Commentary

The article “Mapping Activity Patterns to Quantify Risk of Violent Assault in Urban Environments,” by Wiebe et al., presents a unique and innovative study that pushes research on the contextual drivers of violence a good step forward. The study attempts to trace back the daily routines of adolescent and young adult males through space and time, and relate how daily encounters with different types of environmental factors are associated with the odds of experiencing assault. In doing so, it reveals that victims of assault spend time in different types of places and use different modes of transportation than nonvictims, and that the places where assaults occur are markedly different from the places inhabited by nonvictims. Furthermore, and quite interestingly, the study incorporates an element of time, and reveals that assaults occur shortly after subjects encounter a change in their environment, suggesting there is an induction period from environmental exposure to violence. The focus on the role of time and space in shaping the occurrence of violence is an important addition to our understanding of the dynamics of urban violence. It moves us beyond a focus on static residential neighborhoods, to identify how the different contexts youth encounter throughout their day matter for violence.1 This study fits within a broader effort in the study of “health and place” to document the multiple contexts that persons inhabit and the ways they move across contexts throughout the day, and to gain insight into how activities in different contexts over time can contribute to health-related behaviors.2 Such efforts are motivated by a belief that residential contexts represent only one of multiple contexts that people inhabit throughout the day. Different contexts present unique risks to health, because of the physical and social characteristics of such places, the situational characteristics that arise in different contexts at different points in time, and the collective features of groups of people that emerge within different contexts.3 Several methods have arisen to account for the contribution of such heterogeneity across time and space, including activity space analysis, which examines movement through different contexts and its impact on health, and ecological momentary assessment, which captures microlevel changes in context as people move through their days. The Wiebe et al. study constitutes an example of an activity space analysis, which relies on retrospective reports of activities and movement throughout the day. The study design offers a number of key strengths that improve our confidence in the findings. First, the study uses two types of comparisons to answer two different questions. By comparing cases to controls at the time of the assault, the study reveals average differences in the types of places encountered by young adults at higher risk for assault. At the same time, by conducting a case-cohort difference-in-difference analysis, the study is able to reveal potential short-term triggers for assault, while accounting for fixed sources of confounding, as well as for secular differences between case and control subjects. This is a nice approach to examining the relative contribution of acute versus chronic exposures, and seems particularly appropriate for an outcome with a short induction period, as is the case with violent assaults. Second, by stratifying by time of day and adjusting for day of the week and weather conditions, the study is able to account for key confounders of the association between environmental exposures and assault. Third, the authors are to be commended for the application of a highly detailed measure that captures all contexts encountered by case and control subjects in a single day, thus providing valuable information about the role that mobility across contexts could play in shaping the risk for assault. While the study is novel and significant, several concerns also arise regarding the control group selection, definition of the induction period, the measures used to evaluate environmental exposures, and the analytic procedures used to carry out the described analyses. First, as with all case–control studies, concerns arise about selection bias. In this case, controls were selected in the neighborhoods served by the hospitals where the cases were selected. Such an approach seems like a reasonable way to ensure that the controls represent the source population that gave rise to the case subjects. The exclusion criteria, however, raise a potential issue: control subjects were required to have a landline to participate in the study, while case subjects were not. Yet young adults often rely on cell phones rather than landlines. If such a difference in exclusion criteria between the case subjects and the control subjects relates to the selection of different contexts, this could raise concerns about selection bias. Using a mixed cell phone and landline phone sample to recruit control subjects using random digit dialing would have addressed this concern. Second, the induction period is defined as the 10 minutes before the assault. However, the reasons for selection of that particular time period are unclear. It would have been helpful to have information about a range of periods before the event, to empirically identify the optimal induction period when activities and environmental features may function as triggers for an assault. Third, by the nature of the case–control design, measures of environmental exposures and activities are retrospective. Concerns arise about differential recall bias: case subjects were asked to recollect activities on the day they were assaulted, while control subjects were asked to report on a random day 3 days before the study date. Having to report on every 10-minute period up to 3 days prior can, in itself, lead to recall issues across case and control subjects. Due to the assault, however, case subjects might be more motivated to recall events and exposures prior the event, compared with control subjects (or in contrast, depending on the severity of the assault and ensuing psychiatric problems such as posttraumatic stress disorder, the cases might be considerably less willing to report on events leading up to the assault)—it is unclear how this may have affected study findings. Recently, methods have been developed using GPS devices to have study participants prospectively document movement across space at regular intervals, using ecologic momentary assessments.4–6 Such an approach can provide highly detailed information on the places a person spends time, who they interact with in such places, and the types of physical features and amenities that affect their behavior in each place. These methods avoid recall bias issues, but as Wiebe et al. note, they require a large sample size in the case of rare events, such as violent assaults. Fourth, in the first analysis presented by Wiebe et al., exposures and activities that occurred at the time of the assault are compared with those that occurred throughout the day among the controls. By focusing on one time of comparison, it is difficult to sort out whether differences in environmental exposures are specifically related to the assault incident or reflect broader differences in exposures between case and control subjects. A sensitivity analysis comparing case and control subjects at random points throughout the day might have been helpful to make this distinction. Finally, few confounders are considered in the comparison of environmental exposures associated with the assault between case and control subjects. Concerns thus arise that the observed differences in environmental exposures may actually reflect individual level differences between cases and controls. One could imagine that differences in personality characteristics such as impulsivity could have led case subjects to select certain types of environmental exposures and to become involved in an assault. Despite such limitations, studies such as this break new ground by showing us that it is possible to consider how daily encounters with different features of the environment can shape behavior. More broadly, Wiebe et al. illustrate the value of studies that document daily trajectories across space, and examine the role that location and timing of exposure to different types of hazards and resources has on health. Future investment should focus on prospective cohort studies that incorporate ecologic momentary assessments with a strong spatial and time component. Such a design would overcome some of the concerns about selection bias and recall bias inherent to case–control studies, and lend richer insights into the ways individuals interact with their environment to shape their health status. More importantly, such detailed information will be critical to the design of preventive interventions that address relevant contexts at critical periods when they can have the largest impact. ABOUT THE AUTHOR MAGDALENA CERDÁ is the Vice Chancellor’s Chair in Violence Prevention and an Associate Professor in the Department of Emergency Medicine at the University of California, Davis. Her research focuses on the ways that the urban context shapes violence and substance use, and the emergence of new drivers and forms of substance use. Recent studies include a simulation of the potential impact that investment in access to treatment versus investment in neighborhood-level preventive interventions could have on rates of mental illness, as well as a simulation of the impact that investment in different types of neighborhood-level interventions could have on racial/ethnic inequalities in alcohol-related homicide.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.