A comparison of long-term medium secure patients within NHS and private and charitable sector units in England.

BACKGROUND
In England, forensic psychiatric hospital services are provided at three security levels: high, medium and low. All are publicly funded and similarly regulated, but medium and low secure services are provided in the private and charitable (PCS) sector as well as the National Health Service (NHS). Originally, medium secure hospital services were conceived as for up to 2 years' inpatient stay, but numbers of longer stay patients have been rising. Little is known about their characteristics or whether they differ between NHS and PCS settings.


AIMS
To describe and compare characteristics of long-stay patients in NHS and in PCS medium security hospital units.


METHODS
Data were extracted from clinical records in 14 NHS and 9 PCS hospital units for all patients fulfilling criteria for long stay: having been in high security for more than 10 years or medium security for more than 5 years or in a mix of both for more than 15 years in total.


RESULTS
178 NHS and 107 PCS patients were eligible for inclusion, respectively, 16 and 22% of the total patient populations in these settings. The mean length of stay in a medium or high secure setting was similar: 163 and 164 months. Characteristics of the patients, however, differed between unit type. NHS services admitted more patients from prison and PCS services more from other hospitals. NHS services included a lower proportion of patients with personality disorder or intellectual disability. 'Challenging behaviour' was more prevalent in PCS; a history of absconding was found more often among NHS patients.


CONCLUSIONS
The two systems of service appear to be used differently. More research is needed to explain why patients apparently without behavioural disturbances remain in specialist secure facilities for such a long time and whether their needs are truly being met in the least restrictive environment possible.

This aim of this study was, therefore, to compare patient characteristics of those detained in NHS and PCS facilities. The findings reported here are based on data collated as part of a larger, multi-site study, funded by the National Institute for Health Research (NIHR), which included high and medium hospitals (Völlm et al., 2017). Based on our clinical experience, we hypothesised that long-stay patients in PCS facilities would be (a) more likely to have been transferred from another hospital rather than from prison, (b) have more challenging presentations and (c) more specialised needs.

| Ethical approval
We used routinely collected clinical data anonymised before being sent to us. The study was thus deemed to be a service evaluation by the sponsoring organisation (Nottinghamshire Healthcare NHS Foundation Trust), not requiring additional research ethics approval (https://researchsupport.admin.ox.ac.uk/sites/default/files/researchsupport/ documents/media/defining-research.pdf).

| Definition of long stay
Long stay was defined as a duration of 10 years or over in high secure, 5 years and over in medium secure, or 15 years if patients had stayed in both, high and medium secure care consecutively. This definition is in keeping with the 5-year threshold used in two previous medium secure studies (Shah et al., 2011;Sharma et al., 2015) and was informed by pilot data suggesting that 15% of patients would be captured by this definitiona number large enough to describe meaningfully a subgroup of patients within the forensic hospital population. All patients who met this criterion and who were resident in one of the participating units at the census date of April 1, 2013 were considered 'within scope' for this comparison (for a full description of the selection of patients, see Völlm et al., 2017). LoS was categorised into continuous LoS, which included the total LoS in medium secure care with or without additional time in high security since first admission to such a setting and current LoS, which represented LoS in the current medium secure unit alone.

| Unit selection
At the time of the study, there were 57 medium secure units in England, 34 NHS and 23 PCS units. For this study, we included just under half of each (14 NHS, PCS units), stratified by region, according to the then 10 Strategic Health Authorities. We oversampled at units specialising in particular patient sub-groups, including women and patients with intellectual disabilities. A contact person was established at each unit to assist with the identification of patients fulfilling the LoS length of stay criteria and therefore eligible for inclusion.

| Data collection
A data collection proforma was developed and piloted, covering almost 250 variables grouped into categories, including socio-demographics, admission pathways, LoS, Mental Health Act status, diagnoses, offending histories, risk information and incidents. Data collection was completed by the units and transmitted in anonymised form to the research team. Data return was 100%.

| Data analysis
Data were entered into an SPSS (Statistical Package for the Social Sciences, vs 22.0) file. Descriptive data are reported and comparisons between NHS and PCS patients made. Cross tabulation was used to establish absolute and relative frequencies for categorical variables. Means, medians and standard deviations were calculated for continuously distributed variables. Non-parametric tests (Mann-Whitney U) were administered to establish statistically significant differences for continuous variables as they were non-normally distributed, identified through inspection of boxplots and calculation of Kolmogorov-Smirnov statistics. Chi-squared tests were used for categorical data or Fisher's Exact tests when assumptions were violated.

| RESULTS
Data were obtained for all 285 patients fulfilling criteria: 178 patients from 14 NHS units and 107 patients in 9 PCS units. They represented a percentage of 16 and 22% of the total patient population in these settings respectively (p = .217).

| Patient characteristics
As shown in Table 1, the two groups were similar in gender and mean age distributions. NHS units held a higher proportion of Black British patients than the PCS units (NHS 15.7%, PCS 6.5%, p = .002).

| Length of stay
Total length of stay in a secure hospital was similar between unit types, as shown in Table 2. Length of continuous stay was also similar. Length of stay in the current unit, however, was higher for the NHS patients at 56.0 months (4.7 years) compared to 48.6 months (4.1 years) for the PCS patients. In line with this, a higher proportion of PCS patients had spent less than 5 years in their current unit (NHS 49

| Admission pathway
As shown in Table 3, significantly more long-stay patients currently residing in NHS units had been transferred there from prison (NHS 62%: PCS 42%). There was a slight complementary tendency for proportionately more patients in PCS units to come from each of the other sources rated, but these findings were not statistically significant.

| Diagnostic category
Patients differed between unit types in diagnostic categories (

| Criminal histories
There were no significant differences between the two groups in terms of offender category (violent, sexual, mixed, or non-offender) or most other criminal history variables examined (Table 6). An exception was that the groups differed in terms of age at first violent conviction, the PCS group being younger by an average of nearly 3 years (NHS 21.6 years: PCS 24.5 years). More NHS patients had been convicted for manslaughter at first conviction for a violent

| Risk and incidents
The prevalence of self-harm or suicidal behaviour was similar in both hospital unit types ( Table 7 also shows that more PCS patients were involved in serious incidents or seclusion in the 2 years prior to rating (NHS 25%: PCS 47%). These included serious assaults on staff (NHS 13%: PCS 30%) or others (NHS 15%: PCS 43%). A higher proportion of PCS patients were involved in a serious incident necessitating seclusion in 5 years before ratings (NHS 28%: PCS 47%).

| DISCUSSION
Our study represents the first detailed national comparison between long-stay patients in NHS and PCS medium secure hospitals. We found some clinically important significant differences between the two groups in some of the characteristics examined. We found few differences in socio-demographic characteristics between the two groups, though NHS services were caring for more patients of black and mixed ethnicity. Of potential clinical importance, patients in the private and charitable sector were more likely to be have been transferred from another hospital than from prison for both the original source of admission and admission to the current unit. From the incident data, it was apparent that these PCS patients had presentations which directly challenged staff, sometimes physically hurting them, and had had more criminal convictions following from their inpatient behaviour. In terms of specialised needs, a higher proportion of the private and charitable sector patients had been diagnosed with personality disorder or with intellectual disability.
The finding that NHS long-stay patients were more ethnically diverse warrants further discussion. There are, to our knowledge, no earlier studies specifically investigating this issue, although a number of studies have identified  (Huband et al., 2018), found that being black was associated with shorter length of stay in hospital, so we might have expected to find underrepresentation among people who stay for long periods in secure settings. The issue is likely to be confounded by other factors. Gajwani, Parsons, Birchwood, and Singh (2016), for example, in their study in one service in England found that ethnicity was no longer associated with detention under the MHA after controlling for other variables, including diagnosis. The apparent overrepresentation of black and mixed patients in our NHS sample might therefore be confounded by diagnosis of serious mental illness in these settings.
Pathways of long-stay patients are complex. In line with other research (Lelliott et al., 2001), there was a tendency for the private and charitable sector patients to have been admitted from non-custodial settings, with higher proportions on a civil section. Just under 10% had no convictions at all yet had been in a medium secure hospital bed for lengthy periods of time. This may have reflected persistent challenging behaviours that were never referred to the criminal justice system. This raises ethical issues in setting England apart from other countries where admission of this group without criminal convictions would not be allowed (Völlm et al., 2016). On the other hand, it is not clear whether simply sending such an individual who is regularly assaultive to clinical staff into the criminal justice system or insisting that non-specialist units, which may in practice be no less constraining, continue to care for them are good alternatives. This is an area that would benefit from more research to establish optimal management.
A high proportion of patients in both groups were admitted to their current medium secure unit from another medium secure unit. This suggests that recovery was not always achieved within the original secure hospital unit and that the ideal pathway, moving from more to less secure settings, is not achieved for a substantial proportion of patients. On the credit side, it may suggest that the system is working well as a whole rather than with each unit in isolation.
Diagnostic categories differed significantly with a higher proportion of mental illness in the NHS group and a higher proportion of intellectual disability in the PCS group. Previous research found that between 63-68% of longstay patients have a mental illness and between 18 and 36% have a diagnosis of personality disorder (Kirby, 1997;Reed, 1997). It is possible that the presence of personality disorder and intellectual disability may be associated with longer lengths of stay because of persistent challenging behaviour and slow risk reduction.
Of note is the high proportion of patients, particularly in private and charitable settings, who had sustained convictions for offences while resident in a secure hospital. A significantly higher proportion of PCS patients were also involved in recent serious incidents and seclusion. Other authors have suggested that such behaviours are likely to contribute to extended length of stay (Castro et al., 2002). Lelliot et al. (2001) also showed that patients in privately run units got higher scores on ratings for aggression and antisocial behaviours, while de Taranto et al. (1998) found that patients with few or no convictions for violence had in fact committed many violent acts, but which had not resulted in prosecution. It is not clear, however, why those in the private and charitable sector were involved in more incidents.
Of crucial importance are the rates of patients not involved in violent or antisocial incidents while in hospital, yet still finding themselves in the long-stay group. Why are these patients still placed in their current security level?
Can clinical teams really justify what risks are being contained when less secure alternatives are available, particularly given the pressure from prisons to admit new patients. The converse argument might, however, state that the lack of incidents is commensurate with the level of security containing such behaviour.

| Strengths and limitations of this study
This is the first nationally representative comparison between a sample of NHS and private and charitable sector long-stay patients in specialist forensic mental health inpatient hospital units. A particular strength is the consideration of the whole pathway from first admission to (high or medium) secure care, thereby providing a more realistic estimate of total length of stay. Oversampling of certain groups will, however, have affected diagnostic composition of our sample.
Data collection by units themselves ensured research staff only had access to anonymised data thereby protecting the identities of a highly vulnerable patient group. This also meant, however, a potential lack of uniformity in interpreting the data collection proforma. Due to differences in routinely collected data, it is expected that there was some missing data with regards to incidents.

| Implications
The pathways for long-stay patients need to be contextualised in terms of policy developments concerning mentally disordered offenders, and those requiring secure conditions. It is not clear that the core group of people requiring secure hospital care has changed much since the opening of medium security hospital services in the early 1980s, but their case and risk management has been subject to major changes. perhaps the detention in a medium security bed is no longer appropriate, and they have slipped through the rehabilitation net? It is likely that on-unit behaviour is not the only determining factor in terms of discharge or stepping down a level of security. It may be that the absence of incidents merely indicates that they were appropriately placed and ongoing challenging behaviours are simply well managed. It would be unlikely that their cases had not been independently reviewed, since patients may choose to apply for a review every year, generally with legal representation, and there is a requirement for a formal Mental Health Review Tribunal hearing every 3 years. Further research with these patients may clarify their particular needs.
While there are around 3,200 beds in medium secure care (NHS standard contract https://www.england.nhs. uk/wp-content/uploads/2013/06/c02-high-sec-mh.pdf), pressures on existing resources must shift the focus on to whether current resources are being used appropriately, a debate which took place within high secure care over 40 years ago. Could some of the long-stay patients considered here be managed in conditions of lower security?

| CONCLUSIONS
This study found key differences between long-stay residents in secure services according to whether these were provided directly by the NHS or through the private and charitable health care sector. Among the long-stay residents, NHS units were more likely to take people directly from the criminal justice system, accepting more serious offenders with mental illness who were more likely to be detained under the criminal justice provisions in mental health legislation. Private and charitable sector units were more likely to take civil detention cases, people with intellectual disability among their difficulties and people who continued to present with challenging, including assaultive behaviours while on the unit. Further research would be helpful to determine the extent to which the different use of such units is beneficial for patients, and also the wider community.