Page 111 - Reduction of coercive measures

Multidisciplinary reduction of coercive measures
restrictive rules on the use of a mobile phone or the Internet or limited opportunities to receive visits. These measures are applied to prevent a resident from ending up in a dangerous situation or suffering serious harm in the (near) future, for example as a result of health risks, or social decline. The third subset contains 6 coercive measures resulting from the use of surveillance technology. The fourth and final subset includes 10 coercive measures resulting from the of use of ergonomic material to physically support the resident. An overview of coercive measures and what subset they belong to is displayed in Appendix A. Registrations were updated regularly by direct care staff and permanent unit consultants. Researchers sent regular prompts for updates to take place. Registrations were double checked by the researchers against case files and treatment plans and corrected if necessary.
Statistical analysis
The dataset was structured to contain for each coercive measure per resident, per unit, a variable that indicated whether or not the coercive measure terminated during the intervention period (1 = stopped; 0 = not stopped). Hence, the dataset had a hierarchical structure, with coercive measures (level 1) nested within residents (level 2) who were nested within residential units (level 3). This strategy was chosen to accommodate turnover of clients within care units, but also aligned with the goal of the multidisciplinary expert team to reduce coercive measures, irrespective of which clients were affected. The effect of the program on reduction of restraint use was tested using generalized linear mixed modeling in SPSS version 23. Mixed modeling is a suitable technique for data with a multilevel structure, and correctly takes into account the dependencies of observations coming from the same clusters (in this case, coercive measures applied to the same resident, and residents residing in the same unit) (Hox, Moerbeek, & Van der Schoot, 2017). Given the dichotomous outcome variable, the binary logistic regression model (with logit link function) was used as the specific type of generalized linear mixed model to test the effect of the program on reduction of coercive measures. The multilevel analyses were conducted in four steps. First, an ‘empty’ model specified
109
5