Do Low Rates of Coercion in Catchment Areas Predict Negative Effects for Persons With Severe Mental Disorders? A Register Study From Norway 2015-2017. Part of Reducing Coercion in Norway (ReCoN)
Do Low Rates of Coercion in Catchment Areas Predict Negative Effects for Persons With Severe Mental Disorders? A Register Study From Norway 2015-2017. Part of Reducing Coercion in Norway (ReCoN)
Involuntary mental health care is permitted because it is believed to make people with severe mental disorders (SMD) better and prevent them from getting worse or even dying In this study we will investigate whether low levels of coercion in an area is connected with poorer outcomes in Norway. It can be assumed that too little involuntary care might lead to the opposite outcomes to those intended by the Norwegian Mental Health Act.
The same law applies all over Norway, but the rate of involuntary care varies: there is up to five-fold difference between the catchment areas of the 69 Community Mental Health Centers. The investigators will estimate rates of involuntary care and adjust for age, sex, urbanity and area deprivation. The data source is the Norwegian Patients Registry, and all patients in treatment for a severe mental disorder in 2015 and their use of mental health care until 2018 will be followed.
Model 1 follows all patients who were treated for a severe mental disorder in 2015. The model will test whether the rates of involuntary care in the area they live can predict the length of time to death.
Model 2 follows patients with treatment for severe mental disorders that had no episode of involuntary care in 2015. The model will test whether the rate of involuntary care in their area predicts their use of mental health inpatient care in 2016 and 2017.
Model 3 tests how long time patients with severe mental disorders that received only voluntary care in 2015 remain without a period of involuntary care in 2016-17, as a function of the rate of involuntary care in their area in 2015.
Model 4 estimates changes in the total number of patients with severe mental disorders in the catchment area in 2016-17 as a function of time and the rate of involuntary in 2015.
Model 5 tests whether suicide rates for a catchment area varies as a function of its rate of involuntary care. Because suicides are rare, we will observe the variables over longer time periods, using involuntary care rates from 2015 to 2018 and suicide rates for 2015-2019.
The study was evaluated by the Research Ethics Committee (ref 2018/795), who approved use of registry data, and by the Privacy Ombudsman at Akershus University Hospital (ref 2018-090).
Involuntary mental health care is permitted because it is believed to make people with severe mental disorders (SMD) better and prevent them from getting worse or even dying. This study concerns whether low levels of coercion in an area is connected with poorer outcomes in Norway. Too little involuntary care could be expected to lead to the opposite outcomes to those intended by the Norwegian Mental Health Act.
The same law applies all over Norway, but the rate of involuntary care varies: there is an up to five-fold difference between the catchment areas of the 69 Community Mental Health Centers (CMHC).
This study will use national register data to test whether areas with low rates of involuntary care shows signs of failing to achieve the patient benefits of involuntary care as intended by the Norwegian Mental Health Act.
Data on involuntary care will be retrieved from the Norwegian Patient Registry (NPR) and combined with data on the general population and demographics from Statistics Norway, to study the hypothesized negative consequences of low rates of involuntary care. All of Norway's 21 Health Trusts and their 69 Community Mental Health Centers that provide specialist services are required to submit to NPR on an annual basis complete data of all service use from their hospitals' electronic patient administrative and clinical systems. The NPR conducts extensive data-quality checks, and publishes completeness data for all variables. The registry implemented a unique patient identifier in 2008, so that a patient's care can be followed across time. From the NPR database, information on all specialist mental health treatment activity for patients with set criteria (in our case diagnosis F20-31 and legal status) during a given time period (here 2015-18) can be extracted.
Both involuntary admissions and outpatient compulsion in the form of community treatment orders (CTO) are envisaged to contribute to the aims of the Mental Health Act, such as protection against harm, improvement, restoration of health, and recovery or cure for the patient. The variable of interest is low use of the combination of these two forms of involuntary care. To our knowledge, there is no established way to calculate a combined measure of inpatient and outpatient involuntary care, and it is not generally established how one form of involuntary care is associated with the other. In Norway, 31% of involuntary admissions continued as a CTO in 2018, and although permitted, CTOs are almost never initiated when the patient lives at home. When a patient is under involuntary admissions or a CTO, the care system has substantial influence and control over their treatment, and medication is a particular focus. Regardless of the form of involuntary care, control over treatment can be continued according to necessity criteria in the law in order to prevent deterioration. For these reasons, the rate of persons affected by involuntary admission and/or CTO per 100 000 capita will be used as an indicator of use of involuntary care in the catchment area and hence, the main covariate.
The age span will be patients from 18 to 65, so that the increasing frequency of involuntary care due to lack of capacity due to dementia towards the end of life will not impact the results. The investigators will control for age and sex by indirect standardization, based on Statistics Norway's tables for population in municipalities and city districts.
The investigators will test whether the living condition index and/or urbanity of the municipality (perhaps merged to fewer categories than five) predict rates of involuntary care. A classification of all municipalities into five degrees of urbanity from a previous study will be used. Statistics Norway published an index of living conditions in Norwegian municipalities and City districts in 2000 and 2008, which has not been continued after 2008. This index combines the level or frequency of a variety of social benefit payments, unemployment, education level and lethality, and should be sufficiently up to date. If urbanity or living condition predicts rates of involuntary care, it will be controlled for by estimating a suitable linear regression model. A hierarchical model adjusting for nesting of municipalities within CMHCs and CMHCs within health trusts will be considered, but this adjustment may be reduced based on intraclass correlations. Then the ratio between expected and observed use of involuntary care will be the main covariate, and will reflect the rate of involuntary care per capita, controlled for age, sex and urbanity and deprivation in each catchment area. Each patient in the dataset will be assigned the value of this covariate corresponding to their area of residence in 2015.
The first group of models is assessing associations between the main covariate measured in the index year and survival or change in outcome variables in the following years for the patients included.
Model 1. Are there more deaths among SMD patients from areas with low levels of involuntary care?
The main covariate is rate of involuntary care in 69 areas in 2015, controlled for age, sex, urbanity and deprivation. The unit of analysis is the individual patient with a care episode and a registered SMD diagnosis in 2015. Outcome variable is time to death from the quarter in 2015 with a registered event with a SMD-diagnosis, observed through 2018. The unit of analysis is the individual SMD patient. The statistical method is survival analysis with adjustment for age and sex and adjustment for random effects.
Model 2. Is there increased use of inpatient care for voluntary SMD-patients from areas with low levels of involuntary care?
The main covariate is rate of involuntary mental health care in 69 areas in 2015, controlled for age, sex, urbanity and deprivation. The unit of analyses is the individual SMD-patient with 'voluntary care only' in 2015. The outcome variable is change in mental health inpatient days in the next two years for the included patients. The statistical method is a linear mixed model with random effects for CMHC.
Model 3. Is there decreased time to the next episode of involuntary care for voluntary SMD-patients from areas with low levels of involuntary care? The main covariate is rate of involuntary mental health care in the 69 areas in 2015, controlled for age, sex, urbanity and deprivation. The unit of analyses is the individual SMD-patient with 'voluntary care only' in 2015. Outcome variable are time to a treatment episode with involuntary care observed through 2016 and 2017. The statistical method is survival analysis with death as competing risk with random effects for catchment area.
Model 4 and 5 are ecological models of the catchment areas.
Model 4. Are low levels of involuntary care in a catchment area followed by more SMD patients in the area? The main covariate is rate of involuntary mental health care in the 69 areas in 2015, controlled for age, sex, urbanity and deprivation. The outcome variable is the change in annual number of patients with SMD in 2016 and 2017. The unit of analysis is the catchment area (N=69). The statistical method is linear regression model with time, main covariate and interaction between the two as independent variables.
Model 5. Is low level of involuntary care in an area related to more suicides?
The main covariate is rate of involuntary care in 2015-2018 in the 69 areas, controlled for age, sex, urbanity and deprivation. The outcome variable is the standardized rate of suicides in the area's population regardless of patient status, from 2015 through 2019. The unit of analysis is the 69 catchment areas. Due to low incidence of the outcome, merging of some areas will be considered. The statistical method is correlation analysis.
Dissemination: Study results will be published study international peer-reviewed scientific journals. The number of papers necessary to disseminate results depends of the complexity of results and their interpretation.
Inclusion Criteria:
Exclusion Criteria: