Table 3.

Correlates of Vaccination Status in Multivariate Hierarchical Linear Modeling

Variable, Fixed Effect	Odds Ratio	95% Confidence Interval	PValue	Interpretation: Incremental Impact of Variable^a
^a The logits were calculated for various scenarios from the HLM equations in Supplemental Appendix 2 (available online at http://www.annfammed.org/cgi/content/full/7/6/534/DC1) and then estimated vaccination rates were calculated using the formula: rate = 1/(1+ exponent [−logit]). A spreadsheet was used to calculate incremental impact.
Pneumococcal polysaccharide vaccine
Intercept (γ₀₀)	0.03	0.01–0.17	<.001
Patient-level factors
Older age, years (γ₁₀)	1.03	1.01–1.06	.019	0.7% increase in rates for each year older
White race, ref = minority (γ₂₀)	1.70	1.25–2.32	.001	12% increase in rates for whites vs minorities
Physician- and practice-level factors
Physician reported time for well visit, minutes (γ₀₁)	1.04	1.01–1.07	.015	0.9% increase in rates for each additional minute
Enhanced immunization documentation^† (γ₀₂)	1.53	0.91–2.54	.102	9%–10% increase in rates for enhanced documentation (eg, electronic medical record or flow sheet)
Influenza
Intercept (γ₀₀)	0.04	0.01–0.19	<.001
Patient-level factors
Older age, years (γ₁₀)	1.03	1.01–1.05	.003	0.7% increase in rates for each year older
White race, reference = minority (γ₂₀)	1.60	1.23–2.10	.001	11% increase in rates for whites vs minorities
Physician- and practice-level factors
Practice uses standing orders (γ₀₁)	2.12	1.57–2.87	<.001	17%–19% increase in rates for standing orders
Average observed physician time in examination room, minutes (γ₀₂)	1.04	1.01–1.07	.016	1% increase in rates for each additional minute