A study of psychotropic medication prescribing patterns in prisons in England and Wales

Hypnotics and anxiolytics (4.1)

Hypnotics are generally used to relieve insomnia, while anxiolytics are effective in the relief of acute anxiety, tension and agitation. The group of drugs known as the benzodiazepines are the most commonly used hypnotics and anxiolytics. 27 The main difference between different benzodiazepine drugs is the variation in the half-lives (the period of time for the blood concentration of a substance to decrease by half). Half-lives are clinically relevant, as longer-acting benzodiazepines (e.g. diazepam) are commonly used as anxiolytics, whereas those with shorter half-lives are more often used as hypnotics (e.g. temazepam).

Though benzodiazepines are clinically effective, they carry the risk of physical and psychological dependence. Users can rapidly develop tolerance, meaning that greater doses are required to achieve the same effects. Furthermore, after longer periods of continuous use (> 4 weeks), withdrawal symptoms have been reported, including insomnia, anxiety, flu-like symptoms, stiffness or weakness and possibly seizures. 27,28 Users may misinterpret discontinuation symptoms as signs of relapse. 29 Furthermore, benzodiazepines can be misused for recreational purposes, either alone or in combination with street drugs such as opiates or stimulants. 28,30

In consideration of these risks, NICE has recommended that benzodiazepines are indicated only in cases where symptoms are severe, disabling or causing extreme distress, and then only after non-pharmacological methods have been considered. 31 Benzodiazepines are not recommended for chronic conditions, such as GAD, post-traumatic stress disorder (PTSD) or OCD, except as a short-term measure during crises. 32–34 In particular, it has been suggested that repeat prescriptions should be avoided in patients with a history of substance misuse or marked personality disorder. 27,28 Certain long-acting benzodiazepines (usually chlordiazepoxide) are also used in the management of alcohol withdrawal in primary care. 30

Drugs used in psychoses and related disorders (4.2)

Antipsychotic drugs, also referred to as neuroleptics, are used in the treatment and prophylaxis of schizophrenia, bipolar disorder and other psychoses. They also have a calming effect on agitated or disturbed patients. Antipsychotic drugs have been grouped according to both the pattern and the mechanism of clinical action. 34 The BNF identifies four categories of neurological ‘extrapyramidal’ side effects (EPSEs) associated with older, first-generation (‘typical’) antipsychotic medications (e.g. chlorpromazine or haloperidol): parkinsonian symptoms (e.g. tremors); abnormal face and body movements (dystonia); restlessness; and tardive dyskinesia, a potentially irreversible condition characterised by involuntary facial contortions, commonly involving the tongue, face and jaw. 27 Subsequently, second-generation (‘atypical’) drugs were launched in the 1990s, which avoided EPSEs, making antipsychotics more tolerable. While first-generation antipsychotic drugs are thought to work by interfering with dopaminergic transmission in the brain, second-generation drugs have a ‘rich pharmacology’, meaning that they act on multiple receptors. 35,36 More recently, a third generation of antipsychotic drugs, with different pharmacological properties, has been introduced. 37 For the purpose of this research, second- and third-generation antipsychotic drugs have been grouped together and will be referred to as atypical antipsychotics. NICE has recommended that atypical antipsychotics should normally be used as first-line treatment with newly diagnosed schizophrenia. 38

Clozapine is the only atypical drug found to be superior in efficacy to typical antipsychotics. 39 Though not a first-line response to psychosis, clozapine has emerged as the gold-standard treatment for patients with ‘treatment resistant schizophrenia’,38,40 defined as those whose illness has not responded to at least two different antipsychotic drugs, including an atypical antipsychotic agent. 38

Mood stabilisers are used in the treatment of hypomania, mania and bipolar disorder to reduce the severity of symptoms, to stabilise mood and as a preventative measure. In the UK, lithium, sodium valproate and certain antipsychotics (olanzapine, quetiapine and risperidone) are licensed for the treatment of acute mania. NICE recommends valproate as a first-line treatment for acute episodes of mania, characterised by periods of overactive, disinhibited behaviour. 41 In such cases, randomised controlled trials have shown that valproate has a response rate of 50%. 28 Potential side effects include weight gain and gastric irritation; it is also teratogenic and should not normally be prescribed for women with child-bearing potential. 41 Lithium has a slower onset of action than valproate and antipsychotics, but is regarded as a useful treatment for moderate symptoms of mania. Side effects of certain medicines used to treat mania include tremors, weight gain and gastric irritation. 27

Antidepressant drugs (4.3)

Antidepressants are used mainly to treat symptoms of depression and anxiety and can also be used in combination with antipsychotics and drugs to treat mania, bipolar disorder and psychotic depression. While antidepressants have been found to be beneficial in the treatment and prophylaxis of moderate to severe depression, they are not recommended for mild depression. 42 Unfortunately, the risk of recurrent depression is high; over half of patients who have a depressive episode will go on to have a second episode. 43

In the BNF, antidepressant drugs are divided into four classes: selective serotonin reuptake inhibitors (SSRIs), tricyclic antidepressants (TCAs), monoamine-oxidase inhibitors, and other drugs. 27 TCAs were the first antidepressants to be introduced. There is little difference in terms of efficacy between the different groups of antidepressants. 27 A flexible approach is recommended when choosing antidepressant drugs, taking into account risk of interactions, side effect profiles and patient preferences. NICE has recommended that the newer SSRI drugs (e.g. fluoxetine, citalopram and sertraline) should normally be used first-line as they are safer in overdose and more tolerable than other antidepressants. 42 However, a series of case reports published two decades ago sparked concerns regarding a possible link between SSRIs and suicidal thoughts and behaviour, though this remains to be conclusively proven or disproven. 44–48 While antidepressants are not addictive, it has been known for years that patients may experience unpleasant effects such as dizziness, mood changes, gastrointestinal disturbances and insomnia on reducing/missing or stopping them. 49

Central nervous system stimulants and drugs used for attention deficit hyperactivity disorder (4.4)

This subchapter of the BNF includes amphetamines and other stimulant medicines, mainly used to treat ADHD. In the Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition (DSM-IV), ADHD has been defined as a ‘persistent pattern of inattention and/or hyperactivity–impulsivity that is more frequently displayed and is more severe than is typically observed in individuals at comparable level of development’ (p. 85). Methylphenidate, atomoxetine and dexamfetamine are normally the drugs of choice with regard to ADHD. ADHD is normally diagnosed in childhood; however, adult ADHD is a recognised condition. 50–52 Drugs within this class have a high propensity for diversion, abuse and dependence. 53 Indeed, the effects of methylphenidate have been compared to those of cocaine, although there are distinct pharmacological differences between the two. 54 Thus, where there is a history of substance misuse, prescribers should be cautious about prescribing, and additional monitoring may be required. 27

Historical perspectives

Although psychopharmacology is central to contemporary mental health care, the use of psychotropic drugs in prison has, historically, been controversial. In prisons in England and Wales during the 20th century, as in psychiatric hospitals, psychotropic drugs became an important new tool in doctors’ ‘medical armoury’. 55 The ability of drugs to manage difficult behaviour and create ‘a quiet prison’ was not lost on prison doctors or discipline staff. For many years, doctors were open about using drugs to help make prisoners more manageable and amenable to the regime; for example, as late as 1978, a prison doctor wrote a paper concerning the use of the antipsychotic drug Depixol^® (Flupentixol, Lundbeck) to ‘treat’ individuals with a psychopathic disorder. 55 However, as prisons became more open to external scrutiny, reports began to surface from prisoners and ex-prisoners who claimed to have been drugged against their will, or who had apparently accepted treatment only as a result of coercion and threats. 55 Increasingly, critics of prison health care and patient pressure groups questioned whether or not drugs were prescribed in prisons for disciplinary purposes to control difficult individuals, rather than to treat mental illness. 55

In 1980, the prison service published data concerning the use of psychotropic medicines in prisons as part of its annual report. 56,57 Unfortunately, the style of analysis and presentation of figures made it difficult to derive meaning from the data. 58 The lack of clarity regarding how drugs were categorised, absence of data on dose and aggregation of data across multiple prisons were some of the problems noted; some went as far as to say that the figures were purposely designed to be ‘as misleading as possible’ (p. 250). 59 Notwithstanding their limitations, the data did highlight the wide variation in dosage rates between prisons.

Current approaches and policy

Today, prescribing psychotropic medication in prison remains a complex topic in current clinical practice, undertaken at the boundary of the care/custody interface. In accordance with the principle of equivalence, prisoners are entitled to access pharmacy services and appropriate medicines, as they would do in the wider community. Nonetheless, health-care professionals have argued that certain aspects of prison-based health care, notably around medicines management, may need to be altered in prisons in order to mitigate risk and to safely discharge the prison’s duty of care to prisoners.

Part of the problem is that staff commonly believe that a minority of prisoners may present with exaggerated or fictitious symptoms in order to attempt to acquire medication to misuse or sell. 60–62 This presents a number of safety and security risks, both to individuals taking illicitly obtained medicines and to vulnerable individuals with genuine mental health needs, who may come under pressure to share prescribed medication with others.

Currently, there is no national agreed formulary in operation in prisons in England and Wales. However, in 2011, the Royal College of General Practitioners (RCGP) Secure Environments Group issued Safer Prescribing in Prison, which evaluated the suitability of individual medicines for use in prison. 61 Written from the position that prescribers in prisons need to balance security and safety risks against individual health needs, this document introduced a traffic-light system for medicines based on risk for harm. It classified medicines as either green, indicating lower-risk and first-choice options (including SSRI antidepressants); amber, suggesting that these may be used with caution, or after other choices have been unsuccessful (e.g. antipsychotics); and red, for medicines considered inappropriate in prisons (e.g. temazepam, a short-acting benzodiazepine).

Although a welcome and useful contribution to the literature, the RCGP guidance was written from a prescriber viewpoint and strongly focused on risk management and drug choice. Although this and other policy guidance documents have acknowledged the values of equivalence and patient-centred care, there are indications that achieving the optimum balance between health care and security remains a significant challenge. For example, while medicines are meant to be held in the possession of prisoners as a matter of principle, research has suggested that staff remain cautious about the risks of allowing this. 5,60

In qualitative research studies, patients have complained about discontinuity of prescribing between the community and prison, preoccupation with security and inadequate access to medicines. With reference to mental health in particular, policy clearly states that medication for mental disorder should not be automatically withdrawn on entry into prison without proper clinical assessment. 24 Yet studies have reported that prisoners experience problems with continuing established medication regimes on entry into prisons, causing significant frustration and distress. 62–66 These reports are corroborated by a records-based study in five English prisons, which found that half of all psychotropic medicines prescribed in the community were discontinued in prison, often without evidence of clinical review or other justification recorded in the notes. 67

The inherent complexities associated with medicines management in prisons are further compounded by a lack of access to robust prescribing data. Unlike in the community, high-quality prescribing data are not routinely available from prison-based prescribers. 5 In a thematic report on mental health, the Chief Inspector of Prisons expressed concern that psychotropic medicines may be overused in prisons, in comparison with non-pharmacological interventions such as ‘talking therapies’. 68 Without robust data, however, questions like these are unlikely to be answered.

The Office for National Statistics study

The 1997 ONS survey of psychiatric morbidity in England and Wales provided detailed, high-quality data on rates of psychotropic prescribing in prison. 6 Table 2 provides a summary of the psychotropic prescribing rates reported by the ONS for adult men and women in prisons and young people in young offender institutions (YOIs).

One-fifth of men in prison were prescribed some form of medication acting on the CNS, which included psychotropic medicines and also analgesics, antiepileptics and drugs for substance dependence. Among this group, the most commonly prescribed medicines were antidepressants, followed by hypnotics and anxiolytics, and antipsychotics. Rates of prescribing were higher in remand prisoners than in sentenced prisoners.

Prescribing among women in prison followed a similar pattern, though rates were even higher: women were twice as likely as men to be prescribed CNS medicines. Overall, half of women were prescribed CNS drugs, one in five women received antidepressants and one in 10 received antipsychotics. In YOIs, 10% of young men were prescribed CNS drugs. For young men, unlike for adult men, rates of prescribing were the same among remand and sentenced individuals. In line with those for adult prisoners, rates of prescribing were much higher among young women than among young men; overall, 40% of young female prisoners were prescribed CNS drugs, with hypnotics and anxiolytics being the most commonly prescribed types of drug (14%).

While the ONS study provided some useful data, a number of limitations are noteworthy. First, the data were collected 15 years ago. Owing to a variety of factors, it is likely that psychotropic prescribing will have changed during this time. Second, the ONS surveys were focused on determining the prevalence of psychiatric morbidity, rather than on medicines use. Inevitably, this gives rise to a number of methodological limitations. There is a limited level of detail provided on medicines use, including a lack of data on drug types, individual drugs and doses prescribed. Certain aspects of the study design and analysis could also be seen to create difficulty in interpreting data. For instance, psychotropic drugs used to treat mental illness are often pooled together with other CNS drugs, which have quite different clinical indications (e.g. drugs for substance dependence, epilepsy and analgesics).

Furthermore, the use of antipsychotic medication was used as an indicator of probable psychosis; as the authors acknowledge, this means that associations between diagnosis and medication use are inevitable, but not necessarily accurate. While there is some consideration of prescribing patterns within particular diagnostic groups, this is complicated due to comorbidity. For example, if an individual has more than one diagnosis, it might be unclear which condition is being treated by a particular medication. Such factors prevent us from drawing any firm conclusions regarding the appropriateness of prescribing patterns from these data.

Finally, one of the purported strengths of the study was that, as one of a series of national psychiatric morbidity surveys with similar methods and measures, comparisons could be drawn between different groups, for example with the general population. 70 However, comparing crude prescribing rates does not take into account the substantial differences in the age and sex mixes between prison and community populations (prisoners are, on average, younger and more likely to be male). Furthermore, the way in which the data are presented makes certain comparisons difficult: for example, it is not easy to compare relative rates of psychotropic prescribing and mental illness in prisons and communities.

The east of England study

Until recently, the ONS study was the only published source of robust research data available on psychotropic prescribing patterns in prisons in England and Wales. One major difficulty which has impeded efforts to investigate prescribing has been information technology. Until recently, prisons relied on paper-based clinical records and adopted varied pharmacy data management systems. Thus, previous attempts to collect pharmacy data were time-consuming, small-scale and impractical. 5 In 2011, the roll-out of a single health-care information technology (IT) system (SystmOne, The Phoenix Partnership, Leeds, UK), in common use in community primary care settings, was completed throughout the prison estate, creating a novel opportunity to link prescriptions with individual patients.

In 2010, researchers from the Offender Health Research Network (OHRN) at The University of Manchester conducted a study which tested the feasibility of extracting data from patient clinical records held in SystmOne to establish psychotropic prescribing patterns in prisons. 71 The research was completed in four prisons in the east of England, covering the counties of Bedfordshire, Cambridgeshire, Essex, Hertfordshire, Norfolk and Suffolk. This particular region of the UK was chosen largely for practical reasons: several prisons in the region had already adopted SystmOne and the enthusiasm and support from the local offender health and medicines management team guaranteed access to prisons and resources. Of the four prisons recruited to the study, two accepted both adult and young men, one accepted adult and young women and one accepted adult men only. Three of the prisons recruited accepted both convicted and unconvicted prisoners and the fourth accepted only convicted prisoners.

On census days, local health-care staff used SystmOne to identify and collect prescribing and demographic data on all prisoners with current, valid prescriptions for hypnotic, anxiolytic, antipsychotic, antimanic, antidepressant and/or stimulant medication, as listed in subchapters 4.1 to 4.4 of the BNF. For comparison purposes, we also obtained a large data set on a random sample of CPRD patients in the community, under a free-licence scheme funded by the Medical Research Council (MRC) (now expired).

The findings of the study indicated that one-fifth (20%) of men and almost half (44%) of women in prison were prescribed at least one psychotropic medication. After adjusting for age differences, rates of psychotropic prescribing in prison were 5.5 and 5.9 times higher than in community-based men and women, respectively. We also found marked differences in the individual psychotropic drugs prescribed in prison and community settings.

The study had a number of limitations. Only prisons in the east of England were sampled, a region which accounted for approximately 10% of prisons in England and Wales at the time, thereby limiting generalisability. Furthermore, while the study attempted to collect data on indications for medicines, these were of questionable reliability, having been collected by several different health-care staff. Analyses on costs, doses or drug interactions were not completed; thus, it was difficult to make inferences regarding the appropriateness of prescribing. Despite these limitations, the study allowed us to develop and refine a data extraction methodology and proved the feasibility of using SystmOne for research purposes. It also enabled us to obtain a valuable community data set and provided some interesting preliminary data on prescribing patterns in prisons in one region of England.

Evidence from international studies

Aside from the UK-based literature, a number of studies have been published describing psychotropic prescribing in prison populations internationally, particularly in Europe and the USA. For example, in 2004, a national cross-sectional study of Norwegian prisons reported that 34% of men and 44% of women in prison were in receipt of medication acting on the CNS. 72 No significant sex differences were found with regard to prescribing. Overall, they concluded that the rate of prescribed psychotropic drug use in prison was higher than in the general population but lower than that observed in psychiatric hospital units.

A Swiss study73 compared prescribing outcomes among a sample of prisoners (n = 179) attending primary care consultations over a 3-week period in 1997. The study showed that psychotropic drugs (mostly anxiolytics and hypnotics) accounted for about half of drugs (46%) prescribed at these appointments. Female prisoners were significantly more likely to be prescribed psychotropic drugs than men (77% vs. 56%), although it should be noted that the female sample was very small (n = 22). Notably, the study also recruited a group of community-based patients (n = 701), who attended an ambulatory care clinic (outpatients only) at a local hospital in an urban area over the same period, for comparison purposes. In an effort to control for age-related differences between the populations, comparisons were limited to male patients in prison and the community aged < 39 years of age. An analysis of this subgroup showed that psychotropic medication was prescribed five times more often at appointments in prison than at the hospital clinic.

Several studies have been completed in the USA reporting the prevalence of psychotropic prescribing in prisons. In response to concerns about rising pharmacy costs, Lund et al. 74 undertook a longitudinal analysis of prescribing trends in the Iowa Department of Corrections prison system. The study utilised a combination of prison population data and drug expenditure records over the years 1990–2000 to determine trends in annual expenditure per inmate. Overall, annual expenditure on psychotropic drugs increased dramatically from $291 per 100 inmates in 1990 to $8138 in 2000, a 28-fold difference. Both the volume and the costs of antidepressants, mood stabilisers, hypnotics and anxiolytics all increased. However, while the overall volume of antipsychotic prescriptions showed a slight decrease, expenses increased ninefold. The authors concluded that a shift towards newer (and safer) drugs was largely responsible for the increase in costs, placing additional strain on already scarce resources.

Baillargeon et al. 2,75,76 and Williams et al. 77 conducted a series of studies in Texas Department of Criminal Justice (TDCJ) prisons. Unlike most of the studies discussed previously, these studies focused on prescribing patterns within two particular diagnostic categories, namely prisoners with depressive disorders and prisoners with psychosis. Retrospective data on prescribing patterns were collected from the clinical records of a large sample of TDCJ inmates with depressive disorders (n = 5305) in 1998–9. 2,76 Data were also collected on 3750 TDCJ prison inmates who were (a) diagnosed with schizophrenia and/other psychotic disorders and (b) receiving antipsychotic medication. 75 In TDCJ prisons, medication prescription and compliance data are maintained within computerised medical records systems on all inmates receiving prescribed medication. Individual-level data on prescriptions, clinical characteristics and sociodemographic data were obtained for all participants.

A detailed analysis of the antidepressant prescribing patterns showed that, overall, 47% of inmates with depressive disorders were prescribed SSRIs, 31% received TCAs and 22% received no antidepressant medication. The researchers found that:

• Men and older prisoners (aged > 50 years) diagnosed with depression were less likely to be prescribed antidepressants than their female and younger counterparts.

• Women, young people (aged < 30 years) and white inmates were more likely to be prescribed SSRIs.

• Male sex, older age and being prescribed TCAs were all positively associated with ‘compliance’ scores (i.e. the number of doses taken divided by the number of doses prescribed during the study period).

Furthermore, among prisoners with schizophrenia treated with antipsychotics:

• Eighty-five per cent received first-generation (typical) antipsychotics; this rose to 89% for other forms of psychoses.

• Women, black men and non-violent prisoners were less likely than their counterparts to be prescribed second-generation (atypical) drugs.

The authors concluded that these findings were reflective of both the increased costs of atypical antipsychotics and newer SSRI drugs costs, and also TDJC prescribing policies; at the time, local formularies stipulated that first generation antipsychotics should be attempted first-line. The authors also noted that SSRI use appeared to be lower among prison inmates than among USA community and inpatient populations (no equivalent community data were available for antipsychotics).

Griffiths et al. 78 recently conducted a systematic review of the literature on prescribing psychotropic drugs to prisoners. This included qualitative and quantitative research on prisoners located in a variety of settings, including forensic hospitals, high-security psychiatric wards, prisons and jails. The review generated a final sample of 32 papers. The studies included in the review examined topic areas such as polypharmacy, prescribing errors, high-dose antipsychotic prescribing and prescribing in older prisoners. Eighteen papers originated in the UK: of these, most (n = 12) were from secure forensic hospitals. The key themes identified from the reviewed studies were polypharmacy, high-dose therapy, duration of treatment, documentation and monitoring, and issues associated with the prisoners’ environment. The authors concluded that polypharmacy, use of high doses and poor continuity of care were common among prisoners prescribed psychotropic medicines. Nonetheless, the authors noted that, owing to the heterogeneity of settings in which studies were conducted, these issues might not apply equally to all settings. They also noted that research to date had focused on antipsychotics, rather than on the broader category of psychotropic medicines. Furthermore, where prevalence studies had been conducted, these often lacked comparisons between key population subgroups, such as between sexes, age groups and ethnicities.

Overall, the evidence from international studies of prescribing in prisons is useful, yet far from comprehensive. Studies appear to indicate elevated use of psychotropic medication among prisoners in comparison with the communities from which they are drawn, a finding which could reflect the increased rates of psychiatric morbidity generally observed in such populations globally. 8–12,20,21,79 Some studies also reported noticeable sex differences in psychotropic use. 2,6,73,77 In such instances, women were generally more likely than men to be prescribed psychotropic drugs, which is consistent with the increased psychiatric morbidity in this group.

The most obvious gap in the existing literature relates to studies which provide comparison data between the general prison population and the community or other settings. 72,73 Community data are arguably useful as they allow us to consider the extent to which there is equity and continuity of prescribing between settings. Unfortunately, neither of the studies which made such comparisons adequately accounted for age and sex differences between populations.

A second limitation concerns the use of drug volume and expenditure data in some studies. 72,74 While making use of readily available data is a simple and efficient strategy, prescriptions cannot be matched to individuals; therefore, this method relies on using mean dosages to calculate use/costs per head. Use of individual-level data, as in the TDCJ studies,2,75,76 while more time-laborious, is far superior as it has the potential to better account for the full range of pharmacological, clinical and demographic factors.

The relationship between psychotropic prescribing and psychiatric morbidity

A key issue that is relevant to research in this area, and indeed to this study, is the extent to which patterns of psychotropic prescribing are mediated by the presence of mental illness. More specifically, a question pertinent to the current study is as follows: if rates of psychotropic prescribing are higher among prisoners, is this simply due to increased rates of psychiatric morbidity?

For a number of reasons, this relationship is not straightforward. While it is likely that underlying patterns of mental illness exert an influence on patterns of psychotropic prescribing, we would not expect complete correspondence between the two rates. The first reason for this is that not all patients with mental illness will be treated with psychotropic medicines, or indeed treated at all. Secondly, patients prescribed psychotropic drugs may not necessarily have a mental illness. Psychotropic medications may be prescribed ‘off label’, outside the narrow terms of their licences, to treat conditions other than mental illness. Thus, while we can expect rates of psychiatric morbidity to exert an influence, this is unlikely to fully account for any variation observed in rates of psychotropic prescribing.

Prison sample

Based on our preliminary work in the east of England,71 we expected the overall point-prevalence of psychotropic prescribing in prisons and the community to be approximately 26% and 8%, respectively.

In a comparative study of percentages, the minimum sample size needed to have a 90% chance of detecting this difference as significant (p = 0.05, two-sided) is 90 per group. However, to perform the detailed subgroup analyses needed for this study (e.g. the proportion of atypical vs. typical antipsychotic prescriptions among black male prisoners), a much larger sample is required. Based on our preliminary work, we aimed to recruit 12 average-sized prisons (n = 500) to ensure that the study was adequately powered, even for detailed subgroup analyses. On this basis, we expected to generate prescriptions for 360 hypnotics/anxiolytics, 360 antipsychotics and 840 antidepressants (Table 4).

In total, 11 prisons were recruited to the study. A range of prisons in England and Wales were invited to participate in the study. Following the preliminary work in the east of England, a number of sites had already expressed interest in participation. Others were recruited via direct e-mail enquiries to health-care managers. Recruited prisons were geographically spread across northern England, the south-west of England, London and the Midlands. The sample included three local prisons, three training prisons, two women’s prisons, two prisons for young offenders and an open prison. Each prison type is further explained in Box 1.

Data collection procedure

For logistical reasons, it was not possible to use a single census day across all prisons; therefore, a census day was identified for each individual prison. Census days were dependent on the timing of approvals and site access arrangements; all census dates were between 30 November 2012 and 30 July 2013.

On census days at participating prisons, population figures (stratified by age, legal status and ethnicity) were obtained from each individual establishment using Prison National Offender Management Information System (P-NOMIS) for use as denominators for calculating prevalence rates. At each prison, SystmOne was used to select all patients who met the following inclusion criteria: (1) aged 18 years or over; (2) in prison custody; and (3) in receipt of a valid, current prescription for psychotropic medication. At three prisons, SystmOne was not used for all prescribing. At these prisons, drug charts were searched manually for patients who met the inclusion criteria. For each patient who met the inclusion criteria, a researcher then extracted the relevant demographic, clinical and prescription data from their clinical record and recorded this on a data extraction pro forma.

For each patient, the following data were collected:

• demographic data, including prison, sex, legal status, ethnicity and year of birth

• physical and mental health diagnoses, as recorded on the standard prison health-screening assessment tool.

For each psychotropic prescription identified, the following data were collected:

• drug name

• dose

• formulation

• indication.

To measure the appropriateness of prescribing, we applied the prescribing appropriateness indicator (PAI) to each psychotropic prescription. The PAI is a set of nine standardised, validated indicators83,84 that are designed to be applied to prescribing data held in medical records. For the current study, we used a subset of five of the indicators deemed to be applicable to psychotropic prescribing (see Appendix 4 for PAI operational guidance):

1. The indication for the drug is recorded and upheld in the BNF.

2. The reason for prescribing a drug of limited value is recorded and valid (according to the BNF, the drug is of limited or no clinical value if it is considered by the Joint Formulary Committee to be less suitable for prescribing. The use of the drug may be justifiable under certain circumstances, which should be clearly documented in the patient medical records).

3. A generic (non-branded) product is prescribed if one is available.

4. If a potentially hazardous drug–drug combination is prescribed, the prescriber shows knowledge of the hazard.

5. If the total daily dose is outside the range stated in the BNF, the prescriber gives a valid reason.

In addition, we reviewed relevant NICE guidelines on anxiety, depression, bipolar disorder and schizophrenia34,38,41,85 to identify other appropriate, measurable standards that may be applied to prescribing. As a result, the following measures were also used:

• the proportion of antipsychotic prescriptions prescribed for GAD

• the proportion of patients prescribed more than one antipsychotic medication, without it being documented as a changeover period

• the proportion of benzodiazepine prescriptions issued for longer than 4 weeks, excluding those prescribed for detoxification purposes

• the proportion of benzodiazepine prescriptions prescribed for GAD or panic disorder.

Prison data set preparation

Using Stata version 12 software (StataCorp LP, College Station, TX, USA), the data were entered into two separate data sets (one for patients and one for prescriptions), linked by an anonymised unique patient identifier generated for the purposes of this research. Data were checked for errors made during data collection/entry. Duplicates were identified and excluded. Range checks were performed to identify impossible numerical values. Consistency checks were used to check for invalid combinations of values (e.g. pregnancy and male sex). Where necessary, continuous (e.g. daily dose) or ordinal variables (e.g. ethnicity) were transformed to form additional categorical variables (e.g. under/over BNF maximum; white/non-white).

Community (Clinical Practice Research Datalink) comparison data

For comparison purposes, we used a data set on a random sample of GP-registered patients in the community from the CPRD. This had been obtained at the time of the preliminary east of England study under a free-licence scheme funded by the MRC (which closed in 2011), and hence the data were from 2010.

The CPRD (www.cprd.com) is a computerised database of anonymised longitudinal medical records from primary care. It is the largest and most representative primary care database in the UK, containing health data on over 4 million active patients from around 488 UK primary care practices. A well-regarded and powerful research tool, it has been used extensively for observational research in primary care, generating over 700 peer-reviewed publications to date. Data are collected using routinely collected data via GP electronic clinical records systems. Data sets available to researchers comprise anonymised patient records, including clinical data on diagnoses, prescribing, comorbidity and demographic details.

Clinical Practice Research Datalink data sets are restricted to general practices and patient records that consistently meet predefined thresholds for levels of quality and completeness, supplying data that are fit for use in research. Only patient data from practices that are ‘up to standard’ are included in research data sets. Individual patient records are also checked for validity issues that would render them ineligible for use in research. Previous population comparisons have indicated that CPRD patients are representative of the general UK population with respect to age and sex. 86

Figure 1 provides an overview of the method used to sample CPRD patients. All CPRD patients who were (a) alive, (b) aged ≥ 18 years and (c) registered with a GP throughout 1 February and 30 July 2010 were eligible for inclusion in the study. CPRD provided total patient counts detailing the number of patients prescribed at least one medicine from each of the following medication groups at any time within the 6-month period specified: hypnotics and anxiolytics, antipsychotics, antimanic drugs, antidepressants and CNS stimulants, and/or medicines for ADHD. All counts, including denominators, were stratified by age, sex and UK region.

FIGURE 1.

Clinical Practice Research Datalink sampling flowchart. a, Sample size calculated on the basis of the mean proportion of the total sample represented by other regions of England and Wales (7.3%).

In addition, CPRD supplied anonymised individual patient-level data for a random sample of 30,602 patients who were prescribed psychotropic medicines. We could not obtain individual-patient-level data (prescription details, demographics, etc.) on all patients prescribed psychotropic medicines in the 6-month period specified, as this exceeded the 100,000 case limit set by the CPRD under the terms of the MRC licence. Thus, we obtained aggregate data on all CPRD patients and individual-patient-level data on a randomly sampled subset of CPRD patients prescribed psychotropic medicines for this purpose; we deliberately included all CPRD patients prescribed CNS stimulants listed in BNF 4.4. This sample included:

• 10,000 patients prescribed hypnotics and anxiolytics (4.1)

• 10,000 patients prescribed antipsychotics and drugs for the psychoses (4.2)

• 10,000 patients prescribed antidepressants (4.3)

• all patients prescribed CNS stimulants or drugs used for ADHD (4.4).

Data analysis

Statistical analyses were performed using Stata software version 12. As the CPRD provided individual-patient data on only a sample of patients prescribed psychotropic medicines, the population denominators supplied (total CPRD counts) had to be adjusted (reduced) accordingly. To generate appropriate denominators for calculating point-prevalence rates, the following calculation was applied to each age/sex stratum (by BNF subchapter):

Period prevalence (6-month) counts and adjusted denominators for CPRD point-prevalence prescribing rates are reported in Appendix 5.

Sex-specific point-prevalence psychotropic prescribing rates [percentage and 95% confidence intervals (CIs)] were generated for prison and community samples for each BNF subchapter, i.e. hypnotics and anxiolytics (4.1); antipsychotics and antimanic drugs (4.2); antidepressants (4.3); and stimulants and drugs used for ADHD (4.4). Rates were indirectly standardised for age where appropriate, using the CPRD data set as the standard population.

Subgroup analyses were performed to examine heterogeneity in prescribing rates (percentage and 95% CIs) among particular subgroups, including (a) sex; (b) prison type (e.g. local, training, open); (c) age group; (d) legal status; and (e) ethnicity. Prevalence ratios (PRs) and appropriate univariate analyses were used to test group differences for statistical significance. Chi-squared tests were used for categorical data and t-tests were used for continuous data.

Key outcomes included the proportion of prescriptions issued in prison that met each of the indicators in the PAI. We also identified the drugs most frequently associated with an inappropriate or no recorded indication, hazardous drug–drug interactions and prescribed daily doses outside the range stated in the BNF.

The mean total cost of psychotropic prescriptions per patient was calculated using prices in the most current version of the BNF. 87 As actual purchasing costs were not available, we calculated the cost for a 28-day prescription for each drug at the dose nearest to the World Health Organization’s defined daily dose (DDD). 88 The DDD has been defined as ‘the assumed average maintenance dose per day for a drug used for its main indication in adults. . . It should be emphasised that the defined daily dose is a unit of measurement and does not necessarily reflect the recommended or Prescribed Daily Dose’. 88

Sample

Three prisons participated in this part of the study, comprising one local prison for adult men, one training prison for adult men and a women’s prison. All three prisons had also taken part in the prevalence survey. At selected GP clinics at each prison, researchers aimed to recruit approximately one in four patients who had booked appointments with the GP. To be eligible to participate, patients had to be:

• aged ≥ 18 years

• able to consent for themselves (if in doubt, the prison health-care team would be consulted)

• suitable to be interviewed by a researcher alone (no risk markers); and

• able to understand spoken English.

In total, data on 156 consultations were generated, which represented 78% of the original target (n = 200).

Procedure

At each prison, researchers worked with participating doctors (n = 6) to identify sessions suitable to sample patients (‘research clinics’). These sessions were all within standard primary care clinics, delivered by GPs (akin to GP consultations in community settings). Standard primary care clinics, rather than dedicated clinics for mental health issues, were sampled to ensure opportunities to recruit prisoners with a wide range of health problems, including common mental illnesses and physical health problems (for comparison purposes). To ensure a more representative sample, sessions were sampled from wings housing remand, convicted and vulnerable prisoners (including clinics on dedicated wings for prisoners who, for safety reasons, need to be segregated).

All doctors had received a participant information sheet and had given formal written consent to participate in the study. All clinics took place over the period July 2013 to October 2013. Once rules had been established for choosing suitable clinics (e.g. one doctor requested that no research should be conducted on days where student doctors were in attendance), the days which would be sampled were not always known in advance to the doctor.

Subsequently, a member of the local health-care team with access to clinic appointment lists identified all eligible patients with appointments booked to see the doctor at these sessions. Patients from each session (approximately one in four) were then selected and approached to participate in no particular order, often depending on availability on the wing. Potential participants were each given a participant information sheet to read or, where preferred, this was explained to them verbally. Written consent was obtained from all patients who agreed to participate.

Over the study period, researchers attended prison GP clinics and asked recruited patients to complete pre- and post-consultation questionnaires (see Appendices 1 and 2), either themselves or as a structured interview if preferred/indicated, for example in cases where participants had literacy problems.

Questionnaires were based on those used by Britten et al. 89 in a similar study based in community primary care settings. Pre-consultation questionnaires aimed to collect basic demographic data, and to identify the nature of the main health problem to be discussed and any secondary health problems. In addition, patients were asked about their expectations and the types of help they wished to receive from the doctor. This was achieved by asking them to indicate their agreement with a series of statements on a three-point scale (agree, uncertain, disagree). Statements addressed areas such as diagnoses, prescribing, medication advice, communication and referrals. After the consultation, each patient was asked to complete a post-consultation questionnaire. These aimed to measure help received and satisfaction with outcomes, including any new prescriptions or changes to medicines.

Following consultations, the doctor was also asked to complete a post-consultation questionnaire for each participating patient in the study (see Appendix 3). In all cases, doctors were not aware which patients were participating in the study until after their consultation. This questionnaire was aimed at measuring understanding of patients’ health problems, perceived pressure to prescribe and satisfaction with prescribing decisions. Doctors were also asked to provide details of drugs prescribed, and whether or not they were strictly indicated.

Data analysis

Key outcome measures included:

• the proportion of patients who, prior to the consultation, wish to start, change or stop a psychotropic prescription

• the proportion of patients who, following the consultation, received a prescription; were satisfied with the consultation; and were happy with the prescribing decision

• the proportion of doctors who, following the consultation, were satisfied with the consultation; felt pressured to write a prescription; and felt comfortable with the prescribing decision (including the decision to prescribe nothing)

• the proportion of prescriptions that were unwanted (by the patient); and not strictly indicated (in the doctor’s opinion).

Descriptive statistics (percentages, frequencies and means) were used to analyse questionnaire responses. Differences between consultations involving psychotropic versus other types of medication were also compared using prevalence ratios (PRs). Predictors of prescribing outcomes, satisfaction and pressure to prescribe were also identified.

Sample characteristics

Table 5 provides an age and sex breakdown of the samples drawn from prisons and the wider community (via the CPRD). Overall, 6052 male prisoners and 785 women prisoners were included in the survey. Based on available prison population statistics dated 26 July 2013 (the week of the final census day), this sample represented 7.5% of the male prisoner population and 20.5% of the female prisoner population.

The CPRD supplied aggregate denominator data for all patients who met the inclusion criteria, which totalled 1,418,347 men and 1,463,848 women (see Table 5). Counts of patients prescribed psychotropic medication within a 6-month period in 2010 were also supplied, broken down by age and sex. In addition, we obtained individual-level patient and prescription data on a random subsample of 10,980 men and 19,622 women prescribed psychotropic medication (see Table 5). Figure 2 provides an illustration of the age differences among the prison and community samples. Individuals in the prison sample were clearly younger: 91% of men and 97% of women in prison were aged ≤ 54 years, compared with 52% of men and 47% of women in the community (CPRD) sample.

FIGURE 2.

Sample characteristics, by age and sex.

Table 6 provides a detailed breakdown of the characteristics of the prisoner sample. Eleven prisons were included in the study. Among the nine participating male prisons, three were category C training prisons, three were adult local prisons, two were institutions for young offenders and one was a category D open prison. Two closed women’s prisons were included. Among the prisoners included in the sample, 88.5% were male, 86.3% were sentenced and 82.9% were of white ethnicity. In 10 prisons, primary health-care services were delivered by NHS providers; in one prison, services were delivered by a social enterprise.

Rates of psychotropic prescribing in prison and the community

Table 7 describes psychotropic medicine point-prevalence prescribing rates for patients in the community, stratified by age and sex. Overall, 5.8% of men and 11.8% of women were in receipt of at least one prescribed psychotropic medication on 30 July 2010. Antidepressants were the most commonly prescribed type of medication, prescribed to 4.6% of men and 10.0% of women. Overall rates of psychotropic prescribing were significantly higher among women than men (PR 2.03, 95% CI 1.98 to 2.09); this was a consistent finding across all medication types, except CNS stimulants/medicines for ADHD. Rates of prescribing increased with age across all medication types, except CNS stimulants/medicines for ADHD, which (as expected) were highest among the 18–24 age group.

Table 8 describes psychotropic medicine point-prevalence prescribing rates for prisoner patients, stratified by age and sex. Overall, 16.9% of men and 47.9% of women were in receipt of at least one prescribed psychotropic medication. As with the community sample, antidepressants were the most common type of psychotropic medication prescribed in prison, prescribed to 13.2% of men and 41.1% of women. Antipsychotics were prescribed to 11.7% of women and 4.3% of men. Rates of psychotropic prescribing were significantly higher among women than men (PR 2.84, 95% CI 2.52 to 3.20); this was a consistent finding across all medication types, except CNS stimulants/medicines for ADHD. Rates of prescribing appeared to be lowest in the youngest age group (18–24 years), with the exception of CNS stimulants/medicines for ADHD, which (as expected) were highest among this group.

Figure 3 shows the relationship between psychotropic prescribing and age among patients in prison and the community (the data for these figures can be found in Tables 7 and 8). Among prisoners, psychotropic prescribing appears to peak in the middle age bands. In the community, however, psychotropic prescribing appears to show a linear increase with age.

FIGURE 3.

Prison and community psychotropic point-prevalence prescribing rates, by age group and sex.

Table 9 compares psychotropic prescribing rates in prison and in the community. After adjusting for age differences, overall rates of psychotropic prescribing in prison were four times higher among men (PR 4.02, 95% CI 3.75 to 4.30) and almost six times higher among women (PR 5.95, 95% CI 5.36 to 6.61) than among patients in the community. Even after adjusting for age differences, rates of CNS stimulant prescribing, in particular, were much higher in prisons: 13 times higher among men (PR 12.99, 95% CI 9.48 to 17.80) and 19 times higher among women (PR 19.01, 95% CI 7.07 to 51.10). Women in prison were over seven times more likely to be prescribed hypnotics and anxiolytics than women in the community (PR 7.3, 95% CI 5.6 to 9.4); in comparison, men in prison were only slightly more likely to be prescribed these drugs than men in the community (PR 1.3, 95% CI 1.0 to 1.6). Rates of antipsychotic prescribing were at least 12 times higher among women in prison (PR 12.74, 95% CI 10.30 to 15.76) and almost five times higher among men in prison (PR 4.81, 95% CI 4.21 to 5.50) than among community patients.

Summary: rates of psychotropic prescribing in prison and the community

• Overall, 17% of the men and 48% of the women we surveyed in prison were prescribed at least one psychotropic medicine.

• After adjusting for age differences, psychotropic prescribing rates were four times higher among men and six times higher among women than in the community.

• As in the community, antidepressants were the most common type of psychotropic medication prescribed to prisoners. Overall, 13% of men and 41% of women prisoners surveyed were prescribed antidepressants.

• Rates of CNS stimulant prescribing for ADHD were particularly high among prisoners of both sexes when compared with patients in the community.

• Prescribing rates for hypnotics and anxiolytics were much higher among women in prison, but were only slightly higher among men in prison, than in the community.

Drug type

In total, 1400 (out of 6837) of the prisoners surveyed were prescribed at least one psychotropic medicine on census days. These prisoners were in receipt of 1740 separate prescriptions for psychotropic medicines. In the community sample, 18,927 patients were prescribed at least one psychotropic medicine, generating a total of 27,621 prescriptions. Thus, among individuals prescribed at least one psychotropic medicine, prisoners were prescribed a mean of 1.2 prescriptions (range 1–5), compared with 1.5 prescriptions among patients in the community (range 1–8). Community patients were more likely than patients in prison to be prescribed more than one psychotropic medicine concurrently (33.0% vs. 20.1%; PR 1.64, 95% CI 1.48 to 1.83).

Table 10 describes the breakdown of these prescriptions by BNF subchapter. This shows that two-thirds (67.5%) of all psychotropic prescriptions in prison were for antidepressants, one-fifth (22.1%) were for antipsychotics, 7.5% were for hypnotics and anxiolytics and 2.8% were for CNS stimulants. Antipsychotic and CNS stimulant prescriptions accounted for a similar proportion of all psychotropic prescriptions in prisons and the community. Antidepressants accounted for a larger proportion of prescriptions in prisons, while hypnotics and anxiolytics accounted for a lower proportion, than in the community.

Table 11 describes which individual medicines were most frequently prescribed in prison, compared with the wider community. In prison, diazepam accounted for half (53.4%) of all hypnotic and anxiolytic prescriptions. Among individuals prescribed hypnotics and anxiolytics, prisoners were twice as likely as patients in the community to receive diazepam (PR 2.48, 95% CI 2.10 to 2.93). In the community, temazepam was the third most frequently prescribed hypnotic and anxiolytic and accounted for 18.5% of these prescriptions, but was prescribed only twice in prison. The most frequently prescribed antipsychotic drugs in prison were olanzapine and quetiapine; both of these drugs were more commonly prescribed in prison than in the community (PR 2.12, 95% CI 1.80 to 2.49; PR 2.25, 95% CI 1.90 to 2.67). There were 49 instances of CNS stimulant prescribing in prison. Methylphenidate accounted for 75.5% (n = 37) of prescriptions, while atomoxetine accounted for the remaining 24.5% (n = 12).

Collectively, SSRIs accounted for the greatest percentage of antidepressant prescriptions in prison (45.1%), TCA drugs accounted for 14.9% and other antidepressants constituted the remainder (40.0%). TCAs accounted for a lower proportion of antidepressant prescriptions in prison than in the community (14.9% vs. 26.5%; PR 0.56, 95% CI 0.49 to 0.65). Mirtazapine was the most frequently prescribed antidepressant in prison (35.2%), but accounted for just 8.3% of community antidepressant prescriptions (see Table 11). Among individuals prescribed antidepressants, prisoners were four times as likely as patients in the community to receive mirtazapine (PR 4.26, 95% CI 3.87 to 4.69).

Table 12 shows the median prescribed daily doses for the most frequently prescribed medications in prison. The median prescribed dose for diazepam, the most popular hypnotic/anxiolytic drug, was 10 mg. This represented one-third of the maximum daily dose identified in the BNF for anxiety (30 mg) and 100% of the World Health Organization’s DDD. Median prescribed doses for the two most popular antipsychotics, olanzapine and quetiapine, were 10 mg and 300 mg, respectively. For olanzapine, this represented 50% of the BNF maximum dose and 100% of the DDD. For quetiapine, this represented 37.5% of the BNF maximum dose and 75% of the DDD. The median prescribed dose for mirtazapine, the leading antidepressant, was 30 mg. This represented two-thirds (66.7%) of the BNF maximum dose and equalled the DDD. In all cases except for melatonin, the median dose prescribed in the community was equal to or lower than the median dose prescribed in prison.

Summary: individual drugs

• Patients in the community were more likely than prisoners to be prescribed more than one psychotropic medicine concurrently.

• Antidepressants accounted for a larger proportion of prescriptions in prisons than in the community, while hypnotics and anxiolytics accounted for a smaller proportion.

• Diazepam accounted for half (53.4%) of all hypnotic and anxiolytic prescriptions in prison. While temazepam was the third most frequently prescribed drug in the community, it was prescribed only twice in prison.

• Prisoners prescribed an antidepressant were at least four times as likely to receive mirtazapine as patients prescribed an antidepressant in the community.

• Prisoners prescribed an antipsychotic were twice as likely as patients in the community to be prescribed olanzapine and quetiapine.

• The median doses of antipsychotics prescribed in prison were all ≤ 50% of the BNF maximum dose.

• Median doses of psychotropic drugs prescribed in the community were usually equal to or lower than median doses prescribed in prison.

Costs

Table 13 describes monthly prescription costs for drugs in each BNF subchapter for men and women. Across prisoners as a whole, the mean overall psychotropic prescription cost per patient was £1.47 for men and £12.98 for women; therefore, costs were almost nine times higher among women. Among women, hypnotics and anxiolytics accounted for the highest proportion of total drug costs; in part, this was due to increased use of the liquid form of diazepam in female-only prisons compared with male-only prisons (80% vs. 25%), which is a more costly preparation than standard tablets (93 pence for 28 tablets vs. £133.63 for the oral solution). Indeed, psychotropic prescriptions issued to women were almost 10 times more likely than those issued to men to be for non-standard preparations, for example liquid, depot and/or orodispersible tablets (19.33% vs. 2.0%; PR 9.8, 95% CI 6.38 to 15.19).

Table 14 describes monthly prescription costs for different prison types (male prisons only). The highest mean overall psychotropic prescription cost was in prisons for young men, at £15.40 per patient. This appears to be attributable to the relatively high cost per patient of CNS stimulants (£11.81 per patient): these drugs were relatively expensive compared with other psychotropics and were prescribed more frequently in these establishments. Costs for adult male local and training prisons were relatively similar overall (£6.74 vs. £5.81); while local prisons spent more on hypnotics and anxiolytics (£1.83 vs. £0.04), training prisons spent more on antipsychotics (£3.18 vs. £2.09). The lowest costs were observed in the open male establishment, with an overall cost of £3.97 per patient.

Summary: costs

• The mean overall psychotropic prescription cost per patient was £1.47 for men and £12.98 for women: costs were almost nine times higher among women.

• Increased drug costs among women were partly due to the increased use of the liquid form of diazepam in female prisons compared with male prisons, which is a more costly preparation than standard tablets.

• Prisons housing young men had the highest mean overall psychotropic prescription costs, at £15.40 per patient.

• The lowest costs were observed in the open male establishment, with an overall cost of £3.97 per patient.

Sex

Table 15 compares rates of psychotropic prescribing among men and women in prison. Women in prison were nearly three times as likely as men in prison to be prescribed psychotropic medication (PR 2.65, 95% CI 2.35 to 2.99). In particular, women were relatively more likely to be prescribed hypnotic and anxiolytic drugs (PR 7.84, 95% CI 5.42 to 11.36).

Legal status

Table 16 compares rates of psychotropic prescribing in convicted and unconvicted prisoners. Among men, rates of psychotropic prescribing were higher among unconvicted prisoners (PR 1.33, 95% CI 1.15 to 1.54). Among women, rates of psychotropic prescribing were similar among convicted and unconvicted prisoners.

Prison type

Table 17 compares rates of psychotropic prescribing in different types of prison for adult and young men. Overall, rates of psychotropic prescribing were lowest at the open prison. Prisons for young men also had relatively low rates of psychotropic prescribing; this was consistent across all medication types except CNS stimulants. Local male prisons had the highest rates of psychotropic prescribing, with 22.3% of men prescribed psychotropic drugs overall.

Ethnicity

Rates of prescribing were compared across different ethnic groupings (Table 18). Overall, rates of prescribing were highest in prisoners who reported being of white ethnic origin. White prisoners were twice as likely as prisoners from black and minority ethnic (BME) groups to be prescribed psychotropic medication (22.2% vs. 9.3%; PR 2.38, 95% CI 1.98 to 2.87).

Summary: subgroup analyses

• Among prisoners, women were nearly three times as likely as men to be prescribed psychotropic medication. In particular, women were more than seven times as likely to be prescribed hypnotic and anxiolytic drugs.

• Among prisoners, psychotropic prescribing appears to peak in the middle age bands and decline in older age. In the community, however, psychotropic prescribing appears to show a linear increase with age, with the highest rates in the older age bands.

• Local male prisons had the highest rates of psychotropic prescribing, with 22.1% of men prescribed psychotropic drugs overall. Overall, rates of psychotropic prescribing were lowest in the open prison. Prisons for young men also exhibited lower prescribing rates; this was consistent across all medication types except CNS stimulants.

• Lower rates of psychotropic prescribing were observed in BME groups in prison.

Prescribing appropriateness indicator 1: indication for the drug

Overall, the indication for the drug was recorded and upheld in the BNF in 65.3% of cases (see Table 19). Compared with other psychotropic medications, antipsychotic prescriptions were the least likely to be accompanied by a valid indication in the patient notes, with 51.6% of these prescriptions meeting the PAI standard (PR 0.75, 95% CI 0.67 to 0.83).

In 21.1% of cases (n = 364), an invalid indication was recorded for a psychotropic drug (Table 20). Antipsychotic prescriptions were more likely than prescriptions for other psychotropic medications to be accompanied by an invalid (not indicated) diagnosis in the patient notes (35.4% vs. 17.2%; PR 2.03, 95% CI 1.70 to 2.44). The most common invalid indications recorded for antipsychotic prescriptions were personality disorder (n = 50), anger/aggression (not associated with psychosis; n = 16) and anxiety (n = 12). Invalid indications recorded for antidepressant prescriptions included low mood (n = 96), insomnia (n = 25), anxiety (n = 24) and personality disorder (n = 22).

A greater proportion of quetiapine (46.0% vs. 19.4%; PR 2.39, 95% CI 1.91 to 2.99) and olanzapine (35.6% vs. 20.0%; PR 1.78, 95% CI 1.37 to 2.31) prescriptions were accompanied by an invalid diagnosis than all other psychotropic medicines.

Prescribing appropriateness indicator 2: drugs of limited value

In 99.1% of cases, drugs of limited value either were not prescribed or were justified with a valid reason in the notes (see Table 19). Overall, there were 17 cases in which limited drugs were prescribed; the most common drug was amitriptyline for depression (n = 14).

Prescribing appropriateness indicator 3: generic prescribing

Overall, generic (non-branded) drugs were prescribed in 99.5% of cases (see Table 19). In total, 62 prescriptions were issued for branded medications. The most common of these was Concerta^® XL (Janssen) (a modified release preparation of methylphenidate, a drug prescribed for ADHD), which was prescribed on 27 occasions. In 14.5% of cases in which a branded product was prescribed, a generic alternative was available.

Prescribing appropriateness indicator 4: drug–drug interactions

Overall, a serious (BNF ‘black dot’) drug–drug interaction involving a psychotropic medication was noted in 15.7% of prescriptions (n = 267). In a minority of cases (6.4%), the prescriber recorded the interaction in the notes, thereby showing knowledge of the hazard. In 85.3% of cases, PAI 4 was met, i.e. either there were no potentially serious drug–drug interactions involving psychotropic medicines or the prescriber had shown knowledge of the hazard (see Table 19). Listed below are the most frequently observed interactions and their potential adverse outcomes, as stated in the BNF: note that not all of these potential hazards would necessarily apply to every patient. For example, if the patient did not have epilepsy, lowering of the convulsive threshold would not be hazardous.

• antipsychotics and antiepileptics (convulsive threshold lowered)

• SSRI and non-steroidal anti-inflammatory drugs (increased risk of bleeding)

• SSRI and antiepileptics (convulsive threshold lowered)

• antipsychotics and methadone (increased risk of ventricular arrhythmias due to QT interval prolongation).

Prescribing appropriateness indicator 5: prescribed daily dose

In 92.9% of cases, the total daily dose was within the range specified in the BNF, or, if not, a valid reason was provided. Note that if there was an indication, we used the dose range for that indication (as described in the BNF). If the indication was invalid, or if there was no indication, we used the smallest minimum and largest maximum dose provided of all indicated conditions listed for the drug in the BNF. The standard was less likely to be met if the drug was an antipsychotic rather than another psychotropic drug (80.6% vs. 96.3%; PR 0.84, 95% CI 0.80 to 0.88).

In 7.2% cases (n = 123), the standard was not met. Among these cases, 95% of the doses prescribed were too low (subtherapeutic). By far the most common drug to be prescribed at subtherapeutic doses was quetiapine (36.8%). In over half (57.1%) of cases where quetiapine was prescribed at subtherapeutic doses, it was for an invalid indication. Other drugs commonly prescribed at subtherapeutic doses included trazodone (15.4%) and risperidone (10.3%).

Other standards

In addition, we measured the extent to which standards derived from relevant NICE guidelines on anxiety, depression, bipolar disorder and schizophrenia33,37,40,84 were met. In summary:

• Overall, 9.7% of all patients prescribed antipsychotic medication in prison were prescribed more than one antipsychotic concurrently (compared with 10.7% of community patients). In four of these cases (11.8%), this was documented as a changeover period. By far the most common drug used in combination was olanzapine (n = 17).

• Overall, 2.1% (n = 8) of antipsychotics were prescribed for GAD.

• Overall, 54.2% (n = 71) of prescriptions for hypnotics and/or anxiolytics were issued for longer than 4 weeks. However, in 70.4% of such cases, the drug was prescribed as part of a benzodiazepine detoxification regime.

• Four prescriptions were prescribed for GAD or panic disorder (3.1% of all hypnotic and anxiolytic prescriptions).

Summary: the appropriateness of psychotropic prescribing in prison

• Valid (BNF-listed) indications for psychotropic drugs were recorded in 65.3% of cases.

• The most common invalid indication recorded for antipsychotic prescriptions was personality disorder.

• The most common invalid indication recorded for antidepressant prescriptions was low mood.

• In 92.9% of cases, the total daily dose was within the range specified in the BNF, or, if not, a valid reason was provided. However, one in five (19.4%) antipsychotic prescriptions were prescribed at a subtherapeutic dose too low to treat psychoses.

• Generic (non-branded) drugs were prescribed in 99.5% of cases.

• A serious (BNF ‘black dot’) drug–drug interaction involving a psychotropic medication was noted in 15.7% of prescriptions.

• One in 10 (9.7%) of patients prescribed antipsychotics in prison was prescribed two or more antipsychotic medicines concurrently.

Sample characteristics

In total, 183 prisoners across three prisons were approached to take part in the questionnaire survey. Of these, 156 prisoners agreed to take part, yielding a response rate of 85.2%.

Table 21 describes the characteristics of the 156 patients recruited to the questionnaire survey. Almost two-thirds (62.8%) of the sample were male and the majority of prisoners were sentenced (87.6%). The mean age was 36 years, and almost half (48.3%) of the sample had been in custody at the current prison for < 6 months. Six GPs were recruited to the study, with an equal split of men and women.

Table 22 shows that, out of the 156 consultations considered as part of the study, the patient pre-consultation questionnaire was completed in 152 (97.4%) cases, the patient post-consultation questionnaire was completed in 134 (85.9%) cases, and the post-consultation doctor questionnaire was completed in 137 (87.8%) cases. In 100% of cases, at least one of these questionnaires was completed, and in 80.8% of cases, all three were completed. Reasons for non-completion of patient questionnaires included patient non-attendance at the appointment, patient refusal and the patient being missed by the researcher. The main reasons for non-completion of doctor questionnaires were patient non-attendance at the appointment and lack of time.

Presenting problem

Before the consultation, patients were asked to state the main health problem that they wished to raise with the doctor. Almost one-quarter (23.8%) of patients identified a mental health problem as their main reason for seeing the doctor (Table 23).

Patients were also asked to define secondary reasons for seeing the doctor, if they had any. Taking this into consideration, 29.1% of patients (n = 44) identified a mental health problem as one of their reasons (primary and/or secondary) for seeing the doctor.

Patient expectations

Patients were asked about the types of help they wanted, by indicating their level of agreement with a series of pre-prepared statements. Table 24 describes the proportion of patients who agreed with these statements.

Overall, two-thirds (69.7%) of patients wished to start, stop and/or change their medication (as indicated by agreement with any of the following three statements: I want a prescription for some medication; I want to change the medication I am taking; I would like to be taken off some medication I am taking). Patients who identified mental health as their primary problem were more likely than patients who identified other types of health problems to want to start, stop and/or change their medication (91.7% vs. 62.6%; PR 1.46, 95% CI 1.23 to 1.74). Patients who identified mental health as their primary problem were also more likely to agree with the following statements:

• I want a prescription for some medication (80.6% vs. 56.1%; PR 1.45, 95% CI 1.15 to 1.82).

• I would like the doctor to offer me a choice of treatments (75.0% vs. 55.7%; PR 1.35, 95% CI 1.05 to 1.73).

• I would like emotional support from the doctor (52.8% vs. 26.1%; PR 2.02, 95% CI 1.31 to 3.13).

• I have emotional problems for which I would like help (77.8% vs. 27.0%; PR 2.89, 95% CI 2.04 to 4.09).

• I want to change the medication I am taking (47.2% vs. 25.2%; PR 1.87, 95% CI 1.17 to 2.99).

They were less likely to agree with the following statement:

• I want to be examined for the cause of my condition (27.8% vs. 56.5%; PR 0.49, 95% CI 0.28 to 0.85).

Help received

Patients were asked about the types of help that they actually received, by indicating their level of agreement with a series of pre-prepared statements.

Table 25 describes the proportion of patients who agreed with these statements overall, and, among those, patients who indicated agreement pre consultation. Overall, 60.9% of patients reported a change in their medication post consultation. Of these, 71.8% were happy with the prescribing decision. Where patients had wanted a change in their medicines, rates of satisfaction with prescribing decisions appeared higher when changes were subsequently made (70.5% vs. 50.0%); however, this was statistically non-significant [χ²(1) = 3.34, p = 0.07]. Patients who wished to start, stop and/or change their medication in some way before the consultation were more likely than other patients to report a prescription change (70.1% vs. 32.6%; PR 2.15, 95% CI 1.37 to 3.38). Of those who wanted advice on medicines before the consultation, half (51.6%) reported having received it. Half (51.5%) of all patients agreed that they had participated in decisions about their treatment, although this represented only half (52.5%) of those who wanted to be involved prior to the consultation.

Over one-quarter (26.9%) of patients who wanted help with emotional problems reported having received it. Patients who identified mental health as their primary problem were more likely than patients who identified other types of health problems to report having received the following from the doctor:

• some kind of change in their medication (75.0% vs. 52.9%; PR 1.42, 95% CI 1.07 to 1.88)

• advice on medication (71.4% vs. 49.0%; PR 1.46, 95% CI 1.07 to 1.98)

• help with emotional problems (39.3% vs. 19.6%; PR 2.00, 95% CI 1.09 to 3.67).

Patient satisfaction

Patients were asked to rate their level of satisfaction with the consultation. Figure 4 shows that, in total, 83% of patients reporting being very satisfied (40%) or fairly satisfied (43%) with the consultation. Seventeen per cent of patients reported being fairly dissatisfied (8%) or very dissatisfied (9%).

FIGURE 4.

Patient satisfaction with the consultation.

There were no significant differences in satisfaction rates between the different prisons or doctors. A greater proportion of individuals who identified mental health as their primary problem reported being fairly/very dissatisfied than other patients, although this was a statistically borderline result (34.9 vs. 19.8; PR 1.76, 95% CI 1.01 to 3.08).

Perception of problem

Doctors were asked to record the main and, if applicable, secondary health problems for each patient participating in the study.

Table 26 shows that in two-thirds (63.2%) of cases in which the patient reported a mental health problem (primary or secondary), the doctor also recorded a mental health problem (primary or secondary). In the case of problems other than mental health problems, there was a higher level of agreement between patients and doctors (91.7% vs. 63.2%; PR 1.45, 95% CI 1.13 to 1.86). In eight cases, the patient reported problems other than mental health issues, while the doctor noted a mental health problem.

Prescribing decisions

Overall, doctors recorded 176 separate health problems and issued a prescription for half of these (49.71%). In 62% of cases, doctors thought that patients definitely (41.5%) or probably (20.5%) wanted a prescription (Figure 5). In 26.7% of cases, doctors reported feeling definitely pressured (5.1%) or a little pressured (21.6%) to prescribe (Figure 6). In 94.0% of cases, doctors reported feeling definitely comfortable (72.0%) or fairly comfortable (22.0%) with their prescribing decision, which included decisions not to prescribe (Figure 7).

FIGURE 5.

Did you think the patient wanted a prescription for this health problem?

FIGURE 6.

Did you feel pressured by the patient to write a prescription for this problem?

FIGURE 7.

Did you feel comfortable about this prescribing decision?

In 5.7% of cases (n = 10), doctors stated that prescriptions were not strictly indicated. In all cases, the prescriptions were wanted by the patient.

Doctors were more likely to issue a prescription when:

• They thought that the patient definitely/probably wanted a prescription (69.7% vs. 16.7%; PR 4.2, 95% CI 2.41 to 7.28).

• They felt definite/a little pressure to prescribe (70.2% vs. 42.2%; PR 1.66, 95% CI 1.26 to 2.19).

• They thought that the problem was a mental health problem (73.5% vs. 44.0%; PR 1.67, 95% CI 1.27 to 2.20).

Doctors were more likely to be comfortable with prescribing decisions when:

• They issued a prescription (100.0% vs. 51.06%; PR 1.96, 95% CI 1.67 to 2.30).

• They did not think that the patient definitely/probably wanted a prescription (100.0% vs. 91.0%; PR 1.10, 95% CI 1.03 to 1.17).

• They did not perceive definite/a little pressure to prescribe (98.1% vs. 83.7%; PR 1.2, 95% CI 1.02 to 1.34).

Doctor satisfaction

Doctors were asked how the consultation went. In 96% of cases, the consultation went OK (62%) or very well (34%). In 4% (n = 5) of cases, doctors reported that the consultation had not gone particularly well. There were no significant differences in satisfaction rates between the different prisons or doctors.

Summary: the acceptability of prescribing decisions

• One-quarter (23.8%) of patients identified a mental health problem as their main reason for seeing the doctor.

• Two-thirds (69.7%) of all patients wished to start, stop and/or change their medication. Patients who identified mental health as their primary problem were more likely to want to change their medication in some way.

• The vast majority (83%) of patients were very or fairly satisfied with the consultation.

• Half (51.5%) of all patients agreed that they had participated in decisions about their treatment.

• One-quarter (26.9%) of patients who wanted help with emotional problems reported having received it.

• Patients who identified a mental health issue as their primary problem were more likely to report having received some kind of change in their medication.

• A greater proportion of individuals who identified mental health as their primary problem reported being fairly/very dissatisfied than other patients.

• In 26.7% of cases, doctors reported feeling at least a little pressured to prescribe.

• There were no cases were prescriptions were not strictly indicated or wanted by the patient.

• Doctors were more likely to issue a prescription when they thought that the patient wanted a prescription, they perceived pressure to prescribe and/or the problem was a mental health problem.

Hypnotics

• Chloral hydrate.

• Clomethiazole.

• Flurazepam.

• Loprazolam.

• Lormetazepam.

• Melatonin.

• Nitrazepam.

• Promethazine hydrochloride.

• Sodium oxybate.

• Temazepam.

• Zaleplon.

• Zolpidem tartrate.

• Zopiclone.

Anxiolytics

• Alprazolam.

• Buspirone hydrochloride.

• Chlordiazepoxide hydrochloride.

• Diazepam.

• Lorazepam.

• Meprobamate.

• Oxazepam.

Antipsychotics

• Amisulpride.

• Aripiprazole.

• Benperidol.

• Chlorpromazine hydrochloride.

• Clozapine.

• Flupentixol.

• Haloperidol.

• Levomepromazine.

• Olanzapine.

• Paliperidone.

• Pericyazine.

• Perphenazine.

• Pimozide.

• Prochlorperazine.

• Promazine hydrochloride.

• Quetiapine.

• Risperidone.

• Sulpiride.

• Trifluoperazine.

• Zuclopenthixol.

• Zuclopenthixol acetate.

Drugs used for mania and hypomania

• Asenapine.

• Carbamazepine.

• Lithium carbonate.

• Lithium citrate.

• Valproic acid.

Tricyclic and related antidepressant drugs

• Amitriptyline hydrochloride.

• Clomipramine hydrochloride.

• Dosulepin hydrochloride.

• Doxepin.

• Imipramine hydrochloride.

• Lofepramine.

• Mianserin hydrochloride.

• Nortriptyline.

• Trazodone hydrochloride.

• Trimipramine.

Monoamine-oxidase inhibitors

• Isocarboxazid.

• Moclobemide.

• Phenelzine.

• Tranylcypromine.

Selective serotonin reuptake inhibitors

• Citalopram.

• Escitalopram.

• Fluoxetine.

• Fluvoxamine maleate.

• Paroxetine.

• Sertraline.

Other antidepressant drugs

• Agomelatine.

• Duloxetine.

• Flupentixol.

• Mirtazapine.

• Reboxetine.

• Tryptophan.

• Venlafaxine.