The introduction of CCTV and associated changes in heroin purchase and injection settings in Footscray, Victoria, Australia.

Supplementary material

Interview scheduling

We used data on drug purchasing obtained from the Melbourne Injecting Drug User Cohort (MIX) study, a prospective cohort of 688 people who inject drugs (PWID) recruited between April 2008 and January 2010. Experienced fieldworkers interview participant approximately annually in face-to-face interviews. Interviews are conducted regularly throughout the year, and attempts to contact participants and schedule interviews commence two months prior to their due date (annually from baseline date). Where contact is difficult (no phone, new address, incarceration etc.), regular attempts are made to contact participants after they become overdue, including through secondary contacts provided at recruitment (e.g. family members, friends). Unless confirmed deceased, participants are not removed from the study.Interviews are labelled consecutively for participants regardless of their time separation, and following extended periods without an interview—for example, after two years in prison—interviews may be conducted after a minimum of six months to catch up to original due dates. At 1 August 2013 (the dataset end) 2,152 interviews had been conducted, and median dates for baseline and the first three follow-up interview waves were:

  • July 2009 (Inter Quartile Range (IQR) March 2009 –October 2009), N=688;
  • August 2010 (IQR April 2010 –December 2010), N=516;
  • August 2011 (IQR April 2011 –December 2011), N=440; and
  • July 2012 (IQR March 2012 –October 2012), N=348.

Relevant survey questions

In each interview, participants are asked details about their three most recent heroin purchases (or fewer if they could not recall enough). For each purchase, participants were asked “When was the drug purchased?”, with categorical responses of‘today’, ‘yesterday’, ‘within the last week’, ‘within the last month’, ‘more than a month ago’ and ‘not purchased in the last month’ allowed. To convert purchase events to a time-series for this study, we date-stamped categorical answers to as 0, 1, 4, 19, 45 and 45 days prior to the interview daterespectively (the categories ‘more than a month ago’ and ‘not purchased in the last month’ were grouped together as it was not clear that they represented distinct responses). Participants are also asked the suburbs that the purchases were made in, the settings where the purchases were made and the settings where the purchases were used; we classified the allowed categorical responses to the settings of purchases and use as shown in Table S1.

Table S1: Aggregation of specific responses into response groups for questions about the drug purchase and use settings.

Responses / Grouped Responses
Purchase locations
Street / Street
Mobile Dealer, Car / Pre-arranged meeting
Home, Friend’s house, Other person’s home / House
Other / Other
Use locations
Street, Park / Street
Car / Car
Public toilet / Public toilet
House, Friend’s house, Other person’s home / House
Disused building, Stairwell, Other / Other

Between 18 April 2008 and 1 August 2013 there were 3,736 heroin purchases with a valid response to day of purchase, 3,633 with a valid purchase suburb, 3,633 with a valid purchase setting and 3,628 with a valid use setting. Due to low response numbers for purchases made and used in ‘other’ settings (n=5 and n=131 respectively), these observations were not included in the setting analysis, leaving 3,628 and 3,497 observations of purchase setting and use setting respectively.

Purchases made by local residents

For this analysis, we considered a drug purchase to be local if the suburb it occurred in was in the same LGAas the participant’s residence, as defined by the Department of Transport, Planning and Local Infrastructure (Department of Transport, 2015).The data suggest that there are several drug market hotspots in Melbourne; among the 124 suburbs in which a heroin purchase was reported to have taken place (including Footscray), 11 had more than 70 heroin purchases reported, representing 70% (N=2,550) of all reported purchases. Therefore, although there are multiple suburbs that collectively form each LGA, our definition of a purchase being made locally refers to participants obtaining heroin from their local hotspot.

Other drugs

MIX participants are also asked about recent purchases of methamphetamine, benzodiazepines and other pharmaceutical opioids, but due to insufficient data for Footscray this analysis was restricted to heroin (70% of observations in Footscray were heroin, 20% benzodiazepines, 5% methamphetamine and 5% pharmaceutical opioids). Drug types were not pooled, to avoid bias from inconsistent sampling of different drug markets over time.

Intravenous versus non-intravenous use

As 98% of heroin use reported by the cohort was intravenous, ‘use’ and ‘injection’ of heroin were considered synonymous for this analysis.

Further details

Further details regarding the MIX cohort and questionnaire can be found elsewhere(Horyniak et al., 2013, Scott et al., 2015).