Introduction
In 2013, a report to the United Nations’ Human Rights Council that described the use of coercion in mental health services as torture caused substantial controversy that impacted the psychiatric self-image and fuelled discussions that are still ongoing [1]. Accordingly, all States that had ratified the UN convention for the rights of persons with disabilities (UN-CRPD) should “impose an absolute ban on all forced and non-consensual medical interventions,” and laws should be revised or abolished, accordingly [1]. This view has been criticised from the side of clinicians who argued that mental health laws should be considered as laws protecting the rights of patients, whose abolition could cause severe damage for some patients, for example, with risk of suicide and that full legal capacity can be hampered by severe mental illnesses [Reference Freeman, Kolappa, de Almeida, Kleinman, Makhashvili and Phakathi2, Reference Steinert3]. Though European bodies have claimed to protect the rights of people with mental illnesses and to reduce the use of coercion in several documents since 1983 [4] and several countries have implemented policies accordingly, up to now, none of the EU member states has taken legislative action to abolish the use of coercion completely. On the other side, the UN Committee on the Rights of Persons with Disabilities (UN-CRPD) has continued to express its demands in subsequent General Reports and Comments and Special Reports to all Member States [5], reiterating the initial calls for the total elimination of coercion in mental health care. This was much welcomed by user advocates and human rights activists in several countries, who feel discriminated by Mental Health Laws and have since referred to this “human rights-based approach” [Reference Kelly, Kelly and Donnelly6, Reference Roper, Joffee-Kohn, Edan, Swingler, Gooding and Hamilton7]. In addition, the World Health Organization echoed the Committee’s view in its guidance on community mental health services and its WHO QualityRights Core Training in 2019 [8], again followed by critics from clinicians [Reference Hoare and Duffy9]. The World Psychiatric Association (WPA) has taken an intermediate position, acknowledging the necessity of coercive interventions in some scenarios but also emphasising the urgent need to implement evidence-based interventions to prevent the use of coercion [Reference Herrman, Allan, Galderisi, Javed and Rodrigues10]. Some supreme courts such as the German Constitutional Court and the German Ethics Council have declared in several decisions that the use of coercion is not generally excluded by the UN-CRPD and in some cases is even mandatory to protect the rights of the patient [11, 12].
Thus, there are distinct differences in attitudes towards the use of coercion among those involved in this discussion, and this applies to researchers also. A survey of 76 researchers on coercion from 30 European countries found that 31% believed that it was possible to abolish coercion in their country, while the rest believed the opposite [Reference Birkeland, Steinert, Whittington and Gildberg13]. A convincing practice of the abolition of coercion in psychiatry (or rather in medicine) has never been demonstrated. However, many pre-post studies around the world have shown a significant reduction in the use of coercion after specific interventions [Reference Duxbury, Baker, Downe, Jones, Greenwood and Thygesen14–Reference Korezelidou, Welte, Oster and Mahler16], while high-quality randomised controlled trials, with very few exceptions [Reference Tinland, Loubière, Mougeot, Jouet, Pontier and Baumstarck17] have shown mostly negative results [Reference Steinert, Baumgardt, Bechdolf, Bühling-Schindowski, Cole and Flammer18]. Hence, the most topical question to date is, are we at least on the right track if not to abolish so at least to reduce the use of coercion with the available tools and knowledge [Reference Whittington, Aluh and Caldas-de-Almeida19], as experts and national policies expect?
Real-world ecologic data on the use of coercion in psychiatric services at the regional or even national level have become increasingly available in recent years, thanks to electronic records and the growing capacity for big data analysis. Where such data are collected and reported annually, longitudinal analyses are possible. However, longitudinal data covering the period from the year 2020 are scarcely available so far. Regarding involuntary admissions (IAs), a considerable increase was reported from the Netherlands between the years 2003 and 2023 [Reference Noorthoorn, Gemsa, Broer, Lepping, Nuijen and Hutschemaekers20]. Regarding coercive measures (CMs), a considerable decrease was reported from Ireland between the years 2018 and 2022 [Reference Lucey, Kiernan, Farrelly, Downey and Stepala21].
The aim of our study was to empirically evaluate whether real-world trends in European countries align with international policy frameworks advocating the reduction of coercion in psychiatric facilities. To this end, we collected longitudinal ecologic data (i.e., aggregated nation-wide or regional data) [Reference Morgenstern, Rothman, Greenland and Lash22] in European countries. The data cover the use of IAs, seclusion, restraint, and coercive treatment (CMs) at the national or Federal State level, allowing to analyse trends in the percentage of admissions exposed to these measures and per population.
Methods
Definitions
IA as used here includes any involuntary hospital admission or stay, also detentions after initially voluntary admission. CMs mean additional coercive interventions, including physical or mechanical restraint, seclusion, and involuntary medication. Other coercive interventions such as shielding and involuntary outpatient commitment are mentioned separately where they occur.
Inclusion criteria
We asked members of the European FOSTREN network (Fostering and Strengthening Approaches to Reducing Coercion in European Mental Health Services, www.fostren.eu), which includes researchers from 33 European countries, whether they could obtain data on both IA and CM in their country or a substantial part of it (federal state level). Minimum requirements were (1) datasets should cover at least four years and continue at least until 2020, (2) the number of either hospital admissions or patients exposed to CM and IA should be available, as well as the total number of admissions/patients, and (3) calculation of population-based figures should be possible. Ideally, epidemiological data on the use of CM should cover restrictive practices on all legal grounds, in adult psychiatry (including geriatric and addiction services), outpatient psychiatry, child and adolescent psychiatry, forensic psychiatry, nursing homes, and somatic hospitals. Being aware that such comprehensive data are nowhere available, the minimum requirement was data from adult inpatient psychiatry in the respective country/region.
Data sources
Data sources are displayed in Table 1. For further information see the Supplement.
Data sources
CM, coercive measure; IA, involuntary admission.
Note: Definitions and legal frameworks vary across countries; table summarises coverage to support interpretation rather than enable direct comparison.
Outcomes
The difficulties of collecting and comparing appropriate outcomes in reporting the use of CM have been discussed repeatedly [Reference Janssen, van de Sande, Noorthoorn, Nijman, Bowers and Mulder33, Reference Lickiewicz, Efkemann, Husum, Lantta, Pingani and Whittington34]. To date, based on different legal frameworks, methods of data collecting and reporting vary considerably between and within European countries. This makes comparisons between countries difficult and perhaps questionable [Reference Savage, Lepping, Newton-Howes, Arnold, Staggs and Kisely35, Reference Rains, Zenina, Dias, Jones, Jeffreys and Branthonne-Foster36]. As the aim of this study was not to compare countries but to analyse trends within countries, we did not attempt to harmonise data and outcomes. We present the results for each country with the specific outcomes used. Country-specific limitations in data recording and additional results are displayed in the Supplement.
Analysis
As all data are full surveys covering all psychiatric admissions, no confidence intervals can be calculated. Trend analyses are not appropriate with few data points (years). Instead, we calculated the percentage of change between first year and last year of observation for all outcomes.
Ethical aspects
As this is a study based on a retrospective analysis of publicly available anonymised data, ethical approval was not required in any of the participating countries.
Results
We found experts from six countries who could contribute data according to the inclusion criteria, from Austria, Germany, England, Norway, Switzerland, and Sweden. Results are presented by country using a standardised internal structure. Percentage-based outcomes and measure-specific details are reported narratively within each country to avoid misleading cross-country comparisons due to heterogeneous denominators and scopes of reporting. Table 2 presents population-based rates (per 100,000 inhabitants) for IA and, where available, other CM. Additional country-specific information is presented in the Supplement.
Population-based outcomes
NA, not available due to data limitations.
Notes: Norway: Rates are per 100 K pop>16 years. CM rates could not be calculated using compatible denominators.
a Switzerland reports age-standardised rates; other countries report raw population rates.
Austria
Figure 1 displays the numbers of IA and CM from 2019 to 2022. IA decreased with the beginning of the COVID-19 pandemic in the year 2020 and is on the rise again since, whereas numbers of restrictions are continually rising. Numbers of total admissions show a pronounced decrease in 2020 (see the Supplement, Figure 1). The total numbers of IA increased from 7625 in 2019 to 8615 between 2019 and 2022 (+3%), and the number of CM increased from 24,596 to 25,340 (+13%), while the reference population grew by 1.7% during the same period (see the Supplement, Table 1). Rates per population are presented in Table 1.
IA and CM in Austria 2019–2022 (absolute numbers).
England
The total number of patients exposed to restrictive interventions increased from 12,000 in 2020 to 17,417 in 2024 (+45.14%). For the total number of IA, this increased from 50,893 in 2020 to 52,458 in 2024 (+3.08%). Figure 2 shows the number of patients exposed to CM as well as the total number of IA in England between 2020 and 2024.
Trends in the use of restrictive interventions and all detentions in England 2020–2024 (absolute numbers).
The number of patients in contact with mental health, learning disability and autism services increased from 2,878,636 to 3,790,826 (+31.7%) between 2020 and 2024. The percentage of patients exposed to any CM of all patients in contact with mental health, learning disability, and autism services between 2020 and 2024 increased from 0.42 to 0.46% (+10.22%).
Comparing the data from 2020 to 2023 in England, the population increased from 56,550,000 to 57,690,300 (+2.02%), while the number of patients exposed to CM per 100,000 inhabitants increased from 21.22 in 2020 to 29.24 in 2023 (+37.79%) and the number of IA per 100,000 inhabitants decreased slightly from 90 in 2020, to 88.94 in 2023 (−1.18%). Rates per population are presented in Table 1. See the Supplementary Table for a breakdown of results for each type of restrictive intervention.
Germany
Figure 3 displays the development of the use of IA and CM over between 2015 and 2023. The percentage of IA of all admissions increased by 16.4%. The percentage of admissions exposed to restraint dropped by 20.5%, while the percentage of those exposed to seclusion increased by 78.6%, those exposed to involuntary medication increased by 60%, and the percentage exposed to any kind of CM increased by 7.2%.
Trends in the use of coercion in Baden-Wuerttemberg (Germany) 2015–2023 (admissions exposed to coercive interventions of all psychiatric hospital admissions).
During the observed time period, the population increased from 10,952,000 to 11,339,000 (+3.5%), the N of psychiatric admissions increased from 97,981 to 119,622 (+22.1%) (forensic psychiatry not included). The N of IA increased from 11,300 to 16,189 (+42.5%), and the N of admissions exposed to any kind of CM increased from 7,200 to 9,575 (+33.0%).
Rates per population are presented in Table 1. Including also all cases treated in forensic psychiatry, the number of CM are slightly higher, increasing from 66.2 up to 84.4 (+27.5%).
Norway
Figure 4 shows national rates of events and persons related to various forms of coercion, all of which increased in the time period. The incidence rate of IA rose just over 12% both when counting persons and admissions (from 183.5 and 129.9 to 206.2 and 145.5, repsectively). In absolute numbers, the increase was from 5,550 patients and 7,821 events in 2015 to 6,660 patients and 9,401 events in 2022. For new community treatment orders (CTOs), each year the increase was 4.75% for patients (from 4,744 to 5,294) and 7% for events (from 2,607 to 2,982).
Rates per capita of IA, coercive treatment, community coercion and CM in Norwegian specialist adult mental health services 2015–2022.
The combined use of CM increased by 86.4% (from 15,485 to 28,859). Physical restraints increased by 237.5%, and seclusion by 91.8%, while there was a 8.75% decrease in mechanical restraints. The increase in the number of people subjected to CM was highest for seclusion (147%) and physical restraints (100%). Physical restraints accounted for 33.2% of all CM events in 2015, and this increased to 46.9% in 2022. The corresponding numbers for shielding was 33.2 and 26.8%, respectively, and the proportion of mechanical restraint reduced from 24.4 to 11.9%.
The rate of persons subjected to involuntary treatment/maintenance medication increased by 100.1% in the period (from 1,757 to 3,768). A number of these persons will also be subjected to CTO.
Sweden
Figure 5 shows the development of the use of IA and the most frequently applied CM over the period. The percentage of involuntarily admitted patients of all admitted patients with psychiatric diagnoses increased from 22.0% in 2014 to 24.2% in 2023. The percentage of patients exposed to mechanical restraint were equal, 2.4%, in 2014 and 2023, while patients exposed to seclusion increased from1.6 to 2.3%, and patients exposed to forced medication increased from 3.0 to 4.5% from 2014 to 2023.
Use of coercion in Sweden 2014–2023. Percentage of patients exposed to IA and CM of all patients admitted with a psychiatric diagnosis (F00–F99).
The population increased by 8.4% from around 9,738,000 in 2014 to 10,558,000 in 2023. The registered number of IA patients were nearly equal in 2014 (13,501) and 2023 (13,836), a slight increase of 2.4%. Thus, the number per 100,000 inhabitants decreased by 5.4% from 138.6 to 131.1. The registered number of patients exposed to mechanical restraint decreased during the period by 6.7% from 1,461 (15.0/100,000 inhabitants) in 2014 to 1,363 (12.9/100,000 inhabitants) in 2023, while registered patients exposed to seclusion increased by 38.3% from 959 (9.8/100,000 inhabitants) to 1,326 (12.6/100,000 inhabitants) and patients exposed to forced medication increased by 37.7% from 1,865 (19.2/100,000 inhabitants) to 2,568 (24.3/100,000 inhabitants).
Switzerland
The following data refer to the years 2016 to 2023 (Figure 6). The absolute (raw) number of people involuntarily admitted to inpatient psychiatric treatment rose from 13,671 in 2016 to 18,347 in 2023 (+34.2%). The age-standardised rate rose from 161 to 203 per 100,000 inhabitants (+26.8%).
Absolute numbers of IA and adult inpatient CM in psychiatric hospitals in Switzerland, 2016–2023.
The absolute number of patients subjected to any form of CM, while in inpatient treatment rose from 5,060 in 2016 to 7,886 in 2023 (+55.8%). It appears that measures taken as part of inpatient treatment have risen disproportionately. For limitations, see the Supplement.
Discussion
This study provides a descriptive synthesis of longitudinal, routinely collected ecologic data on involuntary admissions and other coercive measures in six European countries. Given heterogeneity in legal frameworks, data sources, denominators, and periods of observation, the analyses are intentionally descriptive and do not include formal statistical testing of trends. Accordingly, observed changes should be interpreted as patterns within countries rather than as statistically validated increases or decreases, and comparisons between countries should be made with caution. Overall, there was no evidence of a substantial decrease in IA or the use of CM in any of the countries. Our study provides a real-world longitudinal confirmation of policy-practice gaps. The suggestion that a decrease in the use of coercion in inpatient psychiatry is underway must be rejected. Expectations of a reduction in coercive practices in many European countries are grounded in a combination of international policy frameworks and accumulating intervention-level evidence. Over the past decade, initiatives from the UN, WHO, etc., have emphasised rights-based approaches [1, 4, 5, 8], recovery-oriented care [Reference Korezelidou, Welte, Oster and Mahler16, Reference Leamy, Bird, Le Boutillier, Williams and Slade37], and the implementation of “alternatives” to coercion (which entail also preventive measures according to current conceptual framework [Reference Herrman, Allan, Galderisi, Javed and Rodrigues10]). In parallel, controlled and quasi-experimental studies have yielded some evidence that specific interventions such as staff training in de-escalation [Reference Price, Baker, Bee and Lovell38], organisational change programmes [Reference Hirsch and Steinert39], advance directives, and peer-supported models [Reference Tinland, Loubière, Mougeot, Jouet, Pontier and Baumstarck17] can reduce the use of coercive measures under defined conditions. Against this backdrop, longitudinal real-world data offer important contextualisation of whether such policy ambitions and local interventions translate into sustained system-level change.
The trends observed from the descriptive findings are inconsistent across countries. The descriptive results suggest an increase in IA in Germany, Switzerland, and Norway, while there was no substantial change in the other countries. More pronouncedly, there seems to be an increase in the use of CM during inpatient treatment in all countries. This holds true for different types of outcomes, such as absolute numbers, percentages of admissions, or rates per population, while the types of CM used showed different trends, for example in Sweden or Germany, partly due to legislative changes [Reference Flammer, Hirsch and Steinert40]. Generally, trends in the use of CM should be interpreted together with trends in IA. As the use of CM is legally associated with involuntary status, a change in the proportion of IA in all admissions is supposed to be associated with a similar change in admissions exposed to CM. Changes in IA, both in absolute numbers and percentages, are supposed to have their origins in causes outside the psychiatric hospital system. These could be, for example, legislative changes, changes in societal attitudes towards mental illness and dangerousness [Reference Schomerus, Schindler, Sander, Baumann and Angermeyer41], absolute or relative changes in availability of community psychiatric services [42], changes in incidence of mental disorders as observed after the COVID-10 pandemic [Reference Ding, Zhang, Wang and Wang43], or demographic changes with a change in the proportion of people at risk such as those with old age and associated disorders and young male migrants [Reference Della Rocca, Luciano, Bello, Barone, Carbone and D‘Arpa44]. An example is Switzerland, where age-standardised rates are available for IA. While the absolute number of IA increased by 34.2% within the period investigated, the age-standardised rate increased only by 26.8%, showing that an increase of populations at risk attributes in part for the observed increase. Changes in the proportion of admissions exposed to CM, on the other hand, are supposed to have their origins inside psychiatric hospitals, if they deviate from the observed changes in IA. These could be, for example, implementation or neglect of coercion reduction strategies [Reference Steinert, Baumgardt, Bechdolf, Bühling-Schindowski, Cole and Flammer18], changes in number [Reference Boden, Smith and Trafton45] and qualification [Reference Moyo, Jones, Dennis, Sharma, McKeown and Gray46] of staff, or a change in the proportion of patients exhibiting violent behaviour. Violent behaviour is strongly associated with the use of CM [Reference Wullschleger, Chieze, Courvoisier, Hurst, Sentissi and Kaiser47], but there is no consistent evidence that violence has been increasing in psychiatric hospitals [Reference Eisele, Flammer and Steinert48]. The conditions of the COVID-19 pandemic have been shown to be associated with an increased use of coercion [Reference Flammer, Eisele, Hirsch and Steinert49]. While there is a sustained impact on workforce availability and the incidence of mental disorders particularly among youth [Reference Ding, Zhang, Wang and Wang43], the increased incidence of depressive and anxiety disorder is no plausible explanation for a rise in the use of CM, as those disorders are rarely exposed to CM [Reference Flammer, Hirsch and Steinert40]. The available data do not allow for quantitative analyses regarding the possible causes as mentioned. However, these considerations could provide a framework guiding the interpretation of the descriptive findings. For example, in England, the number of IA per 100,000 inhabitants decreased by 0.4% during the observed period, while the number of patients exposed to CM increased by 42.3%. This suggests a change in practice within psychiatric hospitals despite an unchanged admission policy. In Germany, the number of IA per 100,000 inhabitants increased by 36.8%, while the number of admissions involving any form of CM increased by 26.0%. This suggests that the increase in the total amount of coercion has origins outside psychiatric hospitals. However, caution is needed because it is well-known that there is a high variation in the use of coercive practices within countries [Reference Aragonés-Calleja and Sánchez-Martínez50, Reference Lay, Nordt and Rössler51], and a causal elucidation of the reasons of changes therefore is needed not only on a national but rather on a regional level.
Our results are consistent with publications from other countries covering earlier years. Data on involuntary psychiatric detentions in 25 US states from 2011 to 2018 showed a 13% increase in detentions, while the population grew by 4% [Reference Lee and Cohen52]. In British Columbia, IA increased by 66% from 2008 to 2018, while the number of voluntary admissions remained relatively stable [Reference Loyal, Lavergne, Shirmaleki, Fischer, Kaoser and Makolewksi53]. Rains et al. analysed government reports of IA from 22 countries covering the period from 2008 to 2017 [Reference Rains, Zenina, Dias, Jones, Jeffreys and Branthonne-Foster36]. The median rate was 106.4 per 100,000 and most countries showed an annual increase, with some exceptions (Denmark, Norway, Finland, Sweden, and Italy). In England, comparable data are available from 1988 to 2016, showing that detention rates more than doubled within this period [Reference Rains, Weich, Maddock, Smith, Keown and Crepaz-Keay54]. In Northrhine-Westphalia, Germany, IA by public law increased by 39% between 2000 and 2019 per 100,000 of the population, rising from 114.8 to 142.9 (an increase of 24%) [55]. In the Netherlands, IA increased between 2003 and 2023 from 70 to 170 per 100,000 after correcting for population growth [Reference Noorthoorn, Gemsa, Broer, Lepping, Nuijen and Hutschemaekers20].
Regarding the use of CM, there is a paucity of published longitudinal data at the national or state level. In Germany, a decrease in the use of CM was observed between 2004 and 2019, which was almost entirely due to a significant reduction in patients with organic disorders [25]. Välimäki et al. [Reference Välimäki, Yang, Vahlberg, Lantta, Pekurinen and Anttila56] analysed nationwide data from Finland between 1995 and 2014 and concluded that lasting changes in CM have not yet been achieved despite changes in legislation. Vruwink et al. [Reference Vruwink, Wierdsma, Noorthoorn, Nijman and Mulder57] analysed nationwide data from approximately 85% of hospitals in the Netherlands between 1998 and 2019. They found a decrease in seclusion during a nationwide programme to reduce seclusion between 2006 and 2012, but this decrease was not sustained afterwards, with a continuous increase in IA and forced medication instead. Contradicting our reported trends, results from Ireland for the period from 2018 to 2022 were published, showing a 24% reduction in seclusion episodes and a 48% reduction in restraint episodes [Reference Lucey, Kiernan, Farrelly, Downey and Stepala21]. This decrease was attributed to better adherence to rules and regulations.
Our study has several limitations. First, legislation and data recording differ between the included countries, and as expected, allow for comparisons between countries only with great caution. Furthermore, the time periods for which data were available were not identical. No country had data on the use of coercive practices in residential homes, somatic hospitals, or prisons. Some of the used outcomes could not be reasonably calculated in all countries. Due to data privacy requirements, databases on IA and CM in most countries comprise little additional information regarding diagnoses and sociodemographic characteristics to prevent personal identification. More in-depth analyses of subgroups concerned were therefore not possible. Additional methodological limitations in the respective countries are mentioned in the Supplement.
Conclusion
Contrary to recent claims of a reduction of coercion in psychiatry, current ecologic real-world data show no decrease or even an increase in IA and in the use of CM in the European countries where we were able to collect data. The findings have to be interpreted with caution in the light of several limitations and their descriptive nature. Further research should explore the reasons in qualitative analyses, for example, interviews with stakeholders in psychiatric hospitals, courts, and community authorities.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2026.10160.
Data availability statement
Data is available on request by the authors, depending on the country’s legal regulations for the collected data.
Acknowledgements
none.
Author contribution
TS collected data from Germany and drafted the manuscript. EF collected data in Germany and calculated outcomes. SH recruited country experts and aided calculation of outcomes. JS collected data from Austria and calculated outcomes. AB and LK finalised the data from Sweden, collected by LK, and contributed to the manuscript. JR collected the data from Norway and drafted the related text. AHD and KF collected data from England, reviewed and read the manuscript, AH contributed substantially to the revision. DR collected data from Switzerland and reviewed the manuscript.
Financial support
There was no funding for this study.
Competing interests
TS received funding for conducting the registry for coercive measures in Baden-Wuerttemberg and for developing a guideline for avoidance of coercion from the German Association for Psychiatry and Psychotherapy (DGPPN), for the PreVCo Study from the German Innovationsfonds, for a registry on coercive measures from the Federal State of Schleswig Holstein, and for an expert review on coercive measures in Luxemburg from the Ombudsman Luxemburg. SH has receives funding for conducting the registry for coercive measures in Baden-Wuerttemberg and for updating the guideline for avoidance of coercion from the German Association for Psychiatry and Psychotherapy (DGPPN). DR has received royalties for a book on human rights in psychiatry from Springer Nature Publishers. AB, LK, EF, JS, JR, AHD, and KF declare that they have no COI.
Open access
Comments
No Comments have been published for this article.