):1519. doi: 10.1186/s12889-020-09424-8.
Improving the neighborhood environment may help address chronic disease and mortality. To identify neighborhood features that are predictors of health, objective assessments of the environment are used. Multiple studies have reported on cross-sectional assessments of health-related neighborhood features using direct observation. As study designs expand to better understand causation and predictors of change, there is a need to test whether direct observation methods are adequate for longitudinal assessment. To our knowledge, this is the first study to report on the reliability of repeated measurements of the neighborhood environment, and their stability, over time.
The Pittsburgh Hill/Homewood Research on Neighborhood Change and Health (PHRESH) study conducted longitudinal assessments in two low-income, African American neighborhoods at three waves (years 2012, 2015, 2017). The PHRESH audit tool is a modification of earlier validated tools, with an emphasis on environment features relevant for physical activity, sleep, and obesogenic behaviors. Trained data-collector pairs conducted direct observations of a 25% sample of street segments in each neighborhood. At each wave, we audited a sub-sample of street segments twice and assessed reliability using percentage inter-observer agreement and krippendorf's alpha statistics. Stability of these items was assessed as exhibiting moderate or high agreement at every time point.
Across waves, a majority (81%) of the items consistently demonstrated moderate to high agreement except for items such as public/communal space, amount of shade, sidewalk features, number of traffic lanes, garden/flower bed/planter, art/statue/monument, amount of trash, and physical disorder. The list of items with poor agreement includes features that are easy to miss (e.g. flower bed/planter), hard to assess from outside (e.g. public/communal space), or may change quickly (e.g. amount of trash).
In this paper, we have described implementation methods, reliability results and lessons learned to inform future studies of change. We found the use of consistent methods allowed us to conduct reliable, replicable longitudinal assessments of the environment. Items that did not exhibit stability are less useful for detecting real change over time. Overall, the PHRESH direct observation tool is an effective and practical instrument to detect change in the neighborhood environment.
改善邻里环境有助于解决慢性病和死亡率问题。为了确定与健康相关的邻里特征预测因素，人们使用客观的环境评估方法。多项研究已经报告了使用直接观察对与健康相关的邻里特征进行横断面评估。随着研究设计的扩展，以更好地了解变化的因果关系和预测因素，有必要测试直接观察方法是否适合纵向评估。据我们所知，这是第一项报告邻里环境重复测量的可靠性及其随时间推移的稳定性的研究。
匹兹堡山/霍姆伍德邻里变化与健康研究（PHRESH）在两个低收入的非裔美国人社区进行了三次纵向评估（2012 年、2015 年和 2017 年）。PHRESH 审计工具是对早期经过验证的工具的修改，重点是与身体活动、睡眠和肥胖行为相关的环境特征。经过培训的数据收集员对每个社区的 25%街道段进行了直接观察。在每一波中，我们对街道段的一个子样本进行了两次审计，并使用百分比观察者间一致性和克里普多夫α统计数据评估可靠性。评估这些项目的稳定性，看它们在每个时间点是否表现出中等或高度一致性。
跨波，除了公共/社区空间、遮荫量、人行道特征、交通车道数量、花园/花坛/种植床、艺术/雕像/纪念碑、垃圾量和物理障碍等项目外，大多数（81%）项目始终表现出中等至高度一致性。一致性差的项目包括容易遗漏的特征（如花坛/种植床）、难以从外部评估的特征（如公共/社区空间）或可能快速变化的特征（如垃圾量）。
在本文中，我们描述了实施方法、可靠性结果和经验教训，以为未来的变化研究提供信息。我们发现，使用一致的方法使我们能够对环境进行可靠、可复制的纵向评估。那些没有表现出稳定性的项目对于检测随时间的真实变化用处不大。总体而言，PHRESH 直接观察工具是一种有效且实用的工具，可以检测邻里环境的变化。
