Skip to main content

Table 4 Estimates of random coefficients in terms of variance and covariance of time trend parameters among care providers

From: Trends in the use of coercive measures in Finnish psychiatric hospitals: a register analysis of the past two decades

Random coefficientsAny coercive measure
Est (SE)
Seclusion
Est (SE)
Limb restraints
Est (SE)
Forced injection
Est (SE)
Physical restraints
Est (SE)
Var: Intercepts1.14(.186)1.45(.238)1.68(.281)1.31(.227) 0.931(.176)
Cov:Year/Intercept−.018(.011)−.032(.012).0087(.019)−.0071(.013)−.014(.015)
Var: Year.0057(.0011).0056(.0012)*.129(.025).0058(.0013).010(.002)
Cov: Year^2/Intercept.0018(.0015).024(.0018) −.0055(.0020)*.0063(.0019)
Cov: Year^2/Year−.0006(.0001)−.0004(.0001)* −.0004(.00015).00017(.00019)
Var: Year^2.00012(.00002).00013(.00003) .00013(.00003).00011(.00003)
χ2 statistic (p value)51.00 (0.000)47.86 (0.000)27.10 (0.000)40.33 (0.000)33.68 (0.000)
  1. The model with random slopes of quadratic terms did not converge for limb restraints; hence, only a random linear slope model was fitted on this method. The χ2 statistic was based on a generalised Ward test for joint parameter estimates using the MLwiN package. ‡p < 0.001, †p < 0.01, *p < 0.05