Table 3.

Logistic regression model showing unadjusted and adjusted estimates of recurrent lead poisoning case in subsequent years following the cohort year, with comparisons made between Massachusetts and Mississippi and Ohio and Mississippi.

Covariate	No. of cases1	Slope parameters
		MA vs. MS			OH vs. MS			Odds ratio of lead law states vs. control state2
		Estimate	Std. error	p-Value	Estimate	Std. error	p-Value
Case in subsequent year (unadjusted)	429	−1.3188	0.2529	<0.0001	−0.3621	0.2683	0.1771	OR (MA vs. MS): 0.267 OR (OH vs. MS): 0.696
		Adjusted estimates
Case in subsequent year (adjusted full model)3	147	−0.5070	0.4441	0.2535	−0.4997	0.5404	0.3552	OR (MA vs. MS): 0.602 OR (OH vs. MS): 0.607
Gender (female vs. male)	429	−1.3294	0.2536	<0.0001	−0.3768	0.2692	0.1615
Race (others vs. African-American)	346	−1.0210	0.3583	0.0044	−0.1346	0.3398	0.6920
Age at confirmation (others vs. <12 months)	427	−1.3279	0.2540	<0.0001	−0.3489	0.2706	0.1973
Building ownership (others vs. private owner-occupied)	372	−1.4694	0.2857	<0.0001	−0.3356	0.2846	0.2383
Type of provider ordering Test (others vs. private health care)	353	−1.4145	0.3226	<0.0001	−0.2564	0.4710	0.5862
Presence of deteriorated paint (absence vs. presence)	286	−1.2736	0.3084	<0.0001	−0.4157	0.3364	0.2166
Presence of visible paint chips (absence vs. presence)	225	−0.7585	0.4105	0.0646	−0.2987	0.3419	0.3822
Mean floor dust–lead loading	389	−1.3086	0.2695	<0.0001	−0.3650	0.2745	0.1837
Mean sill dust–lead loading	356	−1.4824	0.2862	<0.0001	−0.5449	0.2843	0.0553

¹No. of cases represent 425 distinct children counted as a recurrent lead poisoning—four children were selected as a new case in multiple years, and thus, may have different values for the adjustment factors.

²Odds ratios are calculated as the exponential of the parameter estimates in this table.

³The Hosmer–Lemeshow test chi-square value for the goodness of fit test was 7.7417, p = 0.4591. Model is a good fit to the data, given p > 0.05.