International Journal of Drug Policy - 2014

Volume 25 Issue 3 May 2014

Issue link:

Contents of this Issue


Page 16 of 153

A.N. Martinez et al. / International Journal of Drug Policy 25 (2014) 516–524 517 2000 showed geographic clustering in one small central area of the city.(Davidson et al., 2003) These findings contributed to the devel- opment of government funded overdose response training targeted at heroin users. Among IDUs, migration and mobility have been linked to high risk injecting behaviors in Vancouver (Rachlis, Wood, Li, Hogg, & Kerr, 2008), San Francisco (Hahn, Page-Shafer, Ford, Paciorek, & Lum, 2008; Montgomery et al., 2012), and New York City (Deren et al., 2003). Researchers in Vancouver found that sex workers who moved their working areas away from main streets as the result of local policing efforts were more likely to be pres- sured into unprotected sexual intercourse (Shannon et al., 2008). In Tijuana, mobile IDUs, or those who injected drugs more than 3 kilo- meters from their residence, lived in neighborhoods with less drug activity and were more likely to share needles and get arrested for carrying syringes (Brouwer, Lozada, et al., 2012; Brouwer, Rusch, et al., 2012). We use activity spaces as an analytical framework to study intra-urban mobility among street-based IDUs in San Francisco and expand the conceptual focus of place from static residential spaces to habitually visited locations in people's everyday lives. Geographic information systems (GIS) software allows us to map individual level geographic data that includes habitually visited locations where study participants sleep, use drugs, and hang out during the day. Although there are multiple approaches to mea- suring activity spaces, which include the use of two dimensional ellipses, kernel densities and shortest paths networks, the calcu- lation of distance traveled between locations of regularly visited geographic locations, such as where one sleeps, works, and social- izes with family and friends, is central among all of them (Morency, Paez, Roorda, Mercado, & Farber, 2011; Paez, Mercado, Farber, Morency, & Roorda, 2010; Wong & Shaw, 2011). To examine accessibility of syringe exchange programs (SEPs) using an activity space framework, we use street-based networks to identify the shortest path routes between participant locations. We further examine the associations between distances of activity space routes and health-related outcomes, including HIV serosta- tus, syringe sharing, and overdose. Methods Data collection procedures The Urban Health Study (UHS) was a study of street-recruited IDUs in the San Francisco Bay Area (Kral et al., 2001). In this analysis, we used self-report data from a cross-section of IDUs interviewed in San Francisco between 2004 and 2005 (N = 1084). Recruiting and sampling procedures were based on targeted sampling procedures developed at UHS (Bluthenthal & Watters, 1995) and utilized in many National Institute on Drug Abuse (NIDA) funded studies of IDUs, including the 25-city NIDA Cooperative Agreement in the 1990s (Kral, Bluthenthal, Booth, & Watters, 1998). To summarize briefly, targeted sampling consists of mapping of county and city indicators to identify those areas with an increased prevalence of drug use, which is followed by ethnographic mapping of promising Census tracts, neighborhoods, or other geopolitical centers. Reli- able and up-to-date information about IDUs in a community are obtained through ongoing observations and social contact with individuals knowledgeable about IDU social networks, "copping spots", shooting galleries, and other locales frequented by IDUs. Recruitment is conducted by outreach workers who are familiar with the communities. Eligibility criteria for the study include being aged 18 years or older and drug injection within the past 30 days, as verified by self-report and physical examination for visible signs of recent venipuncture or dermapuncture consistent with drug injection (e.g. "track marks"). The questionnaires were administered by trained and experienced interviewers, who used computer-assisted data collection on laptop computers in the field with Questionnaire Development System (QDS tm ) software (NOVA Research Company, Inc., Bethesda, MD). Individual-level measures Dependent variables. We examine three health-related out- comes: (1) HIV serostatus, (2) syringe sharing, and (3) non-fatal overdose in the past 12 months. Syringe sharing is a standard behavioral measure of HIV risk among IDUs and has been used extensively in socio-epidemiological studies (Kral et al., 2003; Kral, Bluthenthal, Erringer, Lorvick, & Edlin, 1999) Participants were asked to report the number of times they engaged in either dis- tributive or receptive sharing in the past six months. These variables were recoded to a '1' if at least one occurrence of either distribu- tive or receptive syringe sharing occurred in the time period and a '0' if IDUs reported zero occurrences of sharing syringes. Fatal and nonfatal overdose has been previously studied as a major cause of morbidity and mortality among injection drug users (Seal et al., 2001). We defined recent nonfatal overdose as a self-reported over- dose that occurred in the past 12 months prior to being interviewed. Independent variables. Individual-level covariates included cat- egorical measures of gender (male and female), age (under 30 vs. older age), race/ethnicity (white, African American, Latino, other), sources of income (government assistance or illegal means) in the past 30 days, years of injection drug use, and homeless status. Homeless status was ascertained by asking if a participant reported being homeless at the time of interview. Six month dichotomous measures of arrest history, injection and non-injection drug use (heroin, methamphetamine, and crack smoking), trading sex for cash or drugs, and frequency of syringe exchange program use were also ascertained. Residential transience was measured by asking "In the last 6 months, how many other locations have you slept?" For statistical analysis, we added one to the number of additional loca- tions reported in a 6 month time period and divided this total by 180, or the number of nights in a 6 month period. Activity space variables To document locations where participants sleep, hang out, and use drugs, the study questionnaire included the following three questions: (1) "In the last 6 months, what is the intersection near- est to where you most often hang out during the day" (2) "In the last 6 months, what is the intersection nearest to where you most often sleep at night?," and (3) "In the last 6 months, what is the intersection nearest to where you most often use drugs?." All reported street intersections were geocoded as points using the San Francisco street network. All geospatial analyses were performed using ArcGIS software (ESRI, Redlands). Locations where partici- pants usually sleep were also geocoded to the level of the Census tract. We had 90% accuracy in successfully mapping and geocoding reported intersections. Activity space routes To calculate routes between each of the three locations reported by a participant, we used the New Route tool in the Network Ana- lyst extension. Fig. 1 depicts the methodological steps involved in creating the routes. Circles labeled '1' represent geocoded locations where participants usually sleep at night. Circles labeled '2' repre- sent geocoded locations where participants usually hang out during the day. Circles labeled '3' represent geocoded locations where par- ticipants usually use drugs. ArcGIS drew a route to connect each of the three locations. The route begins at the location where partic- ipants usually sleep and ends at the location where participants

Articles in this issue

view archives of International Journal of Drug Policy - 2014 - Volume 25 Issue 3 May 2014