This special issue, titled Towards an Evidence-Based Approach to Pre-trial Detention in Europe, sets out to investigate numerous challenges in the use of pre-trial detention across Europe. In recent years, several European jurisdictions have experienced problems related to the overuse of pre-trial detention (De Vries, 2022; Morgenstern, 2020; Smith, 2022). Several studies have shown that this practice affects the well-being of detainees and their procedural rights, is costly and leads to dysfunctional consequences for the criminal justice system (Fazel et al., 2022; FairTrials, 2016; Open Society Justice Initiative, 2014, Open Society Justice Initiative & UNDP, 2011). Most worryingly, pre-trial detention increases the likelihood of guilt verdicts and the acceptance of guilty pleas (Euvrard & Leclerc, 2017; Lerman et al., 2022). In doing so, pre-trial detention might operate as self-fulfilling prophecy increasing the risk of future incarceration (Wermink et al., 2022).

The causes of the increased use of pre-trial detention across Europe are not easy to pinpoint. Existing legal and criminological literature has shed light on some factors. These include the growing discretion of some criminal justice actors (most notably, prosecutors and judges), who operate as ‘gatekeepers’ in the prison/bail decision-making. Several studies (e.g. Heard & Fair, 2019) point to a lack of access to adequate legal assistance and to insufficient or restricted availability of case materials, thus making it difficult for suspects and defendants to challenge detention orders. These practical hurdles, along with underfunded legal aid schemes, may explain the limited role played by defence lawyers in the preparation of bail hearings (Fair Trials, 2016).

Such state of affairs is not simply worrisome for fair trial and procedural rights. Embedded within the pre-trial detention practices are ungrained biases that infiltrate the choice between detention and bail, challenging the principle of equality (Van den Brink, 2019). Research findings (Heard & Fair, 2019) show that pre-trial decision-making may be influenced by factors such as nationality, ethnicity, race or social class. All in all, evidence confirms the hypothesis that pre-trial detention may be used as an actuarial form of preventive punishment (Stevens, 2012), one that targets specific groups of individuals through incapacitation.

To reduce the dangers of inconsistent decision-making and unfair treatment, several jurisdictions have started to experiment with prediction and risk-assessment tools. Among the most common forms of risk prediction methods are those based on formal, actuarial, and algorithmic methods. The key assumption behind the adoption of these tools is that they would perform better than intuitive methods of risk-assessment in the hands of judges. It has been submitted that a larger resort to data-driven and machine-based predictive tools would also help reducing prison population and increase release on bail. While part of the literature maintains rather emphatically that, in terms of accuracy, statistical methods generally outperform subjective clinical (human) judgments (Chanenson & Hyatt, 2016), recent meta-studies have sought to refute the comparability between prediction methods employed by machine-learning algorithmic tools and the accuracy of human predictions. Some studies even illustrate that no evidence exists that algorithms would be more accurate in predicting recidivism than human beings (Dressel & Farid, 2018).

Be that as it may, the overuse of pre-trial detention keeps raising important questions from the point of view of fundamental rights, especially liberty, proportionality and the presumption of innocence. International human rights instruments constrain the use of pre-trial detention, often described as a last resort. These constraints are mostly procedural in nature (as they concern the grounds on which detention may be ordered). One may wonder whether the detrimental impact of pre-trial detention could effectively be mitigated by a greater adherence to international human rights standards. The Council of Europe’s top court (the European Court of Human Rights) has developed over the years a set of standards to assess the lawfulness (i.e. fundamental rights compliance) of pre-trial detention. These standards have been criticised for not considering properly the root causes behind the overuse of pre-trial detention (Martufi & Peristeridou, 2020a). The European Union (EU), on the other hand, has been quite reluctant to adopt common rules on pre-trial detention, despite several voices raising concerns on the negative impact of a lack of harmonised standards in this area (e.g. Martufi & Peristeridou, 2020b).

Against this backdrop, the present special issue — which builds on an eponymous series of webinars organised by Maastricht and Leiden Universities on March–April 2021 — brings together scholars and practitioners from different backgrounds, in an attempt to develop a multi-disciplinary approach to the study of pre-trial detention. The aim is to unpack both the practice of pre-trial detention and the ‘legal scaffold’ governing its use. To acknowledge the complexity of this phenomenon, one must build on the inputs from various disciplines: law, public policy, criminology and psychology. The contributions collected are therefore heterogeneous in their approach and methodology. While the selection of topical issues addressed by the contributors is far from exhaustive, the tapestry of disciplinary viewpoints collected therein shows that pre-trial detention lies at the intersection of various fields of study. In addition, as pre-trial detention is a social phenomenon, contrasting its possible abuse requires a complete understanding of its multiple normative and empirical facets.

The works reunited in this collection follow a neatly structured narrative, key topics of the contemporary debate: decision-making, risk-prediction, judicial culture and harmonisation. The various themes covered are illustrative of an evidence-based approach to pre-trial detention. The nod to an ‘evidence basis’ is not to be understood exclusively as a claim that future policy should be informed by empirical research. It also refers to a broader trend towards a systematic evaluation of criminal justice practices, which must be rigorously assessed in light of their efficiency (Stevenson, 2018).

A first group of contributions addresses the topic of pre-trial decision-making. The work of Wermink, Light and Krubnik (Wermink et al., 2022) and the article by Dhami and Van den Brink (Dhami & Van den Brink, 2022) deal with the biases and the risks behind the use of pre-trial detention, which may infiltrate judicial deliberations.

The work of Wermink, Light and Krubnik (Wermink et al., 2022) is an analysis of the Dutch pre-trial detention practice underpinned by a unique combination of qualitative and quantitative methods, namely the combined analysis of criminal case data and interviews with judges and prosecutors. By inquiring the case study of the Netherlands, this contribution confirms the long-lasting assumption that nationality disparities influence the application of custodial measures before trial and affect incarceration outcomes after conviction. Through a longitudinal data review based on propensity score analyses, the authors show that foreign nationals are 1.7 to 2 times more likely than Dutch nationals to receive pre-trial detention after arrest. Additionally, persons detained before or during trial are over 50% more likely to receive incarceration as punishment. This explains differences in the sentencing outcomes between foreigners and nationals, with the former more prone to receive higher sentences at the end of trial. In particular, the lack of home address in the host country makes it more likely to be the subject of deprivation of liberty.

In a similar vein, the article by Dhami and Van den Brink (Dhami & Van Den Brink, 2022) provides insights into the decision-making process, through a comparative overview of remand/bail cases in England and Wales and the Netherlands. This contribution efficiently explores current literature concerning the way Dutch and English decision-makers operate, trying to further extract relevant findings from a comparison between the two jurisdictions. Here, the emphasis is on what the authors refer to as the ‘decision performance’ of court-based decision-makers. Comparative findings show that such performance is highly influenced by factors including: the population brought to a remand hearing in the two countries; the available justifications for ordering detention on remand and significant leeway left for risk assessments; the adversarial or inquisitorial nature of criminal procedure; further conditions such as the paucity of information and the pressure and workload of decision-makers. All the factors point to the highly unstructured nature of bail/remand decision-making, thus affording more discretion to the relevant actors. While the degree of discretion may vary across different systems (e.g. in England and Wales, decision-makers mostly rely on the same specified risk predictors, thus reducing inter-individual variability), substantive and procedural legal constraints are per se insufficient to avoid subjective interpretation and ‘heuristic’ decision strategies. Dhami and Van den Brink add on an existing body of literature (Dhami & Van den Brink, 2022) by illustrating that behind an officially stated rationale to deny bail, a court’s reasoning may easily hide further extra-legal reasons. After all, as the authors convincingly show, it is probably unrealistic to expect decision-makers to accurately articulate their reasoning and/or self-report the actual consideration underlying their decisions.

In light of the above, it is fair to conclude that solutions to avoid arbitrary decisions and unequal sentencing outcomes cannot be found exclusively within existing legal frameworks. The question of how to reduce variability of outcomes, biases and disparities is thus the central focus of the debate on automatised risk-prediction methods in criminal justice. Reflecting the increased emphasis on algorithmic predictive tools, a second group of contributions within this special issue deals with the dilemmas raised by the use of actuarial (and IT-driven) methods in the assessment of risks relevant for pre-trial detention. In their contribution, Fazel, Sariaslan and Fanshawe (Fazel et al., 2022) highlight the possible benefits of risk-assessment tools to assist judicial decision-making. Van Dijck (Van Dijck, 2022), by contrast, provides a more critical gaze on the use of artificial intelligence (AI) in relation to risk prediction in criminal justice, advocating for a greater involvement of human oversight on automatised predictive tools (in accordance with EU’s commission proposal for a regulation on AI).

Most notably, Fazel, Sariaslan and Fanshawe (Fazel et al., 2022) discuss the example of OxRec, a risk assessment tool used to predict recidivism for individuals. OxRec provides an evidence-based support to the decision-making moving from the assumption that judges are rarely equipped to conduct comprehensive risk assessments. Fazel, Sariaslan and Fanshawe report several advantages in using OxRec: transparency and consistency of decision-making, better risk communication and potentially anchoring assessment to best quality empirical evidence. Admittedly, the authors are alive to several problems associated with OxRec and other risk-prediction tools. The authors account for several methodological problems: tools based on old and suboptimal methods, or they take too long to complete or require extensive training. In addition, such tools might not take into account recent research regarding which risk factors are important. One example is the ‘young age of first violent incident’ which some tools continue to pinpoint as a key factor, despite recent research revealing this to be scarcely determinant. Fazel, Sariaslan and Fanshawe report that the external validation of OxRec in the Netherlands and Sweden has showed this to be a transparent and modern tool. Importantly, unlike some of its predecessors, the OxRec studies have tested calibration, which is a crucial aspect of a reliable assessment tool. One important problem with assessment tools is their tendency to embed possible biases and discrimination as some indicators rely on racial and socio-economic status of suspects. Fazel, Sariaslan and Fanshawe illustrate that the OxRec shows sufficiently good performance as far as discrimination is concerned. However, authors concede that more research is required to ascertain whether its use can actually improve risk assessment and assist judges in decision-making.

A more critical appraisal of the OxRec and other prediction tools (such as COMPAS) is put forward by Van Dijck (Van Dijck, 2022). In his contribution, the author distinguishes and compares two methods of risk-assessment: a statistical approach which relies on finding patterns in data (OxRec) and a machine learning approach where the machine understands patterns and trains algorithms (COMPAS). A critical difference between the two is that machine learning approaches tend to be a black box, in that they lack transparency as to the algorithm and their results cannot be challenged easily at court. Both instruments are also tested against the requirements of the EU proposal on the Artificial Intelligence Act. An important conclusion is that complex statistical and machine learning tools often do not perform much better than simpler tools. Thus, Van Dijck proposes that future providers should be required to demonstrate convincingly why their tool outperforms significantly simpler models — and why it is better than human assessment. This is an important finding as it challenges the idea that assessment tools would be necessarily better than human decision-making. This is also depicted in Van Dijck’s criticism to the use of human oversight as safeguard: how should this take place if machines are perceived better than humans anyway? There is a need, according to the author, to determine the added value of human oversight in a process taken over by technical tools to improve human decision-making.

If we may conclude that automatised risk prediction tools can be helpful but not a panacea to ensure fair decision-making, further insights may be provided by looking at the criminal justice system as a cultural organism. This posture may offer fertile ground for an exegesis on why an increased tendency exists in the use pre-trial detention within some judicial cultures and not in others. A third group of contributions in this special issue addresses the topic of judicial culture and whether apparent and latent idiosyncrasies of legal systems contribute to the (ab)use of pre-trial detention.

Judicial culture has been identified already as one possible aspect influencing the application of pre-trial detention (Fair Trials, 2016, Hammerschick et al, 2017). However, it remains unclear what judicial culture is and how this influences the application of this measure. The articles by Rogan and Smith, respectively, try to elucidate this issue.

According to Rogan’s study (Rogan, 2022), the low numbers of pre-trial detainees in Ireland are to be explained by a judicial culture that favours bail and liberty over detention. Rogan’s findings attribute to legal frameworks a pivotal role in cultivating a judicial culture that is sceptical and not immediately permissive towards pre-trial detention. The Irish legal framework includes a strong constitutional guarantee in favour of bail that has reportedly established bail at a default position and imbued practitioners with “set of liberal sensitivities” in using the pre-trial detention mechanisms. However, next to the legal framework, practitioners appear to share a common understanding on how pre-trial detention should (not) be used. This is explained to a large extent, according to Rogan’s findings, by the interchangeability of barristers as professionals who may work both for the defence and for the prosecution in different cases. Career prosecutors do not exist per se in Ireland. Accordingly, practitioners in Ireland may be more sensitive to the seriousness of pre-trial detention and have a more cooperative attitude when working with each other. Another contributing factor to the genesis and maintenance of such judicial culture is that the risk of (re)offending as a ground for pre-trial detention was only later included into the normative framework. Interestingly, the risk of (re)offending has been considered difficult to reconcile with the presumption of innocence by several authors (e.g. Ashworth, 2006; Martufi & Peristeridou, 2020a; Weigend, 2014).

A parallel account of the relationship between legal framework and judicial culture can be found in the article by Smith (Smith, 2022), who argues that despite amendments in the English legal framework, judicial practice remained mostly unaffected. This study is a follow-up to previous extensive research conducted in England and Wales during a large project documenting pre-trial detention practices (Cape & Smith, 2016). The 2016 study was followed by significant amendments in the English legal framework, allowing more time to bail decisions, furthering a more sound judicial reasoning and requiring prosecutors to ensure the availability of information relevant to bail. These amendments were expected to form a substantial step towards a more careful use of pre-trial detention. The current study carried out by Smith shows, alas, that this was not so. While the data show high level of awareness of these new legal provisions, these were not implemented as expected. Partly, an explanation could be linked to the low political priority or the lack of clarity regarding the measure to be implemented — e.g. the concept of sufficient time to make a decision on pre-trial detention — but overall, the practice did not change much. For example, decisions on bail were not significantly better reasoned after the provisions entered in force, with about 75% of the respondents to the study indicating that these had limited or no impact. Smith suggests in his conclusion that legal reform may be inadequate to transform judicial practice; this is because changing the legal framework is a top-down and static change that does not affect in a durable manner ‘coal face’ practice. Smith thus proposes to engage directly with practitioners through training and ongoing consultation could better address the issue. This important contribution shows that while legal framework may set the scene for a certain attitude towards pre-trial detention as seen in Rogan’s work (Rogan, 2022), the rabbit hole goes much deeper: a cultural change would require the adoption of soft law measures that are bottom-up and dynamic in nature. This would “plant the seeds” for a long-lasting transformation, which would reshape the attitude of criminal justice practitioners.

Whether because of the diverse legal cultures, the differences of legal framework or the inherent problems with judicial decision-making, pre-trial detention functions very differently across different legal systems. Additionally, as demonstrated in the article by Wermink, Light and Krubnik (Wermink et al., 2022) in this special issue, non-nationals suffer further discriminations. One question that has been lingering for some time is whether harmonisation across the European Union could kick-off a cultural change across Europe, leading to more balanced and sensible use of this measure. Accordingly, in a last group of contributions, the possibility and added value of an EU harmonisation is discussed.

In her contribution, Tralmaka (Tralmaka, 2022) takes issue with the lack of such harmonisation, exploring its impact on mutual trust between judicial authorities in the different Member States. The author takes issue with the style of EU criminal policy-making, which currently follows a piecemeal approach: on the one hand, there is reluctance to harmonise pre-trial detention standards even if the European Arrest Warrant facilitates pre-trial detention by making arrest of suspects more efficient; on the other hand, the safeguards in place within the EU are only marginal. These safeguards either stem from the ECHR or from a piecemeal incidental application of the six EU procedural rights directives to pre-trial detention measures. Tralmaka highlights that these directives have only incidental, minimal or otherwise limited impact on pre-trial detention (see also Martufi & Peristeridou, 2020b). She illustrates this point convincingly by comparing these directives with Directive 2016/800 on procedural safeguards for children who are suspects or accused persons in criminal proceedings: the latter introduced also substantive provisions directly affecting the use of pre-trial detention. Why then not offering similar harmonisation for adult suspects in other EU instruments? Furthermore, Tralmaka offers insight into further research conducted by the NGO Fair Trials showing the extent to which pre-trial detention is overused across Europe. The spotlight of these most recent findings is on ‘flawed decision making’: in this respect, the Charter and the ECHR might be insufficient to address the overuse of this measure as these standards do not really improve the decision-making process per se. Talmaka makes a plea for measures at the EU level that will go beyond the ECHR standards and address the underlying problems of the decision-making process. It is our hope that the various contributions in this special issue (e.g. Dhami & Van den Brink 2022; Wermink et al., 2022; Fazel et al., 2022; Van Dijck, 2022) dealing with the aspects of decision-making process and risk assessment could be an inspiration for future EU measures.

If the data in Wermink, Light and Krubnik (Wermink et al., 2022) and the impact that the lack of harmonisation has on mutual trust explained by Tralmaka (Tralmaka, 2022) were not enough to justify an EU harmonisation of pre-trial detention, one could add the arguments of Wieczorek to the basket (Wieczorek, 2022). In her contribution, Wieczorek looks at the question of harmonisation from a constitutional point of view. She argues that there are two paths in stimulating EU competences in the field of criminal procedure. One is to argue that harmonisation helps mutual recognition in some way (utilitarian rationale); the other one is to claim that certain common standards should exist for deontological reasons as they have value in themselves (deontological rationale). Wieczorek unpacks the arguments that would justify, from a constitutional perspective, the harmonisation of pre-trial detention. Accordingly, the author makes a strong case that at least two aspects of pre-trial detention should be harmonised — both often forgotten within policy and academic discussions — namely compensation of unjust detention and material detention conditions. However, as she shows, the reasons that would justify such harmonisation are more convincingly found in the deontological rationale of harmonisation: we simply want good EU standards in pre-trial detention, even if — from a purely utilitarian standpoint — we might not strictly speaking need them. Eventually, Wieczorek offers a fresh critic of the constitutional framework surrounding EU harmonisation in criminal matters, as her main argument is that the existing legal basis for EU criminal law is simply too narrow, limited and unfit to address the complex matters created by mutual recognition. Essentially, her contribution provides a plea against a dry evidence-based policy solely motivated by utilitarian reasons.

To conclude, with this special issue, it is our ambition to further the scholarship by offering an interdisciplinary and evidence-based account of the practice surrounding pre-trial detention. Admittedly, the drift towards an evidence-based discourse in this field may be seen, at the same time, as a promise and as a threat. This dualism is visible across the different contributions collected in this special issue. We are most grateful to all contributors for sharing their outstanding work and exceptionally bright ideas. We are also indebted to the anonymous reviewers for their voluntary academic service-work to dot the i’s and cross the t’s.