Involuntary admissions for non-affective psychotic disorders in young refugees and peers in Denmark: A population cohort study

treatment using logistic regression, and rate ratios (RR) of further involuntary admissions, total number of involuntary admissions, and days of involuntary care among patients initially admitted involuntarily using Poisson regression. We compared refugees with majority peers (native-born with native-born parent), other migrants, and descendants of non-refugee migrants. Results: Compared with the majority group, refugees, non-refugee migrants and descendants were at increased risk of involuntarily admissions (OR range = 2.12 – 2.69). Differences in sex, age, education, household income and family situation did not explain these disparities. In contrast, the risk of subsequent involuntary care did not differ between groups (RR range = 0.77 – 1.31). Conclusions: The findings highlight the need to review if and why processes of needs detection and voluntary treatment enrolment are less effective for minorities in Denmark. Further studies should investigate the pathways to care across population groups to inform interventions that address disparities.


Introduction
Refugees and other migrants and ethnic minority groups are at increased risk of developing non-affective psychotic disorders (NAPD) compared with native-born populations (Brandt et al., 2019;Dapunt et al., 2017;Hollander et al., 2016).For young refugees, a greater exposure to risk factors for psychosis that often accompanies forced migration may contribute to this finding, such as traumatic life events (Brandt et al., 2019).Simultaneously, refugees share other risk factors for psychosis with other minority groups, including experiences of discrimination and social exclusion, restricted possibilities to participate in society, and social adversity or "social defeat" (Brandt et al., 2019;Dapunt et al., 2017;Selten and Cantor-Graae, 2010).Delayed and suboptimal healthcare may aggravate this disadvantage.Involuntary admissions in particular may be indicative of more acute symptomology and a more problematic relationship between the patient and healthcare provider (Rodrigues et al., 2019;Sheehan and Burns, 2011;Wyder et al., 2015).
There is evidence that some ethnic minorities and migrant communities experience higher rates of involuntary psychiatric admissions, including black populations in the UK and Canada (Anderson et al., 2014), migrants in Sweden, Denmark and Canada (Holmer et al., 2020;Rodrigues et al., 2019;Terhune et al., 2022) and refugees in Denmark and Canada (Norredam et al., 2010;Rodrigues et al., 2019).However, the current evidence base is limited by several factors.First, most studies reporting ethnic variation in involuntary admissions were conducted in the UK.Of 71 included studies in the most recent systematic review on ethnic variation in involuntary admissions (Barnett et al., 2019), 49 were based on UK data.Second, most studies have explored involuntary admissions in populations of prevalent NAPD cases (or for psychiatric disorders more broadly).Knowledge of group differences in involuntary care in incident cases is important as disparities at each stage of contact may have different implications for how to address the issue: while involuntary admissions of patients already in treatment calls for attention to processes within the treatment system, involuntary admissions in the pathway into care places attention instead on the system of needs detection and the process of enrolment into early intervention programs.
According to the Danish Mental Health Act, involuntary admissions are permissible if 1) the prospect of getting better will be exceedingly worsened if the patient does not get treatment ("treatment indication"), or 2) the patient constitutes a danger to either themselves (suicidal) or others (homicidal) ("danger indication") (Brandt-Christensen, 2012).All medical doctors can initiate an involuntary admission, but the measure requires approval by the senior psychiatrist at the receiving ward.Patients are informed both orally and in writing of the reasons for involuntary admission, and attempts to admit the patient voluntarily must be made first.Monitored by the Danish Health Authority, regional healthcare providers have, since 2014, had a strategic focus on reducing the proliferation of involuntary admissions and other coercive measures at psychiatric hospitals (Danish Health Authority, 2021).
In the present study, we investigated whether young refugees, nonrefugee migrants, and descendants of non-refugee migrants were at higher risk of involuntary admissions at or within three months of their first NAPD diagnosis than their majority peers.To focus on the population of youth who at least partially grew up in Denmark, we only included youth already living in Denmark by age 18.Among those who were admitted involuntarily during this period, we additionally examined differences in the risk of: i) having multiple involuntary admissions, ii) the total number of involuntary admissions, and iii) the number of days that patients in each group were hospitalized against their will.These groups differ in their particular migration histories, notably whether or not their families were forcibly displaced and whether they themselves experienced migration or were born to parents who migrated.At the same time, all groups have in common the experience of growing up as minorities in Denmark.In a previous study (de Montgomery et al., 2023), we have shown that the risk of entering treatment for NAPD was 2-3 times higher in these groups compared with the majority group, but that differences between the groups were small, and that individuals from minority groups tended to be older than their majority peers when entering NAPD treatment.If these groups experience delays in entering NAPD treatment, which the higher age at first contact may indicate, we hypothesized for the current study that they would also be more likely to be admitted involuntarily in the initial phase of their NAPD treatment.

Study population
The study utilized nationwide registers linked at the individual level through the unique personal identifier assigned to all residents of Denmark.Information on hospital admissions and discharge diagnosis were obtained from the National Patient Registry (Lynge et al., 2011), purchases of prescribed antipsychotics (Anatomic Therapeutic Chemical classification (ATC) codes N05A, omitting lithium N05AN) through the Danish National Prescription Registry (Wallach Kildemoes et al., 2011), and socio-demographic information through Statistics Denmark.
The study included all youth aged 18-35 years who had a first contact with psychiatric services (inpatient or specialized outpatient) between 01.01.2006 and 31.12.2018 with a NAPD diagnosis (International Classification of Diseases (ICD-10) codes F20-29 (WHO, 2004)) as assigned at discharge (n = 29,783).To focus on incident cases we excluded individuals with prior contacts for NAPD in the preceding three calendar years and individuals who had purchased anti-psychotic medicine within 3-15 months before admission.Purchases of antipsychotic medication in the last three months prior to first admissions were considered as part of the pathway to first admission, as medication may be attempted in primary care before entering inpatient or specialized outpatient care.Individuals who were not living in Denmark during three full calendar years prior to first NAPD treatment were excluded as we could not determine whether these healthcare contacts were incident or prevalent.
Four groups were defined for comparisons.Because grounds of residence data was only available from 1993 onwards, refugees were defined as persons born outside Denmark who either had legally recognized refugee status or were reunified to an individual with refugee status, or who immigrated between 1986 and 1993 from a major refugee sending country.Non-refugee migrants (henceforth 'migrants') were similarly born abroad, but were not identified as refugees.Both refugees and migrants were restricted to those who immigrated before turning 18 to focus on the population of youth who at least partially grew up in Denmark.Descendants of non-refugee migrants were defined as persons born in Denmark to immigrant parents whose origin country was either Turkey, Pakistan or Morocco, as substantial labor migration from these countries in the 1960s and 1970s has led to the presence of sizeable groups of descendants from these origin (henceforth 'descendants').This group provides an interesting comparison for several reasons: they were born in Denmark and should subsequently not face language barriers, their parents came to Denmark mainly as labor migrants rather than as refugees and should differ from the refugee population in terms of their exposure to traumatizing experiences.At the same time, they may share other characteristics with the refugee population such physical and cultural traits that place them at greater risk of facing discrimination and social exclusion.Persons born in Denmark to parents who were also born in Denmark constitute the final comparison group used as the referencegroup throughout the analysis (henceforth the 'majority' group).
The final study population consisted of 11,871 individuals: 665 refugees, 477 migrants, 368 descendants and 10,371 in the majority group (Table 1).

Outcomes
The main outcome was an involuntary inpatient admission proximal to the time of the first in-or outpatient contact where a NAPD diagnosis was given (binary: no/yes).Proximal was defined as the period from the first NAPD contact until three month later, as patients were still considered to be in the initial phase of psychosis treatment during these three months.Most of these involuntary admissions coincided with the very first NAPD contact (between 68 %-77 % across the groups).For ease of reference, we refer to this outcome simply as involuntary admission at first diagnosis.The involuntary admissions were further categorized according to the indication for their admission, i.e. the treatment or danger criteria.
For those who experienced an involuntary admission at first diagnosis, three additional outcomes were measured: i) having more than one involuntary admission until the end of the study period in 2018 (binary: no/yes), ii) the number of involuntary admissions (count variable) and iii) the number of days in involuntary care (count variable).

Covariates
Sex, age, family income level, prior substance abuse, living alone, and education were included as covariates.Sex was coded binary as male/female according to register information (assigned at birth), while age was defined as age at first NAPD diagnosis.Family income level was measured at age 18 and was coded as a binary variable defined as equivalized household income below 60 % of the median as the threshold of living in a "low income" household.Substance abuse was defined as any substance-related inpatient or specialist outpatient contact (ICD-10 codes F10-F19) during the three years prior to their first NAPD contact.An individual was defined as living alone if they were living alone on 31 December during the calendar year prior to first NAPD contact.Education was measured as the highest completed education on 1 October of the year of first contact and was coded as less than upper secondary vs. higher.

Statistical analyses
Logistic regression was used to estimate odds ratios (OR) to compare involuntary admissions at diagnosis across the groups.Simple models were estimated adjusting for sex, age and year of first admission, while full models further included living in a low-income household at age 18, prior treatment for substance abuse, living alone, and educational level.This analysis was also conducted separately for involuntary admissions due to the treatment and danger criteria.Finally, to assess the amount of time in involuntary care among those whose initial contact for NAPD was involuntary, we estimated risk ratios (RR) and 95 % confidence intervals of having more than one involuntary admission, the number of involuntary admissions, and the number of days admitted involuntarily, using Poisson regression.Analyses were conducted using Stata/IC v.16.1.

Results
The proportion of individuals who were admitted involuntarily at first NAPD diagnosis varied across the groups, from 11.3 % in the refugee group to 4.1 % in the majority group (Table 1).Between 54 % and 58 % of involuntary admissions were admitted on grounds of the danger indication, while the remaining were admitted based on the treatment indication (Table 2).
In all groups, most of those admitted involuntarily were men (71-76 % across groups) (Table 1).In general, most NAPD cases were men, but this sex difference was larger in the non-majority groups (67-73 %) than in the majority group (59 %).Living in a low income household at age 18 was most common in the refugee group and least common in the majority group (39 % vs. 17 %), as was the proportion living alone (62 % vs 49 %).Among refugees, a higher proportion had received prior treatment for substance abuse among those admitted voluntarily (16 %) than involuntarily (10 %), while the reverse was found for their majority peers (18 % and 11 %, respectively).
In the adjusted models, refugees, migrants and descendants were all more likely than their majority peers to be admitted involuntarily within three months of their first NAPD diagnosis (Table 2).After adjustment, the odds of involuntary admission were 2.69 (95 % CI: 2.05-3.53)times higher among refugees, 2.43 (1.76-3.36)times higher among migrants, and 2.11 (1.44-3.12)times higher among descendants relative to the majority group.This pattern of findings was similar for both admission indications (danger vs. treatment) (Table 2).Among those admitted involuntarily, there were no significant differences between the groups in terms of the risk of experiencing multiple involuntary admissions (RR range = 0.87-1.31), the total number involuntary admissions (RR range = 0.87-0.99)and the total number of days in involuntary care (RR range = 0.77-0.91)(Table 3).For young refugees, the trend was more frequent involuntary treatments (RR = 1.31, 95 % CI = 0.95-1.81)combined with fewer days of involuntary treatment (RR = 0.84, 95 % CI = 0.57-1.27).

Discussion
Our findings show that young refugees, non-refugee migrants and descendants of non-refugee migrants were at greater risk of being hospitalized involuntarily around the time of their first NAPD diagnosis than their majority peers.However, once in care we did not find group differences in the number of subsequent involuntary admissions or days in involuntary care among those initially hospitalized involuntarily.The observed higher odds for involuntary treatment in NAPD patients for the three minority groups align with the excess risk reported in other studies.For refugees, the odds were almost three times higher compared with their majority peers, which is similar to the excess rates of involuntary detention reported for black groups of psychosis patients in early intervention services in the UK (Mann et al., 2014), but higher than the excess risk for refugees in treatment for psychotic disorders previously reported from Denmark (Norredam et al., 2010).The excess risk for migrants and descendants of migrants was also larger than in previous studies from Sweden and Denmark (Norredam et al., 2010;Terhune et al., 2022).Most previous studies explored the risk of involuntary admission over a longer time period, rather than at first admission, and for broader age ranges, which could explain the differences in rates.This underscores the importance of the initial pathway into care in nonmajority groups' excess risk of involuntary care, and is consistent with our finding of no group differences in the number of further involuntary admissions or days in involuntary care among those initially admitted involuntarily.
The Swedish study by Terhune and colleagues (Terhune et al., 2022) did, like the present study, explore differences at first diagnosis.While they reported lower relative differences, they also reported higher rates of involuntary hospitalizations, especially in the majority group (9.5 % of their majority sample, compared with 4.1 % in our study).This could suggest that the higher relative differences found in our study may be driven by differences in the majority group.According to the treatment standard at the time of our study, first-episode psychosis patients aged 18-35 should enter into treatment in early intervention teams that have been established in all Danish regions during the first decade of this century (Nordentoft et al., 2015).Involving cross-professional teams, this two-year treatment service combines assertive community treatment, antipsychotic medication, sessions for families and social skills training.In a separate study (de Montgomery et al., 2023), we compared the age-specific incidence of NAPD in Denmark with Sweden, where early intervention services have not been widely established.Consistent with expectations, we found that youth entered psychosis-related treatment at earlier ages in Denmark.However, we also found that the age distribution at first NAPD treatment differed by group in Denmark, with the three ethnic minority groups being older than their majority peers when entering treatment.One possible interpretation of this finding is that majority youth benefit more from the early intervention program than non-majority groups, which accentuates the relative differences between these groups in the risk of adverse pathways to care.Whether there are differences by migration background in entering and making use of early intervention services in Denmark should be explored in future studies.Rates of NAPD were also high in Denmark compared ‡ A small number of cases were admitted by both indications and accordingly the sum of N by indication may exceed the N in the total column.with incidence rates reported internationally, which could suggest a lower threshold for entering NAPD treatment and consequently lower rates of involuntary admissions.This would align well with the central aim of early intervention services to increase detection and to intervene during an early stage of disease development (McGorry et al., 2008).If majority youth have easier access to early intervention services and receive treatment with milder symptoms than minority youth, the disease severity would be higher among minority NAPD patients at the initial point of contact.This could contribute to the higher risk of involuntary admissions observed in minority groups.
Compared with other international evidence, the rates of involuntary admissions were in fact low in all groups.In two cohorts of patients presenting with first-episode psychosis at hospitals in the London boroughs of Lambeth and Southwark, overall proportions of patients involuntarily hospitalized when first presenting were 41 % and 24 % (Oduola et al., 2019).A study from Canada found that 26 % of patients with incident psychosis in Ontario who were hospitalized involuntarily within two years of first presentation.While design and population differences may account for some of the difference between these studies and our study, the large difference could suggest that in international comparison rates of involuntary hospital admissions are relatively low in Denmark among first-episode psychosis patients, at least in patients belonging to the majority population.
Several mechanisms have been suggested to explain ethnic variation in involuntary admissions and use of coercive measures, including a greater prevalence of psychotic disorders in certain minority groups, language barriers, absence of family members, and ethnic disadvantages more generally (Barnett et al., 2019).As we only considered the risk of involuntary admissions among patients with a diagnosis of NAPD, the difference in prevalence of psychosis was taken out of the equation.We also found that factors related to socio-economic position, including education and household income, did little to attenuate group differences.While we did observe the highest odds of involuntary admissions in refugees and other migrants (OR range = 2.4-2.6), the odds in nativeborn descendants of labor migrants were not much lower (OR = 2.21, 95 % CI 1.44-3.12).This suggests that language barriers were unlikely to be the most important factor to explain the differences.Other factors that have been suggested include experiences of discrimination and difficulties in establishing productive therapeutic relationships, negative expectations or cultural stigma towards psychiatric treatment, as well as more limited possibilities for family support in navigating the healthcare system (Barnett et al., 2019).
Meanwhile, we did not observe group differences among those hospitalized involuntarily in further involuntary hospitalizations or amount of time hospitalized involuntarily.This could indicate that the mechanisms that lead to ethnic differences in involuntary admissions primarily relate to the initial process of entering treatment, and less to processes once in treatment.Unpacking these mechanisms is an important topic for future research.
Giving explicit priority to the mental health care of ethnic minorities and migrants in national and local health policies will be important to ensure that sufficient attention is given to this issue (McKenzie, 2008).As the populations served by health services are becoming increasingly diverse, the urgency of such a policy commitment will not likely abate.While there is no standard way of translating this policy intention into practice, moving forward will require both engagement of communities and specific assessment of cross-cultural needs in each service encounter (Healey et al., 2017).Cultural competence training of medical staff and other care professionals has for many years been a focus point (Bhui et al., 2007), and has been highlighted by a Guidance document by the European Psychiatric Association (Schouler-Ocak et al., 2015) as crucial to efforts to develop culturally sensitive services.However, the implementation of such training remains insufficient in medical programs across Europe (Sorensen et al., 2019).

Strengths and limitations
A strength of this study is that the data material included all NAPD cases from 2006 to 2018 in each of the comparison groups.This allowed for stringent definitions of groups and incident cases.Through high quality registry data, reliable information on NAPD diagnosis and mode of admission, collected routinely for administrative purposes thus minimizing the risk of recall bias, could be combined with extensive socio-demographic and economic data.A limitation of the study is that we did not have information on whether a given treatment for NAPD coincides with the onset of symptoms, nor whether patients had received prior treatment in primary care.Information on healthcare use abroad was not available, so the study relied on a three year residency restriction to ensure that the three-year washout period was applied uniformly across groups.Given the administrative nature of the data material, we were unable to adjust for confounding factors that were either not recorded or not available in the registers.Factors such as family conflicts, housing difficulties and substance abuse that did not result in service contact may well differ between groups and influence care pathways.After applying sample restrictions, the population of especially non-majority patients involuntarily admitted was too small to allow further disaggregation according to countries or regions of origin or sex.Some misclassification of refugees and other migrants was possible, as data on grounds of residence was only available from 1993 and even then was partially imputed by Statistics Denmark due to missing values.However, as refugee immigration was very dominated by particular refugee-sending countries during this period, while nonrefugee immigrants from these countries were uncommon, it is unlikely that this misclassification would be of a magnitude to alter the patterns reported.

Conclusions
In this study, we provide further evidence that refugees as well as non-refugee migrants and descendants of non-refugee migrants are at higher risk of experiencing involuntary admissions around the time of their first NAPD diagnosis.This excess risk persisted even after adjusting for socio-economic factors and previous treatment for drug-related disorders.Beyond the first involuntary hospitalization, we did not observe group differences in further involuntary hospitalizations or days in involuntary care.This suggests that the initial pathway into care is particularly problematic for some minority groups.Factors related to both patients and providers could explain these differences.In particular, group differences in entry into early intervention programs could entail greater risk of delayed treatment in non-majority groups.Differences between refugees, non-refugees and descendants of non-refugee migrants were negligible, which suggests that the most important factors may be common across these three groups and thus play out within the country of residence rather than prior to arrival.

Role of funding agency
This work was supported by the Swedish Research Council (Grant number 2018-05783).The funding agency had no involvement in the design, analysis and interpretation of data, writing of the manuscript or the decision to submit the article for publication.

Ethics approval
Data processing was approved by the Faculty of Medical Sciences at the University of Copenhagen (case number SUND-2016-65).

Table 1
Characteristics of the study population by group † .Divided by involuntary admissions within three months of first NAPD diagnosis and other.
† Refugees and non-refugee migrants settled in Denmark before age 18, descendants of primarily labor migrants from Turkey, Pakistan and Morocco, and Majority youth born in Denmark to parents born in Denmark.*Not reported due to risks of back identification in cell sizes below 5.C.J. de Montgomery et al.

Table 2
Odds ratios (OR) and 95 % confidence intervals of involuntary admissions during the first three months after first NAPD diagnosis in each group † .Total and by treatment/danger indication ‡ .Logistic regression.All models are adjusted for age, sex and cohort entry year, full model further adjusted for low-income household, prior treatment for substance abuse, living alone, and educational level.†Refugees and non-refugee migrants settled in Denmark before age 18, descendants of primarily labor migrants from Turkey, Pakistan and Morocco, and Majority youth born in Denmark to parents born in Denmark.

Table 3
Relative Risk (RR) and 95 % confidence interval (CI) of multiple involuntary admissions (multiple), total number of involuntary admissions (total), and total days in involuntary care (days) in each group † among those admitted involuntarily at first diagnosis.Poisson regression.
Notes: Risk Ratios were adjusted for age, sex and year of first NAPD diagnosis.†Refugeesand non-refugee migrants settled in Denmark before age 18, descendants of primarily labor migrants from Turkey, Pakistan and Morocco, and Majority youth born in Denmark to parents born in Denmark.C.J. de Montgomery et al.