TABLE 4—

Results of Marginalized Zero-Inflated Negative Binomial Model Assessing the Effects of Early Head Start (EHS) on Mean Increments in Dental Care Visits: Zero Out Early Childhood Caries Study, North Carolina

Variable	Parameter Estimate	Model-Based SE	Empirical SE	OR (95% CI)	RR (95% CI)
Probability of having an excess dental care visit
EHS	−1.23**	0.22	0.25	0.29 (0.19, 0.46)
Propensity score	0.03	0.56	0.56	1.03 (0.32, 3.29)
Needed dental care at baseline	−4.98	7.78	11.45	0.01 (0.00, 66 529.45)
Had a dental care visit at baseline	−1.20*	0.45	0.66	0.30 (0.12, 0.77)
Constant	−0.59**	0.20	0.26	0.55 (0.37, 0.84)
Overall mean number of dental care visits
EHS	0.30**	0.068	0.063		1.35 (1.17, 1.55)
Propensity score	0.30	0.160	0.160		1.35 (0.98, 1.87)
Needed dental care at baseline	0.53**	0.040	0.061		1.69 (1.56, 1.84)
Had a dental care visit at baseline	0.42**	0.089	0.090		1.52 (1.26, 1.83)
Constant	0.31**	0.092	0.090		1.36 (1.13, 1.65)
Random effects variance components
Standard deviation of excess zeros intercept	0.69**	0.13	0.014
Standard deviation of mean model intercept	0.28**	0.061	0.054
Correlation of random intercepts	−0.94**	0.078	0.061
Overdispersion parameter	0.04	0.019	0.023

Note. CI = confidence interval; OR = odds ratio; RR = rate ratio. ORs, RRs, and CIs are based on model empirical standard errors. Models included random effects for each of the 25 EHS program clusters (see Appendix B, available as a supplement to the online version of this article at http://www.ajph.org). Baseline data were collected from 2010–2012; 24-month follow-up data were collected from 2012–2014. The sample size was n = 1178.

*P < .05; **P < .01.