Documenting the Document: The Forensic Hospital Report and Its Knowledge Moves

Abstract

Drawing on case files from a Canadian provincial review board tasked with determining the disposition of persons found ‘not criminally responsible on account of mental disorder’, we explore the role of the forensic hospital report in the production of medico-legal risk knowledges. Through a detailed case study, we show how the report's content and particular material form allow the Board to produce the ‘significantly threatening individual’ – the very thing the Board (and report) are meant to presuppose. We therefore call on scholars to document their documents, and, in the spirit of actor-network theory (ANT), to analytically treat socio-legal objects as active participants in knowledge's creation. By accounting for the ‘knowledge moves’ the hospital report might allow, encourage, or prohibit human actors to make, we hope even ANT sceptics can use these tools to better understand various legal decision-making processes and their effects.

Introduction

In March 2015, 21-year-old RJ took a sledgehammer to the interior of his parents’ residence before setting fire to his father's Mercedes Benz. ¹ The fire not only destroyed the vehicle but also spread to the home, causing an estimated $200,000 worth of damage. When police arrived, RJ's mother told them he came back to the house shortly after setting the fire and asked her if she could ‘turn this off’, pointing to his ears and indicating he was hearing voices. RJ was promptly arrested, at which time he gave a statement claiming the voices in his head told him to set the fire and destroy the home. ²

After being diagnosed with schizophrenia and substance use disorder (related to cannabis use), RJ was also found not criminally responsible on account of mental disorder (NCR) – or what only a few decades ago would have been classified as legally ‘insane’ – under section 16 of Canada's Criminal Code. As mandated by the Code, the special verdict of NCR meant RJ became subject to his province's review board, in this case the Ontario Review Board (ORB), which has jurisdiction over determining if persons found NCR should be indeterminately detained in a forensic hospital, conditionally discharged into the community (with whatever restrictions the hospital deems necessary), or discharged absolutely, meaning they re-enter society as legally free persons. The disposition RJ and every NCR person receives ostensibly hinges on the answer to the statutory question of whether they constitute a ‘significant threat to public safety’, defined broadly in case law and section 672.54 of the Code as ‘a risk of serious physical or psychological harm to members of the public – including any victim of or witness to the offence, or any person under the age of 18 years – resulting from conduct that is criminal in nature but not necessarily violent’. If they are not found to be a significant threat, by law they must receive an absolute discharge. If they are deemed to be a significant threat then they must receive either a detention order or a conditional discharge, whichever is least onerous while still providing adequate protection to the public. There is no equivalent of a warrant expiry when one is found NCR, and these persons therefore remain under the hospital and Board's control indefinitely or until they are no longer found to be a significant threat.

Three years after the index offence, RJ was still under a detention order, having been declared a significant threat to public safety at each of his yearly disposition hearings. At his 2018 hearing, the Board highlighted RJ's unwillingness to abstain from cannabis use and his alleged mental ‘decompensation’ as a result of said cannabis use as reasons why he remained a risk. As is the case at every review hearing for NCR persons in Ontario, the details of RJ's alleged decompensation and those of other events purported to be significant to the Board's assessment of risk were all outlined in a document titled the ‘Administrator's [of the forensic hospital] Report to the Ontario Review Board’ but referred to generically by the ORB as ‘the hospital report’.

Reading RJ's hospital report, hearing transcript, and other associated case materials, along with 25 additional ORB case files from 2018 (which included over 5000 pages of supplementary material), we were struck by the salience of the hospital report in the Board's decision-making processes. As we document below, the report appeared to be critical to the functioning and administration of the Board not only because of its content, but also because of its particular materiality, circulation, and temporal flexibility (parties could of course always add or subtract from the report, if needed). Drawing on insights from the fields of socio-legal studies, science and technology studies (STS), and the anthropology of documents, in this article we place our focus on the socio-legal object that is the forensic hospital report, analytically treating it as an active participant at RJ's ORB hearing. Using the tools provided by actor-network theory (ANT), we show how the report's materiality allows it to serve as a catalyst for certain knowledge translations (see, e.g. Callon, 1986; Latour, 1987), breaking down epistemic barriers by bringing together – or in the original spirit of the sociology of translation, ‘carrying across’ – different and often competing interests and ways of knowing, including those found in the criminal courts, scientific laboratories, psychiatric hospitals, and even on ‘the streets’ (e.g. Valverde, 2003), thus allowing them to pragmatically co-exist.

By treating the hospital report as an active player within the ORB's network, we are able to document the report's own ‘knowledge moves’ – moves or actions that make certain claims, decisions, and concepts within the ORB possible (see Valverde, 2005, 2007). Indeed, our core argument is that the Board's ability to claim a given person is a significant threat depends in large part on the productive interplay between the textual content of the report and its material form and presence within a particular network. Going a step further, however, our main contribution is in showing how that interplay also helps to make possible the very ‘thing’ the hospital report and the human actors engaging with it are meant to presuppose: the significantly threatening individual.

We begin the paper by discussing our methods and access, as well as the ORB's mandate generally. We then describe both the analytical and empirical justification for focusing predominantly on the materiality of the hospital report (and socio-legal objects more broadly), before describing the report's actual content, which we argue renders it a ‘hybrid’ object (see, e.g. Valverde et al., 2005). Finally, we document some of the hospital report's knowledge moves, or again, those conditions of knowledge-possibilities created by the interplay between the report's particular context, content and material form.

Method and Access

The 26 ORB case files we reviewed were obtained through access requests to the Ontario Court of Appeal, as any appeal of a review board's disposition is heard and decided by that province's court of appeal. Thus, unless there is a publication ban, all legal exhibits and documents tendered at the requisite Board hearing become part of the public record (since criminal matters in Canada are public, and the appeal of an NCR disposition is still technically, albeit curiously, considered a criminal matter). The individual case files we documented typically included but were not limited to: verbatim hearing transcripts; medical reports; letters of correspondence between the legal parties, hospitals, and doctors; letters from the NCR person to various staff members or parties; hospital reports to the ORB; the Board's official reasons for their decisions; and the court of appeal decision. All 54 Ontario NCR cases where a disposition was appealed and decided in 2018 were located, and the first 26 cases were examined.

The files were reviewed separately by the authors and iteratively coded over 1 year due to their size and depth of content. It of course must be said that these cases may not be representative of all NCR persons by the very nature of having been appealed. Yet our purpose is precisely not to make generalizable claims about the entire ORB process or the process of legal knowledge production writ large (as if there was such a thing). Instead, we endeavour to highlight, using the ORB and the case of RJ as a contextual example, the theoretical insights gained by socio-legal scholars interested in legal knowledges by taking seriously the roles played by the materiality of objects in legal decisions, roles that are expressly contingent on the specific network of the specific case in which one is asking ‘how do “they” know …’ (e.g. ‘… that RJ is a significant threat to public safety’)?

In that light, RJ's case was chosen by virtue of it being relatively ordinary within our sample, at least qualitatively. Still, we recognize that the forensic hospital report might have different effects even in similar ORB cases, hence why we follow Mariana Valverde (2007) in focusing primarily on the knowledge processes of a single case within a single ‘legal complex’, rather than asking broad questions about ‘law’. What we hope is broadly generalizable, though, is the theoretical toolkit and methodological framing we employ, especially the analytical importance of accounting for the materiality of things and texts when examining the conditions of possibility of legal knowledges and decisions.

The Analytical Toolkit: Documents as ‘Actors’

One of ANT's key insights is to break down, or ‘bracket’ (Riles, 1998, 1999), the human versus non-human and the natural versus social divide that has been a hallmark of modern Western science and culture, albeit a mistaken one (see Latour, 1993). ANT therefore calls on researchers to view non-human entities as potential actors with agential powers (e.g. Callon, 1986; Latour, 1987, 2005), even if we choose to view those powers as simply analytically useful as opposed to ontologically real. We have little interest here in the quasi-philosophical debates over whether objects ‘actually’ have agency. Like Valverde (2007), we prefer to use ANT pragmatically, as a tool to help us think through the conditions of possibility behind certain medico-legal decisions, including how actual people (or institutions (or networks)) lay claim to knowledge. Whether or not we attribute agency to objects, objects do obviously allow, encourage, demand or even forbid humans to say and do certain things. At its most basic, then, what our use of ANT provides is a way of empirically accounting for the possibilities and prohibitions that an object presents within a given actor-network, and how it might constrain or make possible various knowledges and claims regardless of who, or what, we attribute those claims to in the end (see also Cloatre, 2008).

The relatively few contemporary socio-legal studies that engage with ANT typically do so with respect to complex technoscientific artefacts, such as rape kits (Quinlan, 2017; Shelby, 2020), forensic DNA evidence (Kruse, 2016), or Sentencing Information Systems (Hutton, 2012). These studies tend to share the goal of uncovering the contingencies behind the creation of their complex objects of interest, and how these processes come to be taken for granted – or what STS scholars might call ‘black boxed’ (e.g. Latour, 1987) – despite them being fundamentally (and again, the more courageous might say ontologically) the product of messy and conditional actor-network relations. In line with these studies, we also stress the importance of accounting for the typically invisibilized processes related to how even mundane objects, like documents and reports, are assembled, specifically how a myriad of actors with different goals, interests and agendas come together to create what appears to the ORB as the coherent and united hospital report. However, these studies often focus solely on the contingencies in the creation of their objects, and not on the possibilities which that process and the object itself create. What we show here is that by opening the black box, the equally contingent nature of an object's or document's social effects can also be exposed.

As a result of this analytical framework, as well as our object of focus, we merge the tools of ANT with theoretical insights from the anthropology of documents (see, e.g. Hull, 2012; Riles, 2006). In a review of ethnographic studies on bureaucratic documents, Hull (2012) shows how much of this scholarship treats documents as ‘mediators’ that, in the words of Bruno Latour, ‘transform, translate, distort and modify the meaning or the elements they are supposed to carry’ (Latour, 2005: 39, as quoted in Hull, 2012: 253). Documents therefore not only serve ‘as neutral purveyors of discourse’ but also mediate ‘the significance of the signs inscribed on them and their relations with the objects they refer to’ (Hull, 2012: 253).

Admittedly, Quinlan (2020) does place analytical focus on ‘texts’ in her study of the socio-political construction of the rape kit backlog in the United States, and Arboleda and Valverde (2021) examine the numerals 1 through 6 in their analysis of the Columbian estrato system, showing how those numbers in that context ‘acquired a life of [their] own’ (p. 686), including, similar to our study, how the meanings of the numbers transformed as they travelled amongst and between different institutions. Additionally, Cloatre (2008) views the pharmaceutical patent as an actor, shifting the focus of analysis from the patent holder to the patent itself. However, these authors mostly emphasize the productive role of the meanings embedded within the content of their texts and numbers and patents, rather than the role of the physical material via which the content travels. While the content of a text is of course critical to any analysis of its role in knowledge production, ANT shows us how the material form – or medium or vessel or container – of the content also conditions its possible effects.

Thus, our contribution is not only to show how the material form a document takes is crucial to the more self-contained processes of discursive mediation. Rather, we also show how a document's content, materiality and contingent construction work together to help other actors bring new objects and concepts into existence, which in turn allows them to make certain decisions. Instead of focusing on either the content of the document or its material form, by examining both, together and at the same time and in their given context, we show how their interplay makes possible the production of entirely new things and concepts and knowledges. In this way, the forensic hospital report can be said to help ‘produce the world that it purports to describe’ (Best and Walters, 2013), or in the context of the ORB, it can be said to help produce the significantly threatening individual it was created to confirm.

The Ontario Review Board

In Ontario, Canada, review boards tasked with determining an NCR person's disposition must be comprised of at least five members. They have to include two psychiatrists (if only one psychiatrist is present, a psychologist or other mental health practitioner can also serve), a federal judge or lawyer with 10 years’ experience (who usually serves as the Chairperson and runs the proceedings), and a public member. At each annual hearing, Board members typically hear testimony about the NCR person's past year and progress from the person's attending psychiatrist (appointed by the associated forensic hospital) and receive disposition recommendations from lawyers on behalf of the NCR person, the forensic hospital and the provincial attorney general (represented by a Crown prosecutor).

The attending psychiatrist (always) and the NCR person (sometimes) are usually the only persons who ‘testify’ at the hearings, and although these are technically meant to be informal proceedings, taking place in the hospital's boardroom or ORB's offices, they remain steeped in criminal legal procedure and symbolism. The Board does make a point of trying not to refer to the NCR person as ‘the accused’ during hearings (with the occasional slip); however, they are still referred to as the accused in all official documents, itself a curious convention considering a finding of NCR is itself a verdict, and thus an NCR person is no longer accused of anything. The person testifying is also called a witness, can only answer questions being asked of her or him by counsel or Board members, and is even subject to cross-examination. The testimony of these witnesses (but again, usually only the attending psychiatrist) along with the printed hospital report typically comprise the entirety of evidence from which the Board must render its decision. Somewhat tautologically, the attending psychiatrist almost exclusively testifies to what is already written in the hospital report, often quoting the report directly when answering questions.

At RJ's ORB hearing in 2018, the decision that he was a significant threat was reached by the Board unanimously, despite RJ not committing any acts of physical violence in his three years under the Board's supervision, either while in hospital or in public when allowed out by the hospital. The detention order was based on a recent alleged psychotic relapse lasting about a month, which manifested in a belief that a previous assault against his father (which took place prior to the index offence) was justified because RJ heard his father say he ‘would like to sleep with [RJ's] girlfriend’ and that co-patients and staff could presently read his mind. During this time, he also threatened to hit his father with a belt after an argument but did not physically assault him.

The key question at the hearing appeared to be whether RJ's recent psychotic episode was caused by his refusal to abstain from cannabis, a violation of his detention order and the theory put forward by the hospital and Crown counsel (and the one the Board eventually agreed with). This debate formed the crux upon which the Board's judgement hinged, as the defence attempted to show that RJ's mental state could remain stable even when he used cannabis, since abstention from cannabis was the one disposition condition RJ refused to accept; had he accepted and abided by this condition he would have ostensibly been granted a conditional discharge. The hospital, however, sought to causally equate his use of cannabis with a propensity for mental decompensation (i.e. the inability to cope with stressors, leading to a relapse of his schizophrenic symptoms), and by extension imply that he was a significant threat to public safety because he was prone to decompensate.

In the rest of this article, we empirically document the hospital report's role in these ‘knowledge games’, specifically how the hospital (and Board) successfully translated the statutory significant threat question into a question about RJ's cannabis use and mental state.

The Hospital Report and Hybridity

In this section we show how the forensic hospital report includes a smorgasbord of claims and knowledges that arise out of, and may only be possible within, certain epistemic cultures (Knorr Cetina, 2007). In some files, we even see the hospital report refer to and analyze recent case law on who qualifies as a significant threat when justifying its position that the NCR person would no longer meet the legal threshold required to detain them, much to the chagrin of the board members who argued significant threat was also a medical determination. In short, here we recognize and account for the fact that the report itself is networked. Despite it appearing as a uniform, internally consistent document, this is only the result of it being black boxed. We argue the report is actually the result of the ‘summing up of interactions through various kinds of devices, inscriptions, forms etc, into a very local, very practical, very tiny locus’ (Latour, 1999: 16), a locus that allows it to appear as a homogenous, uncontroversial and uncontested document once it reaches the ORB's members.

Immediately exposing itself as a document of hybridity, the title page of RJ's hospital report includes – among basic identifying details such as RJ's name, date of birth, and ORB file number – a list of the ‘Index Offence[s]’ for which he was found NCR, which is situated directly above his ‘Current Diagnosis’ of schizophrenia and cannabis use disorder. In the bottom corner of every page is RJ's Medical Record Number and Contact Serial Number, which can be used by other hospitals to access his complete medical history.

The substantive portion of RJ's hospital report begins by describing the circumstances of the events for which he was found NCR, although it is unclear exactly where the details of these descriptions came from. It then proceeds to describe his entire ‘social history’, beginning with the caveat that because RJ was a ‘poor historian’, often providing ‘contradictory facts’ (e.g. surrounding alleged instances of past parental abuse), this history is ‘primarily from his parent's perspective’ and from what the hospital could obtain from healthcare records. Without acknowledging how or why this information might be relevant to the Board's or even the hospital's assessment of risk, this social history includes details about his family's emigration to Canada as refugees, his parents’ current employment status, details of his mother's pregnancy (e.g. ‘she had been in the hospital on bed rest for the majority of her pregnancy’), any traumatic incidents (including when RJ's grandmother accidentally spilled boiling water on him when he was 13-month-old, with the added detail that ‘For a year after the incident, his mother took him to the church daily for a nun to care for the burns’), his entire school and romantic history, and history of substance use. This is followed by his criminal record and complete psychiatric history. The report then includes all the observations from previous years’ hospital reports that were drafted for his past ORB hearings.

Somewhat surprising given current work on actuarial risk assessments and the ‘new penology’ (see, e.g. Feeley and Simon, 1992; Hannah-Moffat, 2019) is the apparent lack of deference given to actuarial risk assessment methods in RJ's case and our ORB files generally. Although the results of actuarial assessments were included at some point in every hospital report we reviewed (but not in every reporting year, and they were not mentioned at each hearing), they were always far overshadowed by clinical notes, including summaries and verbatim excerpts from the daily logs of nurses and staff members at the hospital, and detailed descriptions of ‘notable incidents’. What is notable or significant, however, is of course determined from the viewpoint of the hospital, or more accurately, the specific but often unknown author(s?) of that section who decides what is relevant from the immense record of notes accumulated each year. Of particular interest, RJ's 2018 report highlights that despite him being granted the ‘privilege’ of living in the community for part of the year, he was brought back into the hospital three times, all ‘as a result of a breach of his ORB disposition secondary to marijuana use’. More generally, RJ's report focuses extensively on his inability ‘to discontinue his marijuana use’ and those (failed) attempts to minimize ‘the long-term implications of continued [marijuana] use on his mental well-being’.

As mentioned, each report ends with the treatment team's disposition recommendation to the ORB. That this recommendation appears last is not insignificant, since, as we discuss below, it allows the report to read as a continuous and self-contained narrative culminating in what should now be the obvious opinion that RJ remains a threatening individual. These opinions are then always immediately followed by multiple signatures from the attending psychiatrist and various hospital officials.

Since the judgement of risk made by the ORB is based on many different types of knowledges produced by a multitude of actor-networks, the key claim that closes out RJ's report, as well as every other report we reviewed (as there was not a single appeal initiated by the Crown) – that the NCR person ‘continues to meet the threshold for significant threat to the safety of the public’ – should be viewed as a hybrid claim, rather than, say, a strictly medical or legal, or even medico-legal claim (Latour, 1993; Valverde, 2003). ³ As Hull (2012) notes in the context of documents generally, in our case certain signs and symbols within the hospital report, such as the ever-present hospital logo and RJ's medical reference number, mediate the messages behind the ostensibly discrete facts contained within it. For starters, this helps render all claims in the report appear as if they have the force and validity attributed to expert medical knowledges. The statement that ‘RJ's cannabis use causes decompensation’ obviously means something very different (possibly nothing) if written on a blank page instead of within a mass of sheets bound under the hospital's logo and the heading ‘Notable Incidents’, or even if it simply came out of the mouth of a lawyer or psychiatrist at the actual hearing.

Indeed, being able to call forth and reference a claim backed by ‘the hospital’, as if that institution and the people within it were a singular entity (see, e.g. Law, 2004), simply by citing a page number or quoting the report is but one important knowledge move sanctioned by our document, moves which we can now begin to describe in detail.

Documenting our Document's Knowledge Moves

Focusing on the case of RJ, in this section we empirically outline some of the specific knowledge moves hospital reports both perform and enable, moves that we argue are crucial to the contingent constitution of the significantly threatening individual in each case. Like Valverde's (2007) analysis of the knowledge moves of appellate courts, we stress the moves outlined here are not exhaustive, nor are they necessarily mutually exclusive; certain knowledge moves may beget or depend on the possibility of other moves/actors. However, here we wish to highlight the moves we think are most likely to be taken for granted by those interested in the processes of legal knowledge production and decision making, predominantly the ones arising from the particular material (rather than purely discursive) qualities of our socio-legal objects of interest.

Citing and Signing: Purifying the Report's Hybridity

Latour (1993) uses the term purification to describe the separation of messy, hybrid entities into binary categories of either the natural or the social, or the human or non-human. In RJ's report, we can see how the hospital correlates social, moral and cultural judgements about RJ's cannabis use with quasi-natural (or at least medico-scientific) claims about mental illness and causality (and note how there is little linguistic issue granting ‘the hospital’ agency in producing this report). This is not to say that cannabis is not really correlated with certain brain states or behaviours defined as mental disorders, and that it might not actually aggravate those symptoms or conditions in certain individuals. What is important here is that within the ORB's context any such correlations never have to be explained via knowledges typically defined as expert or scientific, which has very serious effects on the NCR person's liberty and treatment.

In all the hospital reports we reviewed there was never any reference to a specific scholarly study, medical journal, scientific experiment or clinical trial, nor did we find even passing references to authoritative expert literatures (other than, perhaps, case law) that might justify a specific claim's presence within the report. ⁴ At first glance, this admittedly appears uncontroversial, as we would have no reason to expect medical or legal documents of this kind to include formal, academic-type citations, and it would likely become an entirely different type of document if it did, which itself is a testament to the importance of a document's particular form.

However, this specific style of referencing (or lack thereof) still has important epistemological effects that need to be accounted for. That these types of documents do not require their facts to be explicitly referenced serves an important function in what we are calling the ORB's purifying knowledge move: although the report is made up of hybrid knowledge claims, the contents of the report appear to come from a single entity, the hospital, in turn symbolically linking all claims to medical/therapeutic expertise simply by their inclusion within the report. The report's hybridity, and thus lack of a consistent epistemology and its reliance on non-expert knowledges, would be immediately exposed if it included scholarly references related to the chemical effects of marijuana on persons with schizophrenia alongside citation-less common-knowledge claims, such as the claim that the impending legalization of marijuana in Canada would only serve to increase RJ's risk.

Instead, the hospital report uses a different mechanism to affirm both its veracity and the procedural rigour of its purified creation: signatures. On the last pages of every hospital report are the signatures of various institutional representatives, typically the attending psychiatrist and two administrative figures from the hospital who have little to do with the actual NCR person in question; in RJ's case, the signatures included the hospital's ‘Psychiatrist-in-Chief’ and its ‘Clinical Director, Forensic Psychiatry Program (on behalf of the Executive Vice President of Clinical Programs)’. Even though the hospital is comprised of many different actors with potentially different goals and interests – including occupational therapists, psychiatrists, social workers, security staff and nurses, as well as their associated objects – the three signatures help stabilize the hospital as a ‘united front’ (Moore, 2007), in turn masking any contestations, alliances, or capitulations that may have taken place between the various actors.

As Scoles (2018) notes in the context of legal contracts, signatures are the ‘collective achievements’ of multiple human and nonhuman actors. She shows how the ‘penned signature acted as a moment of stability, which had pulled into agreement the jostling, competing and over-lapping actor-networks that were mobilized to align all the relevant material constituents’ (p. 280). In the hospital report, the signatures stabilize the networked entity known as the hospital and render all the observations and claims within the report on equal footing, allowing the notation that RJ's room was ‘messy’ to be seen as relevant to risk and his medical history. However, just as Biagioli (2006) reminds us to pay attention to the authors not present on the front page of scientific articles, the presence of the attending psychiatrist's signature and simultaneous absence of the signatures of the everyday nurses and staff who actually observe the incidents described in the report further invisibilizes the messy and contingent nature of the hospital's actor-network.

The most salient example of the effects of these purifying knowledge moves can be seen in the report's linking of RJ's marijuana use to mental decompensation. The causal link is never explained in any form within the report, much less through citation practices familiar to academics and most experts of medicine, science and even law (see Strathern, 2006; Valverde, 2007). Furthermore, the hospital report always treats cannabis/marijuana as a homogenous substance, never discussing the potential differences between strains, dosages, or methods of ingestion, and how those differences might manifest differently in a particular individual. RJ tries to explain much of this himself to the Board during his testimony, telling the Board members that when he uses cannabis he only uses ‘less than half a gram’ and that if he uses ‘during the day, there's a strain called sativa, which is more energetic to the body. And during the night it would have previously been Indica, which is more, I guess, sedating to the body’. He uses his experiential knowledge to defend his position that he can safely manage his marijuana use, and that it can actually help rather than hinder his mental state by aiding his anxiety and sleep – a position which, as will be shown below, can be discredited by the Board by the sheer fact that it was not recorded in the hospital report.

Despite the hospital report not including any citations, the report can itself be, and indeed is cited by the Board, which occasionally copies from it at length in its official disposition decision, known as the Reasons for Disposition (Reasons). For example, one NCR person's official Reasons stated:

As [the attending psychiatrist] put it, [the NCR person's] baseline level of risk is high. Accordingly, the monitoring that is required includes a detention order to admit [the NCR person] as it is likely that he will continue to use marijuana and be at risk of a rapid deterioration in his mental state (see Hospital Report (unnumbered pages); ‘Re-offence Scenario’; ‘Risk Management Plan’). [italics in original]

Putting aside the fact that it assumes mental ‘deterioration’ is a risk factor, it is possible this reference was simply included to help interested parties locate a specific claim within the pages of the report. However, it can also have the added effect, even if unintended by any human actor, of lending further credibility to the claim the doctor stressed at the hearing. The Board can then essentially justify its decision by repeating a claim from the hospital report and then instill it with added epistemological clout by affirming its facticity through a reference to that very same report.

The report therefore becomes a document to be cited rather than a document that needs to cite. In RJ's case, the Board cites the hospital report to support the claim that his marijuana use makes him a significant threat. However, if one were to look up that claim or other cited claims within the report it will simply restate the claim or observation as a brute fact given his past behaviour, as opposed to explaining its correlation with risk (or even mental illness). In a circular fashion, and to reiterate, a given claim's presence within the report both grants the claim its relevance and subsequently confirms it in both worlds; it is relevant by the very nature of it being mentioned in the report and it is in the report because it is relevant.

Through the report, the rules of the ORB's knowledge game have been changed: the Board no longer has to be explicitly concerned with the vague and unanswerable question of whether RJ will harm another person upon release, but, just like the hospital, it can equate that concern with the question of whether cannabis will cause RJ to mentally decompensate. It is now RJ's mental state that is at issue at the ORB hearing, and not his potential to commit harm (the statutory legal question). The hospital is thus successful in its knowledge translations (Callon, 1986), rendering the Board's legal interests synonymous with its (ostensible) medical interests. Since decompensation is contrary to the hospital's (again ostensible) treatment goals, through the report's translation scheme it also becomes contrary to the Board's public safety goals.

Invisibilizing (Human) Observations: The Report as a ‘Super Witness’

In her ANT-inspired analysis of early forensic rape kits, Quinlan (2017, 2020) describes the kits as ‘technoscientific witnesses’, ones that are seen as far more valid and reliable than the actual victim who literally witnessed and experienced the assault. In a similar way, the hospital report to the ORB can also be viewed as a special type of witness. Like the human ‘witnesses’ in ORB proceedings, the report can be called forth to testify, even if it cannot technically speak out loud. Like Quinlan's rape kits, however, the report's materiality and non-humanness also affords it the ability to perform certain knowledge moves not available to human actors. In ORB hearings we often see parties referring to – or we might even say deferring to – the hospital report as if the statements and observations within it were indisputable facts. The daily observations of nurses and staff detailed in the report cannot be questioned because the human persons who made the observations are typically not present at the hearing, or there is no record of who actually made the observation in the first place. ⁵

The effects of the report's ability to invisibilize actual human witnesses become most salient in those cases where the hospital's instability is temporarily exposed. For example, in one hearing it came out that the attending psychiatrist was actually on vacation during a significant event described in the hospital report, when the NCR person was placed in seclusion. After being questioned about the person's time in seclusion, the doctor admits they ‘can only describe [the NCR person's] presentation as per what I read in the chart because I was on vacation in that period of time. When he was secluded, it was my last day of working and when I came back from vacation, he just got trialled out of seclusion’. Still, the doctor's account of events – which had now of course been exposed as another unknown human's account of events – becomes beyond questioning through their deferral to an alternate, unanswerable witness, in this case ‘the chart’ as it was summarized in the hospital report. Importantly, and despite the doctor's own admission, the facticity of the details in the report/chart remained unchallenged.

By (anonymously) placing staff observations in the report's particular form, it allows the doctor to reference incidents that they themselves were not directly privy to and without necessarily mentioning who actually witnessed them. ⁶ A common physician's phrase at ORB hearings is therefore: ‘As outlined in the hospital report …’. What this partially discursive, partially material knowledge move does, however, is once again black box the processes (i.e. networks) involved in documenting the observation that the doctor is referring to, helping yet again to render those observations beyond questioning. As Latour notes, things that are black boxed appear stable because they have hidden potential instances of ‘uncertainty, people at work, decisions, competition, [and] controversies’ (Latour, 1987: 4). Although the doctor's interpretation of what some significant event means to the risk posed by an NCR person can be debated by the parties at the hearing, the black-boxing of the report means the facticity of the descriptions of the event itself remains stable.

In an exchange between RJ's attending psychiatrist and his defence counsel, his counsel tries to tease out whether his recent acute psychotic episode was really the result of his marijuana use, or if it was the result of an administrative error in the prescription of an opiate painkiller (which, by the doctor's own admission, RJ should not have received as it could exacerbate his symptoms) during an unrelated shoulder surgery. His lawyer points out that the psychotic episode took place shortly after the surgery; however, the doctor attaches RJ's decompensation back to his conscious choice to ingest cannabis by referencing the sequence of observations outlined in the report. As the doctor notes in their testimony,

Yes, we noticed that [RJ's] decompensation got worse after the administration of Dilaudid [the opiate painkiller], but there is ample evidence that [RJ] was already becoming unwell prior to having the surgery, as listed again on page 28 and 29 of the hospital report, which highlights the sequence of events prior to having the surgery […] and then those events speak to the fact that his mental status was deteriorating, was fragile.

Yet the accuracy and relevance of those descriptions in the report are never questioned, and indeed cannot be questioned; they are facts that can only be interpreted. After the above quote, RJ's lawyer drops this line of questioning and moves on.

In this case, the black-boxing of the report also worked against RJ in another way. RJ's own account of events can be discredited and rejected simply because it is not present in, or conflicts with, the account documented within the hospital report. In explaining why they rejected RJ's testimony that his cannabis use was not causally related to the index offence, in their official Reasons the Board defers not to any human expert or witness but to the hospital report:

The panel finds that [RJ's] evidence, especially his evidence regarding his usage of cannabis, is not credible. His testimony at the hearing does not accord with statements he previously made, as set out in the Hospital Report about the fact that he had used cannabis just prior to the commission of the index offence.

Therefore, it appears only those statements made by RJ that get recorded in the hospital report are elevated to the same fact-like status of the hospital's own observations.

The ‘Reports Multiple’: Increasing Stability Through Material Mediation

Despite the report being copied and distributed to all parties prior to each annual hearing, these copies do not stay materially consistent as they circulate between actors, even if the actual content remains the same. Similar to Irene van Oorschot's (2014) analysis of judicial file work in a Dutch court, where ‘successfully navigating the information in the case file relies on material work, like highlighting, coding, juxtaposing and summarizing practice’ (p. 454), the ability to physically alter hospital reports in real time allows for certain knowledge moves to be made. Coinciding with how we view significant threat as being constituted with the aid of the hospital report (rather than being contained within it), Van Oorschot shows how ‘The “object of knowledge” – the case – as such interactively emerges and is not in any simplistic way “contained” in the case file’ (p. 454, emphasis in original).

At a nominal level, RJ's hospital report is referred to by many different names as it moves from and within the hospital to the ORB hearing. At times it is referred to as the ‘Administrator's Report to the Ontario Review Board’ (the title of the report as it appears on the front page), the ‘Hospital Report’ or even just ‘the report’ (how it is referred to generically at the ORB's hearing and in its documents), Exhibit no. 1 (when it is officially introduced as evidence in the hearing), and even as ‘the hospital's evidence’, which again makes the hospital appear as a singular actor rather than a collective assemblage of human and non-human actors. When it enters the ORB's network as an exhibit, one of the report copies also receives a stamp indicating its exhibit number and the date it was received as such, instantly turning it into a piece of legal evidence.

As the stamp indicates, these different ways of talking about and labelling the report are not merely discursive conveniences, but they often coincide with physical changes to what we might more accurately now call the ‘reports multiple’ (see Mol, 2002). Like Van Oorschot's judicial case files, most ORB case files, including the hospital reports we reviewed, were also visibly marked up, whether it be with marginal notes, underlines, and/or highlights. The original hospital report that is copied and shared with all parties therefore becomes physically different reports as it (they) is (are) mobilized by other actors within the network. During his testimony, RJ's attending psychiatrist even refers to highlights he made in his own copy of the report when he states, ‘Again his, [RJ] has, has a view and an opinion about the use of cannabis which differs from the treatment team. It's our opinion that, and as I had highlighted [in the hospital report] and I’ll just point to where it's highlighted’.

Somewhat paradoxically, the ability to edit or even cross-out sentences, paragraphs and previous facts in a report is a knowledge move that helps to maintain the report's veracity, even when it was the result of a specific claim in the report being exposed as factually false. In RJ's hearing, the entire proceedings had to be paused when it was realized the dates of some drug-screening urinalysis tests were incorrect. After taking a brief recess to allow the doctor to consult with the nursing team about these urine screens (seeing as the nurses were not present at the hearing but they were the ones who actually conducted and recorded the screenings), it was confirmed by the doctor that several tests listed in the report were never conducted, including a false overreporting of non-existent tests that were positive for cannabis. After apologizing for the error, the doctor told the parties, ‘So the, with regards to the list on Page 30 of the hospital report […] the June 27^th and 28^th which says 2017 is actually an error; that should be completely off the report’. The remedy for this? As the Chairperson of the proceeding states, ‘So essentially the error, cross off the first line June 27, 28, and the second line of May should actually be 2018?’ By crossing off and changing the erroneous dates, the rest of the report's validity remains intact. The observations in the report, and even those in that particular section, can hold on to their fact-like status even as the possibility of false observations within the report has been explicitly exposed. With those dates crossed-out, the hearing then carries on as if the error was never made or noticed.

Tacking on Pages: The Report's Temporal Knowledge Moves

Here we show how the hospital report can be viewed not simply as a biography of a person's risk factors and medical and life history, but, in line with ANT, also as a biographer of risk (see Kruse, 2016). Individual risk factors present in the current year's report can derive new meaning from their contextual placement within a years-long – and often decades-long – historical narrative of the NCR person's time under ORB and hospital supervision, and even their entire life history prior to the index offence. The report therefore serves as a temporal mediator that creates the correlations between the historical and contemporary risk factors that it then seeks to affirm. The current, crucial observation that ‘RJ continues to use marijuana’ has (negative) meaning in this context only through its association with past events where his marijuana use was said to be correlated with his mental state, and where his mental state was said to be correlated with physical violence, as documented by observations and statements in the hospital report.

The way the report is arranged, structurally and stylistically, therefore has important effects that influence if and how certain temporal claims about risk can be made. Interestingly, the observations and significant events from all previous years’ hospital reports were always included as part of the current hospital report, as opposed to being kept as separate documents. While a ‘new’ report is drafted for each hearing, it is really more of an update, typically with only a few additional pages pertaining to the current year tacked on, along with a new cover page. This allows the report and, by proxy, judgments of significant threat to take the form of an ongoing historical narrative, one that enables the parties to perform what we are calling its temporal knowledge moves. For example, parties to a case are able to claim that an NCR person remains a significant threat simply because they were found to be a significant threat in some other past reporting year and the relevant conditions or issues from that year, as outlined in the report, remain present. This is a key move in the Board's ongoing construction of the significantly threatening individual, as it allows the NCR person to be judged against her or his past self rather than some archetypal or statutorily defined threatening person.

At RJ's 2018 hearing, the Crown counsel asked the attending psychiatrist, ‘Now when you compare this year's report with last year's, I think you characterize that as a step back, a significant step back in his progress under your care and support, is that fair to say?’ The doctor agreed with that interpretation and further stated that ‘the difference between the 2016 report and the 2017 [hospital] reports […] also highlights the impact of the use of cannabis has had on his mental health […]’. The doctor goes on to say that, during the previous reporting year, RJ ‘wasn’t using cannabis, his urine drug screens were clean, he was focused, he was motivated, he returned to school’. Thus, a lack of motivation, focus and clean urine screens become risk factors simply by relating them to his better motivation, focus, and urine screens in a past reporting year, and without any explanation of their actual association to risk or mental decompensation, let alone the definition of risk outlined in statute and case law. The doctor even admitted that ‘we were at the point where the treatment team was actually considering a recommendation for a Conditional Discharge, when he had a relapse’. Again, the doctor does not need to explain cannabis's medical or pharmacological (or even psychological) link to risk, as it is enough to note he had a ‘relapse’.

At the same time as the report allows for risk to be measured by comparing multiple years, the way the report is compiled also allows the hospital to appear temporally stable. If each year's report was a totally separate document that only included that specific year's events, those reports could of course still be compared with each other. However, the discrete packaging of the documents could have the effect of exposing some of the hospital's own contingencies and instabilities, such as changes in the NCR person's doctors, nursing staff, or reporting procedures, which could reduce its epistemic authority. The generic reference to either ‘the hospital’ or ‘the administrator’ that titles each annual document makes it seem like the hospital that controls the NCR person from year to year is a completely stable entity, even if the hospital's and the specific NCR person's nurses, doctors, administrators, and security staff can and often do change, or even if the NCR person is moved to a new ward, with little-to-no reference as to how that might affect the report's creation. If the report were titled ‘Dr So-and-So's Report', it could risk exposing that united front presented by the hospital's network when, for example, the eponymous doctor changes, or even when they simply go on a holiday.

Conclusion

Through our pragmatic application of ANT, we see that the forensic hospital report serves as an important player in the ORB's knowledge games, actively facilitating translations between the various parties to the risk assessment hearing. If, as ANT posits, knowledges and facts (and even power) are produced via the stabilization of networks of actors – for as Annemarie Mol (2002) reminds us, ‘To be is to be related’ (p. 54, emphasis in original), and as Foucault (1990) told us long ago, power is ‘strictly relational’ (p. 95), exercised between actors rather than by actors – then we should also be attuned to how objects contribute to the processes of stabilization, even if we simply wish to ask ‘how does this human, in this case, claim to know x?’

As we hope to have shown, forensic hospital reports are not merely documents that contain discrete facts and observations that are then applied and interpreted by human actors. Rather, its material presence also contributes to manufacturing the very concept of the threatening individual that its discrete facts are assumed to be presuppose. More practically, then, these documents have serious effects on the legal status of NCR persons through their ability to allow the Board to claim they are a significant threat to public safety based on facts that were never necessarily significant nor threatening.

Of course, the ORB's network also includes many other documents and objects not explicitly discussed here, including those from other jurisdictions such as civil statutes and public health acts/forms (see Shaw et al., 2023), which may themselves influence the creation of the forensic hospital report and the risk-knowledges we claim it helps produce (hence the importance of the network). However, at our level of analysis, it is clear that without the particular construction of the hospital report, certain claims and decisions rendered by the all-powerful Ontario Review Board may not have been possible.

Although we stressed the potentialities of the physical ‘ink and paper’ report, since that was the form the report took in our case files, we recognize that different knowledge moves would likely be elicited if the document were to change to a new material form (e.g. if it remained solely as an electronic document, viewed and transported digitally). The materially contingent nature of knowledge production we highlight here would again be consistent with the observations of Van Oorschot (2014), who found that judges struggled to ‘see their cases’ when the filing system of their court changed from paper to digital. The judges’ knack for constructing the case out of the mass of physical documents at their disposal had always relied on more traditional processes of ‘material mediation’, such as highlighting and annotating via pen and paper – processes that could no longer be completed in the same way when documents started coming across their computer screens as opposed to their literal desks. The effect was a different conceptualization of the case altogether.

In RJ's hearing, we see a similarly contingent construction via the hospital report: the content of the report and its material form work together to mediate each other as well as the significantly threatening individual already presumed to be ‘out there’. Thus, we believe scholars interested in analyzing legal decisions and knowledges, and in particular their conditions of possibility, should be attuned to the productive interplay between the discrete claims and facts existing within knowledge networks, and the specific materiality of the socio-legal objects and actors that not only carry them, but through that process also help to produce the very facts and things they carry. In particular, we foresee opportunities to merge this analytical framework with other programmes related to risk assessment, such as the new penology and the purported rise of actuarial methods within legal complexes. If nothing else, even a pragmatic use of ANT can help expose if and when we fall into the trap of assuming that some conception of risk is pre-given, instead reminding us that predictive tools and their related objects do not simply tell us if someone is a risk, but that they also help to bring the risky individual into existence.