Application of a nonrandomized stepped wedge design to evaluate an evidence-based quality improvement intervention: a proof of concept using simulated data on patient-centered medical homes

Table 1 Results for the three-level ordinal model with timing of implementation, linear time trends, respondent-level, and site-level characteristics

Parameter	Model 1	Model 2	Model 3	Model 4
	Coeff (SE)/OR	Coeff (SE)/OR	Coeff (SE)/OR	Coeff (SE)/OR
Intercept, γ ₀₀₀	−1.22 (0.04)/0.29**	−0.821 (0.074)/0.440**	−0.874 (0.144)/0.417**	−1.235 (0.063)/0.291**
Threshold 2, δ ₂	1.294 (0.006)/3.647**	1.294 (0.006)/3.647**	1.294 (0.006)/3.647**	1.294 (0.006)/3.647**
Threshold 3, δ ₃	2.345 (0.009)/10.437**	2.345 (0.009)/10.437**	2.345 (0.009)/10.437**	2.345 (0.009)/10.437**
Time-level predictors
QTR, γ ₀₁₀	0.005 (0.005)/1.005			0.005 (0.005)/1.005
TREAT, γ ₀₂₀	0.265 (0.048)/1.303**			0.262 (0.048)/1.300**
QTR^aTREAT, γ ₀₃₀	0.049 (0.007)/1.051**			0.049 (0.007)/1.051**
Respondent-level predictors
Gender (ref: MALE)
FEMALE, γ ₁₀₀		−0.007 (0.018)/0.993		−0.007 (0.018)/0.993
Race/ethnicity (ref: WHITE)
NONWHITE, γ ₂₀₀		−0.013 (0.013)/0.987		−0.013 (0.013)/0.988
Site-level predictors
Site Size (ref: SMALL)
MED, γ ₀₀₁			−0.085 (0.204)/0.918	−0.040 (0.083)/0.961
LG, γ ₀₀₂			0.122 (0.171)/1.130	0.049 (0.069)/1.050

ISSN: 1471-2288