Primary prevention of lead poisoning in children: a cross-sectional study to evaluate state specific lead-based paint risk reduction laws in preventing lead poisoning in children

Table 4 Multiple logistic regression showing unadjusted and adjusted estimates of slope parameters and the odds ratio of observing addresses with subsequent cases after identification of an index case, for lead law states (Massachusetts and Ohio) versus control states (Mississippi)

Covariate	# Addresses¹	Effect Estimates
		Lead law states (MA and OH) vs. control state (MS)			Odds ratio of lead law states (MA/OH) vs. control state (MS) (95% CI)²
		Estimate	Std. error	p-value*
Address with subsequent case(s) (Unadjusted main effects model)	292	-0.5546	0.2749	0.0434	0.57 (0.34-0.98)
		Adjusted estimates
Address with subsequent case(s) (Adjusted e stepwise regression main effects association with all variables controlled for in the model)^3,4	115	-1.5626	0.4806	0.001	0.21 (0.08-0.54)
Covariate
Year building built (pre-1950 vs. newer)	150	-1.3864	0.4090	0.001
Building type (Single family vs. Multi-unit)	182	-0.7102	0.3642	0.051
Building ownership (Private, owner-occupied vs. other)	184	-0.8670	0.3440	0.012
Floor Dust-Lead Loading (mean)	191	-0.6696	0.3283	0.041
Sill Dust-Lead Loading (mean)	171	-0.9873	0.3469	0.004
Median Household Income in County (median)	292	-0.6559	0.3945	0.096
Mean Household Size in County (mean)	292	-0.8508	0.3485	0.015
Poverty in County (%)	292	-0.4180	0.4044	0.301
CAPI in County (%)	292	-0.5805	0.2988	0.052
Households in County with High School Graduates (%)	292	-0.7308	0.4004	0.068
Non-whites in County (%)	292	-0.5961	0.4144	0.150
Pre-1950 homes in County (%)	292	-0.3011	0.5540	0.587
Rentals in County (%)	292	-0.5503	0.2758	0.046

¹#addresses represent 292 distinct addresses that had sufficient blood lead data for assessing the potential for subsequent cases following the index case. For the adjusted estimates, the n next to each covariate represents the number of addresses for which that information was available. For the final adjusted model, the N represents the addresses that had both main effects and covariate information available.
²Odds ratios are calculated as the exponential of the corresponding slope parameter estimates in this table. Lead law state = 1 and control state = 0.
³Results presented shows the final stepwise model adjusting for all covariates listed; the main effects variable, address with subsequent cases, continued to be the best fit to the data. Intercept and county indicator parameters were forced into the stepwise model.
⁴The Hosmer-Lemeshow chi-square value of the goodness-of-fit test was 4.4971, p = 0.4803. Model is a good fit to the data, given p > 0.05.
*P-value ≤0.05 implies statistical significance.

ISSN: 1476-069X