Research and analysis

The impact of drug-related law enforcement activity on serious violence and homicide: A systematic review

Published 27 March 2025

Authors

RAND Europe

Elle Wadsworth, Mafalda Pardal, Lucy Strang, Laura Atuesta, Fin Oades, Emily Hutton, Eric Sevigny, Emily Lawson

Preface

This report presents the findings from a systematic literature review on the impact of drug-related law enforcement activity on serious violence and homicide.

The report was prepared for the UK Home Office to inform work undertook to implement the Drugs Strategy ‘From Harm to Hope: a 10-year drugs plan to cut crime and save lives’, published under the 2019 to 2022 Johnson Conservative government (HM Government 2021).

RAND Europe is an independent not-for-profit policy research organisation that aims to improve public policy and decision making in the public interest through research and analysis. This report has been peer-reviewed in accordance with RAND’s quality assurance standards.

For more information about RAND Europe or this document, please contact:

Lucy Strang
RAND Europe, Westbrook Centre
Milton Road, Cambridge CB4 1YG
United Kingdom

Tel. +44 1223 353 329

Email: lstrang@randeurope.org

Executive summary

The UK Home Office commissioned RAND Europe to systematically review the global literature in English, Dutch, Spanish and Portuguese on the impact of drug-related law enforcement activity on serious violence and homicide between 2011 and 2024. The research was commissioned to inform work the Home Office and other government departments undertook to implement the most recent Drugs Strategy ‘From Harm to Hope: a 10-year drugs plan to cut crime and save lives’, published under the 2019 to 2022 Johnson Conservative government (HM Government, 2021). Part of the Drugs Strategy related to stepping up the level of law enforcement on illicit drug-related activity.

This review addresses the following research question in the global literature: What is the impact of drug-related law enforcement activity on serious violence and homicide?

To explore this question, we examined 5 sub-questions:

1. Are certain types of drug-related law enforcement activity effective in reducing serious violence and homicide?
2. What aspects of drug-related law enforcement activity make them more/less likely to be effective in reducing serious violence and homicide?
3. Do certain types of drug-related law enforcement activity have unintended consequences, such as worsening adverse serious violence and homicide outcomes?
4. When drug-related law enforcement activity generates adverse serious violence and homicide outcomes, what are the factors that lead to these unintended consequences?
5. How do serious violence and homicide outcomes relating to drug-related law enforcement activity differ by country and local areas, and why?

This report reviews the evidence on 8 types of drug-related law enforcement activities (Chapter 3):

• selective enforcement for a specific area or group
• leadership removal
• attack or removal of a drug trafficking organisation member
• arrests or charges
• illegal crop eradication
• drug seizures
• military interventions
• multi-jurisdictional anti-drug coordination and resourcing

It also explores possible barriers to effectiveness and contributing factors that may help such activities reduce violence (Chapter 4). The report then analyses differences between and within countries regarding these interventions and their outcomes (Chapter 5).

The report concludes with reflections and implications from this review’s findings (Chapter 6), as follows:

Overall, the available evidence suggests that drug-related law enforcement activities are of limited effectiveness in reducing violence. Indeed, more studies demonstrated an association between drug-related law enforcement activities and increased violence than decreased violence. Selective enforcement tactics appeared the most promising in their capacity to reduce violence, although the evidence base covered in this review is limited.

Passive drug-related law enforcement activities, such as increasing police presence in known drug market areas, appear promising in reducing violence. However, less evidence is available on the effectiveness of these interventions than on active law enforcement activities.

The causal mechanisms of violence reduction are under-explored in the literature. However, several studies discussed supply disruptions, focused deterrence and positive relationships between police and communities as potential success factors.

Barriers to the effectiveness of violence-reduction efforts included the resilience of drug markets, the cultural significance of violence in some drug trafficking organisations, and law enforcement’s limited resources.

This review did not identify any UK-based evidence – most research came from the Americas. While most law enforcement activities in this review also occur in the UK, the results are not directly replicable in a UK setting.

Evidence on the relationship between drug-related law enforcement and serious violence and homicide over the last decade is lacking. What was previously effective (or ineffective) in reducing violence may yield different results now.

More evidence is needed on the effectiveness of drug-related law enforcement activities in retail-level markets or prison settings in reducing violence.

Relevant agencies planning and implementing drug-related law enforcement activity should consider the risk of increased violence, particularly for interventions for which available evidence suggests a strong association (for example, leadership removal and seizures).

Future UK research on drug-related law enforcement and violence could focus on interventions that may reduce violence, such as selective enforcement, and whether the findings presented can be validated.

1. Introduction

In 2022, the UK Home Office commissioned RAND Europe to systematically review the literature on the impact of drug-related law enforcement activity on serious violence and homicide between 2011 and 2024. The research was commissioned to inform work the Home Office and other government departments undertook to implement the most recent Drugs Strategy ‘From Harm to Hope: a 10-year drugs plan to cut crime and save lives’, published under the 2019 to 2022 Johnson Conservative government (HM Government, 2021). Part of the Drugs Strategy related to stepping up the level of law enforcement on illicit drug-related activity.

The enforcement activities outlined in the Drugs Strategy include the County Lines Programme, which targets gangs and organised criminal networks in the illicit drug market through dedicated mobile-phone lines or other communication forms (HM Government, 2023). Other interventions are directed at tackling middle-market distributors by disrupting financial and communication channels and expanding the use of technology in detecting drug smuggling into prisons.

Alongside these efforts is a recognition of the need to better understand the evidence base on the relationship between drug-related law enforcement activity and serious violence and homicide. This understanding depends on identifying the nature and extent of the available evidence, the gaps in the evidence base, and the conclusions we can draw to target such activities effectively.

1.1 Research questions

To this end, we undertook a systematic review of the relevant literature to answer the primary research question: What is the impact of drug-related law enforcement activity on serious violence and homicide?

We also considered 5 component sub-questions:

1. Research question 1 (RQ 1): Are certain types of drug-related law enforcement activity effective in reducing serious violence and homicide?
2. Research question 2 (RQ 2): What aspects of drug-related law enforcement activity make them more/less likely to be effective in reducing serious violence and homicide?
3. Research question 3 (RQ 3): Do certain types of drug-related law enforcement activity have unintended consequences, such as worsening adverse serious violence and homicide outcomes?
4. Research question 4 (RQ 4): When drug-related law enforcement activity generates adverse serious violence and homicide outcomes, what are the factors that lead to these unintended consequences?
5. Research question 5 (RQ 5): How do serious violence and homicide outcomes, relating to drug-related law enforcement activity, differ by country and local areas and why?

Given this review’s focus on the impacts of drug-related law enforcement on serious violence and homicide, we did not collect evidence on other goals or impacts of drug-related law enforcement’s activities (for example, harm reduction or other drug-related offences).

1.2 How the review was conducted

To address the above research questions, we conducted a systematic review identifying and assessing the evidence on the impact of drug-related law enforcement activities on serious violence and/or homicide.

We defined the key concepts involved as follows:

• ‘drug’ includes all illegal substances (for example, cocaine, heroin)
• ‘law enforcement’ includes any law enforcement branch/unit and the military (when the military has a policing role)
• ‘activity’ includes any law enforcement presence, activity or intervention
• ‘serious violence’ includes knife crime, gun crime and homicide, as defined in the Serious Violence Strategy (HM Government, 2018)

Annex A provides a full description of the methodology used in this review. We included any primary empirical study of any research design and geographic scope in this review that was:

• focused on drug-related law enforcement’s impact on serious violence and/or homicide
• published in English, Spanish, Portuguese or Dutch
• published in 2011 onwards
• published as an academic journal article, grey literature or a PhD dissertation

Our review includes literature published from 2011 onwards, as a previous systematic review on this topic covered evidence published up to January 2011 (Werb et al., 2011). We identified and included articles using academic databases and grey literature sources published from January 2011 to December 2022.

After expanding our search terms to include terminology present in relevant articles but not in our original searches (for example, ‘leadership’, as detailed in Annex A), we repeated our literature searches in January and March 2023. We repeated the searches for a final time in February 2024. We built search terms from the overarching research question and included terms relating to drugs, law enforcement and serious violence/homicide. We screened articles for relevance at 2 stages, in line with systematic review practices – the first focused exclusively on the title and abstract, while the second drew on the full-text review. We extracted data from the final set of 33 eligible articles using a data-extraction tool developed by the research team for this review. Two researchers independently conducted all screening and data extraction to ensure consistency and replicability. These 2 researchers discussed any discrepancies to reach a consensus, or if necessary, a third researcher made the final decision.

We conducted a quality assessment of the identified literature using 2 adapted checklists (CASP, 2022; Clark et al., 2014), accounting for the broad research designs included in the review. The adapted checklist assessed the quality of the study’s aims, methodology, research design, data collection, ethics, analysis and how well the discussion reflected the findings. Two researchers assessed the studies on quality across 12 criteria, using the following 3 categories: ‘No’ (coded as ‘0’), ‘Yes’ (coded as ‘1’) and ‘Unsure’. The 2 reviewers discussed any discrepancies or ‘Unsure’ entries until resolved. We then calculated each study’s total score and used it to divide them into 3 colour-coded categories: ‘Low’ (0 to 4; red), ‘Medium’ (5 to 8; amber) and ‘High’ (9 to 12; green). As we conducted the quality assessment to help interpret the strength of the evidence, rather than exclude articles based on quality, we retained all low-quality studies in the review.

1.3 Studies included in the review

Following the application of the screening and inclusion criteria, we included 33 studies in the review:

1. The searches yielded 1,749 results
2. After removing duplicates, 1,316 remained
3. Following the title-and-abstract search, 155 remained
4. After the full-text review, we included 33 studies.

See Annex A’s PRISMA flowchart for the full breakdown of searches.

1.4 Limitations

A few limitations to this systematic review are worth highlighting. First, some evidence about the effectiveness of drug-related law enforcement activity may not be available for analysis in the literature due to security or privacy issues. This unavailability may limit the ability to fully answer some of our research questions since we rely exclusively on data captured in the reviewed studies. Second, we chose the languages included in this review based on the study team’s skills and an early scoping exercise identifying the most common languages in which the most relevant literature was published, which could have introduced bias. However, additional languages beyond English also broadened our review’s geographic reach beyond the previous review on this subject (Werb et al., 2011). Third, Chapter 6 describes the lack of studies analysing data from the past 10 years (possibly due to a publication lag) that prevented the team from reviewing the most recent data. Finally, most studies examined law enforcement activities at the ‘higher’ drug supply chain levels, for example, drug trafficking organisations (DTOs) or crop eradication. In contrast, less evidence exists on the ‘lower’ drug supply chain levels, such as street-level dealing.

1.5 Structure of this report

This report begins by outlining the studies included in the systematic review (Chapter 2). Chapter 3 addresses research questions (RQ) 1 and 3, reviewing available evidence on 8 drug-related law enforcement activities:

• selective enforcement for a specific area or group
• leadership removal
• attack or removal of a DTO member
• arrests or charges
• illegal crop eradication
• drug seizures
• military interventions
• multi-jurisdictional anti-drug coordination and resourcing

Chapter 4 addresses RQ 2 and RQ 4, exploring potential contributing factors to these activities that can reduce violence and barriers to their effectiveness. Chapter 5 then addresses RQ 5 by analysing differences between and within countries regarding these interventions and their outcomes. Finally, Chapter 6 concludes with a discussion of this review’s reflections and implications.

2. Overview of studies included in the systematic review

This review included 33 articles, more than twice the number covered by the previous systematic review (n=15, Werb et al. 2011). Of these 33, most were journal articles (n=28), although a few (n=5) PhD dissertations also met the inclusion criteria. Table 1 below summarises the reviewed studies’ key characteristics. Most studies used quantitative methods, such as regression analysis, as a research design. Those reviewed studies that relied on qualitative methodologies drew on interviews and ethnographic research, with one study using mixed methods.

Table 1: Overview of the reviewed studies’ key features

Key features	Categorisation	N (%)
Document type	Journal articles	28 (85%)
	PhD dissertations	5 (15%)
Study design	Quantitative	30 (91%)
	Qualitative	2 (6%)
	Mixed methods	1 (3%)
Geographic location	North America	23 (70%)
	South America	8 (24%)
	Europe	2 (6%)
	Asia	1 (3%)
Drug-related law enforcement activities	Selective enforcement	7 (21%)
	Leadership removal	7 (21%)
	Attack or removal of a DTO or drug gang member	4 (12%)
	Arrests or charges for drug law violations	4 (12%)
	Crop eradication	6 (18%)
	Drug seizures	3 (9%)
	Militarised interventions	3 (9%)
	Multi-jurisdictional anti-drug coordination and resourcing	1 (3%)
Serious violence and/or homicide related outcome measured	Homicide	19 (58%)
	Homicide and other violent crimes (for example, aggravated assaults, non-fatal shootings)	13 (39%)
	‘Violent crime’, broadly defined	2 (6%)
Quality assessment	High quality	18 (55%)
	Medium quality	8 (24%)
	Low quality	7 (21%)

Notes:

1. Since some studies included multiple interventions and countries, the percentages do not necessarily total 100% (n=33).

2.1 Most reviewed evidence focused on interventions in the Americas

Most studies reported on drug-related law enforcement interventions in North America, particularly Mexico (n=14) and the United States (US) (n=9). An additional 6 studies analysed interventions based in South America: Colombia (n=4), Bolivia (n=1), Peru (n=1) and Brazil (n=2). Only 2 reviewed studies focused on European developments: the Netherlands (n=1) and Denmark (n=1). One study focused on Asia (Philippines, n=1). The geographic distribution of the studies this systematic review covers is more diverse than the 2011 review (Werb et al., 2011). For instance, 87% of studies included in the 2011 review were in the US, compared to only 27% in the current review. The current review also used an extensive list of search terms to capture international literature (including those in Spanish, Dutch and Portuguese). In contrast, Werb et al. (2011) used more restrictive, US-focused search terms.

2.2 The studies analysed a range of drug-related law enforcement interventions

The reviewed studies provided insights into the effectiveness of various drug-related law enforcement interventions. We identified 8 different types of interventions:

1. Selective enforcement
2. Leadership removal
3. Attack or removal of a DTO or drug gang member
4. Arrests or charges
5. Crop eradication
6. Drug seizures
7. Military interventions
8. Multi-jurisdictional anti-drug coordination and resourcing

Table 2 below provides additional information on these interventions.

Table 2: Types of drug-related law enforcement interventions analysed in the reviewed studies

Type of drug-related law enforcement intervention	Broad intervention definition	Number of studies focusing on this intervention1	Countries where this intervention was examined
Selective enforcement	A sustained intervention on a specific small area (for example, neighbourhood, drug-dealing hotspot) or group, usually involving a suite of activities such as enhanced enforcement, heightened police presence and community involvement	7	Brazil, US, Netherlands
Leadership removal	Arresting or killing a high-ranking DTO or gang member	7	Mexico, US
Attack or removal of a DTO member	Attacking or removing a DTO or drug gang member who is not a leader or high-ranking member	4	Philippines, Mexico
Arrests or charges	Drug-related arrests or charges that did not target the leader of a drug-related gang and the altercation was not violent	4	Denmark, Mexico, US
Crop eradication	The forced destruction or substitution of illegal crops (for example, coca and opium poppy) by law enforcement	6	Bolivia, Colombia, Peru, Mexico
Drug seizures	The confiscation of illegal drugs from drug trafficking organisations by authorities	3	Mexico
Military intervention	Employing the armed forces to fight DTOs, including deploying anti-drug aircraft, setting up checkpoints and conducting patrols	3	Mexico
Multi-jurisdictional anti-drug coordination and resourcing	Federal funding to increase anti-drug policing initiatives at the local level1	1	US

Notes:

1. Other interventions in this review may also involve additional federal funding for anti-drug policies. However, this category is for studies focused exclusively on the effects of the funded initiatives.

Different strategic options underlie these intervention types. For instance, some prioritise targeting the leadership of drug trafficking organisations or gangs, while others target particular areas or groups of people or encompass general arresting/charging practices. The interventions also focus on different time points in the illicit supply chain, from crop eradication to drug seizures and arrests. Different law enforcement actors are also involved, ranging from local police units to federal police to the armed forces. Chapter 3 reports the findings for each intervention type in more detail.

2.3 Homicide rates were the primary outcome assessed in the literature

This review includes multiple outcomes under ‘serious violence and homicide’, including knife crime, assault, gun crime, manslaughter and assassination, as reflected in our search strategy. Of the included studies:

• 19 used ‘homicide’ as the measured outcome, such as the number of monthly homicides – drawing on data from local and federal government databases
• 13 included homicide with other violent crimes (for example, aggravated assaults, non-fatal shootings)
• 2 used ‘violent crime’ as the outcome without defining the specific crimes included

2.4 Despite the variation, most included studies were high quality, according to our quality assessment checklist

Most of the included articles reported high-quality studies, comprising 18 high-quality, 8 medium-quality and 7 low-quality studies. Low-quality studies included journal articles with opaque methodologies, rendering the findings difficult to interpret and assess. Out of a possible 12, studies scoring 0 to 4 were ranked ‘Low’, those scoring 5 to 8 were ranked ‘Medium’ and those scoring 9 to 12 were ranked ‘High’. Chapter 3 reports and discusses each study’s low, medium or high quality.

Most studies clearly articulated their aims and used an appropriate research design to address them. The majority also used data deemed appropriate for answering the research questions. Approximately half the studies had a transparent methods section replicable from the information given, and two-thirds had a discussion reflective of the findings. Most of the quantitative studies reviewed did not use a randomisation method in the sample selection. However, this is likely due to the study types, based on retrospectively evaluating existing interventions (that is, non-randomised). No qualitative study considered how the relationship between the researcher and participant could bias the study findings. Finally, only one included study had mentioned ethical issues, for example, obtaining approval from an ethics committee.

3. The effectiveness and unintended consequences of drug-related law enforcement activities

Key messages:

• the evidence on the effectiveness of selective enforcement strategies in reducing serious violence is mixed but points more towards reducing violence: 4 studies showed a reduction in violence, 3 showed no change and one showed an increase
• most studies found an association between leadership-removal interventions and increased violence
• four studies found an association between attacking or removing a DTO or drug gang member and increased violence
• three out of 4 studies examining the relationship between drug arrests/charges and violence found an association between drug arrests and increased violence
• the evidence suggests a strong association between illegal crop eradication and increased violence
• data from Mexico points to an association between drug seizures and increased violence, though this slightly decreased depending on the type of drug seizure.
• three studies in Mexico found an association between anti-drug military interventions and increased violence

This chapter presents our review’s findings on drug-related law enforcement activities’ effectiveness and unintended consequences in reducing serious violence and homicide. To do so, we provide insights relevant to 2 of the central research questions underpinning this study:

RQ 1: Are certain types of drug-related law enforcement effective in reducing serious violence and homicide?

RQ 3: Do certain types of drug-related law enforcement activity have unintended consequences, such as worsening adverse serious violence and homicide outcomes?

This chapter introduces the reviewed evidence on each type of drug-related law enforcement activity identified in this review (see Table 2).

3.1 Selective enforcement

Overall, the evidence on the effectiveness of selective enforcement strategies in reducing serious violence is mixed but points more towards reducing violence: 4 studies showed a reduction in violence, while 3 showed no change and one showed an increase.

Of the 7 identified articles examining a relationship between selective enforcement (that is, a drug-related law enforcement intervention for a specific area or group) and violence, 4 studies found an association between selective enforcement and reduced violence (Corsaro et al., 2012; Corsaro, 2013; Israel de Souza, 2019; and Tealde, 2019).

Two studies evaluated the Favela Pacification Program (FPP) in Rio de Janeiro, Brazil (Israel de Souza, 2019; Tealde, 2019). The FPP was a government intervention assigning a pacifying police unit (PPU) to selected favelas under DTO control to remove the DTO while remaining physically present in the favela to retain control. First, Israel de Souza (2019) conducted an ethnographic study to examine pacification in selected favelas in Rio de Janeiro, Brazil, between 2012 and 2018. Based on ethnographic qualitative analysis, the results indicated that residents of the favelas felt positive about pacification as they could allow their children to play outside without fearing their entanglement in violence. Residents also reported welcoming the arrival of pacification in 2009, as favelas were under the control of ‘outside’ drug dealers, not their local community. However, residents described a challenging adjustment period, and the police faced difficulties settling into the community. Nonetheless, residents reported it being a peaceful intervention once the police settled. However, violent-crime rates rose again when pacification ended. According to residents, the intervention reduced violence in the favelas while active, but violence resumed once the intervention ended. Our quality assessment deemed this study ‘medium’ quality. While the study included thorough descriptions of the aims, intervention, findings and discussion, the study lacked some methodological details. However, this may be partly due to ethnographic studies’ typical reporting style.

Second, Tealde (2019) evaluated the FPP’s effect on violence in Rio de Janeiro, Brazil, between 2007 and 2010, based on homicide rates per 100,000 inhabitants. The author examined 37 pacified favelas, dividing them into 2 intervention groups and one control group. The 2 intervention groups comprised favelas pacified before September 2010, previously under the control of one of 2 DTOs (Comando Vermelho [CV] and Amigos dos Amigos [ADA]). The control group comprised favelas under CV’s control and not yet pacified by September 2010 (though pacified later). Using event-study regression analysis, the author demonstrated mixed results: homicides significantly decreased in pacified favelas previously under CV’s DTO but increased in ADA favelas. The author estimated the program to have saved approximately 34 lives: an estimated 64 minus 30 lost due to displacing violence to other favelas without pacification. However, since our quality assessment deemed this study ‘low’ quality, the results should be interpreted cautiously. The author did not provide a transparent or replicable methodology, providing insufficient detail on the population assessment, comparator group and outcome measurement. The author also declared the possibility of a contaminated control group, yielding low confidence in this paper.

Two studies evaluated the Drug Market Intervention (DMI) in High Point, North Carolina, US (Corsaro et al., 2012; Corsaro, 2013). The High Point DMI concentrated police in targeted areas, using threats of sanctions to reduce drug-related re-offending and involving community residents in the initiative (for example, local community leaders speaking to known dealers to express their disapproval of drug markets). Both studies evaluated the impact of High Point DMI on violent crimes, which included homicides, rape, sexual offences, assaults and robberies. The first (Corsaro et al., 2012) examined the relationship between implementing the High Point DMI and violent crimes in targeted census blocks and High Point generally between 1998 and 2008. The study compared targeted neighbourhoods to non-targeted areas pre- and post-intervention. Difference-in-difference panel regression estimations indicated that violent crime significantly decreased in targeted neighbourhoods post-implementation compared to the rest of the city (a 13 to 18% decrease). The results suggested that this violence was not simply displaced, as no increased violence was evident in the surrounding areas. When examining within-group trajectories of areas with ‘chronically high’, ‘moderately stable’ and ‘negligible’ violent crime, results indicated a significant (17%) reduction in violent crime in the ‘chronically high’ areas. The study found no change in violence levels in the ‘moderately stable’ group, but a small yet statistically significant increase in the ‘negligible’ group, suggesting the intervention’s biggest impact was on the most violent areas.

The second study (Corsaro, 2013) examined the relationship between the High Point DMI and violent crimes in 4 targeted neighbourhoods between 1998 and 2009, comparing pre- and post-intervention levels in each. Interrupted time-series analysis results demonstrated mixed evidence. Compared to pre-intervention levels, one neighbourhood demonstrated a strong significant association between the intervention and reduced violence (a 19.9% reduction), one demonstrated a modest but significant association (a 13.5% reduction), and 2 showed no association. When the authors pooled and analysed the rest of the city, they found no significant change in violent offences. Thus, both studies found that the High Point DMI had mixed results depending on the neighbourhood/area. Our quality assessment also deemed both studies as ‘high’ quality. Both provided a thorough and transparent study description, including a sufficiently rigorous analysis section and a robust interpretation of the findings.

We also reviewed 3 studies that found no evidence of an association between selective enforcement and change in violence (Corsaro & Brunson, 2013; Klement & Blokland, 2021; Lawrence, 2023). Corsaro and Brunson (2013) examined the relationship between implementing the ‘pulling levers’ law enforcement intervention and violent crimes in an open-air drug market in Peoria, Illinois (US), between 2006 and 2010. The intervention comprised a threat-sanction approach that warned offenders of the sanctions imposed if they re-offended. The study measured violent crime based on robberies, aggravated assaults and homicides. The authors also examined multiple additional outcomes beyond this review’s scope, including property and drug-and-disorder crimes. Interrupted time-series analysis results indicated no statistically significant change in violent crimes when comparing pre- and post-intervention levels. The authors also conducted a telephone survey among residents familiar with the Peoria intervention, asking about their perception of its impact. Most (66%) respondents considered the intervention to have had no impact. As with the Corsaro (2013) and Corsaro et al. (2012) studies, our quality assessment deemed this study ‘high’ quality.

Klement and Blokland (2021) examined the relationship between the ‘whole of government’ approach in the Netherlands and the charges for violent crimes committed by Outlaw Motorcycle Group (OMCG) members and support club members before and after the intervention between 2009 and 2015. Interrupted time-series analysis results demonstrated no significant changes in the rate of violent crime criminal charges among OMCG members and support members post- versus pre-intervention. The ‘whole of government’ approach involved multiple government agencies conducting interventions to make membership less desirable and rewarding, such as the zero-tolerance stance on any law-breaking activity committed by an OMCG member. However, since the study was unclear about the specific interventions included and did not clearly define ‘violent crime’, our quality assessment deemed it ‘medium’ quality.

Lawrence (2023) examined the effect of introducing a bicycle police patrol unit in 2021 on total crime, violent persons crimes and shootings in the Kensington area of Philadelphia (US) between 2019 and 2023. Using a micro-synthetic control method, the author concluded there was no significant association between the bicycle police patrol unit and total crimes after one or 2 years. However, when the data were disaggregated by crime type, the bicycle police patrol unit was associated with a significant increase in violent person crimes and shootings 2 years in the treatment area compared to the synthetic control. Both violent person crimes and shootings increased in the first year, but this increase was not significant. The author concluded that any changes observed were seen gradually and that COVID-19 could have been a factor in the outcomes observed as it impacted and restricted policing activities in Philadelphia, including the bicycle unit. Our quality assessment classed this study as ‘high’ quality. The author provided a thorough study account detailing the aims and methods, and the conclusions accurately reflected the findings.

3.2 Leadership removal

There is mixed evidence on the effectiveness of leadership-removal strategies in reducing serious violence. However, most studies found an association between leadership-removal interventions and increased violence.

Seven studies examined the relationship between leadership removal and violence, defining leadership removal as arresting or killing a leader or high-ranking DTO or drug gang member. Six of these studies found an association between leadership removal and increased violence (Calderon et al., 2015; Del Rio, 2022; Dickenson, 2014; Lindo & Padilla-Romo, 2018; Phillips, 2015; Vargas, 2014). However, 3 of these studies also found evidence of decreased violence (Lindo & Padilla-Romo, 2018; Phillips, 2015; Vargas, 2014), as did a study on gang leadership arrests (Burke, 2022). This variation depended on the short- versus long-term impacts of leadership removal, the rank of the DTO member captured, spillover violence into neighbouring regions, and indirect impacts on the number of criminal organisations.

Calderon et al. (2015) explored municipalities in Mexico, where a leader or lieutenant of a DTO was captured or killed by law enforcement (referred to by the authors as a ‘kingpin strategy’) between 2006 and 2010. The study examined the relationship between the kingpin strategy and violence, using homicides among males aged 15 to 39 as a proxy for DTO-related homicides and homicides among the general population to measure violence. Negative binomial regression analysis showed that capturing a DTO leader was significantly associated with increased DTO-related homicides in the first 6 months post-capture in the municipality where the capture occurred. The average number of monthly homicides was approximately a third higher post-capture than pre-capture. However, there was no difference in the long term (6 to 12 months post-capture). Regarding the effect of leadership removal on neighbouring areas, the average number of monthly DTO-related homicides did not increase. However, monthly DTO-related homicides among the general population did increase 6 to 12 months after the DTO leader’s capture. Areas where law enforcement strategies arrested or killed DTO leaders showed higher rates of homicides than areas without similar law enforcement strategies. However, the study found no evidence that the capture of a DTO lieutenant was related to increased violence. Our quality assessment deemed the study ‘high’ quality. The authors provided a clear and detailed description of the research steps from the literature review to conclusions, providing sufficient information to replicate their approach.

Lindo and Padilla-Romo (2018) examined the effect of law enforcement’s arrest or killing of high-ranked DTO leaders in Mexico for each DTO between 2001 and 2010, using monthly municipality-level homicide data to measure violence. The authors only included the first DTO-leader capture in each DTO between 2001 and 2010, not subsequent leader captures for that DTO in the same period. Econometric analysis results demonstrated that capturing or killing a DTO leader contributed to a significant and sustained increase in violence in the municipalities where the capture/kill occurred and smaller but still-significant spillover effects for non-neighbouring municipalities (where the DTO leader had a presence). The authors also assessed the validity of the results by examining whether the homicide rates deviated from expected levels (based on pre-capture/kill levels). The results were not significant, supporting the conclusion that the kingpin strategy determined the change in the homicide rate. However, although violence increased in the targeted and non-neighbouring municipalities, violence decreased in neighbouring municipalities (see sections 4.2 and 4.3). Our quality assessment deemed this study ‘high’ quality. The authors provided a clear and transparent study description, a rigorous analysis and a discussion accurately reflecting the findings.

Phillips (2015) examined the relationship between arresting or killing a DTO leader and state-level violence in Mexico between 2006 and 2012, using the monthly number of DTO-related homicides to measure violence. Time-series negative binomial regression analysis demonstrated that a state’s total number of past DTO-leader arrests or killings was significantly associated with increased long-term violence (DTO-related homicides 6 to 9 months afterwards). Each additional arrest or kill was associated with a 5% increase in violence. Interestingly, this study also found evidence of decreased violence in the short term; arresting or killing a DTO leader in the previous 3 months was significantly associated with a 12% reduction in drug-related homicides. Additional stratified analyses showed that only a DTO leader’s arrest, not their killing, led to reduced violence (up to 6 months after arrest). Moreover, Phillips (2015) stratified the analysis by the DTO’s rank, showing that only removing the mid-level – not the top-level – dealer was significantly associated with reduced violence. Our quality assessment deemed this study ‘high’ quality. The author provided a transparent study description, a rigorous analysis, and a discussion that accurately reflected the findings. However, the study lacked a clear discussion of limitations.

Del Rio (2022) analysed the effects of 2 DTO-leader arrests (referred to as a ‘high-value target strike’) on DTO-related violence in Tijuana, Mexico, between 2012 and 2017. The author examined the arrest of El Ingeniero, the leader of the Arellano Felix Organization (AFO) DTO, in 2014 and of ‘El Chapo’, the leader of the Sinaloa Cartel, in 2016. The number of monthly homicides measured the violence, using intentional homicides with a firearm as a proxy for DTO-related violence. Interrupted time-series AutoRegressive Integrated Moving Average analysis demonstrated that arresting both DTO leaders was significantly associated with increased monthly homicides in Tijuana after the arrests. El Ingeniero’s capture was associated with an additional 9 homicides per month, and El Chapo’s with an extra 50 homicides per month. Our quality assessment deemed this study ‘high’ quality. The author provided a thorough and detailed study description, analysis and interpretation, including a limitations section.

Dickenson (2014) examined the relationship between government interventions to capture or kill a DTO leader with changes in violence. Violence was measured using state-level monthly homicides in Mexico between 2006 and 2010. Time-series negative binomial regression analysis results demonstrated that state-level monthly homicide rates significantly increased after removing a DTO leader (remaining stable after 12 months), showing a greater relative increase after a DTO leader’s killing than their capture. Results also indicated that the effect was more pronounced in states where the DTO was based than where the leadership removal occurred. An additional government strategy was also modelled, with the government offering a bounty (Mex$30m – equating to approximately £1.5m in 2010) for the DTO leader. Violence consequently increased but to a lesser degree. Our quality assessment deemed this study as ‘high’ quality. However, one of this review’s co-authors has previously tried to replicate Dickenson’s 2014 study but could not reproduce the results. The co-author re-analysed the findings and demonstrated a more modest increase in violence, casting doubt on the validity of the original results (Sevigny & Greathouse, 2018).

Vargas (2014) used mixed methods to analyse 2 case studies comparing the consequences of arresting leaders of 2 drug gangs in a Chicago neighbourhood in the US on violent crime. Violent crime was measured using the monthly number of homicides, aggravated batteries and aggravated assaults involving firearm use per 1,000 residents between 2006 and 2012. The author examined 2 drug gangs: ‘Latin Kings’ and ‘22 Boys’. Ordinary least squares regression through difference-in-difference modelling showed that arresting the 22 Boys’ leader was significantly associated with increased violent crime in their territory in the first month after the arrest and a significant decrease in violent crime in the second month. There were no statistically significant changes in violent crime from months 3 to 6. This study’s qualitative results indicate that violence increased after the 22 Boys’ leader’s arrest due to adjacent rival gangs attempting to occupy 22 Boys’ territory. In contrast, the results showed no evidence of an association between the arrest of the Latin Kings’ leader and violent crime in the first 6 months post-arrest. Neither the Latin Kings’ nor the 22 Boys’ leaders’ arrests affected violence in neighbouring territories or led to a long-term (6-month) change in violence rates. We classified this study as ‘high’ quality. The author described the study’s qualitative and quantitative aspects in enough detail to replicate it and undertook a thorough analysis.

Burke (2022) examined gang leadership arrests on gang-related shootings (both fatal and non-fatal) in 5 drug markets in the Westside of Chicago between 2010 and 2019 in the US. Negative binomial regression analysis results demonstrated that one arrest of a drug gang leader was significantly associated with a 20% reduction in the incidence rate of gang-related shootings (fatal and non-fatal) at 6 months. The authors examined the relationship at 3, 6, 9 and 12 months, finding that the relationship was significant at 6, 9 and 12 months (albeit with a diminishing effect over time) but not at 3 months, suggesting a delay and decay in the leadership arrest’s impact on violent incidents. Our assessment deemed the study as ‘medium’ quality. Despite its rigorous analysis, including robustness checks, the study’s discussion section did not thoroughly reflect the findings or address the study’s limitations.

3.3 Attack or removal of a member of a DTO or drug gang

Four studies found a positive association between attacking or removing a DTO member and violence; no studies found that it decreased violence.

Atuesta and Ponce (2017) examined the relationship between public security-force interventions and private violence (violence between DTOs), where public security-force interventions included killing or detaining DTO members and private violence was measured as the rate of criminal members deaths per 100,000 inhabitants in Mexico between 2006 and 2011. The authors conducted a mediation analysis, where ‘fragmentation’ (the division of larger DTOs into a greater number of smaller DTOs) explained the relationship between law enforcement and violence, demonstrating that confrontation with law enforcement (detainment and killings) significantly increased DTO member deaths. In the short term, this association was seen directly and indirectly through fragmentation (by increasing the number of DTOs). However, the results suggest that the association’s medium- and long-term effects reduce over time. Using the number of DTOs as a ‘fragmentation’ mediator suggests that while the association between law enforcement and violence is strongest in the short term, violence will increase in the long term if law enforcement increases the number of DTOs. Our assessment classified this study as ‘high’ quality. The authors provided a detailed and transparent study account, from the literature review to the conclusions.

Atun et al. (2019) examined the relationship between the Philippine National Police (PNP) anti-drug campaigns and related homicides between 2016 and 2017. The anti-drug intervention (Project Tokhang) included house visits from the police to dissuade consumers from drug-taking, random drug tests, and armed raids at houses considered sites of drug consumption or trafficking. Homicides analysed in this study included those of a violent nature where the victim had a history or alleged history of drug consumption, had drug-dealing links or was on a drug ‘watch list’, or where the homicide was linked to the anti-drug intervention. The study used police and media reports to attribute drug-related homicides to the anti-drug intervention. According to the authors, more than half the analysed drug-related homicides occurred in the context of a police operation, the details of which were reported by the media and police. The remaining homicides resulted from unknown assailants, presumed to be from anti-drug vigilante action. The study’s descriptive analysis of media documentation suggested that the drug-related homicide rate mirrored police operations during the study period. For example, drug-related homicides increased soon after Project Tokhang’s launch, decreased after its temporary suspense, and increased again following its relaunch. It is worth noting that the authors aggregated drug-consumer deaths with those of people involved in the drug trade. Our assessment classified the study as ‘medium’ quality. Although the authors clearly described the aims, research design and findings, they did not adequately document the data collection, intervention or analyses. Indeed, the study is unclear about police-conducted killings and whether they were a direct part of the intervention or an unintended consequence.

Osorio (2015) examined the escalation of drug violence (including shootings, homicides and torture) in Mexican municipalities between 2000 and 2010, following violent law enforcement tactics, including attacking or killing a DTO member. Using spatiotemporal regression analysis, Osorio (2015) examined non-violent (arrests and seizures) and violent law enforcement tactics. While both significantly increased the levels of violent events between DTOs, violent law enforcement tactics appeared to escalate violence more than non-violent ones. The results also demonstrated evidence for ‘spatial diffusion’, where spatial diffusion is the increasing conflict between DTOs migrating over to the immediate neighbouring vicinity. See sections 3.4 and 3.5 below for details on non-violent law enforcement tactics. Our quality assessment deemed this study as ‘medium’ quality. Although the author thoroughly described the interventions, analysis and discussion, the findings, limitations and methods were not clearly described or easy to follow.

Rios (2013) examined the relationship between drug-related government prosecution to reduce DTOs and targeted executions in Mexico between 2006 and 2010. The author examined casualties and arrests generated by law enforcement operations against DTOs, defined as law enforcement ‘confrontations’. Ordinary least squares regression modelling demonstrated that each confrontation between law enforcement and traffickers significantly increased the number of drug-related targeted executions by 0.98, concluding that violence increased after the commencement of law enforcement operations. However, the author did not define the specific law enforcement operations involved or separate them from conflicts between DTOs. Indeed, the explanation for the intervention (‘confrontations’) was unclear. The author noted that it was impossible to untangle different categorisations of interventions in the data (confrontation between law enforcement and DTOs versus confrontation between DTOs). Therefore, the author stated the intervention classification system was “far from perfect”. This study’s results should be interpreted cautiously since we deemed it ‘low’ quality.

3.4 Arrests or charges

Of the 4 studies examining a relationship between drug arrests/charges and violence, 2 found an association between drug arrests and increased violence. One study found an association with increased violence as well as an association with decreased violence depending on the charges analysed. One study found an association with decreased violence.

Moeller and Hesse (2013) examined the relationship between the number of charges for drug law violations and serious violence in Copenhagen, Denmark, between 2000 and 2009. They measured serious violence using the number of charges for homicide, attempted homicide and grievous bodily harm. Drug law violations were not specified further as the Danish crime statistics did not record and disaggregate the violations by offence or drug type. Fixed-effects regression analysis results indicated that drug law violations were significantly associated with increased violent offences in the subsequent year across all Danish municipalities. Our quality assessment classified the study as ‘low’ quality, as the data did not specify the type of offence within drug law violations. Moreover, the methodology and analysis sections were unclear, as were the study dates (2009 and 2010 were both mentioned) and focus area (the study uses data from Copenhagen, Denmark, but the discussion focused on a specific area, Christiania).

Osorio (2015) examined the relationship between violent and non-violent law enforcement tactics and the number of violent events between DTOs. This included shootings, homicides and torture in municipalities in Mexico between 2000 and 2010. As also described in section 3.6 below, the author demonstrated that non-violent law enforcement tactics, namely drug arrests, were significantly associated with increased violence. The study did not specify the type of offence, only that they were arrests of a DTO member. The author also demonstrated evidence for spatial diffusion (the spread of violence), where the resulting violence from the law enforcement tactics increased conflict between DTOs, which spilt over to the immediate neighbouring vicinity.

Buggs (2018) examined the relationship between drug-related law enforcement interventions and gun violence in Baltimore, US, between 2003 and 2017. In this case, the interventions included drug possession and trafficking arrests and major drug busts. A drug bust was defined as ‘major’ if 5 or more individuals were arrested, one or more individuals faced federal charges or the charges were drug gang-related. Time-series negative binomial regression analysis results indicated that drug possession arrests were significantly associated with a “slight” increase in non-fatal shootings during the following month. However, when the author applied a spatial lag to examine neighbouring police precincts, drug possession arrests were associated with decreased non-fatal shootings in the main precinct the following month. Moreover, drug trafficking arrests were associated with increased non-fatal shootings 3 and 5 months after the arrests. The author concludes that these drug law enforcement interventions (drug possession and trafficking arrests and major drug busts) did not reduce gun violence. Our quality assessment classified the study as ‘high’ quality – the only reviewed PhD thesis we classed as such. The author provided a thorough study account detailing the aims and methods, and the conclusions accurately reflected the findings.

Burke (2022) examined the relationship between search warrants for illegal drugs and firearms and arrests for gang members and the rate of gang-related mass shooting events (per 1,000 residents) in districts in Chicago, US, with drug markets between 2010 and 2020. Using linear fixed-effects models, findings demonstrated a significant association between the use of search warrants and a reduction in gang-related mass shootings, as well as a significant association between arrests and a reduction in gang-related mass shootings. Our quality assessment classified the study as ‘high’ quality. The author provided a thorough study account detailing the aims and methods, and the conclusions accurately reflected the findings.

3.5 Crop eradication

The available evidence suggests a strong association between crop eradication and increased violence.

Of 6 identified studies examining the relationship between eradicating illegal crops (for example, coca) and violence, 5 found an association between crop eradication and increased violence (Acevedo, 2015; Degiovanni, 2013; Grisaffi et al., 2021; Marin Llanes, 2022; Muniz-Sanchez et al., 2022), although Grisaffi et al. (2021) also reported decreases in violence. Only one other study (Fisher & Meitus, 2016) found an association between crop eradication and decreased violence.

Two studies evaluated ‘Plan Colombia’, a collection of interventions funded by the Colombian and US Governments to tackle the cultivation, processing and distribution of illegal drugs in Colombia. Plan Colombia intended to weaken DTOs by removing their main source of finance. Both studies evaluated the eradication of illegal coca crops within Plan Colombia (Acevedo, 2015; Degiovanni, 2013). Acevedo (2015) conducted an econometric evaluation to examine the crop eradication programme’s short- and long-term effects on violence, where violence was measured by attacks between guerrillas, clashes with guerrillas and government, as well as civilian and combatant deaths at the municipality level between 1999 and 2005. Results demonstrated that the municipalities experiencing coca eradication through Plan Colombia were significantly associated with increased violence in the long term. An increase of 1% of the total municipality area (in hectares) experiencing coca eradication was associated with a 22% increase in guerrilla attacks, a 24% increase in clashes between government and guerrillas, a 22% increase in combatant deaths, and a 16% increase in civilian deaths. In the short term (30 days), municipalities experiencing coca eradication through Plan Colombia were also associated with increased violence; an increase of 1% of the total municipality area (in hectares) experiencing coca eradication was associated with a 23% increase in civilian casualties. However, the authors did not find significant effects of the coca spraying on guerrilla attacks, clashes, or combatant causalities post-eradication compared to pre-eradication. We classified this study as ‘medium’ quality in our assessment. The study provided a detailed account of the interventions, findings and discussion; however, the authors were not transparent on the methods and analysis nor the study’s limitations.

Degiovanni (2013) examined the effect of Plan Colombia on violence at the municipality level in Colombia between 1999 and 2010. Plan Colombia examined the effects of crop eradication, destroyed narcotics labs, and law enforcement attacks, measuring violence by the homicide rate per 100,000 inhabitants. Spatial economic modelling results indicated that coca eradication was significantly associated with increased violence in municipalities where the coca eradication took place. Authors demonstrated that drug enforcement had first- and second-order effects on violence, where first-order effects stem directly from the government and DTOs’ expenditures on the drug war. Second-order effects emerge from the DTOs’ retaliation against government anti-drug activities. The homicide rate increased by 1% from first-order effects and 4% from second-order effects. We classed this study as ‘low’ quality in our assessment: the author did not provide a clear account of data collection or a discussion of the limitations, and the findings were difficult to interpret.

Marin Llanes (2022) examined the relationship between the implementation of the National Comprehensive Program for the Substitution of Illicit Crops (PNIS) in Colombian municipalities and the killing of social leaders (local activists who represented the interests of their local communities), 2005 to 2019. The PNIS was a substitution programme (substituting illegal crops with legal crops) with an agreement between the Colombian Government, the Revolutionary Armed Forces and the community. Using difference-in-difference modelling, authors demonstrated a significantly higher rate (a 481% increase) and probability (that is, how likely something is to happen, a 122% increase) of social leader homicides in municipalities with the PNIS program compared to those without between 2005 and 2019. We classed this study as ‘high’ quality in our assessment. The author provided a clear and detailed study description sufficient for replication.

Muniz-Sanchez et al. (2022) examined the relationship between the eradication of poppy and cannabis crops and violence in north-Mexican municipalities, reporting the highest number of homicides where the victim or offender was a DTO member (defined by authors as ‘executions’) between 2006 and 2011. Bayesian hierarchical space-time model analysis showed that eradicating illegal crops was significantly associated with increased violence in northern Mexico. Moreover, the effect was stronger with poppy-crop eradication than with cannabis crops. The authors suggested the difference was due to the higher price of opium gum and that DTOs trading in poppy crops were generally more violent than those trading in cannabis. We classed this study as ‘high’ quality in our assessment. The authors provided a thorough and rigorous analysis of the intervention and a discussion that reflected their findings.

Grisaffi et al. (2021) examined the relationship between coca control and development strategies (including crop eradication) and violence in Bolivia and Peru. However, the authors did not specify the period of reference. Qualitative analysis (type unspecified) showed that after introducing a non-coercive model to limit coca production and transition growers to another crop, Bolivia experienced a reduction in violence. In Peru, the authors reported that there was an increase in violence and instability after the introduction of a forced crop eradication of coca. This took the form of cycles of forced eradication, coca replanting and violence. We classed the study as ‘low’ quality in our assessment. The study had several shortcomings: the authors mentioned that interviews, focus groups and other qualitative research methods were conducted but did not detail data collection, recruitment or analysis. Moreover, the study’s location and period were unclear. The authors mention some specific areas in Bolivia, but whether their analysis is exclusively focused on those is unclear. Therefore, the results of this study should be interpreted with caution.

Fisher and Meitus (2016) examined the relationship between the aerial eradication of illegal crops by the Uribe government in 2004 and 2005 and guerrilla violence (kidnappings, terrorism, assassinations) at the department level in Colombia. Negative binomial fixed-effects regression analysis demonstrated that an increase in aerial eradication was significantly associated with reducing total violence in Colombia the following month. However, when the authors analysed each type of violence separately, there was no longer a significant association between eradication and assassination or terrorism, only kidnappings. We classed this study as ‘medium’ quality in our assessment. The authors provided a thorough description of their analysis and clearly discussed their findings; however, the methods section lacked enough detail to be replicable.

3.6 Drug seizures

Data from Mexico points to increased violence, though evidence suggests a slight decrease depending on the type of seizure.

The 3 studies examining drug seizures in Mexico found an association with increased violence (Calderon, 2012; Castillo et al., 2020; Osorio, 2015). However, examining the relationship between 6 types of municipal-level drug seizures and executions (measured as DTO-related homicides) in Mexico between 2007 and 2010, Calderon also found evidence of reduced violence in the same study associated with particular types of drug seizures/executions. Poisson regression analysis results demonstrated a significant increase in homicides after cocaine and maritime vehicle seizures. The author also reported a significant reduction in homicides after seizures of weapons, land vehicles, aerial vehicles, heroin, opium gum and opium labs. However, the decrease in homicides was negligible. Our quality assessment classed this study as ‘low’ quality because it was a dissertation section that was neither clear nor transparent about the intervention or analysis. It also gave no clear statement of findings or in-depth discussion reflecting them. Other analyses included in Calderon (2012) were captured in another study in this review (Calderon et al., 2015). Note that we classed Calderon (2012) as a ‘low’ quality study but Calderon et al. (2015) as a ‘high’ quality study; the quality difference is likely due to the academic peer review process and iterations between 2012 and 2015 to strengthen the article.

Osorio (2015) examined the relationship between violent and non-violent law enforcement tactics – including drug confiscation – and the number of violent events between DTOs, including shootings, homicides and torture in Mexican municipalities between 2000 and 2010. Using spatiotemporal regression analysis, the author examined non-violent and violent law enforcement tactics. The non-violent law enforcement tactics, namely confiscating illegal drugs, were significantly associated with increased violence. The author also demonstrated evidence for spatial diffusion (the spread of violence), where the resulting violence from confiscating illegal drugs increased conflict between DTOs, spilling over to the immediate neighbouring vicinity.

Castillo et al. (2020) examined the relationship between scarcity and violence in Mexico between 2006 and 2010 using Colombian authorities’ cocaine seizures to measure scarcity and monthly homicides per 100,000 inhabitants to measure violence. Poisson fixed-effects regression modelling results demonstrated that cocaine shortages (measured as a function of drug seizures) were significantly associated with increased homicides in Mexico. The effect of cocaine seizures on violence varied depending on the municipalities’ distance from the US border. The association was strongest in those closest to the US border and weaker towards south Mexico, where seizures had no association with violence. The authors also explored DTOs as a variable in the analysis, finding that cocaine seizures had a more pronounced effect on violence when 2 or more DTOs were present in the municipality. We classed this study as ‘high’ quality in our assessment, as the authors’ clear and detailed description and rigorous analyses were sufficient for replication.

3.7 Military intervention

All studies examining a relationship between anti-drug military interventions and violence found an association between military interventions and increased violence. However, one study found mixed results when examining individual regions in Mexico.

Martinez and Phillips (2022) examined the relationship between military interventions (defined by the authors as confrontations between DTO members and the Mexican army) and DTO-related violence at the municipality level in Mexico between 2007 and 2011. The authors used the monthly homicide rate of males aged 15 to 29 to measure DTO-related violence. Mixed-effects linear regression analysis results demonstrated that a single additional military intervention significantly increased the monthly homicide rate by 19%. However, controlling for the number of DTOs in the analysis attenuated the increase, suggesting that DTO competition was a key driver of violence alongside military interventions. We classed this study as ‘high’ quality in our assessment, as the authors described the study and analyses sufficiently to replicate and conducted rigorous analyses.

Espinosa and Rubin (2015) examined the relationship between military interventions and violence in Mexico between 2006 and 2010. They defined military interventions as confrontations between DTO members and the Mexican army that resulted in at least 3 deaths and measured violence using the monthly homicide rate per 100,000 inhabitants. Using Rubin’s Causal Model analysis and propensity score matching, the authors compared municipality regions that received a military intervention to regions that did not. The authors demonstrated that military interventions were associated with increased violence overall. However, examining individual regions revealed variation: 6 regions showed increased violence; 2 showed reduced violence; and 10 showed no significant change. The authors note that their analysis only assesses violence one year post-intervention and that the results may change after a period of stability between the military and the DTO members. We classed this study as ‘medium’ quality in our assessment, as the authors presented a thorough study, including detailed and rigorous analyses; however, the discussion lacked sufficient interpretation of the findings.

Flores-Macias (2018) examined the relationship between anti-drug military operations and state-level homicide rates in Mexico (no specific calendar period indicated). The anti-drug operations included road checkpoints, established military bases in DTO territories, aircraft deployment and vehicle patrols. Difference-in-difference modelling using synthetic controls suggested that violence was steadily decreasing before the drastic increase in militarised operations under Calderón’s presidency began in 2006. Once anti-drug military operations began, violent crime increased twice as fast in Mexican states with anti-drug militarisation than in states without. Our quality assessment classed this study as ‘low’ quality since the author did not include sufficient detail in the methods section. Though suggesting they used subnational data for state capacity, homicide and kidnapping rates, they did not describe the data sources. The results of this study should therefore be interpreted cautiously.

3.8 Multi-jurisdictional anti-drug coordination and resourcing

One study examined the relationship between federal funding for multi-jurisdictional coordination and resourcing for anti-drug policing initiatives on the one hand and violent crime on the other. It found no evidence of a reduction in violence.

Cox and Cunningham (2021) examined the relationship between the Edward Byrne Memorial State and Local Law Enforcement Assistance Grant Program (EBMGP) and city-level violent crime in the US between 1982 and 2009. The EBMGP was allocated to state and local government units, agencies and organisations to tackle drug consumption and drug-related violent crime. A large proportion of the EBMGP funding was used to create multi-jurisdictional drug task force units to prosecute, detain and rehabilitate drug offenders. This study measured violent crime using aggregate counts of murder, manslaughter, rape, assault and robbery, comparing outcomes in cities that received EBMGP to cities that did not. Linear regression analysis through difference-in-difference modelling indicated no evidence that the EBMGP influenced total violent crime in recipient cities compared to non-recipients. The EBMGP was associated with reduced robberies when violent crime was disaggregated; however, there was no association between EBMGP and rates of reported assaults. Our quality assessment deemed this study as ‘high’ quality. The authors provided a clear and detailed study description, rigorous analyses and a discussion of limitations.

4. Features of drug-related law enforcement activity contributing to their effectiveness in reducing violence

Key messages:

• the authors theorised that drug-related law enforcement activity aspects that increased an activity’s effectiveness in reducing serious violence included disrupting the drug supply (either through organisational or economic disruption), building trust between the government and the community and focused deterrence
• some studies showed no change in violence; suggested reasons for violence levels remaining the same after drug-related interventions include drug markets’ resilience, the cultural significance of violence in organised crime groups and limited police funding
• suggested factors contributing to increasing serious violence included DTOs’ fragmentation and destabilisation, increased competition, retaliation, distrust of the government, and DTOs punishing locals for collaborating with the government

Based on the reviewed literature, this chapter focuses on understanding what factors might contribute to the varying effectiveness of drug-related law enforcement activities in reducing serious violence and crime. The findings address the following research questions:

RQ 2: What aspects of drug-related law enforcement activity make them more or less likely to be effective in reducing serious violence and homicide?

RQ 4: When drug-related law enforcement activity generates adverse serious violence and homicide outcomes, what are the factors that lead to these unintended consequences?

4.1 Aspects of drug-related law enforcement activities that may increase their effectiveness in reducing serious violence and homicide

The following sections highlight aspects of drug-related law enforcement activities that authors have suggested may increase their effectiveness in reducing serious violence and homicide based on the available evidence. These theories were not empirically tested but discussed as potential explanations for the decreased violence demonstrated in the reviewed studies.

4.1.1 Interventions reduced violence by disrupting the drug supply through organisational or economic disruption

Three studies theorised that the reduced violence demonstrated in their respective analyses was due to drug supply disruption (Burke, 2022; Fisher & Meitus, 2016; Phillips 2015). For example, Burke examined leadership removal in 5 drug markets in Chicago, theorising that the reduction in fatal and non-fatal drug-related gang shootings after a leader’s arrest was because it cut the local drug supply’s source. The author argued that the drug gang could no longer meet demand, thus unable to fight for an increased market share, creating organisational and economic disruption. Nevertheless, other studies hypothesised and/or demonstrated cases where drug supply disruption increased violence (see Castillo et al., 2020, section 4.3).

4.1.2 Interventions reduced violence by deterring offenders from re-offending

Three reviewed studies evaluated the US ‘pulling levers’ intervention (Corsaro et al., 2012; Corsaro, 2013; Corsaro & Brunson, 2012). The ‘pulling levers’ intervention derives from deterrence theory, which assumes that humans are rational and will adapt their current behaviour when presented with the consequences of their potential actions. For example, Corsaro et al. (2012) theorised that the success of focused deterrence principles could explain the reduced violence in High Point, North Carolina (US) after implementing ‘pulling levers’. In this interpretation, the threat of criminal sanctions changes behaviour, and the criminal justice system effectively produces these threats. However, although ‘pulling levers’ appeared successful in targeting high-risk offenders, the authors noted that a large proportion of crime is committed by those with less extensive criminal histories (that is, not deemed ‘high-risk’, thus not targeted by the intervention). This caveat potentially limits its effectiveness.

4.1.3 Interventions reduced violence by building trust between the government and community members

Two qualitative studies dived deeper into the relationships between law enforcement and the local communities that interventions targeted (Grisaffi et al., 2021; Israel de Souza 2019). Both studies concluded that reduced violence following drug-related law enforcement activities was due to building trust and relationships between those implementing the intervention and those in the community. For example, Israel de Souza conducted an ethnographic study of the pacification intervention in Rio de Janeiro’s favelas, proposing that the Pacification Police officers settled into the community when they arrived in the favela, posing a minimal threat to others and subject to minimal threat themselves. It reportedly took time for the police to settle into the community before the author arrived in the favelas, but their presence appears to have deterred violence once they did. However, violence resumed once pacification disbanded. Thus, the author concluded that the pacification effects did not last because the police were only stationed in the favelas and thus had limited impacts on broader drug-related criminal networks.

Grisaffi et al. described a similar trust-building intervention examining Bolivia’s crop eradication/substitution programme. The author concluded that Bolivia’s success in reducing violence and limiting coca production was due to participatory development in drug crop regions via land titling and state investment. This participatory approach put local knowledge and engagement at the heart of the strategy and without the coercion of Peru’s programme. However, Bolivia’s programme went beyond coca elimination and addressed the causes of coca cultivation (for example, poverty), going beyond trust building.

4.2 Challenges ensuring effectiveness in reducing serious violence and homicide

This section highlights aspects of drug-related law enforcement activity and drug markets theorised to explain the absence of any change in violence or displacement due to drug-related law enforcement activities. These theories were not empirically tested but discussed as potential explanations for the authors’ findings.

4.2.1 Interventions did not have a measurable impact on serious violence due to the resilience of drug markets and the cultural significance of violence in organised crime groups

Two studies examining selective enforcement found no change in violence levels after their respective interventions. Both theorised the lack of evidence was due to the drug markets’ resilience or the violence used within them. For example, Klement and Blokland (2021) examined the Netherlands’ ‘whole of government’ intervention in reducing violence committed by OMCGs and their support members. The authors proposed that the cultural relevance of violence within OMCGs may explain the absence of any evident association. OMCGs were said to be synonymous with violence – part of a cultural group identity that cannot easily be changed.

4.2.2 One intervention had no measurable impact on serious violence or homicide due to limited police funding

One study examined multi-jurisdictional anti-drug coordination and resourcing through the Edward Byrne Memorial State and Local Law Enforcement Assistance Program (EBGMP) and found a non-significant relationship between EBGMP implementation and violent crime (Cox & Cunningham, 2021). The authors proposed that police funding in the cities with/without the EBGMP might explain the absence of an association, suggesting that the concomitant funding may have been only a modest proportion of the overall police budget and thus insufficient to make a significant difference. Moreover, the authors suggested that other funding may have been distributed to cities that did not receive the EBGMP, effectively evening out the resources for tackling violent crime.

4.2.3 Interventions reduced violence in certain areas by displacing the violence in other areas

Four studies theorised that the reduced violence demonstrated in their respective analyses was due to its displacement rather than overall reduction. For example, Lindo and Padilla-Romo (2018) demonstrated that municipalities neighbouring the area where a leadership removal occurred experienced reduced violence. They suggested this was because the leader’s municipality draws violent activity from neighbouring areas. Three other studies exploring drug seizures, drug arrests and selective enforcement also described violent-crime displacement from one area to another (Buggs, 2018; Calderon, 2012; Tealde, 2019). Calderon’s 2012 study, examining the relationship between violence and different types of drug seizures, theorised the lack of change was because DTOs find it more profitable to avoid government interventions by relocating. Thus, the only type of drug seizures associated with increased violence were those where the DTOs could not relocate, that is, cocaine seizures from specific seaports in Mexico.

4.3 Factors reported to increase serious violence and homicide after implementing a drug-related law enforcement activity

The following section highlights aspects of drug-related law enforcement activity theorised to contribute to increasing serious violence and homicide. These theories were not empirically tested but discussed as potential explanations for the increased violence demonstrated in the authors’ studies.

4.3.1 Interventions increased violence by fragmenting DTOs and creating more competition

Some have theorised that breaking a DTO’s monopolistic control would reduce violence (see section 4.1). However, 4 studies conducted in Mexico demonstrated that attacking/removing a DTO leader or member was associated with increased violence, concluding that the DTOs’ fragmentation could explain the increase (Atuesta & Ponce, 2017; Osorio, 2015; Phillips, 2015; Rios, 2013). Atuesta and Ponce measured fragmentation based on the number of DTOs, concluding from their analysis that an increased number of DTOs after a DTO member is detained or killed heightens the competition between them, potentially explaining the increase in violence. Osorio (2015), Phillips (2015) and Rios (2013) proposed the same theory, describing the increased violence after a DTO member or leader’s removal as being due to the DTOs’ fragmentation.

Three other studies examining interventions outside of removing a DTO leader or member also theorised that DTO competition could explain increased violence. Castillo et al. (2020) examined cocaine seizures and violence, theorising that additional money flow into the cocaine trade through scarcity (and therefore increased prices) would trigger violence between DTOs, especially in areas important for drug trafficking. Martinez and Phillips (2022) examined military interventions and violence, finding that DTO competition (measured by the number of DTOs) was a significant variable in their analysis, suggesting that DTO competition was a key driver of violence. Similarly, Moeller and Hesse examined increased violence after drug law violations in Copenhagen, Denmark, proposing that increased competition between criminal groups and conflicts settled through violence could explain the increased violence following law enforcement activities.

4.3.2 Interventions increased violence by disorganising and destabilising DTOs or drug gangs

Other studies showed that organisational and economic disruption of DTOs or drug gangs could reduce violence (discussed in section 4.1). However, 5 studies found increased violence following a DTO or gang leader’s arrest or killing due to that group’s, or drug markets’, subsequent disorganisation/destabilisation (Calderon et al., 2015; Dickenson, 2014; Lawrence, 2023; Lindo & Padilla-Romo, 2018; Vargas, 2014). For example, Vargas examined violence after 2 gang leaders’ removal in Chicago, US, concluding that the spike in violence after the 22 Boys’ leadership arrest was due to the successor’s disorganisation, triggering violent competition within 22 Boys. Conversely, there was no change in violence after the Latin King’s leader’s arrest because the structured organisation ensured immediate replacement. Lindo and Padilla-Romo examined the effect of leadership removal in the municipalities where it occurred, in neighbouring municipalities, and in non-neighbouring municipalities, where the DTO had a presence. Violence increased in the area the leader was removed from and in non-neighbouring areas with a DTO presence. This finding demonstrates that leadership removal can destabilise a DTO in other areas, not just those where removal occurs. Moreover, Lawrence examined violence after the implementation of a police bicycle unit. Violent person crimes and shootings were significantly higher 2 years post-intervention, theorised by the author to be due to the disruption of the drug markets in the treated areas (that is, 3 drugs markets in Philadelphia).

Del Rio (2022) described a similar association, reporting that the arrest of 2 DTO leaders in Tijuana, Mexico, revealed the DTOs’ weakness and invited violence within and between DTOs to find a successor. However, Del Rio compared El Chapo’s 2016 arrest to his earlier 2014 arrest, concluding that the latter did not lead to violence because El Chapo’s co-founder succeeded him. Since the co-founder was considered an equal, the DTO was not perceived as weak. All studies that reported increased violence after leadership removal – particularly after a leader’s killing rather than arrest – attributed the destabilisation to competition for leadership, attempts to prove qualification for leadership or efforts to protect strategic territories from other DTOs (Calderon et al., 2015; Del Rio, 2022; Dickenson, 2014; Lindo & Padilla-Romo, 2018; Vargas, 2014). Interestingly, multiple studies demonstrated that the increase in violence was greater when a DTO leader was killed rather than arrested. The authors suggested this might be because leaders can return after detention or continue offending while in prison.

4.3.3 Interventions increased ‘disciplinary’ violence as DTOs looked to punish locals for collaborating with the government

Two studies demonstrated increased violence after crop eradication programme implementation, theorising that DTOs punishing community members for collaborating with law enforcement/government officials might be one reason for the increase (Acevado, 2015; Marin Llanes, 2022). For example, Marin Llanes examined the impact of crop eradication on the killing of social leaders in Colombia, concluding that social leaders represent ‘peace-building’ exercises between communities and law enforcement. The study’s results suggested that armed groups or DTOs oppose the expansion of crop eradication programmes and thus remove social leaders in their territories. Acevado concluded that because the government did not respond to guerrilla violence after crop eradication in the short term, the violence primarily focused on civilians. The authors hypothesised that DTOs punished farmers for collaborating with the government.

4.3.4 Interventions may have increased violence by creating further conflict between DTOs and the government to retaliate or retain territory

Four studies demonstrating increased violence after a drug-related law enforcement activity proposed that retaliation and conflict with the government or law enforcement conducting the intervention explained the increased violence (Acevado, 2015; Calderon, 2012; Degiovanni, 2013; Flores-Macias, 2018). Two studies evaluating crop eradication interventions concluded that while coca eradication initially weakens the guerrillas/DTOs by destroying their product and thus their source of income, they use violent conflict and retaliation against the government to regain coca fields and control (Acevado, 2015; Degiovanni, 2013). Flores-Macias evaluated anti-drug military operations, theorising that the increased violence seen after implementing these military operations could be explained by the DTOs adjusting their violence to retaliate accordingly. As the military is likely to use greater violence/force than the police or governments, the DTOs must increase and match their violence to retain their territory and survive.

4.3.5 Interventions may have increased violence in illegal crop territories or areas important for drug trafficking by disrupting supply chains

Castillo et al. (2020) demonstrated increased violence in Mexican municipalities near the US border after drug supply disruption through cocaine seizures. This finding contrasts with the studies described in section 4.1, which theorised that drug supply disruption decreased violence (Burke, 2022; Fisher & Meitus, 2016; Phillips, 2015). Castillo et al. proposed that the reduced supply following drug seizures would increase the cost of cocaine and thus DTOs revenues, assuming that cocaine is an inelastic good (where demand is static after a price change). DTOs would thus use violence to compete for and secure the increased revenue, especially in the areas important for trafficking, such as municipalities near the US border. The authors also theorised that if the DTO leaders do not oversee seizures, they could assume the shortage is due to DTO member error and create violence through punishment. Another study examining crop eradication and violence theorised that eradication causes product- and future-income loss, weakening a DTO against its rivals (Muniz-Sanchez et al., 2022). The authors theorised that DTOs re-impose their power through violence to avoid losing their territory. The authors also demonstrated that the increase in violence was more significant after poppy-crop eradication than after cannabis-crop eradication. The authors suggested this might be due to several factors, including the higher price of opium gum, the more violent nature of the DTOs trading poppies and the greater risk of growing poppies, generating more conflict in retaining territory and power.

4.3.6 One intervention may have increased violence by generating distrust in the government

Grisaffi et al. (2021) conducted 2 case studies comparing crop eradication/substitution programs in Bolivia and Peru, demonstrating that violence increased after US-backed crop eradication in Peru. The authors concluded that crop eradication in Peru was erratic and characterised by deep distrust between the farmers and the government. Consequently, the farmers defended their land from the government rather than defending the coca for financial gain.

4.3.7 One intervention may have increased violence through direct police action

One study examined the Philippine National Police anti-drug campaign and the resulting increase in deaths, concluding that the police were responsible for over half of the drug-related murders in the analysed period (Atun et al., 2019). The authors argued that many homicides resulting from police operations occurred because the police did not comply with operational project rules. Many homicides occurred in private homes, which the police should not enter without a warrant. The authors proposed that if compliance were improved, the number of deaths caused by the police would have been lower.

5. Cross- and within-country differences in drug-related law enforcement activities and violence and homicide

Key messages:

• most studies identified from the US evaluated selective enforcement of serious violence and homicide
• all studies identified from Mexico evaluated a national strategy to fight DTOs
• most studies from South America evaluated national crop eradication programmes
• neither of the 2 European studies demonstrated drug-related law enforcement activities’ effectiveness in reducing violence
• in the study from the Philippines, violence appears directly linked to the intervention, that is, the police conducted the violence

This chapter focuses on understanding how serious violence and homicide outcomes differ by country and local areas in the reviewed literature. The findings address the following research question:

RQ 5: How do serious violence and homicide outcomes relating to drug-related law enforcement activity, differ by country and local areas and why?

As Chapter 2 highlighted, most studies included in our review were based in North America, particularly Mexico and the US. The remainder were based in South America (including Colombia, Bolivia, Peru and Brazil) and, to a lesser extent, in Europe (the Netherlands and Denmark) and Asia (the Philippines). Although the studies were located in multiple countries, all were published in English.

5.1 Most identified studies from the US evaluated selective enforcement

Most of the studies in the US evaluated the effectiveness of selective enforcement as a drug-related law enforcement activity (Corsaro & Brunson, 2013; Corsaro et al., 2012; Corsaro, 2013; Lawrence, 2023). Compared to studies elsewhere, those in the US predominantly found evidence that interventions were either associated with reducing serious violence and homicide or with no change in violence. In contrast to most of those in Mexico and South America, the included US studies assessed local-level interventions conducted in cities or neighbourhoods.

5.2 All identified studies from Mexico evaluated a national strategy against organised crime

All included studies from Mexico evaluated interventions from a national strategy to fight organised crime. While the studies examined the impacts of municipality- or state-level activities, the interventions were national. Most studies in Mexico examined the removal or attack of a DTO leader or member as a drug-related law enforcement activity (Atuesta & Ponce, 2017; Calderon et al., 2015; Del Rio, 2022; Dickenson, 2014; Lindo & Padilla-Romo, 2018; Osorio, 2015; Phillips, 2015). Studies examining DTO leader/member removal found mixed evidence of effectiveness. These included increases, decreases or no changes in violence depending on the status of the removed DTO member, whether it comprised their arrest or killing, whether the study examined spillover effects in neighbouring areas, and the length of the period studied. A leadership-removal strategy’s success or failure appears complex, depending on more than simply ‘removing’ a DTO or gang leader to reduce violence.

5.3 Most identified studies from South America evaluated national crop eradication programmes

All Colombian studies evaluated federal crop eradication programmes at the department (similar to province/state) or municipality level. There was little evidence that crop eradication reduced violence. One study found an association between crop eradication and reduced violence. However, once the violence was disaggregated by ‘assassinations’, ‘terrorism’ and ‘kidnappings’, only ‘kidnappings’ significantly reduced, driving the reduction in ‘total’ violence (Fisher & Meitus, 2016). One Bolivian study reported reduced violence through community-based methods, crop substitution and addressing farmers’ underlying issues (for example, poverty). However, the authors were unclear about how they reached this conclusion using their qualitative methodology; therefore, we deemed the study’s quality ‘low’. The remaining studies on crop eradication found that interventions increased violence via DTO conflict with the government and/or civilians or between DTOs to regain territory.

5.4 Neither European study demonstrated drug-related law enforcement activities’ effectiveness in reducing violence

The 2 European studies in our review reported that drug-related law enforcement activities were ineffective in reducing serious violence or homicide (Klement & Blokland, 2021; Moeller & Hesse, 2013). The study in the Netherlands did not find evidence of a change in violence after implementing a whole-government zero-tolerance approach to OMCGs. However, since this approach included multiple interventions, it is unfeasible to disentangle which interventions changed violence outcomes, and which did not, based on the information in the reviewed study. Arrests for drug law violations were associated with increased violent offences the following year across all municipalities in Copenhagen between 2000 and 2009 (Moeller & Hesse, 2013), mirroring the results of a study examining drug arrests in Mexico (Osorio, 2015). Both studies suggested increased violence due to the disruptive effect on DTO/gangs, which heightened competition.

5.5 Violence in the study from the Philippines appears directly linked to the intervention itself

Only one identified study addressed drug-related law enforcement activity in Asia, evaluating a broad drug-related Philippine National Police intervention targeting residents consuming drugs and involved in the drug trade. However, the authors highlighted the data limitations for this analysis, suggesting that the government did not produce reliable reports. This study also differed from all other studies in the review because the intervention involved killing drug traffickers, even if this was not officially stated or acknowledged. Moreover, the intervention also targeted those consuming drugs rather than just targeting drug traffickers.

6. Reflections and implications

In this systematic review, we sought to understand the impact of drug-related law enforcement activity on serious violence and homicide. This report describes the evidence on the effectiveness of 8 different types of law enforcement activity across the globe, exploring the aspects that may increase the effectiveness in reducing violence and differences between and within countries regarding the interventions and their outcomes. However, this review only evaluated drug-related law enforcement activities on serious violence and homicide; it did not focus on interventions not specifically drug-related or those pursuing other aims. Thus, this review does not include other drug-related law enforcement activities’ aims, such as reducing drug use, reducing non-violent crimes and reducing supply. Likewise, it does not include other unintended consequences of drug-related law enforcement activities, such as environmental damage from crop eradication.

The available evidence suggests that drug-related law enforcement activities are of limited effectiveness in reducing violence. Indeed, more studies demonstrated an association between drug-related law enforcement activities and increased violence than decreased violence. These findings broadly reflect those of the earlier systematic review (Werb et al., 2011), which found that increasing drug law enforcement was unlikely to reduce drug market violence alone and risked exacerbating it. Of the 8 types of law enforcement activity identified in this review, selective enforcement tactics implemented locally appeared the most promising in reducing violence, although the evidence base is limited.

Passive drug-related law enforcement activities appear promising, but there is less available evidence on these interventions’ effectiveness than active law enforcement activities. Studies investigating passive interventions, such as distributing warnings to high-risk offenders or increasing police presence in known drug market areas, demonstrated reduced violence (Corsaro et al., 2012; Corsaro, 2013; Israel de Souza 2019). However, the studies in this review predominantly involved active interventions (for example, arrests, killings and illegal crop eradication), perhaps due to the dominance of studies conducted in Mexico during President Calderón’s administration, which implemented an aggressive policy against DTOs. Passive interventions in the selective enforcement category showed promising results. Therefore, it would be beneficial to evaluate such interventions – such as increasing police presence, distributing warnings to high-risk offenders (and issuing charges on re-offence) or building trust between police and community members – in different settings, including at the local level in the UK.

The causal mechanisms for violence reduction are under-explored in the literature. However, possible contributing factors discussed in several studies included supply disruptions, focused deterrence and positive relationships between police and communities. Studies also flagged the displacement of violence (rather than a reduction in overall level) as a potential caveat to promising findings. Barriers to violence-reduction efforts included drug markets’ resilience, the cultural significance of violence in some DTOs, and limited law enforcement resources.

We did not identify any UK-based evidence in this review. As with the previous systematic review on this topic (Werb et al., 2011), no included studies were based in the UK. The studies on arrests and charges conducted in Western settings may be the most transferrable to the UK. The evidence from arrests and charges in Europe and the US demonstrated either no change in violence or an association with increased violence. The evidence from selective enforcement in the US demonstrated primarily positive results in reducing violence, with one demonstrating no change or an increase in violence depending on the outcome. However, the law enforcement activities more commonly used in the UK had less evidence in the current review, that is, arrests and seizures were underrepresented compared to leadership removal or illegal crop eradication.

Likewise, some law enforcement activities evaluated in this literature have limited applicability to the UK. For example, we can draw lessons from studies examining illegal crop eradication where forced aerial crop eradication appears to increase violence rather than reduce it. However, since coca (about which most of the evidence relates) is not grown in the UK, evidence from coca eradication may not be directly transferable to crops more likely found in the UK. Similarly, we can draw lessons from studies examining drug seizures suggesting that cocaine scarcity increases violence. Indeed, cocaine scarcity in the UK could raise the price (assuming no compensatory reduction of purity), and thus rapacity among gangs may trigger violence. However, the studies in this review examined drug seizures in producer countries (for example, Colombia) with violence in trafficking countries (for example, Mexico). Since the UK is a consumer country, this mechanism may not translate.

There is a lack of evidence on the relationship between drug-related law enforcement and serious violence and homicide over the last decade. The current review includes studies published between 2011 and early 2024. However, the studies predominantly analysed data up to 2012. Only 8 of the 33 studies in this review analysed data from the last decade. Moreover, what was effective (or ineffective) in reducing violence then may have different results now. To illustrate, the studies in Mexico analysed data during President Calderón’s time (between 2006 and 2012). Under President Calderón, the Mexican Government began an aggressive policy towards DTOs, including increasing military deployment and focusing on leadership removal (rather than, for example, illegal crop eradication or drug seizures). The impact of leadership removal outside the context of President Calderón’s administration may provide different results. Indeed, Mexico’s current president (since 2018) reportedly no longer prioritises leadership removal as a strategy (Quackenbush, 2019).

The last decade has also seen substantial changes in the UK drug trade, affecting which drug-related interventions may successfully reduce local violence. For example, the types of drugs on offer have changed since 2012. The UK has seen the introduction of ‘spice’ and other new psychoactive substances (NPS) (Stephenson & Richardson, 2014). While the Psychoactive Substances Act 2016 banned all psychoactive substances (except those with exemptions, for example, alcohol), spice and other NPSs are still a concern in specific communities, such as prisons. Moreover, buying and selling drugs has changed in the past 10 years. Social media and darknet marketplaces are now commonly used to promote and purchase drugs. While there is research examining the use of darknet marketplaces and violence (for example, Barratt et al., 2016), we identified no studies examining the effect of online drug-related law enforcement activities on serious violence and homicide.

More evidence is needed on the effectiveness of drug-related law enforcement activities in retail-level markets or prison settings in reducing violence. Most studies included in this review analysed the impact of drug-related law enforcement activities on DTOs or organised crime groups. Evidence on the impact of drug-related law enforcement activities on lower-level drug gangs, such as those involved in County Lines at the retail level or in prison settings, is lacking. These efforts must be supported by evaluative research to better understand where and how to focus police resources and reduce violence while avoiding or minimising unintended consequences. Relevant agencies planning and implementing drug-related law enforcement activities for County Lines or prison settings should consider the risk of increased violence, particularly for those interventions for which available evidence suggests a strong association (for example, leadership removal and drug seizures). In addition, policymakers can also draw on the proposed reasons for the reductions in violence in this review, that is, focused deterrence, increased police presence, and building trust between the government/police and community members (seen in selective enforcement, for example, Corsaro et al. (2012), Corsaro (2013), Corsaro & Brunson (2012)). For a systematic review specifically on focused deterrence strategies, please see Braga et al. (2019).

Overall, there is limited evidence for the effectiveness of drug-related law enforcement activities in reducing violence and, conversely, more evidence demonstrating these interventions’ unintended negative consequences. However, it is challenging to draw firm conclusions about the implications for the UK and examine causality among the relevant variables based on this review, particularly given the various interventions and settings studied. Numerous avenues for future research could address the evidence gaps identified in this report. Indeed, future UK research on drug-related law enforcement and violence may productively focus on interventions shown to reduce violence, such as selective enforcement, and test whether the findings presented in this review can be validated. Furthermore, in planning and implementing drug-related law enforcement activities, agencies should consider the risk of increased violence, particularly for those interventions for which available evidence suggests a strong association (for exampe, leadership removal and drug seizures).

7. References

Acevedo, M.C. (2015) Essays in the Political Economy of Conflict and Development. Harvard University, Graduate School of Arts and Sciences.

Atuesta, L. & Ponce, A. (2017) ‘Meet the Narco: Increased Competition among Criminal Organisations and the Explosion of Violence in Mexico.’ Global Crime 18(4): 375–402. Doi: 10.1080/17440572.2017.1354520

Atun, J.M., Mendoza, R.U., David, C.C., Cossid, R.P.N. & Soriano, C.R.R. (2019) ‘The Philippines’ Antidrug Campaign: Spatial and Temporal Patterns of Killings linked to Drugs.’ International journal of Drug Policy 73: 100–111. Doi: 10.1016/j.drugpo.2019.07.035

Barratt, M.J., Ferris, J.A. & Winstock, A. R. (2016) ‘Safer scoring? Cryptomarkets, social supply and drug market violence.’ International Journal of Drug Policy, 35, 24–31. Doi: 10.1016/j.drugpo.2016.04.019

Braga, A. A., Weisburd, D. & Turchan, B. (2019) Focused deterrence strategies effects on crime: A systematic review. Campbell Systematic Reviews, 15(3)

Buggs, S.A.L. (2018) Evaluating Efforts to Reduce Gun Violence in Baltimore: Drug Law Enforcement, Cure Violence, and Focused Deterrence. Johns Hopkins University.

Burke, P.J. (2022) ‘Reducing Organized Criminal Violence through Leadership Removals: Evidence from the Drug Wars on Chicago’s Westside, 2010–2019.’ Journal of Crime and Justice. Doi: 10.1080/0735648X.2022.2072366

Calderon, G. (2012) Three Essays in Development Economics. Stanford University, Department of Economics and the Committee on Graduate Studies.

Calderon, G., Robles, G., Diaz-Cayeros, A. & Magalino, B. (2015) ‘The Beheading of Criminal Organizations and the Dynamics of Violence in Mexico.’ Journal of Conflict Resolution 59(8): 1455–1485. Doi: 10.1177/0022002715587053

Castillo, J.C., Mejia, D. & Restrepo, P. (2020) ‘Scarcity without leviathan: the violent effects of cocaine supply shortages in the Mexican drug war.’ The Review of Economics and Statistics 102(2): 269–286. doi: 10.1162/rest_a_00801

Clark, A. K., Wilder, C. M. & Winstanley, E. L. (2014) ‘A systematic review of community opioid overdose prevention and naloxone distribution programs’. Journal of addiction medicine, 8(3), 153–163. Doi: 10.1097/ADM.0000000000000034

Corsaro, N. (2013) ‘The High Point Drug Market Intervention: Examining Impact across Target Areas and Offense Types.’ An International Journal of Evidence-based Research, Policy, and Practice 8(4): 416–445. Doi:10.1080/15564886.2013.814613

Corsaro, N. & Brunson, R.K. (2013) ‘Are Suppression and Deterrence Mechanisms Enough? Examining the “Pulling Levers” Drug Market Intervention Strategy in Peoria, Illinois, USA.’ International Journal of Drug Policy 24(2): 115–121. Doi: 10.1016/j.drugpo.2012.12.006

Corsaro, N., Hunt, E.D., Hipple, N.K. & McGarrell, E.F. (2012) ‘The Impact of Drug Market Pulling Levers Policing on Neighborhood Violence: An Evaluation of the High Point Drug Market Intervention.’ American Society of Criminology, Criminology and Public Policy 11(2): 167–199. Doi: 10.1111/j.1745-9133.2012.00798.x

Cox, R. & Cunningham, J.P. (2021) ‘Financing the War on Drugs: The Impact of Law Enforcement Grants on Racial Disparities in Drug Arrests.’ Journal of Policy Analysis and Management 40(1): 191–224. Doi: 10.1002/pam.22277

CASP (Critical Appraisal Skills Programme) (2022) ‘CAPS Checklists’. As of 17 April 2023

Degiovanni, H.B. (2013) Three Essays on Crime. West Virginia University.

Del Rio, J. (2022) ‘Do high value target strikes reduce cartel-related violence? An empirical assessment of crime trends in Tijuana, Mexico.’ Trends in Organized Crime, 1–24. doi: 10.1007/s12117-021-09444-9

Dickenson, M. (2014) ‘The Impact of Leadership Removal on Mexican Drug Trafficking Organizations.’ Journal of Quantitative Criminology 30(4): 651–676. doi: 10.1007/s10940-014-9218-5

Espinosa, V. & Rubin, D.B. (2015) ‘Did the military interventions in the Mexican drug war increase violence?’ The American Statistician, 69(1), 17-27. doi: 10.1080/00031305.2014.965796

Fisher, D.G. & Meitus, A.A. (2016) ‘Uprooting or Sowing Violence?: Coca Eradication and Guerrilla Violence in Colombia.’ Studies in Conflict and Terrorism 40(9): 790–807. doi: 10.1080/1057610X.2016.1239430

Flores-Macias, G.A. (2018) ‘The Consequences of Militarizing Anti-Drug Efforts for State Capacity in Latin America: Evidence from Mexico.’ Comparative politics 51(1): 1–20. doi: 10.5129/001041518824414647

Grisaffi, T., Farthing, L., Ledebur, K., Paredes, M. & Pastor, A. (2021) ‘From Criminals to Citizens: The Applicability of Bolivia’s Community-Based Coca Control Policy to Peru.’ World Development 146: 1–14. doi: 10.1016/j.worlddev.2021.105610

HM Government (2018) ‘Serious Violence Strategy.’ As of 21 April 2023

HM Government (2021) ‘From Harm to Hope: A 10-year Drugs Plan to Cut Crime and Save Lives.’ As of 24 February 2023

HM Government (2023) ‘County Lines Programme overview.’ As of 17 April 2023

Israel de Souza, S.E. (2019) Expiration Date: Mega-Events and Police Reform in Rio De Janeiro’s Favelas. The University of Notre Dame.

Klement C. & Blokland, A. (2021) ‘Preventing outlaw biker crime in the Netherlands or just changing the dark figure? Estimating the impact of the Dutch whole-of-government approach on outlaw biker crime using interrupted time series analysis.’ European Journal of Criminology. doi: 10.1177/14773708211024845

Lindo, J.M. & Padilla-Romo, M. (2018) ‘Kingpin Approaches to Fighting Crime and Community Violence: Evidence from Mexico’s Drug War.’ Journal of Health Economics 58: 253–268. doi: 10.1016/j.jhealeco.2018.02.002

Marín Llanes, L. 2022. ‘The Killing of Social Leaders: An Unintended Effect of Colombia’s Illicit Crop Substitution Program.’ International Journal of Drug Policy 101: 1–11. doi: 10.1016/j.drugpo.2021.103550

Martinez, I. & Phillips, M. (2022) ‘The perfect storm. An analysis of the processes that increase lethal violence in Mexico after 2006.’ Trends in Organized Crime 25: 58–83. doi: 10.1007/s12117-021-09410-5

Moeller, K. & Hesse, M. (2013) ‘Drug Market Disruption and Systemic Violence: Cannabis Markets in Copenhagen.’ European Journal of Criminology 10(2): 206–221. doi: 10.1177/1477370812467568

Muñiz-Sánchez, V., Fuerte-Celis, P., Méndez-Ramírez, K. 2022. ‘The killing fields. A Bayesian analysis of crop eradication and organized crime violence in Mexico.’ Spatial Statistics 47:100553. doi: 10.1016/j.spasta.2021.100553

Osorio, J. (2015) ‘The Contagion of Drug Violence: Spatiotemporal Dynamics of the Mexican War on Drugs.’ The Journal of Conflict Resolution 59(8): 1403–1432. doi: 10.1177/0022002715587048

Phillips, B.J. (2015) ‘How Does Leadership Decapitation Affect Violence? The Case of Drug Trafficking Organizations in Mexico.’ The Journal of Politics 77(2): 324–336. doi: 10.1086/680209

Quackenbush, C. (2019) ‘There Is Officially No More War. Mexico’s President Declares an End to the Drug War Amid Skepticism.’ Time, 31 January 2019. As of 17 April 2023: https://time.com/5517391/mexico-president-ends-drug-war/

Rios, V. (2013) ‘Why did Mexico Become so Violent? A Self-Reinforcing Violent Equilibrium caused by Competition and Enforcement.’ Trends in Organised Crime 16(2): 138–155. doi: 10.1007/s12117-012-9175-z

Sevigny, E.L. & Greathouse, J. (2018) ‘The Effect of Mexico’s Kingpin Strategy on Drug-Related Violence: A Systematic Review and Meta-Analysis.’ Presented at the International Society for the Study of Drug Policy Conference, Vancouver, Canada.

Stephenson, G. & Richardson, A. (2014) ‘New Psychoactive Substances in England: A review of the evidence.’ Home Office. As of 17 April 2023

Tealde, E. (2019) ‘The impact of police presence on drug-trade-related violence.’ Economia 20(1):61–81. doi: 10.1353/eco.2019.0008

Vargas, R. (2014) ‘Criminal Group Embeddedness and the Adverse Effects of Arresting a Gang’s Leader: A Comparative Case Study.’ Criminology 52(2): 143–168. doi: 10.1111/1745-9125.12033

Werb, D., Rowell, G., Guyatt, G., Kerr, T., Montaner, J. & Wood, E. (2011) ‘Effect of drug law enforcement on drug market violence: A systematic review’. International Journal of Drug Policy 22(2), 87–94. Doi: 10.1016/j.drugpo.2011.02.002

Annex A: Full literature-review methodology

To address the research questions outlined in the full report, we conducted a systematic review to identify and assess the evidence on the impact of drug-related law enforcement activities on serious violence and/or homicide. We included studies in this review if they met the criteria outlined below.

Eligibility criteria

We included studies of any research design and geographical scope that were:

1. Focused on the impact of drug-related law enforcement on serious violence and homicide.
2. Published in English, Spanish, Portuguese or Dutch.
3. Published in or after 2011.
4. Reported in an academic journal article, grey literature or a PhD dissertation.

We excluded studies that were:

1. Focused on decriminalising illegal substances, that is, decriminalisation was the ‘law enforcement activity’.
2. A review rather than an original research study.
3. Reported in an editorial, book, commentary, letter, protocol or guidelines.
4. Focused on only one element of the research question, that is, law enforcement or serious violence.

Searches

We searched the following academic databases:

• Web of Science
• Scopus
• Social Science Abstracts
• OAISTER
• National Criminal Justice Reference Service Abstracts (NCJRSA)
• Criminal Justice Abstracts (CJA)
• ProQuest Dissertations & Theses A&I

We targeted the following grey literature sources:

• International Society for the Study of Drug Policy (ISSDP)’s Grey Literature Bibliography
• the European Monitoring Centre for Drugs and Drug Addition (EMCDDA)
• the Scientific Research and Documentation Center (WODC) – the knowledge institute for the Netherlands’ Ministry of Justice and Security
• the UK Government’s homepage (GOV.UK)

We included articles published from 2011 onwards because a previous systematic review on this topic detailed evidence published up to January 2011 (Werb et al., 2011). We chose the selected databases to allow us to search in English while including Spanish, Portuguese and Dutch materials to broaden the geographic scope while keeping within the study team’s language skills.

We conducted the first search to include articles from 1 January 2011 to 20 December 2022. After consulting 2 expert advisors, Professor Laura Atuesta and Professor Eric Sevigny, we identified additional relevant literature missed in the first search. Therefore, we repeated the search with additional terms (see the ‘Second search’ section below). The latter included articles from 1 January 2011 to 30 January 2023. The third search included articles from 1 January to 15 March 2023. The final search included articles from 15 March 2023 to 2 February 2024.

Search terms

First search: 20 December 2022

Table A1 details the search terms chosen. Each database varies in the search terms’ style or requirements and thus may differ slightly for each. The respective databases search the title and abstract for the search terms outlined in Table A1 (where ‘TI’ = ‘Title’ and ‘AB’ = ‘Abstract’). The search terms displayed in Table A1 are for the National Criminal Justice Reference Service Abstracts, showing how we divided the search terms into sections representing each part of the research question (S1 to S6). ‘S1’ and ‘S2’ searched terms relating to drugs, requiring one of the search terms shown. ‘S4’ included search terms relating to law enforcement, and ‘S5’ included search terms relating to serious violence. We combined these sections into ‘S3’ and ‘S6’ to create the final search terms, where a search term from each of ‘S3’, ‘S4’ and ‘S5’ must be present. ‘S7’ narrows the search from January 2011 to December 2022.

Table A1: Search terms for the National Criminal Justice Reference Service Abstracts

Set number	Search terms
S1	TI(“drug related” OR “drugs related” OR “drug market” OR “supply side” OR “drug business” OR “drug cartel” OR “drug deal”OR “drug sell” OR (drug AND (“organized crime” OR “organised crime”)) OR “drug ring” OR “drugs supply” OR “drug war” OR “drug smuggl” OR “drug traffic” OR “drug lab” OR “drug production” OR “drug manufactur” OR “drug distribut” OR “drug transport” OR “drug runn”) OR AB(“drug related” OR “drugs related” OR “drug market” OR “supply side” OR “drug business” OR “drug cartel” OR “drug deal” OR “drug sell” OR (drug* AND (“organized crime” OR “organised crime”)) OR “drug ring” OR “drugs supply” OR “drug war” OR “drug smuggl” OR “drug traffic” OR “drug lab” OR “drug production” OR “drug manufactur” OR “drug distribut” OR “drug transport” OR “drug runn*”) OR SU(“Drug law offenses” OR “Drug related crimes” OR “Drug smuggling” OR “Black market” OR “Drug manufacture/production” OR “Drug business”)
S2	TI(“illicit drug” OR “illicit substance” OR “illegal drug” OR “illegal substance” OR “controlled drug” OR “controlled substance” OR narcotic* OR Heroin OR Fentanyl OR Cocaine OR Crack OR Methamphetamine* OR Amphetamine* OR Cannabis OR Marijuana OR MDMA OR “synthetic drug” OR “designer drug” OR “psychoactive” OR “hallucinogen”) OR AB(“illicit drug” OR “illicit substance” OR “illegal drug” OR “illegal substance” OR “controlled drug” OR “controlled substance” OR narcotic* OR Heroin OR Fentanyl OR Cocaine OR Crack OR Methamphetamine* OR Amphetamine* OR Cannabis OR Marijuana OR MDMA OR “synthetic drug” OR “designer drug” OR “psychoactive” OR “hallucinogen”) OR SU(“Controlled drugs” OR “Illicit chemicals” OR Cocaine OR Crack OR Methamphetamines OR Heroin OR Hallucinogens OR “MDMA (designer drug)” OR “Designer drugs”)
S3	S1 OR S2
S4	TI((enforc* OR disrupt) AND (militar OR police* OR policing OR “patrol officer” OR counternarcotic OR “counter narcotic” OR “counter drug” OR “anti-drug agen” OR “antidrug agen” OR arrest* OR (drug* AND (bust* OR raid* OR sweep* OR “crack down” OR crackdown)) OR “drug seizure” OR “drug interdiction” OR “drug intercept” OR “supply reduction” OR deterrence OR patrol OR “border patrol” OR “border force” OR “Customs and Border Protection” OR “Customs Service” OR (eradicat* AND (crop* OR suppl* OR drug)) OR “drug task force” OR “drug taskforce” OR “regional task force” OR “interagency task force” OR “public security” OR “law enforcement” OR “tactical diversion squad” OR “SWAT team” OR “tactical unit” OR “Special Weapons and Tactics” OR surveill)) OR AB((enforc OR disrupt) AND (militar OR police* OR policing OR “patrol officer” OR counternarcotic OR “counter narcotic” OR “counter drug” OR “anti-drug agen” OR “antidrug agen” OR arrest* OR (drug* AND (bust* OR raid* OR sweep* OR “crack down” OR crackdown)) OR “drug seizure” OR “drug interdiction” OR “drug intercept” OR “supply reduction” OR deterrence OR patrol OR “border patrol” OR “border force” OR “Customs and Border Protection” OR “Customs Service” OR (eradicat* AND (crop* OR suppl* OR drug)) OR “drug task force” OR “drug taskforce” OR “regional task force” OR “interagency task force” OR “public security” OR “law enforcement” OR “tactical diversion squad” OR “SWAT team” OR “tactical unit” OR “Special Weapons and Tactics” OR surveill*)) OR SU(“Drug enforcement administration” OR “Law enforcement” OR “Drug eradication programs” OR “Drug Interdiction” OR “Drug enforcement officers” OR “Drug law enforcement” OR “Drug law enforcement units” OR “Crime specific countermeasures” OR “Military role in drug law enforcmnt” OR Deterrence OR “Police-military cooperation”)
S5	TI(violen* OR homicid* OR kill* OR murder* OR manslaughter OR “violent crime” OR execution OR executing OR executed OR assassinat* OR assault* OR “violent death” OR slaying OR “gun violence” OR “gun crime” OR “drug violence” OR “gun death” OR “firearm violence” OR “firearm death” OR “knife crime” OR “knife violence” OR “knife death” OR “knife attack” OR knifing* OR bludgeon* OR strangulat* OR tortur* OR shooting* OR stabbing* OR “wrongful death”) OR AB(violen OR homicid* OR kill* OR murder* OR manslaughter OR “violent crime” OR execution OR executing OR executed OR assassinat* OR assault* OR “violent death” OR slaying OR “gun violence” OR “gun crime” OR “drug violence” OR “gun death” OR “firearm violence” OR “firearm death” OR “knife crime” OR “knife violence” OR “knife death” OR “knife attack” OR knifing* OR bludgeon* OR strangulat* OR tortur* OR shooting* OR stabbing* OR “wrongful death*”) OR SU(Homicide AND “Drug related fatalities”)
S6	S3 AND S4 AND S5
S7	S6 Limits: Publication date: 20110101-20221220

Second search: 30 January 2023

As detailed above, we conducted a second search after consulting 2 expert advisors. The second search included terms that appeared in relevant articles but were not in our searches, for example, ‘leadership’. Table A2 shows the new search terms in bold.

Table A2: Search terms for the National Criminal Justice Reference Service Abstracts

Set number	Search terms
S1	TI(“drug related” OR “drugs related” OR “drug market” OR “supply side” OR “drug business” OR “drug cartel” OR “drug deal”OR “drug sell” OR narco OR “organized crime” OR “organised crime” OR “criminal organ” OR kingpin OR cartel* OR “Los Zetas” OR “La Familia Michoacana” OR “Sinaloa Cartel” OR “drug ring” OR “drugs supply” OR “drug war” OR “drugwar” OR “drug ring” OR “drugs supply” OR “drug smuggl” OR “drug traffic” OR “drug lab” OR “drug production” OR “drug manufactur” OR “drug distribut” OR “drug transport” OR “drug runn”) OR AB(“drug related” OR “drugs related” OR “drug market” OR “supply side” OR “drug business” OR “drug cartel” OR “drug deal”OR “drug sell” OR narco OR “organized crime” OR “organised crime” OR “criminal organ” OR kingpin OR cartel* OR “Los Zetas” OR “La Familia Michoacana” OR “Sinaloa Cartel” OR “drug ring” OR “drugs supply” OR “drug war” OR “drugwar” OR “drug ring” OR “drugs supply” OR “drug smuggl” OR “drug traffic” OR “drug lab” OR “drug production” OR “drug manufactur” OR “drug distribut” OR “drug transport” OR “drug runn”) OR SU(“Drug law offenses” OR “Drug related crimes” OR “Drug smuggling” OR “Black market” OR “Drug manufacture/production” OR “Drug business”)
S2	TI(“illicit drug” OR “illicit substance” OR “illegal drug” OR “illegal substance” OR “criminal drug” OR “controlled drug” OR “controlled substance” OR narcotic OR Heroin OR Fentanyl OR Cocaine OR Crack OR Methamphetamine* OR Amphetamine* OR Cannabis OR Marijuana OR MDMA OR “synthetic drug” OR “designer drug” OR “psychoactive” OR “hallucinogen”) OR AB(“illicit drug” OR “illicit substance” OR “illegal drug” OR “illegal substance” OR “criminal drug” OR “controlled drug” OR “controlled substance” OR narcotic OR Heroin OR Fentanyl OR Cocaine OR Crack OR Methamphetamine* OR Amphetamine* OR Cannabis OR Marijuana OR MDMA OR “synthetic drug” OR “designer drug” OR “psychoactive” OR “hallucinogen”) OR SU(“Controlled drugs” OR “Illicit chemicals” OR “Designer drugs”)
S3	S1 OR S2
S4	TI((enforc* OR disrupt* OR prohibit) AND (militar OR police* OR policing OR “patrol officer” OR counternarcotic OR “counter narcotic” OR “counter drug” OR “anti-drug agen” OR “antidrug agen” OR arrest* OR (drug* AND (bust* OR raid* OR sweep* OR “crack down” OR crackdown)) OR “drug seizure” OR “drug interdiction” OR “drug intercept” OR “supply reduction” OR deterrence OR patrol OR “border patrol” OR “border force” OR “Customs and Border Protection” OR “Customs Service” OR (eradicat* AND (crop* OR suppl* OR drug)) OR “drug task force” OR “drug taskforce” OR “regional task force” OR “interagency task force” OR “public security” OR “security force” OR “law enforcement” OR “tactical diversion squad” OR “SWAT team” OR “tactical unit” OR “Special Weapons and Tactics” OR surveill OR “high value target” OR (leader AND (decapitate* OR beheading)))) OR AB((enforc OR disrupt* OR prohibit) AND (militar OR police* OR policing OR “patrol officer” OR counternarcotic OR “counter narcotic” OR “counter drug” OR “anti-drug agen” OR “antidrug agen” OR arrest* OR (drug* AND (bust* OR raid* OR sweep* OR “crack down” OR crackdown)) OR “drug seizure” OR “drug interdiction” OR “drug intercept” OR “supply reduction” OR deterrence OR patrol OR “border patrol” OR “border force” OR “Customs and Border Protection” OR “Customs Service” OR (eradicat* AND (crop* OR suppl* OR drug)) OR “drug task force” OR “drug taskforce” OR “regional task force” OR “interagency task force” OR “public security” OR “security force” OR “law enforcement” OR “tactical diversion squad” OR “SWAT team” OR “tactical unit” OR “Special Weapons and Tactics” OR surveill OR “high value target” OR (leader AND (decapitate* OR beheading*)))) OR SU(“Drug enforcement administration” OR “Law enforcement” OR “Drug eradication programs” OR “Drug Interdiction” OR “Drug enforcement officers” OR “Drug law enforcement” OR “Drug law enforcement units” OR “Crime specific countermeasures” OR “Military role in drug law enforcmnt” OR Deterrence OR “Police-military cooperation”)
S5	TI(violen* OR homicid* OR kill* OR murder* OR manslaughter OR “violent crime” OR execution OR executing OR executed OR assassinat* OR assault* OR “violent death” OR slaying OR “gun violence” OR “gun crime” OR “drug violence” OR “gun death” OR “firearm violence” OR “firearm death” OR “knife crime” OR “knife violence” OR “knife death” OR “knife attack” OR knifing* OR bludgeon* OR strangulat* OR tortur* OR shooting* OR stabbing* OR “wrongful death” OR decapitat or behead) OR AB(violen OR homicid* OR kill* OR murder* OR manslaughter OR “violent crime” OR execution OR executing OR executed OR assassinat* OR assault* OR “violent death” OR slaying OR “gun violence” OR “gun crime” OR “drug violence” OR “gun death” OR “firearm violence” OR “firearm death” OR “knife crime” OR “knife violence” OR “knife death” OR “knife attack” OR knifing* OR bludgeon* OR strangulat* OR tortur* OR shooting* OR stabbing* OR “wrongful death” OR decapitat or behead*) OR SU(Homicide AND “Drug related fatalities”)
S6	S3 AND S4 AND S5
S7	S6 Limits: Publication date: 20110101-20230130

Third and fourth search: 15 March 2023 and 27 February 2024

Comments on an early draft of this report suggested that our search may have missed some relevant sources, as indicated by previous reviews in this area (Sevigny & Greathouse, 2018). We therefore conducted a third search, expanding the search terms to ensure complete coverage of relevant materials. A fourth and final search was conducted to capture the most up-to-date literature prior to publication. Table A3 shows the new search terms in bold.

Table A3: Search terms for the National Criminal Justice Reference Service Abstracts

Set number	Search terms
S1	[STRICT] TITLE,ABSTRACT(“drug related” OR “drugs related” OR “drug market” OR “supply side” OR “drug business” OR “drug cartel” OR “drug deal”OR “drug sell” OR narco OR “organized crime” OR “organised crime” OR “criminal organ” OR kingpin OR cartel* OR “Los Zetas” OR “La Familia Michoacana” OR “Sinaloa Cartel” OR “drug ring” OR “drugs supply” OR “drug war” OR “drugwar” OR “drug ring” OR “drugs supply” OR “drug smuggl” OR “drug traffic” OR “drug lab” OR “drug production” OR “drug manufactur” OR “drug distribut” OR “drug transport” OR “drug runn” OR “war on drugs” OR “drug related violence” OR “cocaine suppl*”)
S2	[STRICT] TITLE,ABSTRACT(“illicit drug” OR “illicit substance” OR “illegal drug” OR “criminal drug” OR “illegal substance” OR “controlled drug” OR “controlled substance” OR narcotic OR Heroin OR Fentanyl OR Cocaine OR Crack OR Methamphetamine* OR Amphetamine* OR Cannabis OR Marijuana OR MDMA OR “synthetic drug” OR “designer drug” OR “psychoactive” OR “hallucinogen”)
S3	S1 OR S2
S4	[STRICT] TITLE,ABSTRACT(“drug seizure” OR “cocaine seizure” OR “drug interdiction” OR “drug intercept” OR “supply reduction” OR arrest* OR deterrence OR patrol* OR “border patrol” OR “border force” OR “Customs and Border Protection” OR “Customs Service” OR “drug task force” OR “drug taskforce” OR “regional task force” OR “interagency task force” OR “public security” OR “security force” OR “law enforcement” OR “tactical diversion squad” OR “SWAT team” OR “tactical unit” OR “Special Weapons and Tactics” OR surveill) OR TITLE,ABSTRACT(eradicat AND (crop* OR suppl* OR drug)) OR (TITLE,ABSTRACT(enforc OR disrupt* OR prohibit* OR effect) AND TITLE,ABSTRACT(militar OR police* OR policing OR “patrol officer” OR “patrol officers” OR counternarcotic* OR “counter-narcotic” OR “counternarcotics” OR “counterdrug” OR “counter-drugs” OR “anti-drug agen” OR “antidrug agen”)) OR TITLE,ABSTRACT(drug* AND (bust* OR raid* OR sweep* OR “crack down” OR crackdown)) OR IF(“Drug enforcement” OR “Drug war”)
S5	TITLE,ABSTRACT(violen* OR homicide* OR killing* OR murder* OR assault* OR “gun violence” OR “gun crime” OR “drug violence” OR “gun death” OR firearm violence” OR “firearm death” OR “knife crime” OR “knife violence” OR “knife death” OR “knife attack”)
S6	S3 AND S5 AND S5
S7	S6 Limits: Publication Date: 20110101-20230315

Selection process

We conducted a structured search to identify relevant studies addressing our review’s research question(s). We used RAND’s Knowledge Services – an extensive, up-to-date resource providing access to over a hundred online journal and literature databases and over 1,000 individual academic journals. A research librarian ensured that the search was sufficiently comprehensive; she conducted the search, retrieved the articles, transferred them to a library in EndNote reference management software and removed duplicate studies.

Two independent reviewers, supported by a third, independently screened the articles. All 3 reviewers first screened 15 article titles and abstracts, removing records that did not meet the inclusion criteria. The 3 reviewers then met and compared categorisations to ensure consistency before proceeding. The 2 reviewers continued to screen the remaining titles and abstracts. The third reviewer screened a sample of the remaining abstracts (every twenty-fifth article) to ensure consistency and made a final decision where there were any disagreements or when the 2 reviewers were undecided.

The 2 reviewers then retrieved the full-text versions of the included articles and completed a second eligibility screening. For articles in Spanish, Dutch and Portuguese, another reviewer screened the full-text articles for inclusion. At the full-text stage, the 2 reviewers indicated the reason(s) for any exclusions. When the 2 reviewers were unsure about an article, a third reviewer made the final decision. After finalising the included articles, 2 reviewers met to ensure eligibility, consistency and coherence. Expert advisors Professor Laura Atuesta and Professor Eric Sevigny were contacted to identify any articles missed from the final list. The second search followed the process detailed above. We then condensed the process for the third and fourth search, with one reviewer screening the articles at the title-and-abstract stage and a second reviewer confirming included articles and sampling a selection of excluded articles. The 2 reviewers then screened the full-text articles for eligibility and continued the same procedure as the previous searches.

We used Cohen’s kappa to calculate inter-reviewer agreement and disagreement at the full-text stage. Cohen’s kappa measures the agreement level between 2 reviewers between ‘0’ and ‘1’, where ‘0’ represents no agreement and ‘1’ represents complete agreement. We aimed to achieve a kappa of 0.6 or above, indicating a ‘substantial agreement’ (McHugh, 2012). Including a third reviewer screening samples for consistency mitigated the risk of disagreement if we fell below 0.6.

Data collection process

Two reviewers independently extracted the data from the final articles using a prepared data-extraction tool in Microsoft Excel. Two reviewers extracted data from 5 included articles. We used these first 5 articles to pilot-test the extraction tool’s efficacy and usability to ensure its fitness for purpose. Reviewers met to compare their extraction and adapted the tool according to feedback. We made the following amendments to the extraction tool presented in the research strategy:

• moved the instructions from a separate sheet to the title row for ease of use
• removed ‘Endnote reference’ and ‘DOI’ due to redundancy
• added additional detail to ‘Location’
• removed ‘Research design’ due to redundancy
• added additional columns to the methods section, including ‘Analysis’, ‘Data sources’ and ‘Any other details’
• condensed the ‘Findings’ columns into 4 columns instead of 6

After amending the extraction tool, the reviewers piloted the updated tool with 3 articles and discussed the process. Once confident with the extraction tool and achieving extraction consistency, each reviewer extracted data from half the remaining articles. The third reviewer checked independent extraction tools to ensure consistency and agreement. After extracting data from all included articles, the tools were merged. The 2 reviewers discussed any discrepancies until resolved.

PRISMA flowchart

Table A4 displays the study-identification process for databases and grey literature based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flowchart.

Table A4: PRISMA flowchart of the search strategy

Stages	Search 1: 20 December 2022	Search 2: 30 January 2023	Search 3: 15 March 2023	Search 4: 27 February 2024	Total across all searches
Recorded identified1	844	124	621	160	1749
Records screened for title and abstract2	577	124	497	118	1316
Records screened for full text3	81	23	22	29	155
Included studies	18	7	6	2	33

Notes:

1. 433 duplicate studies were excluded across the searches at this stage.
2. 1161 studies were excluded across the searches at this stage.
3. 122 studies were excluded across the searches at this stage.

Data items

The primary extracted data related to the impact of drug-related law enforcement activity on serious violence and homicide, where:

• ‘Drug’ included all illegal substances (for example, cocaine or heroin)
• ‘Law enforcement’ included any law enforcement branch/unit and the military (when the military has a police role)
• ‘Activity’ included any law enforcement presence, activity or intervention
• ‘Serious violence’ included knife crime, gun crime and homicide, as defined in the Serious Violence Strategy (HM Government, 2018)

Other data collected in the extraction tool included the following:

• reference details (for example, author(s), article title, document type)
• methods (for example, study location, study period, data sources and analysis)
• outcomes categorised using PICOS (for example, population, intervention, comparator, outcome and study design)
• findings (for example, evidence on research questions)
• quality assessment (see the section below)

Quality assessment

We conducted a quality assessment of the identified literature using an adapted Critical Appraisal Skills Programme (CASP) Qualitative Checklist (CASP 2022) and an adapted checklist from Clark et al. (2014). Since the review included qualitative, quantitative and mixed methods studies, we used a quality assessment suitable for all research designs. Clark et al. (2014) created their quality checklist to suit both qualitative and quantitative studies, so we combined the checklists to cover quality in all research designs.

The questions included in these checklists assessed the quality of the:

• aims
• methodology
• research design
• data collection
• ethics
• analysis

In the CASP checklist, the quality assessment options are: ‘Yes’, ‘No’, and ‘Can’t tell’. The quality assessment options in the Clark et al. checklist are: ‘0’ and ‘1’ (where ‘0’ = ‘No’, and ‘1’ = ‘Yes’). To compare quality scores across both checklists, we removed ‘Can’t tell’ from the CASP checklist and assigned a score of ‘1’ for ‘Yes’ and ‘0’ for ‘No’, with an ‘Unsure’ section for those requiring further reviewer discussion. We calculated a total score for comparison across studies.

We conducted the quality assessment alongside the data extraction, including it in the extraction tool. As with the data extraction, 2 reviewers assessed quality. Two reviewers first assessed the quality of 5 included articles to pilot-test the quality assessment’s efficacy and usability to ensure its fitness for purpose. Reviewers met to compare their quality assessments and adapted the tool according to feedback. Based on the quality assessment presented in the research strategy, we made the following amendments:

• moved instructions from a separate sheet to the title row for ease of use
• removed ‘Sample size is given’ due to redundancy
• added additional detail to the data collection and analysis stage to capture the reproducibility of methods
• added additional columns to examine the quality of the discussion and limitations section

After amending the quality assessment, the reviewers piloted the adapted quality assessment with 3 articles and discussed the process. Once confident with the quality assessment and consistency in scoring between the 2 reviewers, they assessed the remaining articles across the 12 criteria. Each reviewer assessed half the remaining articles. The third reviewer checked independent assessments to ensure consistency and agreement. After assessing all articles, the tools were merged. The 2 reviewers discussed any discrepancies or ‘Unsure’ entries between the 2 reviewers until resolved.

We calculated the total score for each article for cross-study comparison and divided them into 3 colour-coded categories for a more user-friendly format: ‘Low’ (0 to 4; red), ‘Medium’ (5 to 8; amber), and ‘High’ (9 to 12; green). We retained all low-quality studies in the review, conducting the quality assessment not to exclude articles but to help interpret the strength of the evidence.

Synthesis methods

Due to the predicted heterogeneity and small study pool, we conducted no quantitative analyses, providing a narrative review instead.