The authors thank the editors of Addiction for the opportunity to respond to the critique of our findings by Mr Graham Mooney . Mr Mooney raises several criticisms of the internal and external validity of our findings , to which we respond in detail below.
- Mr Mooney states that variations in number of patients treated or socio-demographic characteristics of the surrounding neighborhoods among the 13 methadone treatment centers (MTCs) included in our study would have biased our findings. This criticism misses several key points of our microecological method, which compares crime slopes (the change in crime counts per unit area with increasing distance from the site) across sites. In the microecological method, crimes per unit area are counted, not crimes per population. Thus numbers of patients in clinics or visitors to the other sites are not relevant. Moreover, in the microecological method, a crime slope is constructed separately for each site, so each site's neighborhood serves as its own control for socio-demographic characteristics and other potentially confounding variables. It is the crime slopes (i.e. change in crime counts with increasing distance from the site), not crime counts, that are compared across sites.
- Mr Mooney suggests potential bias from a lack of controlling for differences in neighborhood socio-demographic variables between the clinic and comparison sites. However, comparison sites other than hospitals (which were too few in number for matching) were matched to the clinic sites based on neighborhood socio-demographic variables and baseline crime rates.
- Mr Mooney states that crime data were collected in 1999–2001, while the telephone survey of MTC characteristics was conducted more recently. The telephone survey of MTC administrators was conducted 3–4 years after the 1999–2001 study period, but focused on MTC characteristics during the study period. Most of the administrators surveyed had been with the same clinic since before the study period, so we believe that the survey data were fairly accurate. In any case, we do not see how even substantial errors in MTC characteristics would negate our basic finding, which is based on the slope of crime counts.
- Mr Mooney believes that the choice of convenience stores through Switchboard.com might have biased the findings and that there was no validation of convenience store status during the study period. We confirmed by telephone survey that each of the convenience stores used as a comparison site in the study was open during the study period. This validation was conducted at the time of the study, not ‘a decade after’ (the time of publication).
- Mr Mooney notes that other foci of crime, such as alcohol sales outlets or open-air drug markets, could have been located near study sites and biased the findings. It is not clear to us how such a bias would invalidate our findings. If such an outlet were close to the MTC it would tend to inflate the surrounding crime counts: the opposite of our finding. If such outlets were located further away from the methadone clinics, they should cause a spike in crimes further away, which was not observed in our study.
- Mr Mooney notes that there may have been biases in the reporting of crimes, such as ‘time-sensitive crime control’ or increased law enforcement activity around some study sites. Even if such differences exist, it is not clear how they would bias our findings. For example, he points to increased law enforcement around hospitals in Baltimore City. Such increased policing could potentially result in increased reports of crimes or in decreased reports of crimes. Our data show no concentration of crime around the hospital comparison sites and no spike in crimes at the edge of hospital campuses. In any case, we protected against confounding MTC-related crimes and hospital-related crimes by geoding each clinic and hospital site by taking a global positioning system (GPS) reading at their front door. Thus, while hospitals and MTCs may have been on the same campus, the study sites were different for each. We excluded from the study the one clinic that could not be differentiated from the hospital because it was located within the main hospital building. Finally, temporal variations in law enforcement, caused by police ‘sweeps’ and other temporary changes in law enforcement, are unlikely to affect our findings because we included all reported crimes over a 3-year period.
- Mr Mooney notes that crimes were not disaggregated by type, even though ‘drug addicts most commonly commit acquisitive crimes to fund their addiction’. We agree with the later statement, but had to analyze crime counts in the aggregate because there were insufficient numbers of crimes of any individual type.
- Mr Mooney alludes to the idea that limiting the area of analysis to ‘concentric buffers of 100 m’ misses the spatial impact of MTCs. However, we initially analyzed buffers of 300 m from the sites. As noted in the paper, we plotted these data with a generalized additive model (GAM) with a spline term. Using this method we found that the vast majority of the variation in crime counts was occurring in the inner 100 m around sites, i.e. there were no spikes in crime further from the MTC sites. Moreover, any increases in crime further than 100 m from the sites would further support the conclusion that crimes do not cluster near MTCs. We are not aware of any plausible mechanism whereby MTCs (or any other type of site) would only increase crime more than 300 m away (the distance evaluated in this study), but have no affect on crime closer to the site.
Finally, Mr Mooney makes two statements not related directly to our study methodology: that treatment for opiate dependence typically takes years to achieve remission and that treatment providers do not share information about long-term treatment outcome. Whether or not these statements are true, they have no bearing on the validity of our findings. A distinction of our microecology approach is that it is not based on the treatment outcome for individual patients or the MTC as a whole, but rather on the spatial distribution of crime around MTCs. Therefore, we continue to believe that our findings are valid and should be considered in discussions of the impact of MTCs on neighborhood crime.
Declarations of interest
SB is medical director of the Baltimore VA Hospital's methadone maintenance clinic; otherwise there are no interests to declare.
This work was supported by the Substance Abuse Policy Research Program, Robert Wood Johnson Foundation; the Center for Substance Abuse Treatment/Substance Abuse and Mental Health Services Administration; and the Intramural Research Program of the National Institutes of Health/National Institute on Drug Abuse.
Susan J. Boyd1, Li Juan Fang2, Deborah R. Medoff2, Lisa B. Dixon3 & David A. Gorelick4
Department of Psychiatry, University of Maryland School of Medicine, Baltimore, MD, USA,1
Department of Psychiatry, Division of Services Research, University of Maryland School of Medicine, Baltimore, MD, USA,2
Center for Practice Innovations, Department of Psychiatry, Columbia University, New York, NY, USA3 and
Intramural Research Program/National Institute on Drug Abuse/National Institutes of Health, Baltimore, MD, USA.4