A Longitudinal Study on Remand Psychiatric Pravelence

Acta Psychiatr Scand 2000: 102: 19±25 Printed in UK. All rights reserved

Copyright # Munksgaard 2000
ACTA PSYCHIATRICA SCANDINAVICA ISSN 0902-4441

A longitudinal study of prisoners on remand: psychiatric prevalence, incidence and psychopathology in solitary vs. non-solitary con®nement
Andersen HS, Sestoft D, Lillebñk T, Gabrielsen G, Hemmingsen R, Kramp P. A longitudinal study of prisoners on remand: psychiatric prevalence, incidence and psychopathology in solitary vs. non-solitary con®nement. Acta Psychiatr Scand 2000: 102: 19±25. # Munksgaard 2000. Objective: To compare two levels of stress (solitary con®nement (SC) and non-SC) among remand prisoners as to incidence of psychiatric disorders in relation to prevalent disorders. Method: Longitudinal repeated assessments were carried out from the start and during the remand phase of imprisonment. Both interviewbased and self-reported measures were applied to 133 remand prisoners in SC and 95 remand prisoners in non-SC randomly selected in a parallel study design. Results: Incidence of psychiatric disorders developed in the prison was signi®cantly higher in SC prisoners (28%) than in non-SC prisoners (15%). Most disorders were adjustment disorders, with depressive disorders coming next. Incident psychotic disorders were rare. The difference regarding incidence was primarily explained by level of stress (i.e. prison form) rather than confounding factors. Quantitative measures of psychopathology (Hamilton Scales and General Health Questionnaire) were signi®cantly higher in subjects with prevalent and incident disorders compared to non-disordered subjects. Conclusion: Different levels of stress give rise to different incidence of psychiatric morbidity among remand prisoners. The surplus of incident disorders among SC prisoners is related to SC, which may act as a mental health hazard.

H. S. Andersen1,2,3, D. Sestoft1, T. Lillebñk1, G. Gabrielsen4, R. Hemmingsen2, P. Kramp1
Clinic of Forensic Psychiatry, Ministry of Justice, Copenhagen, 2Psychiatric Department, Bispebjerg University Hospital, Copenhagen Hospital Cooperation, 3 Psychiatric Department, Rigshospitalet, University of Copenhagen, Copenhagen Hospital Cooperation and 4 Institute of Theoretical Statistics, Copenhagen Business School, Copenhagen, Denmark
1

Key words: prisoners; solitary con®nement; longitudinal; psychiatric morbidity; adjustment disorder; prevalence; incidence Henrik Steen Andersen, Centre for Crisis and Disaster Psychiatry, Psychiatric Department, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen é, Denmark Accepted for publication March 15, 2000

Introduction

Solitary con®nement (SC) in prisons has attracted substantial interest in the media with comparatively less research on SC and its consequences. SC may be used as a disciplinary means, to avoid collusion (threathening witnesses, destroying evidence and co-ordinating statements) or it may be voluntarily demanded by the prisoner. The literature (1±8) does not agree as to the mental health effects of SC. Studies on remand prisoners (1±3) points to SC as a stress factor giving rise to a surplus of symptoms. Studies on sentenced prisoners have either agreed on this conclusion or

have found no differences between the two prison forms (4±8). Substantial differences regarding administration of justice and practical implication exist in different countries. International comparisons of research is further complicated by diverse and insuf®cient methodology in most studies. In a longitudinal study of randomized samples of Danish prisoners remanded in custody this present paper aims to: (i) Compare incidence of psychiatric morbidity in a group of prisoners in SC with that of a group of prisoners in non-SC. 19

Andersen et al. (ii) Relate incidence to prevalence (i.e. prior to imprisonment) of psychiatric disorders. (iii) Relate prevalence and incidence of psychiatric disorders to quantitative measures of psychopathology.
Material and methods Sample
Table 1. Number of examinations in solitary con®nement (SC) and non-solitary con®nement (non-SC). Drop-outs are subjects not examined at the following examination due to release from the prison, sentence or exclusion from the study SC 1. Examination (after 1±2 days) 2. Examination (3 weeks) 3. Examination (2 month) 4. Examination (3 months) 5. Examination*** (4 months) 133 37 13 6 2 34 11 4 1 Non-SC* Drop-out** 62 13 3 3 Non-SC 95 48 14 4 2 Drop-out 47 34 10 2 Total 228 119 38 14 5

The prison setting has been described in Andersen et al. (9). Subjects were two random samples of prisoners on remand in SC and non-SC, respectively. From lists of all new prisoners subjects were chosen by the research team the day after imprisonment using a randomization list. It was decided in advance whether to include a prisoner from SC or from non-SC. Only prisoners on remand were included. Age was from 18 to 60 years. Both females and males were included. Prisoners who did not speak Danish ¯uently (i.e. subjects not born in Denmark) were excluded, as the psychometric instruments demanded full knowledge of Danish. Subjects were only included once. Subjects were not paid for participating. Participation was voluntary and written and oral informed consent was obtained from all subjects after a thorough information of the purposes and implications of the study. The SC group consisted of 133 subjects; 139 were approached and 96% agreed to participate. The nonSC group consisted of 95 prisoners. Approaching 112 gave a participation rate of 85%. On the only obtainable measures (age and gender) the refusers did not differ from the included subjects. Index examination. Participants in both groups were examined by one of the two psychiatrists (D.S. and H.S.A.) after imprisonment (181=79% on day 2; 36=16% on day 3; 11=5% on days 4±6). The interview (1K±4 h) was performed in a semistructured way. Each subject was followed-up by the same interviewer. The following day cognitive tests were carried out by a psychologist (T.L.). Subsequent examinations. The second and the following interviews were similar (K±3 h). The second examination took place after 3 weeks of imprisonment on remand. Thereafter subjects were examined every month. Subjects that were released or sentenced were not examined further. Prisoners transferred from SC to non-SC were examined once in non-SC. The falling number of examinations over time was due to release, sentence or transferral from SC to non-SC. The number of examinations in the groups is shown in Table 1. To obtain reliability on evalua20

* Subjects examined after transferral from SC to non-SC. ** Subjects examined after transferral from SC to non-SC are excluded from further examinations. *** Two subjects from the SC group were examined until 9 months of imprisonment (10 examinations) and two subjects from non-SC were examined until 7 months of imprisonment (8 examinations).

tions joint interviews on psychopathological issues (n=11) were performed throughout the study.
Interview

Present State Examination, 10th edn (PSE-10). PSE interviews were performed at every examination. The PSE-10 (10) results were used primarily as the foundation of ICD-10 diagnoses (11). Scores distinguished between the month before imprisonment and time since imprisonment. All diagnoses were made by consensus between the two psychiatrists after discussion of each subject. To be diagnosed as having a new disorder during imprisonment the disorder had to be new or a severe exacerbation of an already existing disorder. Hamilton Anxiety and Depression Scales (HAS & HDS) (12, 13). HAS and HDS were repeated at all examinations. Both covered the time since imprisonment on the ®rst examination and the last 3 days on the following examinations. Abuse/dependence interview. A semistructured interview was constructed and administered on ®rst examination with comprehensive coverage of previous and present abuse and dependence of alcohol, cannabis, opioids, stimulants, hallucinogens and sedatives/hypnotics. ICD-10 dependence diagnoses were made on basis of this interview. Sociodemographic interview. This semistructured interview was constructed for the purpose and performed on the ®rst examination. It covered childhood experiences, schooling and education, previous and present work experiences and

Prisoners on remand housing, previous and present criminality, previous and present somatic disorders and previous psychiatric disorders. Wechsler Adult Intelligence Scale (WAIS) (14). Six subtests of the WAIS was performed on the ®rst examination in order to derive IQ scores with subgrouping of performance and verbal scores.
Questionnaires

The study has been declared and approved by the Danish Scienti®c Ethical Committee and the Danish Data Protection Agency.

Results

General Health Questionnaire, 28-item version (GHQ-28) (15, 16). GHQ was administered at every examination. The bimodal scoring procedure was used. The GHQ-28 covered the last 2 weeks on all examinations. Eysenck Personality Questionnaire (EPQ) (17) The EPQ was administered at the ®rst examination. No time period for this questionnaire was speci®ed.
File records

From the Danish Register of Criminality all previous imprisonments, charges and sentences from each subjects were reported. From the Institute for Basic Psychiatric Research, Department of Psychiatric Demography all previous psychiatric admissions were registered (18). Medical ®les from the prison were examined for each subject.
Statistics

Statistics were computed using the Genstat 5, release 3 (19). Comparisons on continous background data were tested for signi®cance with Mann±Whitney's rank sum test. On categorical data x2 tests or exact test were applied on nominal scales and Kendall's tau on ordinal scales. The in¯uence of confounding factors on incidence of psychiatric morbidity was tested by stepwise multiple regression analyses with incident disorders as the dependent variable. Potentially confounding factors were introduced one by one with test of signi®cance of both the confounding factor and SC vs. non-SC. A three-way analysis of variance (prison form/prevalent disorders/incident disorders) was used when comparing subjects without disorders, with prevalent and with incident disorders on quantitative psychopathological data (HAS, HDS and GHQ-28). A signi®cance level of 5% was chosen with trends between 5% and 10%. Reliability was measured as inter-rater reliability (agreement between the two interviewers in 11 joint interviews). Reliability was measured as correlation between the two interviewers.

Sociodemographic data were not distributed evenly among SC and non-SC subjects. Most differences pointed to the non-SC group as the more dysfunctional group (9). Reliability tests in joint interviews did not point to systematic bias between the two interviewers. Mean of differences was 0.46 on the HAS: P=0.676; correlation: 0.902: P