The use of seclusion and mechanical restraint is probably the oldest problem for psychiatric institutions and nowadays is more controversial than ever. The liberation of the “insane” in Paris by Philippe Pinel during the French Revolution is considered by textbooks of history and psychiatry to be the start of modern psychiatry. Yet, at the beginning of the 21st century, seclusion and restraint with belts are still used by most hospitals worldwide in the care of persons with acute mental illness (1).
There is a consensus among stakeholders and political and patients’ organizations that the least restrictive measure should be applied whenever a coercive action is considered (2–6), and this opinion is also expressed in the German guideline on therapy for aggressive behavior (7). However, what is the least restrictive of undesirable alternatives? The appropriate approach is to ask the patients. In some studies, patients were shown pictures of coercive measures and asked to rank them by extent of felt discomfort (8,9); in others, patients who had experienced a coercive measure were retrospectively interviewed about their feelings toward the measure and its violation of human dignity (10,11). We conducted the first randomized controlled trial comparing seclusion and mechanical restraint (10). As a primary outcome, we used the Coercion Experience Scale (CES) (12), a measure that we had constructed to assess restrictions of human rights and subjective distress. We found no significant differences between seclusion and mechanical restraint in that study.
Yet seclusion and restraint not only cause acute distress during the experience but also may exert long-term effects—a kind of traumatization—including self-stigmatization, traumatic memories, negative attitudes toward psychiatric treatment and psychiatric institutions, and even symptoms of posttraumatic stress disorder (PTSD) (13). Studies from Finland and New Zealand found that both psychotic symptoms themselves and the experience of hospitalization, particularly coercive measures, induced posttraumatic symptoms (14,15). Patients who are psychotic and who have a history of sexual or physical abuse have a significantly increased risk of being subjected to seclusion or restraint during psychiatric hospitalization and, thus, of being retraumatized, with a negative impact on their mental health (16). So far, no sound evidence has indicated whether one kind of coercive measure is more traumatic than another.
The first purpose of this study was to reassess the views of the patients in the original study through a follow-up interview about one year after use of the coercive measure. The second purpose was to explore whether symptoms of PTSD related to having experienced the coercive measure had emerged and to compare PTSD symptoms among patients who experienced seclusion or restraint.
The original study (10) comprised 102 patients, 60 of whom had experienced seclusion and 48 of whom had experienced mechanical restraint. Twenty-six of those patients were randomly assigned to seclusion (N=12) or mechanical restraint (N=14). A total of 76 patients had to be excluded from randomization and were included in the cohort arms (seclusion [N=48] and mechanical restraint [N=28]). No significant differences in outcomes were detected between the randomly assigned groups nor between patients in both the randomized and cohort groups who experienced seclusion or restraint.
When consent to study participation was obtained, it included participation in a follow-up interview. Patients left a telephone number for contact at follow-up. If the patient could not be contacted, further attempts were made on at least three different days and three different times of day for three weeks, for a maximum of 27 attempts. In case of repeated failure, public telephone files were searched for possible changed numbers. Patients were excluded if contact was not possible by these measures.
Data were recorded through semistructured interviews by telephone or in person by a Ph.D. student (MB) who was not part of the staff and who had not been present on the ward at the time of the coercive measure. In all cases, the interviewer clarified that the patients’ assessments should relate specifically to the index coercive measure about which they had been interviewed as an inpatient. Interviews took place between May and December 2006.
Items from the CES (12) were the main outcome measure. The CES assesses subjective distress during coercive measures by means of a self-rated questionnaire. The CES was developed as a validated instrument that should serve as an outcome of coercive measures, allowing comparison of the subjective negative impact of different interventions. The core idea of the CES is to investigate patients’ subjective experiences of applied coercive measures with respect to restriction of personal autonomy and human rights and degree of suffering. (In the final construction of the CES, some items were added that were not available at the time of the follow-up interview.)
Patients are asked to assess their experiences for violation of human dignity, restriction of freedom to move, restriction of autonomy, use of coercion at the beginning of the measure, and restriction of interpersonal contact. These items are assessed through a 5-step scale, from 1, little, to 5, extreme. Subsequently, they are asked to rate their subjective experience for the same items except the first by using another 5-step scale, with 1 indicating partly agreeable, partly disagreeable; 2, moderately disagreeable; 3, disagreeable; 4, very disagreeable; and 5, extremely disagreeable (“the worst I ever experienced”).
In addition, we used an analog scale (range 0–100) to assess the total stress experienced during the coercive measure.
To assess symptoms of PTSD, we used the Impact of Event Scale–Revised (IES-R) (17). The IES-R is a widely used instrument assessing the presence and severity of posttraumatic symptoms. The incidence of each symptom within the last seven days is recorded on a 4-step scale, with 0 indicating never; 1, rarely; 3, sometimes; and 5, frequently. There are three subscales: intrusion (N=7 items), avoidance (N=8), and hyperarousal (N=7). Possible scores range from 0 to 110. The presence of PTSD is estimated by a regression formula. Regression values ≥0 render the diagnosis probable. We used the German version of the IES-R, which yielded sufficient test values, with a diagnostic sensitivity of the regression formula of .76 and a specificity of .88 (18).
Further information with regard to medication, sociodemographic data, and subjective feelings related to the experienced coercive measure was obtained.
Differences in categorical variables were calculated by using chi square and Fisher’s exact tests. Because the variables were not normally distributed, differences in continuous variables were examined by the Mann-Whitney U test. The level of significance was determined as p<.05. To correct for multiple pairwise tests of CES items, the Bonferroni method was adopted.
In total, 60 (59%) patients from the original sample were available for reinterview—52 by telephone and eight face to face. Thirty-one patients had experienced seclusion, and 29 had experienced mechanical restraint. Reasons for dropout were loss to follow-up (N=18, 18%), retraction of informed consent (N=13, 13%), no recollection of the coercive measure (N=8, 8%), continuous severe psychopathological symptoms making reasonable interview participation impossible (N=2, 2%), and language difficulties (N=1, 1%). A comparison of the original data of patients who did or did not drop out revealed no significant differences in terms of gender, age, number of previous hospitalizations, and scores for the Clinical Global Impression and the Positive and Negative Syndrome Scale (PANSS) at the beginning of the coercive measure. Only PANSS scores at the end of the measure had been significantly higher among the patients who dropped out.
A total of 33 (55%) patients were male, and the mean±SD age was 39.6±12.5 years; 38 (63%) patients had a diagnosis of a schizophrenic disorder, 14 (23%) were diagnosed as having bipolar disorder (manic), and the remainder had other diagnoses. Most coercive measures were carried out because of threatening behavior (N=39, 65%), followed by danger to self and danger to property, with no significant differences between secluded and restrained patients. However, secluded patients received less medication during the measure than restrained patients (p<.01). Sociodemographic data for the original and the follow-up samples did not significantly differ (10). In total, 47 (78%) patients had not experienced coercive measures since the index measure; nine of the secluded and four of the restrained patients had experienced subsequent measures, mostly (p<.05) of the same kind.
The time between the coercive measure and the follow-up interview was 515±222 days; the time between the first interview and the follow-up interview was 484±233 days. There was no significant correlation between time since first interview and the score for any CES item.
A total of 21 (68%) patients who experienced seclusion and 12 (41%) who experienced mechanical restraint retrospectively judged the measure as justified. The others, however, complained about inhuman practice, arbitrariness, or disproportionality.
The results of the CES items from the same patients in the original and follow-up studies are displayed in Table 1. The values for seclusion remained roughly unchanged, whereas those for restraint increased remarkably. Significant changes were observed for violation of human dignity, restriction of freedom to move, experience of restriction of freedom to move, experience of restriction of autonomy, coercion at beginning of measure, and experience of coercion. Although the original study found no significant differences between CES ratings for seclusion and restraint except for restriction of freedom to move, CES ratings for mechanical restraint were significantly more negative on six of the nine items in the follow-up study.
Table 1CES scores among patients who experienced seclusion or mechanical restraint, by original or one-year follow-up studya
| Add to My POL
|Original study (N=102)||Follow-up study (N=60)|
|Seclusion||Mechanical restraint||Seclusion||Mechanical restraint|
|Violation of human dignityb||2.5||1.5||2.7||1.5||2.5||1.5||3.8||1.3||<.01|
|Restriction of freedom to moveb||2.5||1.5||3.6||1.4||<.01||2.6||1.3||4.1||.8||<.01|
|Experience of restriction of freedom to movec||2.6||1.5||3.1||1.6||2.7||1.4||3.9||.9||<.05|
|Restriction of autonomyb||3.3||1.4||3.8||1.2||2.8||1.4||3.7||1.1|
|Experience of restriction of autonomyc||2.7||1.4||3.1||1.4||2.5||1.2||3.7||1.2||<.01|
|Coercion at beginning of measureb||2.2||1.5||2.6||1.4||2.3||1.4||3.6||1.2||<.01|
|Experience of coercionc||3.0||1.5||2.6||1.5||2.7||1.3||3.8||1.1||<.05|
|Restriction of interpersonal contactb||2.3||1.5||2.1||1.6||2.2||1.4||2.5||1.5|
|Experience of restriction of interpersonal contactc||2.1||1.3||2.1||1.5||2.0||1.3||2.7||1.5|
Mechanical restraint had lasted 12.0±10.5 hours and seclusion had lasted 6.0±6.0 hours (p<.01).The patients overestimated the mean duration of seclusion by 7.9±19.0 hours (median=1.3) and underestimated the duration of mechanical restraint by 1.3±11.6 hours (median=1.5), both nonsignificant differences. There was no significant correlation between patients' estimation and actual duration of seclusion. The correlation between estimated and actual duration of mechanical restraint was significant (r=.52, p<.01). Estimates of duration of coercion by patients grouped by actual duration (up to four hours, between four and eight hours, or over eight hours) were compared. No significant differences were found among patients who experienced seclusion. Significant differences were found among patients who were subjected to mechanical restraint (p<.01). Patients who were restrained for more than eight hours underestimated the duration by 5.9±6.0 hours (median=4.0).
A total of 18 (58%) patients who experienced seclusion and 13 (45%) who experienced mechanical restraint judged the duration to be justified. A total of 34 (57%) patients indicated that they would have liked more contact with staff during the measure. When asked what alleviated distress during the measure, patients most frequently mentioned contact with staff (N=12, 39%, seclusion; N=20, 68%, mechanical restraint). Having personal objects nearby also alleviated distress (N=7, 22%, seclusion; N=3, 11%, mechanical restraint). Thirty-five patients (58%) indicated that the measure had some helpful effects, predominantly by calming them down.
The feelings reportedly experienced most frequently during the measure were helplessness (N=43, 72%), tension (N=38, 63%), being at the mercy of others (N=38, 63%), rage (N=35, 58%), fear (N=34, 57%), desperation (N=29, 48%), anger (N=29, 48%), and disappointment (N=28, 47%). Differences between seclusion and restraint were generally small and not significant: horror was the only feeling that varied between the two groups, with a trend toward significance (N=14, 48%, restraint; N=8, 26%, seclusion; p<.10).
Estimated total stress was 52.4±31.7 (median=50) among patients who experienced seclusion and 62.1±33.4 (median=75) among patients who experienced mechanical restraint. The differences in favor of seclusion were not significant.
The total scores for the IES-R were −2.9±1.5 (median=−3.9) for secluded patients and −3.0±1.4 (median=−3.0) for restrained patients. The subscores revealed no significant differences. Two patients who experienced restraint and one who experienced seclusion had a total score ≥0, indicating event-related PTSD.
This follow-up study shed new light on our randomized controlled trial comparing the subjective distress of seclusion and mechanical restraint (10). The main result of the original study was that no significant difference could be detected between seclusion and restraint. However, in this follow-up study, in which nearly 60% of the original participants were reinterviewed after about 18 months, patients retrospectively assessed mechanical restraint significantly more negatively. Assessments of seclusion remained rather stable, whereas the negative impact of mechanical restraint increased considerably over time. There was even a tendency to rate mechanical restraint worse than seclusion on items for which seclusion is definitely worse from an objective perspective, such as restriction of interpersonal contact.
Sample selection bias is not a probable explanation. Patients lost to follow-up largely did not differ from those who were included, and, most notably, we used only the data of the patients who participated in the follow-up study for comparison with the original study. Thus the differences between the first and second interviews are probably valid and must be interpreted as a serious statement about the distress and detrimental psychological effects associated with use of mechanical restraint. These data confirm the results of other studies of the use of mechanical restraint versus seclusion (8,11,19).
Although the first interview had been conducted during the inpatient stay and shortly after the coercive measure, the follow-up interview was conducted much later by a completely independent interviewer, with patients being generally in much better mental condition than was the case at the first interview. It seems that recall of the coercive experiences was generally still rather strong, and indignation about the use of mechanical restraint had increased as the patients’ lives had become more stable. This was the case even though a considerable proportion of patients judged the measure justified—albeit less in the case of mechanical restraint (41%) than seclusion (68%). It is well known from other studies that some patients retrospectively accept coercive measures as appropriate (20–24).
This study raises points in favor of the use of seclusion. Nevertheless, seclusion also causes considerable subjective distress, which at the time of the intervention was judged equally as upsetting as the use of mechanical restraint (10). No significant differences in negative feelings about either measure, including feelings of helplessness, were found in the first or the follow-up interviews. Many of the patients reported strong negative feelings toward the coercive measure, with small or no differences between those who experienced seclusion or restraint.
Nevertheless, patients were obviously capable of retrospectively judging the situation in a differentiated manner. Even though patients expressed many negative feelings, no less than 58% reported that the measure had some positive effects, mainly calming them down. This result can be seen as a contribution to the discussion about whether coercive measures are akin to therapy, as suggested by Sailas and Fenton (25), or are purely safety measures that indicate a failure of therapeutic approaches (26,27). Thus, although such coercive measures are primarily measures of safety and not therapeutic in nature, they may have some concomitant positive therapeutic effects in some cases, with the major flaw, however, of causing severe psychological distress in many cases.
Patients’ reports of what alleviated the experience of a coercive measure are helpful for clinicians. Interpersonal contact was reported as most helpful; it can be established during seclusion by visits from staff at fixed intervals. Second, and perhaps not sufficiently appreciated by staff during the everyday routine, keeping personal objects nearby seems to be important. With regard to the overestimation of time in seclusion, as a consequence of this study we placed clocks in the seclusion room in clearly visible locations and advised staff, through internal guidelines, to repeatedly explain to patients how long the measure is expected to take and the conditions under which it will end.
Notwithstanding the strong negative assessments of seclusion and, even more, of mechanical restraint, the study provided only minor evidence that these measures had specific psychological consequences in terms of PTSD. Although 40% of the original sample had been considered at risk for the development of PTSD, according to the IES-R only three patients in the follow-up sample (N=2, mechanical restraint; N=1, seclusion) had developed symptoms sufficient to suppose the presence of PTSD (18). Including the eight patients who were excluded due to lack of recollection, which suggests that the episode did not have a negative impact, those three patients represent 4% of the 68 patients in the sample. It can be questioned whether a coercive intervention fulfills the A criterion of PTSD diagnosis at all; however, at least some patients subjectively experience coercion as a life-threatening situation.
The incidence of PTSD after coercive measures may have been underestimated in this follow-up study. One reason for the rather low numbers may be that PTSD existed but had abated in the course of the 18 months between interviews, which would correspond to the most frequent natural course of PTSD (28). Another reason may be that the comprehensive interviews conducted after exposure prevented the development of PTSD. The interviews undertaken during the study were far more extensive than debriefings in routine clinical practice, and thus the finding of low incidence could be, in part, a research artifact.
This study had several limitations. The role of medication in patients’ subjective experiences remains unclear. The use of medication before seclusion or restraint did not differ significantly in terms of drug or dosage. However, during the measures, the seclusion group received significantly less additional medication (none, in most cases) than the restraint group. This finding applies only to the patients who were excluded from randomization. Less use of medication among patients in the seclusion group may be due to the fact that patients in this group exhibited less severe violent behavior and less severe psychopathological symptoms. Information about the extent to which additional coercion was used to dispense the medication was not recorded. But it is possible that patients experienced medication as a part of the coercive procedure and attributed their subjective feelings not only to the mechanical measure but also to the concomitant medication. This may have negatively influenced perceptions of mechanical restraint. Also the finding that contrary to patients who experienced seclusion, patients who experienced mechanical restraint tended to underestimate the duration of their restraint might be caused, in part, by sedating medication. On the other hand, taking into account the generally worse assessments of mechanical restraint, there was no further evidence that medication was helpful in reducing the psychological distress caused by coercive measures. Further, 13 patients, nine from the seclusion group and four from the restraint group, had experienced subsequent coercive measures between the initial interview and the follow-up interview. This could have biased their assessments, because it may have been difficult for them to differentiate between their experiences of different events. Another bias can result from the fact that among the patients who were excluded from randomization, those who experienced mechanical restraint had shown more severe violent behavior and psychopathological symptoms than those who experienced seclusion (10), which may have influenced their assessments.
Finally, this study was conducted at only one hospital. The conditions of seclusion and of mechanical restraint were standardized as much as possible by hospital guidelines, but certainly the conditions differ among hospitals in many aspects, such as architecture, devices used, frequency and quality of staff contact, and use of medication. Thus the possibility cannot be excluded that under different conditions, patients’ assessment of coercive measures might have differed. Generalizability is, therefore, questionable to some extent.
In our first study, assessments of seclusion or restraint were carried out only a short time after patients experienced a coercive measure. In contrast, this follow-up study of assessment of coercive measures 18 months later provides some evidence that seclusion, a less restrictive alternative, may be associated with less psychological distress. Nonetheless, seclusion was also associated with considerable distress and long-lasting negative feelings.
The incidence of enduring symptoms of PTSD was lower than expected. The study provides insight into patients’ experiences during coercive measures and provides some ideas about how such measures can be made more tolerable for patients if other alternatives have failed. Most important are maintaining contact with staff during the measure and taking into account the patient’s personal needs. This includes keeping one’s own property, availability of a clock, having darkness during sleeping hours, and eating and using a toilet under conditions consistent with human dignity.