In times of austerity, it is important for models of mental health care to be efficient as well as effective. One strategy is to ensure that care is appropriately matched to clinical need, reserving more intensive interventions for patients with more severe symptoms. Although models of care that incorporate treatments of varying intensity, such as stepped care (1–3), have entered mainstream mental health care practice in the United Kingdom, it is unclear how such developments have translated to health care in jails.

In the United Kingdom, primary care organizations within the National Health Service hold responsibility for both jail and community health care services. Many offenders have complex and multiple health needs, with higher rates of mental illness and substance misuse than in the general population (4–7). Outside custody, offenders have chaotic patterns of contact with health services; thus imprisonment represents a valuable public health opportunity to engage a socially excluded group. Aspects of the jail setting may, however, complicate provision of effective mental health care, including drug and alcohol withdrawal regimens, uncertainty about release dates, and security measures. Notably, levels of distress and suicide rates are highest during the first month of imprisonment (8,9). Yet there are few published studies of how appropriately offenders are triaged to different levels of mental health care during early custody.

A prospective study was conducted to examine the relationship between the severity of psychiatric symptoms and the level of health care intervention received during the first month in jail. The null hypothesis was that no difference would be found in the clinical, demographic, or criminological characteristics of inmates with mental illness who received different levels of mental health care or substance misuse interventions in custody.

Setting and participants

Individuals entering custody in five English jails between April 2006 and February 2007 were recruited. All jails accepted individuals directly from court, and roughly half had not been convicted. In England, all new inmates undergo mandatory, brief standardized health screening on their first day in custody to detect immediate general medical and mental health needs, followed by a detailed health assessment the next day. Screening includes questions addressing previous psychiatric treatment, use of prescribed psychotropic medications, substance misuse, and risk of self-harm or suicide. On the basis of health screening, referrals are made to relevant primary care, mental health, or substance misuse services for further assessment and treatment.

Inmates who did not speak English, had been transferred from another jail, lacked capacity to give consent, or whom jail staff considered unsafe to interview were excluded. After participants received a full description of the study, written informed consent was obtained. Ethical approval was given by the Thames Valley Multisite Research Ethics Committee. Participants were screened for mental illness (N=3,079) with the Prison Screening Questionnaire (PriSnQuest) (10). Participants screening positive (N=1,097) were invited for full clinical interviews with the Schedule for Affective Disorders and Schizophrenia (SADS) (11) to diagnose psychiatric illness. Of these, 81% (N=887) were subsequently available and willing to be interviewed.

All participants who screened positive using PriSnQuest and who were diagnosed with the SADS as having a current mental illness were identified (N=524). We successfully located and accessed clinical records for 78% of these individuals, giving a final sample size of 409. [A flowchart depicting the sampling strategy is available as an online data supplement to this article.] Tests for attrition bias found no significant differences between participants with accessible records and those with nonaccessible records in terms of race-ethnicity, age, index offense, and primary psychiatric diagnosis. We were, however, more successful in accessing women’s records than men’s (N=127, 98%, versus N=282, 71%; χ²=41.6, df=1, p<.001).

Interviews

Interviews to diagnose mental illness and substance misuse and to assess symptom severity were conducted by trained researchers within one week of participants’ entry into custody. Demographic and criminological information was also collected. SADS was used to diagnose psychiatric illness, and diagnoses were grouped into three hierarchical categories: schizophrenia or any other psychosis, major depressive disorder, and other mental illness (including minor depressive disorder, general anxiety disorder, obsessive-compulsive disorder, and phobias). The Michigan Alcoholism Screening Test (MAST) (12) and the Drug Abuse Screening Test (DAST) (13) were used to diagnose alcohol and drug misuse, respectively. Both the number and severity of symptoms were assessed, as recommended by current U.K. guidance for using stepped-care models within mental health care (14). Severity of psychiatric symptoms was assessed with the expanded, 24-item version of the Brief Psychiatric Rating Scale (BPRS-E) (15) and the 12-item General Health Questionnaire (GHQ-12) (16). Severity of symptoms associated with substance misuse was assessed with MAST and DAST total scores, which address lifetime and current alcohol and drug use. Training in conducting the assessment was facilitated by an experienced consultant psychiatrist, with regular training updates throughout the study to ensure consistency and competence.

Clinical records review procedure

Clinical records (paper and electronic), including all health care and substance misuse treatment records available in the jail, were retrieved for each participant and reviewed to determine interventions received during the first month of custody. Following a standard review protocol, we identified, recorded, and coded all consultations, assessments, treatment interventions, and care planning events documented in clinical records. Participants’ clinical records were accessed at weekly intervals for up to one month or until they left the jail, whichever was sooner.

Individuals were allocated to one of three hierarchical groups according to the highest level of mental health intervention received during their first month in jail: secondary care intervention (including contact with a psychiatrist, psychologist, or member of the specialist mental health team), primary care intervention (including contact with the medical officer, general practitioner, or general nursing staff), and no intervention. Among individuals with co-occurring disorders (mental illness and a substance misuse problem as diagnosed with the MAST or DAST), interventions received from substance misuse services during the first month in custody were also noted, including contact for detoxification with medical or nursing staff or substance misuse team workers.

Statistical analyses

Analyses were conducted with Stata, version 10, for Windows (17). Descriptive statistics (means and percentages) were used to describe the characteristics of participants allocated to each intervention level. Associations between dichotomous variables were examined by chi square tests and odds ratios. For continuous and normally distributed variables, differences between group means were tested with t tests (two-tailed). A p value of <.05 was considered significant. To complement p values from t tests, effect sizes were estimated with Cohen’s d (the standardized difference between two means) with the metan command in Stata. Logistic regression modeling techniques were used to identify multivariate predictors of level of intervention in jail, with demographic, clinical, and criminological characteristics as independent variables.

Sample characteristics are described in Table 1. Two-thirds (69%) of participants were men, and the mean age of the sample was 33 years. Fifty-three percent had not been convicted. Overall, 20% of participants had a diagnosis of schizophrenia or any other psychosis, 60% had major depressive disorder, and 20% had other types of mental illness. Eighty-four percent had either a drug or alcohol misuse problem.

Level of mental health care intervention

Demographic, clinical, and criminological characteristics of the groups that received primary mental health care, secondary mental health care, and no intervention are summarized in Table 2. During their first month in custody, 133 individuals (33%) received secondary mental health care, 181 (44%) received primary mental health care, and 95 (23%) received no intervention.

We first compared individuals who received no intervention with those who received a primary care intervention, excluding those who received secondary mental health care (Table 2). Individuals treated in primary care were more likely to be female (odds ratio [OR]=7.07, 95% confidence interval [CI]=3.19–15.68), to have a diagnosis of depression (OR=2.01, CI=1.20–3.36), and to have received one of the following medications immediately before custody: antidepressant (OR=2.4, CI=1.30–4.60), hypnotic or anxiolytic (OR=3.16, CI=1.54–6.50), or detoxification medication (OR=4.28, CI=1.81–10.14). Compared with individuals treated in primary care, those who received no intervention were more likely to have a diagnosis of psychosis (OR=2.43, CI=1.23–4.77), to not be convicted (OR=1.77, CI=1.06–2.96), or to be charged with or convicted of a violent index offense (OR=1.85, CI=1.02–3.34). No significant associations were found between level of intervention and symptom severity.

Next, we compared individuals who received primary mental health care with those who received secondary mental health care, excluding those who received no intervention (Table 2). Individuals who received secondary care were more likely to be male (OR=1.66; CI=1.03–2.67), to have a diagnosis of psychosis (OR=3.34, CI=1.81–6.17), to report any previous contact with mental health care services (OR=2.32, CI=1.4–3.87), to report being in contact with mental health care services immediately before custody (OR=2.05, CI=1.15–3.64), and to report being prescribed antipsychotics immediately before custody (OR=4.10, CI=1.97–8.55). Compared with those treated in secondary care, those treated in primary care were more likely to have been discharged from custody within a week (OR=9.53, CI=2.13–42.62), to have a diagnosis of mental illness other than depression or psychosis (OR=2.14, CI=1.14–4.04), and to report being prescribed a hypnotic or anxiolytic (OR=1.70, CI=1.00–2.92) or detoxification medication immediately before custody (OR=2.90, CI=1.50–5.60). Individuals treated in secondary care had higher BPRS-E scores (t=–2.96, df=296, p=.003), indicating more severe symptoms; however, the effect size was small (d=.35).

All variables that had a statistically significant association (p<.05) with level of intervention (as indicated in Table 2) were entered into an ordered logistic regression model, with level of mental health care intervention as the ordinal dependent variable (hierarchically defined as none, primary, or secondary). The following variables remained significant (p<.05): gender, major depressive disorder, duration in custody, prescribed antipsychotic use before custody, and BPRS-E total score. The overall model was significant (χ²=52.26, df=13, p<.001).

Substance misuse intervention for co-occurring disorders

A total of 344 (84%) participants had a coexisting drug or alcohol misuse problem. Overall, 180 (52%) individuals with co-occurring disorders received an intervention from substance misuse services within one month of jail arrival (Table 3).

Participants with a drug misuse problem were more likely to receive an intervention than those with an alcohol misuse problem alone (OR=3.67, CI=1.91–7.05). Symptoms of drug misuse (total DAST score) were significantly higher in the group that received an intervention (moderate effect size, d=.59). No significant differences were found in symptoms related to alcohol misuse (MAST scores) between those who did or did not receive an intervention. Intensity of psychiatric symptoms was not associated with receipt of an intervention from substance misuse services.

Receipt of an intervention from substance misuse services was more likely among men (OR=1.91, CI=1.18–3.11) and among white individuals (OR=2.26, CI=1.15–4.44). Receipt was less likely among those with a violent index offense (OR=.58, CI=.34–.96); however, this difference was not significant when adjustments were made for alcohol misuse. No significant group differences were found with respect to legal status and duration in custody. Individuals who had been in contact with community drug treatment services immediately before custody were more likely to receive an intervention in jail from substance misuse services (OR=2.86, CI=1.84–4.44). Participants who had received inpatient drug detoxification in the past were less likely to receive an intervention while in custody (OR=.47, CI=.31–.71). With regard to prescribed medication, individuals prescribed detoxification medication prior to custody were more likely than those not prescribed detoxification medication to receive an intervention from substance misuse services (OR=3.56, CI=1.90–6.67).

All variables that had a statistically significant association (p<.05) with receipt of a substance misuse service intervention (as indicated in Table 3) were entered into a binary logistic regression model, with receipt of an intervention from substance misuse services (yes or no) as the binary dependent variable. Only ethnicity, inpatient drug detoxification (lifetime), and prescribed detoxification medication immediately before custody remained significant (p<.05). The overall model was significant (χ²=54.12, df=9, p<.001).

We found that level-of-care decision making in early custody was not consistently linked to the clinical characteristics of patients with mental illness in jails. We noted a significant preference for treating psychosis in secondary rather than primary care and for treating illnesses other than psychosis and depression in primary rather than secondary care. However, 27% of inmates with schizophrenia or other psychosis and 19% of those with major depressive disorder received no treatment at all during their first month in custody. Furthermore, differences in symptom severity between those who received different levels of mental health intervention were either nonsignificant or not clinically meaningful.

Narrative reviews of the literature have highlighted several limitations of models of jail mental health care in the United Kingdom, including poor triage systems, limited resources, lack of specialized services, and inadequate primary mental health care (18–20). Notably, the wider literature suggests that such problems are not unique to the United Kingdom (21–23). Jails may lack the resources to cater for the full range and level of need in the population. Although spending has increased over the past few decades in the United Kingdom, funding for offenders with mental illness arguably remains low in relation to the level of need (24). Furthermore, although secondary mental health care “in-reach” teams have become well-established in U.K. jails, primary care has arguably remained underdeveloped in comparison, leaving these teams overstretched (18,19).

In line with the wider literature on offender health (7,25), a diagnosis of mental illness with a coexisting drug or alcohol misuse problem was the norm rather than the exception in this sample. Just half of participants with co-occurring disorders received an intervention from substance misuse services. Those who misused drugs were significantly more likely than those who misused alcohol alone to be treated. The literature also notes a comparative neglect of people with alcohol dependence and a lack of funding for alcohol-specific treatment services in jail. A recent thematic review concluded that despite high levels of unmet need, specific assessment and support services for alcohol users were limited and inconsistent, especially for those without concurrent drug problems (26). This is concerning, especially given that alcohol misuse is associated with a wide range of health and social problems, including mental illness, homelessness, and violent offenses (7).

In this study, psychiatric diagnoses were made on the basis of direct assessments of individuals with standardized tools that were validated for research purposes. This strategy allowed systematic measurement of symptoms at a set time point for every participant. However, it would be unrealistic to expect complete concordance with diagnoses made by other mental health care professionals responsible for these patients. Differences of opinion could result from the types and timing of assessments, supporting information available, or differences in skills and training. Many offenders have complex needs, and determining psychiatric diagnoses in this client group may be difficult, even by qualified professionals with unlimited resources. We recognize that our diagnoses are time dependent and primarily based on patient self-report; however, use of standardized measures increased the reliability of diagnoses.

Measurements of symptoms were taken during the first week of custody to ensure the correct temporal sequence between symptoms and health care service response. A longer period of observation, while interesting, would have introduced additional complexity to the analysis, arguably without commensurate benefit. The first month in custody represents the most critical period for the identification of mental illness; health screening processes are focused at reception into custody, and a third of all suicides among prisoners in England and Wales occur during the first week of custody (9). Nonetheless, the findings cannot be generalized beyond this period.

This study depended on information about service contacts documented in clinical records of individual patients. Unrecorded contacts would not have been captured. In some cases, data collection in participating jails was completed before the introduction of easily accessible electronic records systems. Reliance on paper records, which are often held in multiple locations within a single jail, resulted in some loss to follow-up. Nonetheless, records were located for 78% of participants who met inclusion criteria, and no evidence was found of attrition bias with respect to key variables (except gender) in terms of differences between individuals who were or were not included in our final sample.

Although health care is not the primary function of a jail, incarceration provides an opportunity to detect, diagnose, and treat mental illness in a socially excluded population that is often reluctant to engage in health care services outside of custody. Nonetheless, in the current climate of financial austerity, mental health service providers need to consider how the health care needs of offenders can be met within existing, if not reduced, resources. The findings of this study suggest that during the critical initial period of custody in jail, inmates with mental illness are not always allocated to the appropriate level of health care intervention and are sometimes not treated at all. At best, this suggests that other clinical or organizational factors may have been influential; at worst, care decisions may have been made arbitrarily after inadequate triage.

For care in early custody to be efficient as well as effective, triage mechanisms and coordination between primary and secondary care services must be improved to ensure that resources are used wisely and that patients are allocated to appropriate types and levels of intervention. Optimistically, others have suggested that increasing cost-consciousness could encourage more radical and innovative approaches in psychiatry (27). Notably, stepped-care service models (1–3), which offer a range of treatment options that vary in intensity, have been integrated into mainstream care pathways in the United Kingdom for the treatment of common mental illnesses such as depression (14) and anxiety (28). These service models have also been recommended for jail health care services (29).

Offenders often present a complicated clinical picture, especially within the uncertain, unfamiliar, and frequently inhospitable environs of a jail setting. The findings of this study indicate that the optimum service configuration has yet to be found. Meeting this challenge is essential if communities and individuals are to benefit from the wider socioeconomic and public health rewards at stake.

The study was funded by the National Forensic Mental Health Research and Development Programme (grant MRD12/87).

The authors report no competing interests.