Abstract
Judges generally make pretrial release decisions based on the risk that a defendant will miss court appearances or pose a danger to the community; for criminal cases initiated in state courts, this decision-making is guided by state bail statutes, with judicial discretion within statutory limits. Individuals with mental illnesses may be disproportionately exposed to pretrial detention, exposing them to the psychiatric, medical, and legal consequences of incarceration. The purpose of this literature review was to find out if and how mental illness is related to missed court appearances or rearrest while individuals are awaiting trial. This review identified a limited number of empirical studies examining the association between mental illness and these outcomes. The results of the studies were variable. Although some studies found that mental illness variables predicted failure to appear, the relationship was more prominent for women than men. Substance use and criminal history appear to be more consistent predictors than mental illness. The data on the relationship between mental illness-related variables and rearrest are also mixed. Further research is needed to understand if or how mental illness affects decisions on pretrial detention.
The physical and emotional harms of incarceration have been well documented. The most frequently encountered carceral settings involve pretrial detention in local jails, which puts individuals at risk for drug or alcohol withdrawal and overdose, homicide, and suicide.1 Individuals with mental illnesses, in particular, are at higher risk for self-harm, suicide, and sexual victimization than those without mental illnesses in correctional settings.2,–,4 Individuals with mental illness also may be disproportionately exposed to pretrial detention through higher jail admission rates and longer lengths of stay compared with those without mental illness.5,–,7
Pretrial detention not only leads to immediate loss of liberty and disrupts income, housing, and families but also can have later legal consequences, particularly for individuals charged with misdemeanors who are not facing prison time. Defendants detained in jail on a misdemeanor charge, compared with those released to the community, are more likely to plead guilty, to be convicted at trial, to have longer incarceration sentences, and to be arrested for future misdemeanors and felonies within 18 months after an initial bail hearing.8 They may plead guilty merely to get out of jail, and convictions may make it more difficult for them subsequently to access jobs and housing.8
Judges generally make decisions on pretrial release based on the risk that the defendant will fail to appear at a court hearing or of being dangerous to the community. At the state level, where the vast majority of criminal court cases are adjudicated, this decision-making is guided by state bail statutes, with judicial discretion as allowed by statute. Guidance on the role of mental illnesses in pretrial release decisions varies by state. For example, certain states allow or require a court to consider a defendant’s mental condition when determining either eligibility for pretrial release or when setting the terms of release. Seven states explicitly permit consideration of mental illnesses in determining eligibility for pretrial release,9,–,15 including Massachusetts, where mental illness can be considered when a defendant is charged with any crime “involving physical force or abuse.”13 Terms of release may include the amount of bail or mandated treatment for mental illness or substance use. This may lead to pretrial incarceration if conditions of release create such insurmountable obstacles that the defendant is guaranteed to remain in custody (e.g., excessive bail) or the defendant fails to meet the conditions of release, resulting in revocation of bail.
Given the consequences of pretrial detention and ongoing efforts to divert individuals away from jail, some jurisdictions have reformed their bail laws to limit the types of charges for which pretrial detention is allowed.16 Additionally, some jurisdictions have adopted pretrial risk assessment tools, which aim to inform release decisions (and reduce overuse of pretrial detention) by providing objective evidence about a defendant’s risk of failure to appear and threat to public safety.17
Mental illnesses may also be factored into pretrial decisions through pretrial risk assessment instruments. The use of pretrial risk assessment tools is common; a recent study found that 48 percent of jurisdictions reported use of a pretrial risk assessment tool.18 The most commonly used tool is the Virginia Pretrial Risk Assessment Instrument, which includes substance use (but not mental illness) as a predictive variable.19 A review from 202020 that looked at six pretrial risk assessment instruments showed good to excellent predictive validity; only one of the tools in this study (Colorado Pretrial Assessment Tool) assessed for “past or current mental health treatment” (Ref. 21, p. 5). This tool also assessed substance use. Four tools assessed substance use only (and not mental health); one assessed neither substance use nor mental health.20
Other tools include the Connecticut Pretrial Risk Assessment Point Scale, which has a combined item asking about “history of substance abuse or mental health issues” (Ref. 22, p. 6). The Ohio Risk Assessment System-Pretrial Assessment Tool (ORAS-PAT) lists “mental health issues” as an item that can be checked off under “other areas of concern.”23 Notably, a report from the Council of State Governments about improving the pretrial process for people with mental illness recommends using “mental health status” in a pretrial risk instrument with extreme caution, as there is a lack of evidence that all persons with mental illnesses are at risk for pretrial failure.27
It remains unclear how judges operationalize these statutes or utilize information from risk assessment instruments in practice. In a 2019 survey on the importance of various factors in assessing pretrial risk, 63 percent of the sample of 42 judges felt that the defendant’s mental condition was extremely or very important in the decision to release or detain during the pretrial period; however, the study did not indicate if judges would be more or less likely to detain individuals based on mental condition.25 Data from Canada suggest that courts sometimes exhibit a reluctance to release individuals with mental illness and that mental illness can often be conflated with unpredictability and risk in the community.26
Existing data are mixed about how judges consider mental illness. Some data suggest that judicial stereotyping about mental illness, particularly related to genetic essentialist biases toward mental illness, negatively affects judicial views on sentencing (i.e., leading judges to express more support for restrictive sentencing, deterrence, and incapacitation).27 In contrast, data from federal sentencing indicate that judges tend to give sentences below the guideline range for individuals with mental illnesses; judges may believe those with mental illnesses are less culpable or are at higher risk of adverse effects from incarceration than those without mental illness.28 Similarly, a 2018 study that evaluated Mississippi judges’ perceptions of dangerousness and treatment needs found that attitudes toward mental illness were generally positive; most participants, including judges, had favorable views about mental health treatment and an accurate understanding of mental illness.29
It is not clear whether defendants with mental illnesses are more likely to miss court dates. A review of the criminal histories of 240 patients recently discharged from inpatient psychiatric facilities found that failure to appear was a common charge (along with probation violations, officer obstruction-related charges, and disorderly conduct). Given the lack of a comparison group (defendants without mental illness) and additional data, it is not clear what might be driving failure to appear among this population.30 As described by Pope et al.,31 in light of studies documenting no-show rates for mental health appointments that range from 10 to more than 60 percent, one might anticipate that people with mental illnesses would be more likely to miss court dates as well. Risk factors for nonappearance in court may be similar to those for missed mental health appointments: time in the community (i.e., between hospital discharge and first appointment or between release and first court date), substance use, psychosocial impairment, poor access to transportation, competing work and family obligations, and mistrust of criminal legal and mental health systems.31,32
Policies governing pretrial release should be rooted in empirical evidence and should seek to balance protecting communities from violence, safeguarding the liberties of individuals who are not yet convicted of crimes, and protecting vulnerable populations from the harms of jail. To date, no review has examined the literature on whether mental illness per se predicts failure to appear, rearrest, or pretrial dangerousness among people released to the community during the pretrial period. The purpose of this scoping literature review is to better understand how and whether mental illness is related to these outcomes.
Methods
Our methods were generally consistent with PRISMA guidelines for a scoping review (PRISMA-ScR). Although we did not prospectively register a protocol, methods were defined prior to the search, with some modifications in light of early results, as detailed below.
Eligibility criteria included studies that examined the relationship between mental illness and outcomes of pretrial court nonappearance, pretrial risk of dangerousness (violent behavior), or pretrial detention versus release. Only English language studies were included. We excluded studies outside the United States (given differences across legal systems), studies that focused on youth (given the different legal systems and different factors resulting in court nonappearance or dangerousness), and studies that focused on substance use.
To search for articles, combinations of key search terms were entered into four databases: PubMed, PsycInfo, Web of Science, and HeinOnline. Proquest was initially probed, but a preliminary search revealed many articles that were either duplicates or unrelated. These databases were chosen as they broadly covered the major areas of science, psychiatry and psychology, and social science in which relevant articles were deemed likely to appear. Search terms included bail, mental illness, mental health, pretrial detention, pretrial release, pretrial risk, failure to appear, and court nonappearance. The same combinations of key words were used across all four databases, with the exception of “pretrial detention and mental illness,” which was not used in Hein Online, given high numbers of results and many duplicates. Searches in PubMed, American Psychological Association (APA) PsycInfo, and Web of Science revealed a relatively small number of articles, so no year cutoffs or additional filters were applied. Certain searches in HeinOnline, however, revealed a much larger number of titles. When that occurred, the search was filtered to include only “article” as the type of result and limited to articles published in the United States after the year 2000 in “peer-edited” journals (defined by Hein Online as a journal that is edited by professionals in the field).
Searches were completed by April 2023. Titles were screened, and those not relevant to bail, mental illness, pretrial detention, failure to appear, or pretrial recidivism were excluded. One author (EN) screened the titles and abstracts to decide whether the study met inclusion criteria.
Sources of evidence included empirical studies that examined the relationship of mental illness to the outcomes of pretrial court nonappearance, pretrial risk of dangerousness (violent behavior), or pretrial detention versus release. We did not include other data sources, such as law reviews, expert opinions, or policy documents.
Figure 1 illustrates the identification and screening of articles. Entering the search terms above revealed 1,482 titles (including titles that may have shown up in more than one search or database). Titles and abstracts that did not clearly address the role of mental illness in pretrial release, failure to appear in court, or pretrial dangerousness were excluded. After excluding those articles, a total of 48 were reviewed more closely. Of those, 44 studies were excluded based on the criteria above, leaving four articles. These 44 articles are listed in an online supplement. One of the four remaining articles was a 2011 meta-analysis of studies on pretrial outcomes that examined 13 studies (although ultimately included only six for final synthesis). These 13 citations were also reviewed individually: three articles were identified that examined mental illness as a predictor of pretrial outcomes, two of which were not previously identified and were added to this review. Thus, in total, we identified five individual studies that empirically examined mental illness as a predictor of a pretrial outcome. These articles are summarized in Table 1.33–37 The references of these five studies were also screened; this did not yield any additional articles for inclusion.
Figure 1. Flow diagram for identification of studies.
A spreadsheet was created with variables that were chosen with other authors and revised based on experience as articles were added. Only one author did the data extractions. Data extracted from the articles included: study aims, study site or jurisdiction, sample size, charge type, portion of the sample with mental illnesses (if reported), definition of “mental health” or “mental illness” used, rates of failure to appear and rearrest, and correlations between mental illnesses and outcomes.
Results
We begin by summarizing each study’s methods, including how each study defined and assessed mental illness, as there were distinct differences across studies. The findings are then summarized according to outcomes of failure to appear (i.e., missed court appearances), rearrest, and any pretrial failure (a combined measure). Although we initially aimed to identify studies that examined the relationship between mental illness and violent behavior, no such studies were identified.
Study Methods and Definitions of Mental Illness
Two studies included dichotomous variables indicating the absence or presence of a “mental illness”33 or “mental health problem”34 and whether or not the individual had received treatment. The first of those reports, a 2003 study of 622 individuals in Connecticut, examined effects of mental illness and substance use on outcomes of bail type, amount of bond, likelihood of being released on bond, and failure to appear.33 Data were collected from courts’ judicial information systems (electronic data), some original case files, and follow-up background checks. Variables included whether the individual had a “mental illness and/or substance abuse problem” (Ref. 33, p 11) (yes or no) and whether the individual had received treatment for a “mental or substance abuse problem” (Ref. 33, p 11) (none, previous treatment, or current treatment).
The second of these studies sought to identify the relationship between pretrial risk factors and pretrial failure (defined as either missing a court date or being rearrested for a new offense), with the aim of developing the Virginia Pretrial Risk Assessment Instrument.34 The study included 1,971 Virginia cases. Data were collected from personal interviews with the defendants, criminal history and court records, criminal justice supervision records, and references provided by defendants. Data included “current mental health problem” (yes or no) and “current physical or mental health treatment” (yes or no) (Ref. 34, p 27).
The other three studies included constructed variables to define mental illness. Lowenkamp et al.35 examined the association of 63 items with failure to attend court appearances and commission of a new offense while released. The sample included 342 individuals across multiple jurisdictions. Data collection integrated structured interview questions and a self-report questionnaire. When possible, conclusions were verified through review of the defendant’s case file. Authors reported that they used a “Medical and Mental Health” construct that included four items, although these items were not specified.
Two studies by Gehring et al. in 201436 and 201837 examined the same sample of 266 pretrial individuals in Hamilton County, Ohio. The sample included 163 men and 103 women who were arrested between January 11, 2010 and February 26, 2010. Data were collected by interviewing pretrial defendants using the Hamilton County Inventory of Need Pretrial Screening Tool. Outcomes included failure to appear for scheduled court appearances and arrest for a new offense in the community. The earlier Gehring et al.36 study included an additional combined outcome measure of any pretrial failure.
In their 2014 study, Gehring and colleagues36 created three different “mental health” scales from a list of potentially relevant variables: one for the whole sample, one for men only, and one for women only. Each scale comprised different but sometimes overlapping combinations of variables (e.g., history of psychiatric hospitalization, prior psychiatric diagnosis, history of taking psychiatric medications). Items were only included in the respective scales if they predicted at least one outcome (failure to appear in court, rearrest, any pretrial failure) for that group (entire sample, men, women) (see Table 1 for the full list of variables for each scale). The scale for women, for example, included a history of being hospitalized for a “mental health problem,” having seen things or heard voices that were not there, having been diagnosed with a mental illness, homelessness because of mental illness, attention deficit-hyperactivity disorder (ADHD), bipolar disorder, and posttraumatic stress disorder (PTSD).
Study Characteristics
Failure to Appear (FTA)
Rearrest
Any Pretrial Failure
Similarly, in the later study, Gehring summed five dichotomous items to create one cumulative “history of mental illness” scale: ever seen a counselor, psychologist, or psychiatrist; been in a “mental health hospital or in a mental health unit”; seen things or heard voices that were not really present; taken any prescribed medication for a psychological problem or to help them feel better emotionally; and been diagnosed with a mental illness (Ref. 37, p. 123).
Failure to Appear
Four studies33,35–37 evaluated mental illness variables specifically as predictors of failure to appear (FTA) for scheduled court dates (see Table 2). The fifth study had one combined outcome variable that included whether someone missed a court appearance or was rearrested, without separating the outcomes.34 The studies showed mixed findings. Two studies found no relationship between mental illness and FTA.33,35 The two studies using the Ohio sample did find that the “mental health” or “history of mental illness” scales predicted FTA, although with gender differences. In the 2014 study, “mental health” variables predicted FTA for women and men (each scale used different items), but stronger correlations were seen for women (bivariate correlation (Pearson’s r, one-tailed) .51 (p ≤ .01) at four and six months versus .14 (p ≤ .05) at four months and .21 (p ≤ .01) at six months for men). 36 The “history of mental illness” scale that comprised five items in the 2018 study strongly predicted FTA for women (bivariate correlation .31, p < .01) but did not predict FTA for men.37
Notably, three of these studies found that substance use predicted failure to appear.35,–,37 Two studies utilized substance use scales comprising several variables, including items such as financial, marital, and emotional problems related to substance use and substance use being involved in the current offense.36,37 The third found that any drug use history and severity of reported problems from drug use predicted failure to appear.35
In the second study, substance use predicted FTA for the entire sample and for women.36 For the entire sample, the correlations for substance use were slightly stronger (r = .23, p ≤ .01 at four months; r = .27, p ≤ .01 at six months) than for the “mental health scale” variables (r = .22, p ≤ .01 at four months; r = .20, p ≤ .01 at six months).36 Criminal history was also consistently a stronger predictor of FTA than mental illness. One study coded “mental illness and/or substance abuse problem” as one variable, finding no correlation with FTA (Ref. 33, p. 11).
Rearrest
Three studies35–37 evaluated mental illness variables as a predictor of rearrest during the pretrial period (see Table 3). Lowenkamp et al.,35 using the four-item “Mental and Medical Health” construct, found no relationship. The Ohio studies that utilized multi-item scales found that these scales predicted rearrest for women, but not for men.36,37 The correlations for women, although significant, were weak. The 2014 study found bivariate correlations of .26 (p ≤ .01) at four months and .21 (p ≤ .05) at six months36; the 2018 study found a correlation of .17 (p < .05).37 Notably, for the entire sample in the 2014 study, the “mental health scale” did not predict rearrest; substance use, criminal history, and employment and financial status were all stronger predictors of rearrest than mental health.36 A similar trend was found for men, where both substance use and criminal history predicted rearrest, but mental health did not. For women, however, mental health was a stronger predictor than criminal history and substance use. The 2018 study, using path analyses, found that, for women, childhood physical and sexual abuse both led to a history of mental illness, which in turn contributed to substance use, which then contributed to rearrest (and to failure to appear).
Any Pretrial Failure
Two studies34,36 evaluated mental illness as a predictor of “any pretrial failure,” defined as FTA or rearrest without separating the two outcomes (see Table 4). One found that “current mental health problem” and “current physical or mental health treatment” did not predict pretrial failure (Ref. 34, p. 27); the other found that the “mental health scale” did predict failure, although correlations were stronger for women (bivariate correlations of .44 (p ≤ .01) at four months and .40 (p ≤ .01) at six months), as compared with men (no significant correlation at four months; .01 (p ≤ .05) at six months).36
Discussion
Overall, the literature contains a small number of empirical studies examining the relationship between mental illnesses and failure to appear in court and rearrest. The results of the studies were variable. Potential explanations for this variability include small sample sizes, problematic approaches for identifying the presence of mental illness, lack of stratification by type and severity of mental illness, absence of measures of treatment, and variable follow-up times. Furthermore, even those studies that supported a relationship found that relationship to be weak in comparison to other variables. In regard to failure to appear in court, only two studies (out of four) found that mental illness-related variables predicted failure to appear, and the relationship seemed to be more prominent for women compared with men. In contrast, both substance use and criminal history appear to be more consistent predictors of failure to appear. The data on the relationship between mental illness-related variables and rearrest also appear to be mixed; one study showed no relationship, and two studies, which used the same sample, supported a relationship only for women.
Furthermore, no studies looked specifically at violent behavior in the pretrial period. Although studies measured recidivism, this may not necessarily represent a good proxy for dangerousness. For example, rearrest may have been precipitated by nonviolent offenses (including failure to abide by the conditions of bail) and thus not indicative of a danger to the community. Data from New York State (outside of New York City) in 2022 related to rearrests within 180 days of arraignment indicate that, out of the total number of releases, 11 percent of defendants were rearrested for nonviolent felonies and eight percent for misdemeanors, as opposed to three percent for violent felonies.38
The small number of studies that exist have substantial limitations. First, in some studies, it is not clear how exactly mental illness was assessed, for example, through self-report versus record review. Each approach has problems of its own. Self-report data can be inaccurate, because individuals may want to deny a history of mental illness for fear of stigma or alternatively may be motivated to feign a history of mental illness to access preferred services or more desirable housing units in correctional settings.39,40 Additionally, definitions of mental health or mental illness were not always clear or precise. For example, the Virginia study used “mental health problem” but did not indicate how “problem” was defined (i.e., if this reflected a previous diagnosis versus active symptoms) or which specific questions were asked of the defendant to assess the presence of a current mental health problem.28 Notably, mental health seemed to predict outcomes, albeit weakly, when more granular definitions were used, as in the two Ohio studies based on the same sample.36,37
Furthermore, studies generally used dichotomous variables (i.e., the absence or presence of a mental illness or a mental health problem) without specific diagnoses or more granular detail (other than the Ohio studies), which may affect results. Different illnesses may affect someone’s ability to appear in court or likelihood of recidivism in different ways. Two individuals with the same diagnosis might differ in their risk of missing a court date or posing a risk to the community, depending on the acuity and severity of their symptoms and whether they are receiving treatment. For example, someone experiencing acute, severe psychotic symptoms of an underlying chronic psychotic disorder may have significant thought disorganization that makes it more difficult to focus on a court date and navigate the necessary means to get to court. A defendant with a chronic psychotic disorder who is receiving treatment and experiencing remission of symptoms may not have the same difficulties. These relationships might further be complicated by substance use disorders, which are often comorbid with mental illnesses and predictive of FTA and rearrest in several of the studies above. Data also show that substance use predicts recidivism following release from prison to a greater extent than mental illness does.41 Future studies should examine specific diagnoses and symptom patterns, comorbidities, and whether or not defendants are receiving effective treatment.
None of the studies examined specific charge severity. Risk for missing a court date or posing a danger to the community might be different based on if an individual is charged with a misdemeanor, a nonviolent felony, or a violent felony. An additional limitation is the potential dilution of the relationship between mental illness and the measured outcomes by virtue of selective decisions about release. If judges were often accurate in identifying defendants with mental illnesses who were at high risk of FTA or rearrest and decided to detain those individuals, those with mental illnesses not detained would have a lower risk of FTA or rearrest, thus obscuring the true relationship between mental illnesses and primary outcomes.
Overall, there is limited evidence supporting the use of mental illness as a factor in deciding whether to release or detain a defendant prior to trial. Retaining defendants with mental illnesses is problematic, particularly given the harms of jail and the downstream legal consequences for those charged with misdemeanors (i.e., being more likely to plead guilty, to be convicted, have longer incarceration sentences, and to commit future crimes).1,8
This scoping literature review has limitations. Certain key words were not used, including mental disorders, bond, or bail jumping, which could have yielded further results. A more systematic assessment of bias was not conducted. In addition, only one review author selected studies and extracted data. Although a second reviewer would have been optimal, the criteria for inclusion were sufficiently clear that it is unlikely that the articles included would have been different, even if a second person had been involved. Finally, only a small number of studies were identified, making it difficult to draw firm conclusions about the relationships among mental illness, failure to appear, and rearrest; this emphasizes the important need for additional research.
Studies focusing on substance use were excluded. Although substance use disorders and substance use are often comorbid with other mental illnesses, not all substance use is associated with a mental illness. For our purposes, we were specifically interested in mental illnesses and their impact on pretrial behavior. This is consistent with both the literature and standard pretrial risk assessments, which identified mental illnesses and substance use as distinct variables.
Further research, both qualitative and quantitative, is needed to understand if or how mental illness affects pretrial detention decisions. This research should ideally follow defendants until final case disposition. Individuals with mental illnesses may be more likely to be remanded to jail because of their mental illnesses (while also being at greater risk of the physical and emotional harms of jail) without convincing evidence that pretrial detention is necessary to ensure that they appear in court or to prevent recidivism or violence in the community. Furthermore, if future research does demonstrate a relationship between mental illness and failure to appear or pretrial dangerousness, policies and practices should focus on how to optimize success for individuals with mental illnesses in the community, for example, by leveraging pretrial community-based services and programs as opposed to resorting solely to detention.
To this end, some jurisdictions have adopted collaborative courts, such as drug courts and mental health courts, that connect individuals to community-based treatment and supervision, rather than rely on pretrial detention. Although this serves as one way to divert individuals from jails, some have noted the potential harms of these programs, including perpetuating false ideas that mental illnesses, in and of themselves, are frequent drivers of criminal behavior (as opposed to socioeconomic factors like poverty, homelessness, and unemployment). These programs may also perpetuate racial inequities; data show that White people are over-represented in mental health courts whereas Black and Latinx populations are over-represented in correctional settings.42,43
In sum, this scoping review found limited and inconsistent data on the relationship between mental illness variables and pretrial outcomes. These findings suggest a need for additional rigorous research before these variables can appropriately serve as the basis for decisions about pretrial disposition.
Footnotes
Disclosures of financial or other potential conflicts of interest: None.
- © 2025 American Academy of Psychiatry and the Law
