Characteristics of Adolescents Subjected to Restraint in Acute Psychiatric Units in Norway: A Case-Control Study

Mechanical and pharmacological restraints, seclusion, and physical holding are interventions used in inpatient child and adolescent mental health services to prevent patients from harming themselves or others or from damaging property. The use of restraints is controversial. Therapeutic benefits, risk of injury, and legal and ethical issues are areas of concern (1), and opinions vary, from viewing restraint as unnecessary harm to considering it as a potentially useful therapeutic intervention (2). Both extremes are less often advocated in recent literature (3,4). Most research on restraint in psychiatry is performed with adult patients. Several studies indicate that restraint is frequently used in child and adolescent mental health services (3), and two studies have found higher rates of restraint use among hospitalized children and adolescents than among adults (3,5).

However, few studies have examined differences between restrained and nonrestrained patients (5). According to Dorfman and Mehta (6), there is a particular lack of studies on adolescents in acute psychiatric units. De Hert and colleagues (3) systematically reviewed the literature on prevalence, indications, predictors, and outcomes in regard to restraint use among children and adolescents between 2000 and 2010. Six of the seven studies involved both children and adolescents. Four of the studies were from the United States (7–10), two were from Australia (11,12), and one was from Europe (13). The sample in the European study was from Finland and included all child and adolescent psychiatric inpatients during a chosen day in early 2000. Multivariate analyses were conducted to find factors associated with the use of holding, mechanical restraint, seclusion, and time-out. Aggressive behavior was the strongest factor associated with all kinds of restraint practice. Psychosis, suicidal behavior, and older age were associated with seclusion and mechanical restraint. Younger age, attachment disorder, and autism were associated with holding (13). In Finland, pharmacological restraint is not used for children and adolescents; time-out is used, and it is defined as different from seclusion.

Delaney and Fogg (14) concluded that adolescents who were at greatest risk of being restrained during brief inpatient treatment were characterized by more use of inpatient services, guardianship arrangements, special education placement, and history of suicide attempts. Few studies have reported the risk factors and predictors of restraint use for adolescent patients (3). Furthermore, small samples, use of different core concepts, and poor descriptions of the inpatient services make it difficult to compare results across studies. Variations in national legislation and practices in the use of restraint are also challenges to cross-national validity of research findings.

The aim of this retrospective case-control study was to investigate whether restrained and nonrestrained adolescents in Norwegian acute psychiatric inpatient units have different social, mental health, and treatment characteristics.

Methods

Setting

We included all Norwegian acute psychiatric inpatient units (N=16) for adolescents age 13–17 that were authorized for use of restraint during the entire data collection period—January 1, 2008, through December 31, 2010. The units had a total of 126 beds (mean±SD=7.4±2.9). According to the Mental Health Regulation, restraint can be applied only in units that are authorized for involuntary admissions. If restraint has to be used in nonauthorized units because of acute situations, the unit must consider transferring the adolescent to an authorized unit. Another two acute units were not included because they were not authorized for involuntary admissions. All included units accepted around-the-clock emergency admissions throughout the study period. Each unit had a defined and unique catchment area.

Procedure

The study had a retrospective case-control design and included all 288 inpatients who experienced restraint from January 1, 2008, through December 31, 2010. For each patient included, another inpatient (from the same unit and time period) who had not experienced restraint during the admission was randomly selected and included as a control. Secretaries in each unit selected each control patient by closing their eyes and pointing to a name in a list of patients sorted according to time of referral during the three-year inclusion period. The selected patient was included as a control if he or she had not experienced restraint. This procedure was repeated until 288 control patients were selected.

Data collection

The study was based on routine data collected from electronic patient records and from restraint protocols (printed version). All psychiatric institutions in Norway are obligated by law to have a restraint protocol (predefined by the Norwegian Directorate of Health) in which each restraint episode is registered. The protocol describes the type and duration of the episode and the reason for it and is checked regularly by an independent and authorized control commission. Data regarding age, gender, social characteristics (immigrant background, living arrangements, and involvement with child protection services), mental health characteristics (ICD-10 diagnosis and Children’s Global Assessment Scale [CGAS] score), and treatment characteristics (number of admissions, length of stay, and involuntary referral) were collected from patients’ records.

Data were collected during a nine-month period from August 2011 to May 2012. The first author visited all the institutions and collected demographic and clinical data. For patients with more than one admission in the three-year period, data were collected from the most recent admission. The number of admissions was recorded for each patient.

The Regional Committee for Medical and Health Research Ethics South and East and the Privacy Protection Officer of Oslo University Hospital approved the study.

The Norwegian Mental Health Care Act

The purpose of the Norwegian Mental Health Care Act is to ensure that mental health care is provided in a satisfactory manner (15). Young people between 16 and 18 years of age have the same rights as adults (over 18 years) regarding voluntary or compulsory admission to mental health care. Parents or other guardians must consent to medical care for those under age 16. All admissions by parental consent are registered as voluntary admissions, even if the young person does not agree to being admitted. An independent and authorized control commission assesses all admissions of adolescents over age 12 who do not agree to the admission.

Sections 4 through 8 of the Mental Health Care Act specify and regulate the legal types of restraint during hospitalization: mechanical restraint, pharmacological restraint, seclusion, and physical holding. Mechanical restraint refers to strapping a patient to a bed with mechanical devices. Pharmacological restraint refers to single doses of medications with the intention of “calming or sedating the patient” in an acute situation. Seclusion is defined as all situations in which the patient is isolated in a locked room without staff present. Physical holding refers to a technique in which staff members restrain a patient physically without any tools. According to the Mental Health Care Act, restraint should be used only when absolutely necessary (principle of necessity) to prevent the patient from self-harming or harming others or from severely damaging property. Restraint cannot be used as treatment. Restraint can be used during both voluntary and compulsory admissions. However, mechanical restraint and seclusion cannot be used if the patient is younger than 16.

Measures

Ethnicity was coded as immigrant background or not. An immigrant was defined as a person with two foreign-born parents. Living arrangements were coded at the time of most recent admission in four categories: living with both parents (biological or adoptive), living with one parent (with or without step-parent), living in foster care or an institution, and other. Unless stated otherwise, data were collected from the most recent admission during the study period.

Diagnoses were made by the local clinical team by using the ICD-10 (16). Patients were grouped into one of six mutually exclusive groups by their principal diagnosis: psychotic disorders (ICD-10 F20–F29), affective disorders (F30–F39), neurotic and stress-related disorders (F40–F48 and F94), eating disorders (F50), externalizing disorders (substance use, F10–F19; personality disorders, F60–F69; hyperkinetic disorder, F90; conduct disorders, F91–F92; and tics, F95), and none of these diagnoses.

Global psychosocial functioning was assessed with the CGAS (17), used routinely by the clinicians at admission (most recent admission). The CGAS is used to measure general functioning, and possible scores range from 1, needs constant supervision, to 100, superior functioning. CGAS scores were divided into three groups (1–39, 40–49, and 50–100). Interrater reliability of the routine use of the CGAS was found to be moderate (intraclass correlation coefficient=.61) in a large Norwegian study of clinicians in outpatient child and adolescent mental health services (18).

Involuntary referral was coded yes if the patient had any involuntary referral during the study period. Current involvement with child protection services at the most recent admission was coded as yes or no. Length of stay was indicated in number of days.

Statistical analyses

For comparisons of restrained and nonrestrained adolescents, we used cross-tabulation with chi square tests. To analyze the impact of the social characteristics (immigrant background, living arrangement, and role of child protection services), mental health characteristics (ICD-10 diagnosis and CGAS score), and treatment characteristics (number of admissions, length of stay, and voluntary or involuntary referral) on the risk of being restrained, we applied a two-step binary logistic regression analysis. The dependent variable was being restrained at least once during the three-year period. In the first step, we performed a separate regression analysis for each variable described above, including age and gender (unadjusted analysis). In the second step, we performed three binary logistic regression analyses with three different models; age (continuous) and gender were used as control variables in all of them. In model 1, age and gender were included as independent control variables and social characteristics were included as independent variables. In model 2, we added mental health characteristics to the variables in model 1. In model 3, we added treatment characteristics to the variables in model 2. No significant interactions were found between the variables analyzed in model 1. For each independent variable, the variance inflation factor was below 2, indicating that multicollinearity did not invalidate the regression analysis. The significance level was set at .05. Analyses were conducted with SPSS, version 18.0.

Results

The sample comprised 288 patients who had experienced at least one restraint episode during the study period, and 288 patients who had not. For all 576 patients, the mean±SD age at the most recent admission was 16.1±1.65 years (range 10–21), and 62% (N=358) were girls. Two-thirds of the sample (N=380, 66%) were 16 years or older. Fifty patients (9%) had an immigrant background. Of these, 81% (N=40) were born in Africa or Asia.

Table 1 and the unadjusted analysis in Table 2 show that the restrained and nonrestrained groups differed significantly on several variables. The restrained adolescents more often had an immigrant background, lived in foster care or an institution, and had involvement with child protection services. The restrained group had larger proportions of patients with psychotic, eating, and externalizing disorders. The restrained patients also had lower psychosocial functioning (mean CGAS scores). Furthermore, restrained patients had more and longer admissions and were more often involuntarily referred.

Results from adjusted logistic regression analyses from models 1, 2, and 3 are shown in Table 2. In model 3, lower psychosocial functioning, immigrant background, and treatment characteristic variables (number and length of admissions and involuntary referral) were significant predictors of being restrained during admission.

In contrast to the unadjusted model, diagnostic status was not a significant predictor of restraint use in the adjusted analyses (models 2 and 3). We tested this further (data not shown) and found that the same conditions (psychotic, eating, and externalizing disorders) predicted restraint use in a modified model 2 that did not include psychosocial functioning (CGAS). The two models (models 2 and 3) that included mental health and treatment characteristics showed that psychosocial functioning was a strong predictor of use of restraint.

Discussion

In this large nationwide case-control study of use of restraint among adolescent inpatients, we found several characteristics that discriminated restrained from nonrestrained adolescents. The adjusted univariate regression analysis showed that inpatients who experienced restraint had lower psychosocial functioning scores (CGAS), were more often from an immigrant background, were more likely to be involuntarily referred, and had more and longer hospital stays. The odds ratios for these variables were high, indicating that they were strong predictors of restraint.

CGAS score was the only mental health characteristic that predicted the use of restraint in the final multivariate regression analysis. This indicates that psychosocial functioning is a more important predictor than diagnosis per se. This finding is consistent with a Finnish study showing that use of mechanical restraint was significantly associated with low general functioning (CGAS) (13,19). According to Leidy and colleagues (20) and Siponen and colleagues (4), diagnoses are frequently found to be important factors in predicting restraint, even in multivariate analyses. Our results were the same in regard to diagnosis, but when CGAS was included in the model, diagnosis was no longer a significant predictor. In line with two previous studies (6,21), we found that adolescents who experienced restraint were more likely to have an immigrant background. Adolescents with ethnic-minority backgrounds are more likely than native-born adolescents to have traumatic experiences (22). These experiences may increase the risk of restraint because the adolescents may become more distressed by a new ward environment and because staff members are predominantly from the majority population. Feelings of insecurity or mistrust may lead to aggressive behavior that necessitates restraint. Staff members’ lack of familiarity with the adolescent’s cultural background may also lead to misperceptions that patients are dangerous and frightening (6). Two U.S. studies found that adolescents from ethnic-minority groups had reduced access to outpatient mental health services (23,24). If this is the case, youths from minority groups may have more serious symptoms on hospital admission, increasing their risk of being restrained.

Like former studies, our study found that multiple admissions and longer stays predicted the use of restraint (13,14,25,26). Being involuntarily referred during the study period was a strong predictor of being restrained. Among adolescents who were involuntarily referred, the odds of being restrained were more than seven times higher. A U.S. study found that admission status was a strong independent predictor of being restrained (27). Patients who are involuntarily referred are likely to have a serious mental health condition, such as psychosis, associated with a high risk of self-harming or harming others and are also likely to actively refuse voluntary treatment. These conditions may increase the risk of aggression during admission.

Age did not predict restraint in the unadjusted model (Table 2). A systematic review from 2002 of restraint use among children and adolescents concluded that patient age was the strongest predictor of restraint (1). In a more recently published study, younger age was the only characteristic that predicted restraint use among adolescents (28). We could not study the effect of age in a proper way, because the rules of admission to treatment are different for adolescents under 16 years old.

In contrast to other studies, gender did not predict restraint use in our study. Three studies have found that rates of restraint were significantly higher for males than for females (4,12,14), whereas others have found that girls were more often subjected to restraint than boys (2).

In our study, living in foster care or an institution predicted restraint use in the unadjusted model. Three previous studies had similar findings (13,14,29). When adjusted for social and mental health characteristics, analyses showed that adolescents living with one parent (with or without step-parent) were less likely to be restrained than those living with both parents. When treatment characteristics were added to the model, living arrangement did not predict restraint use. Having contact with the child protection services predicted restraint use in the unadjusted analysis in our study but not in the adjusted analysis.

The decision to use restraint may result from a complex interaction of patient-related, staff-related, and structural characteristics (30). In our study, we focused only on the patient characteristics.

Limitations and strengths

Diagnoses were made by hospital clinicians; we were not able to test the reliability of the diagnoses. Our data on restraint use were from routine reports by hospital clinicians, and the reliability and validity of these reports were not tested. CGAS reliability was not measured for this study. Other studies indicate that reliability is moderate. However, lower reliability would have led to an underestimation of the relationship between CGAS scores and use of restraint.

Our study included all adolescent patients who experienced restraint in all relevant services in Norway during a three-year period. The study was based on well-defined concepts of restraint. All data were manually collected by the first author. Clinicians are required by law to document the use of restraint in a specific protocol. These protocols are controlled by an independent and authorized control commission.

Conclusions

Several characteristics predicted the use of restraint in adolescent acute psychiatric units: lower CGAS score, immigrant status, multiple admissions and longer stays, and involuntary referral. Future studies should focus on the benefits of using these patient characteristics to identify the group of adolescents who are at risk of experiencing restraint during admissions.