Seclusion Reduction in Dutch Mental Health Care: Did Hospitals Meet Goals?

In recent decades, the use of coercion—in particular, the use of seclusion—in Dutch mental health care has been a topic of public debate. Use of seclusion is controversial and regularly attracts media attention. Seclusion appears to be used more frequently and for longer periods in the Netherlands, compared with other countries, whereas other measures, such as forced medication and restraint, appear to be used less frequently (1–3). Cross-national comparisons have been difficult because of the lack of reliable data. Data on frequency and duration of coercive measures were not nationally registered in the Netherlands before 2006.

Dutch psychiatric hospitals have been engaged in projects to reduce coercive measures since 1998 (4,5). A wide range of new care methods has been developed to reduce the use of restraint and seclusion. These initiatives were supported by the Dutch government with €35 million in funding between 2006 and 2012. Implementation of change was organized at the hospital level. GGZ Nederland, the Dutch mental health umbrella organization, has supported exchange programs that organize quarterly meetings, allowing hospitals to learn from each other. Apart from a general policy to reduce the use of restraint and seclusion, no goals were specified beforehand. Hospitals participated in quantitative (6–8) and qualitative (9) research aimed at identifying good practices. These were categorized into five clusters (9): therapeutic engagement, prevention and deescalation, evaluation and reflection, collaborative care, and consumer involvement.

Changes included adaptation of frequent and structured team briefings, risk assessment (6), individualized crisis management, evaluation of support activities, feedback of data on coercive measures, and deescalation training (9). These changes enabled culture shifts in psychiatric units aimed at different ways of thinking and working, expressed in concepts, such as “from control to contact,” “from reactive to proactive,” “from routine to reflection,” “from closed to open,” and “from informing to participating” (9). In addition, the building layout was changed in many hospitals; for example, single-person bedrooms, comfort rooms, family rooms, and low-threshold access to nurses in the ward or behind accessible counters, rather than in nurse stations, were introduced (10).

Measuring effects of these approaches was challenging, given a lack of complete and reliable data. It was also difficult to extrapolate trends from available Health Care Inspectorate data (11). Therefore, five psychiatric hospitals, in collaboration with GGZ Nederland and the Health Care Inspectorate, set up the Argus register in 2006 (12,13). Beginning in 2007, a growing number of hospitals were included. In 2010, data were gathered from 17 hospitals, constituting 67% of the catchment area in the Netherlands. In 2011, data were gathered from 29 hospitals and seven psychiatric wards of general hospitals, or 93% of the catchment area. Participation in the register became mandatory in 2012 for all psychiatric facilities. In 2013, the Argus register received data from 66 hospitals, which together make up 75% of all facilities and cover 100% of the catchment area. By 2014, all 87 psychiatric hospitals and psychiatric wards in general hospitals with involuntary admissions used the Argus data set.

In 2006, GGZ Nederland set a seclusion reduction target of 10% per year (14). A 2012 report based on inspectorate data showed that seclusion increased until 2006 and decreased afterward (11). This study used 2008–2013 data in the Argus register to evaluate whether the objective of 10% per year was achieved. It also assessed whether seclusion was replaced by other measures, such as forced medication, an important political question in the Netherlands (15). Finally, we examined to what extent hours in seclusion per admission hours was associated with ward or hospital type and with specific patient profiles.

The following research questions were addressed: What were the trends in number and duration of seclusion incidents and secluded patients between 2008 and 2013? Were other coercive measures being substituted for seclusion? To what extent was seclusion determined by type of psychiatric disorder or type of ward or hospital?

Methods

Coercive Measures

The register monitors the use of seven types of coercive measures (12,13). “Seclusion in a high-security room” is defined as the seclusion of a patient for care and treatment in a specifically designated room that has been approved by the health authority as a seclusion room. “Seclusion in a low-security room” is defined as secluding a patient for care and treatment in a specifically designated room not necessarily approved by the health authority. This is a low-stimulus single room furnished with a bed, table, chair, wardrobe, and washing facilities. “Seclusion in any other area” refers to a seclusion room that does not meet any health authority building criteria (10). “Physical restraint” is defined as restricting a patient in his or her movement options, either by using mechanical devices or by physically keeping hold of the patient. “Parenteral (forced) medication” is defined as forced medication administered intramuscularly or intravenously to the patient. “Forced hydration and/or nutrition” is defined as fluids or food orally forced or parentally administered. “Other coercive measures” include those with a therapeutic intent other than the above-mentioned measures (for example, a mandatory stay in the patient’s bedroom). For each coercive measure, the date, start time, and, except for forced medication, end time are recorded. The patient’s level of resistance is recorded in three categories: on own request, no clear objection, and clear verbal or physical objection.

Data used in our study were sent anonymously to the national Argus register and were not traceable to an individual patient. At the Argus register, the data are checked for reliability and completeness. Reliability for the data used in our study was determined by comparing the data on coercive measures with data from other sources, such as Health Care Inspectorate letters, medical charts, and reports to courts; kappa values ranged from .64 for forced medication to .92 for seclusion (13). Completeness was checked by linking to patient background data in the Dutch health care finance system. When more than 5% of the measures could not be linked to patient background data, hospitals were obliged to check their sources and redeliver data. In addition, results of the analysis were shared with the project managers and ward staff, who were asked to check the accuracy of the data. If questions remained, site visits and further checks were made. All hospitals participated in these procedures (N=66 in the final year).

Number and Duration of Incidents and Admissions

An incident was defined as a period in which a seclusion or restraint measure was used after not being used for at least 24 hours. We used this definition for two reasons. First, when a measure has not been used in the previous 24 hours, a new decision by the psychiatrist in charge is required. Second, in the Dutch context, seclusion generally occurs in sequences of several discrete episodes. For forced medication, the event rate—the number of events per day—was calculated (16). For all incidents, the separate decisions and the reports to the Health Care Inspectorate are important numerators. The numbers were used in annual policy reports to the Dutch government and the Health Care Inspectorate. In analyzing trends for seclusion and restraint, two calculations were used, the number of incidents per number of admissions and the duration of the incident per the duration of the admission. The number of incidents per admission is important in wards with many admissions and mainly short-term incidents. The number of hours of coercive measures and the number of events per duration of the admission are relevant to specialized wards, such as geriatric or forensic wards in which admissions are longer (13).

Patient Background Data

For number of patients admitted and admission hours, hospital administrative departments provided data collected during routine admission procedures. The database covered patient characteristics (13), such as gender, date of birth, marital status, ethnicity, and diagnosis. The ward types were categorized into six groups (10): psychiatric ward in a general hospital, admission ward, child and adolescent ward, long-stay ward for the care of persons with chronic mental illness, geriatric ward, and forensic ward. Hospitals were classified as urban or rural by using data on housing density (17). They were classified as small if they had fewer than 50 beds, as medium if they had up to 400 beds, and as large if they had 400 or more beds (10,13).

Analyses

Trends in the number and duration of seclusion incidents were characterized with counts and with chi-square, Student’s t, or nonparametric Wilcoxon U tests when appropriate. The substitution of other types of coercive measures for seclusion was analyzed in a sample of 25 hospitals with complete and readily linkable data for 2011, 2012, and 2013. Incidents of enforced fluids or nutrition were not included, because they were rare. To determine the extent to which seclusion was related to type of psychiatric disorder or type of ward or hospital, we used multilevel logistic regression analysis stratified per year (18). The outcome variable was duration of seclusion incident per duration of admission, with admission days as the exposure variable (19). Model fit was inspected at each step of the analysis by means of McFadden’s R² (20). A patient’s admission was identified as level 1, the patient as level 2, and the ward as level 3. This analysis took into account size differences between hospitals, their patient mix, the nature and size of specific wards, and the location of the hospital (urban or rural).

All procedures performed were in accordance with the ethical standards of the National Research Committee as obtained in 2006 and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Results

Number and Duration of Seclusion Incidents

Table 1 summarizes the seclusion findings from 2008 to 2013, a period during which the number of hospitals providing data to the Argus register increased to an estimated coverage rate of nearly 100%. Both the seclusion rate and the (mean and median) duration of seclusion incidents decreased during this period. The seclusion rate fell from 11.8% in 2008 to 7.0% in 2013, and the median duration of seclusion incidents decreased from 92 to 16 hours. At the level of the individual hospital, absolute change varied between a decrease of 93% and an increase of 112% (Figure 1). A greater than tenfold difference was found between hospitals in the duration of seclusion incidents per admission duration (mean for 2013 of 4.8; 95% confidence interval=.8–10.3). Of the 24 hospitals with a decrease in seclusion hours per admission over the years that reduction was evaluated, 19 were engaged in the seclusion reduction program for at least three years; most of the increase was observed in hospitals that started in the program at a later point. Because of the limited number of participating hospitals during the first years (2008–2010), it is unclear whether the reduction in the seclusion rate and incident duration reflected an overall reduction or the inclusion in the later samples of hospitals with low rates.

However, more than 93% of all admitted patients after 2011 were included in the register, allowing a more sound comparison, which showed a mean annual reduction in seclusion rates of 9% and a reduction in mean and median duration of seclusion incidents of 8% and 32%, respectively.

Substitution

An important question is whether the reduction in seclusion may have led to the substitution of another type of coercive measure. In a sample of 25 hospitals, the number of admissions remained relatively stable between 2011 and 2013, whereas the number of admission hours and the mean and median duration of admissions decreased substantially (Table 2). In absolute numbers, seclusion incidents and physical restraint incidents remained constant, whereas the initially low number of incidents of forced medication increased by 81%. The total number of coercive interventions increased significantly. The duration of seclusion incidents was compared with the duration of admissions, with the analysis correcting for the observed substantial decrease in admission hours. This analysis indicated a significant decrease in seclusion incidents per 1,000 admission hours, compared with a significant increase in forced medication incidents per 1,000 days. These findings suggest that forced medication was partly substituted for seclusion. This was confirmed by a rank-order analysis that showed an inverse proportional relationship in the ratios for each hospital between seclusion and forced medication in 2011 and 2012 but not in 2013. The total number of days on which coercive interventions occurred decreased. The number of incidents of physical restraint per 1,000 admissions remained constant, and thus no substitution of physical restraint occurred.

Table 3 presents data on the characteristics of secluded patients. In a multilevel logistic regression analysis, patient characteristics were related to the duration of seclusion adjusted for the duration of admission in 2011, 2012, and 2013. The multivariate analysis showed younger age, male gender, and stays on admission wards or forensic wards and having a psychotic or bipolar disorder were associated with higher rates of seclusion, whereas having an adjustment disorder or depression—and in 2011 having schizophrenia or Asperger’s syndrome—was associated with lower rates. For urban location and hospital size, no clear picture emerged. These characteristics were associated with both higher and lower rates of seclusion in 2011–2013. The final model showed McFadden’s R² of approximately .23, implying a reasonably good model fit (20).

Discussion

Between 2008 and 2013, hospitals that submitted data to the Argus register experienced a yearly decrease of approximately 16% in the mean duration of seclusion incidents and a yearly decrease of 30% in the median duration of seclusion incidents, adding up to an overall decrease of above 50%. Seclusion incidents per admitted patients decreased by 41% overall, from 11.8% in 2008 to 7.0% in 2013. This figure is lower than that in a study conducted in 2002 in 12 Dutch hospitals, which found that 15.7% of all admitted patients were secluded (2). Thus some of the reduction in seclusion use likely occurred before 2008.

The reduction in seclusion seemed to be accompanied by an increase in forced medication, suggesting substitution of one measure for another. However, longitudinal analysis of all coercive measures in two previous studies showed that the decrease in seclusion was greater than the increase in forced medication (22,23). Our study seemed to confirm this finding, because use of enforced medication remained well below use of seclusion despite a clear rise in number of enforced medication events (Table 2), although not at the individual patient level. To some extent, it may be concluded that Dutch practice is changing in line with international treatment policy (21,24,25).

Substantial differences between hospitals were found in the duration of seclusion per the duration of admission. Such differences were found in other studies over different time frames (3,13,16,19). In one of these studies, a top-down coordinated approach to seclusion reduction was associated with fewer seclusion hours per admission hours (3). This observation was confirmed in several reports of the Health Care Inspectorate, which found that seclusion reduction was related to a focused investment in change to the least restrictive clinical practice (26,27). In our study, we observed both a decrease and an increase over the years in the duration of seclusion per the duration of admission. On the one hand, this may have reflected differences in the number of years that the hospitals were actively committed to seclusion reduction programs. On the other hand, there were large differences in the size of hospitals; patient profiles (for example, patients with drug intoxication were more prevalent in some regions); and number of specialized wards, which means a difference in patient populations. In addition, a national policy of bed reduction led to a sometimes steep reduction in a hospital’s total admission hours, which may have obscured a reduction in seclusion hours. Hospitals that admitted patients with characteristics associated with a likelihood of seclusion, such as having a psychotic disorder or a forensic problem, may have had higher seclusion rates than other hospitals. In future benchmarking, it is advisable to correct for such differences. Most of the variation between hospitals was not explained by the data. The recent development of “high and intensive care” (25), which is in line with international developments in this area (26), aims at fostering more uniformity, adherence to evidence-based practice guidelines, and sustainability of provision of the least restrictive acute care.

A number of observations can be made on the basis of our findings. Hospitals participating in reduction efforts for a longer time reported greater reductions in use of seclusion than hospitals with shorter participation periods. In the United States, where the use of coercive measures has been addressed since the 1980s, research shows that a substantial cultural change among employees requires at least five years (30), which is confirmed by our study. Also, qualitative research into best practices in coercion reduction in the Netherlands (9) showed that creation of alternatives to coercive measures requires a change in organizational structure and culture, stable and motivated management, and support at all levels of the organization.

There are some limitations in using data gathered from hospitals with varying periods of participation in coercion reduction efforts. In addition, the register’s coverage expanded over the years. The 2013 data represent the complete catchment area. The trends reported in Table 1 represent 67% of the catchment area, and the findings presented in Tables 2 and 3 represent 93%. It is not inconceivable that hospitals that joined the Argus register at a later stage also implemented reduction project activities later and therefore achieved less substantial reductions. It is also possible that they had achieved lower seclusion rates before they started reporting data to the register. A second important limitation was that before 2012 we had no legal means at our disposal to oblige hospitals to deliver—or redeliver—data. Some hospitals could not be included in the database, which may have led to a selection bias. A third limitation was that we could not include contextual variables in the analysis, such as information about personnel, ward management, or leadership, which may have led to some of the unexplained variation. A final potential limitation was the limited coverage of specialized clinics for patients with addictions or developmental disabilities or for forensic patients. Thus the findings primarily reflect the main body of general psychiatry. Before 2010, national implications of the findings are limited.

Conclusions

On a national level, the number and duration of seclusion incidents in Dutch psychiatric hospitals decreased from 2008 to 2013. Change occurred in half the hospitals in the study. These hospitals are halfway along in the trajectory toward a reduction in seclusion, in line with the previously set goals, even though the use of medication may have been substituted for the use of seclusion. The hospitals with no substantial reduction have a long way to go. Given the stage of this health care innovation in the Netherlands, continuing reduction efforts and frequent scientific evaluation are crucial in meeting the objective of the least possible coercive care in Dutch psychiatric hospitals. We recommend following the U.S. “six-core strategy plan to reduce seclusion and restraint” (31), a systematic and nationwide coordinated approach, supported by quantitative and qualitative research and timely feedback.