Implementation of evidence-informed practice through central network actors; a case study of three public health units in Canada

Table 3 Regression model to predict the EBP behavior scores over time based on their friendship patterns in the subgroup who sought information from highly engaged central staff

		Regression coefficients
Respondents	Friends with someone who is:
Baseline
Not highly engaged	Not highly engaged	9.9(1.2)
	Highly engaged, but not central	−3.5(2.1)
	Highly engaged central	−0.8(1.7)
Highly engaged, but not central	Not highly engaged	0.1(3.4)
	Highly engaged, but not central	−3.3(2.4)
	Highly engaged central	−1.8(2.7)
Highly engaged central	Not highly engaged	6.1(3.2)
	Highly engaged, but not central	0.4(4.3)
	Highly engaged central	5.6(2.4)*
Change from baseline
Not highly engaged	Not highly engaged	−0.7(1.1)
	Highly engaged, but not central	2.3(1.8)
	Highly engaged central	3.3(1.4)*
Highly engaged, but not central	Not highly engaged	−2.0(3.2)
	Highly engaged, but not central	2.6(2.3)
	Highly engaged central	5.7(2.3)*
Highly engaged central	Not highly engaged	8.1(3.2)*
	Highly engaged, but not central	3.0(3.8)
	Highly engaged central	4.8(2.0)*
Random effects variance
Health Units		0.6(1.5)
Individuals		29.2(5.4)
Residual		21.8(3.0)

1: The coefficients for all levels at baseline represent the difference from the first row
2: The coefficients for all levels at ‘change from the baseline’ represent the difference from their baseline
*:p < 0.05