Erratum: "Low-Level Environmental Lead Exposure and Children's Intellectual Function: An International Pooled Analysis".

The source of the errors in this article, which were limited to the Boston cohort, was twofold. Specifically, performance IQ data were used instead of full-scale IQ data for children in the Boston cohort. In addition, when converting log-transformed 6-, 18-, and 24-month blood lead measurements, a value of 1 was added instead of subtracted, making the blood lead values in the Boston cohort higher than they should have been. Consequently, there were small errors in the peak, early-childhood, and lifetime lead values in the Boston cohort. In turn, these errors resulted in small errors in the associations between blood lead levels and full-scale IQ for the Boston cohort and, thus, the pooled analysis as a whole. Accordingly, we reanalyzed the data using full scale IQ data and corrected blood lead levels for the Boston cohort and provide the corrected data and results in the revised text.

There also were small changes in the model values used as the a priori criterion for identifying the blood lead metric that was the strongest predictor of IQ. We originally relied on the to select the blood lead metric for the primary analysis. However, the for concurrent () and early childhood () blood lead levels were very similar ( and 0.643, respectively) in the revised analysis. In both the original and revised analyses, the coefficient for concurrent blood lead levels had a smaller standard error and higher level of significance than early childhood blood lead levels. In the revised analysis, the standard error for concurrent blood lead and IQ was 0.530 and the level of significance was . In contrast, the standard error for early childhood blood lead and IQ was 0.597 and the level of significance was . We therefore retained concurrent blood lead level as the metric of choice and clarified this in the revised text.

Importantly, the U.S. EPA and other investigators concluded that the data errors had only minor effects on the estimates and did not change the overall results or alter the scientific conclusions about the effects of low blood lead concentrations on cognitive function in young children (Crump et al. 2013; U.S. EPA 2013; Kirrane and Patel 2014).

The authors regret the errors.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Table 1

Characteristics of the children and of their mothers in the pooled analysis ().

Characteristic	Value
Child characteristics
Femalea	669 (50.2)
Birth weightb (g)	3,286±503
Gestation at deliveryb (weeks)	39.6±1.9
Birth orderc	2.0 (1–5)
Blood lead concentration (μg/dL)c
Concurrent	9.7 (2.5–33.2)
Peak	18.0 (6.0–47.0)
Early childhood	12.7 (3.5–34.5)
Lifetime average	11.9 (3.6–34.5)
Peak blood lead concentration <10μg/dLa	258 (19.4)
Peak blood lead concentration <7.5μg/dLa	118 (8.9)
IQb	93.4±19.6
Age at IQ testingb (years)	6.9±1.2
Maternal characteristics
Age at deliveryb (years)	25.4±5.4
Maternal IQb	88.2±18.5
Education at deliveryb (grade)	11.1±2.8
HOME scoreb	37.0±8.4
Marrieda	896 (67.3)
Smoked during pregnancya	453 (34.1)
Alcohol use during pregnancya	278 (21.2)

HOME Score was standardized to preschool test. Early childhood blood lead concentration was defined as the mean of 6- to 24-month blood lead tests. Lifetime average blood lead concentration was defined as the mean of blood lead tests taken from 6 months through the concurrent blood lead test.

No. (%).

.

Median (5th and 95th percentiles).

Table 2

Characteristics of 1,333 children and their mothers in seven cohort studies of environmental lead exposure and IQ.

Characteristic	Boston (n=116)	Cincinnati (n=221)	Cleveland (n=160)	Mexico (n=99)	Port Pirie (n=324)	Rochester (n=182)	Yugoslavia (n=231)
Percent femalea	60 (51.7)	108 (48.9)	73 (45.6)	50 (50.5)	174 (53.7)	89 (48.9)	115 (49.8)
Birth weightb (g)	3,412±510	3,144±457	3,199±498	3,254±432	3,393±502	3,226±506	3,328±526
Gestation at deliveryb (weeks)	40.0±1.8	39.6±1.7	39.6±1.2	40.2±1.1	39.9±1.7	39.1±1.8	39.3±2.9
Birth orderb	1.6±1.0	2.6±1.4	2.2±1.1	1.8±0.9	2.0±1.1	2.4±1.4	2.6±1.7
IQ test	WISC-R	WISC-R	WPPSI	WISC-S	WISC-R	WPPSI	WISC-III
IQ scoreb	119.3±14.9	87.0±11.4	86.7±16.2	107.8±11.0	106.0±13.7	84.9±14.4	74.2±13.3
Age at IQ testing (years)	10	7	4.8	7	7	6	7
Blood lead concentrationsc
Concurrent blood lead	5.4 (0.8–12.7)	7.5 (3.5–20.0)	14.2 (7.0–28.5)	7.0 (3.0–16.5)	13.0 (6.0–24.0)	4.0 (1.5–12.0)	15.9 (4.7–47.8)
Peak blood lead	10.4 (4.1–25.0)	17.9 (9.0–38.0)	18.0 (9.0–34.0)	15.0 (6.0–40.0)	27.0 (15.0–46.0)	9.0 (3.5–23.3)	23.8 (7.6–61.5)
Early childhood	6.1 (1.3–16.0)	12.0 (6.6–26.6)	13.4 (7.9–24.8)	11.4 (4.3–26.8)	20.5 (11.0–33.3)	5.8 (2.4–13.1)	14.1 (4.3–44.0)
Lifetime mean	5.6 (2.2–13.0)	11.7 (5.8–24.9)	14.5 (8.1–25.3)	10.6 (4.5–21.3)	18.6 (10.8–30.2)	5.5 (2.4–12.8)	15.8 (5.6–49.3)
Peak blood lead <10μg/dLa	55 (47.4)	23 (10.4)	11 (6.9)	20 (20.2)	0 (0.0)	103 (56.6)	46 (19.9)
Peak blood lead <7.5μg/dLa	28 (24.1)	1 (0.4)	1 (0.6)	8 (8.1)	0 (0.0)	69 (37.9)	11 (4.8)
Maternal characteristics
Age at delivery (years)b	30.5±4.2	22.7±4.3	22.2±3.8	27.1±5.9	26.0±4.2	24.8±6.6	26.6±5.1
Race (nonwhite)a	5 (4.3)	197 (89.1)	69 (43.1)	NA	NA	130 (71.4)	NA
Maternal IQb	124.2±16.2	75.2±9.4	73.4±13.2	93.4±11.9	94.4±11.0	81.1±12.6	87.3±14.8
Education at delivery (grade)b	15.2±2.0	11.2±1.4	10.6±1.6	11.4±3.5	10.6±1.0	12.2±2.0	8.8±3.9
HOME scoreb	50.5±3.5	32.7±6.2	38.1±6.7	36.8±6.7	42.3±4.6	31.9±6.3	30.4±6.8
Marrieda	107 (92.2)	30 (13.6)	82 (51.2)	88 (88.9)	298 (92.0)	60 (33.2)	231 (100)
Tobacco use during pregnancya	29 (25.0)	111 (50.2)	128 (80.0)	6 (6.1)	79 (24.6)	41 (22.6)	59 (25.5)
Alcohol use during pregnancya	61 (52.6)	31 (14.0)	75 (46.9)	6 (6.1)	82 (25.3)	9 (5.5)	14 (6.1)

NA, Not applicable. HOME score was standardized to preschool scale. Concurrent blood lead tests taken at 5 years of age were used as the concurrent blood lead test for the Boston cohort and the IQ test was done at 10 years. Test scores of children in the Yugoslavia cohort are low because of adjustments in adapting tests where no standardization existed; rather than deriving appropriate analogues, some culturally driven items were removed, resulting in lower scores.

No. (%).

.

Median (5th and 95th percentiles).

Table 3

Concurrent blood lead concentration and mean IQ scores by characteristics of children and their mothers ().

Covariate	No.	Median concurrent blood lead (μg/dL) (5th–95th percentiles)	IQ±SD
Child
Female	669	9.0 (2.4–31.4)	94.2±18.6
Male	664	9.9 (2.6–35.7)	92.7±20.4
Birth weight (g)
<3,000	359	10.0 (2.2–28.7)	88.8±18.3
3,000 to <3,500	519	9.9 (2.4–34.2)	93.9±19.8
≥3,500	455	9.1 (2.8–34.7)	96.6±19.7
Gestation at delivery (weeks)
<38	144	8.9 (3.1–37.9)	84.0±19.4
38 to <42	1,071	9.8 (2.5–33.2)	94.3±18.9
≥42	115	10.0 (3.2–24.8)	97.1±22.9
Birth order
1	479	9.0 (2.1–32.6)	97.2±19.6
2	407	10.0 (2.6–31.4)	93.9±19.7
≥3	446	10.0 (3.0–36.9)	89.0±18.7
Maternal
Race (only U.S. cohorts)
White	278	7.9 (1.3–22.0)	101.9±21.3
Nonwhite	401	7.1 (2.8–21.5)	84.9±12.8
Age at delivery (years)
<25	650	10.5 (3.0–32.0)	89.6±17.3
≥25	683	9.0 (2.1–34.7)	97.1±20.9
Maternal IQ
<85	618	10.0 (2.9–32.0)	83.3±15.0
≥85	715	9.0 (2.1–34.3)	102.2±18.9
Education at delivery (grade)
<12	710	12.0 (4.1–35.5)	90.4±18.8
12	397	8.7 (2.4–34.3)	91.1±17.8
≥12	226	5.5 (1.1–15.2)	107.1±19.4
HOME score
<30	276	9.4 (3.0–43.0)	77.9±14.9
30 to <40	561	10.0 (2.8–32.2)	88.4±15.4
≥40	496	9.5 (2.0–22.0)	107.8±16.5
Married
Yes	896	10.0 (2.7–37.5)	96.6±21.0
No	436	8.1 (2.4–22.0)	87.0±14.4
Prenatal smoking
Yes	453	11.5 (3.2–33.2)	89.7±17.3
No	876	8.7 (2.2–33.6)	95.3±20.4
Prenatal alcohol ingestion
Yes	278	10.1 (2.2–25.0)	100.1±20.2
No	1,035	9.5 (2.7–34.3)	91.8±19.1

Table 4

Mean unadjusted and adjusted changes in full-scale IQ score associated with an increase in blood lead concentration (log scale), from the 5th to 95th percentile of the concurrent blood lead level at the time of IQ testing.

Blood lead variable	Unadjusted estimates [β (95% CI)]	Adjusted estimates [β (95% CI)]	Blood lead concentration (5th to 95th percentile, μg/dL)	IQ deficits [5th to 95th percentile (95% CI)]
Early childhood	–3.51 (–4.83 to –2.19)	–2.21 (–3.38 to –1.04)	3.5–34.5	5.0 (2.4–7.7)
Peak	–5.04 (–6.44 to –3.65)	–2.86 (–4.10 to –1.61)	6.0–47.0	5.9 (3.3–8.4)
Lifetime average	–5.58 (–6.98 to –4.18)	–3.25 (–4.51 to –1.99)	3.6–34.5	7.4 (4.5–10.2)
Concurrent	–4.84 (–5.98 to –3.69)	–2.65 (–3.69 to –1.61)	2.4–33.1	7.0 (4.2–9.7)

Adjusted for site, HOME score, birth weight, maternal IQ, and maternal education. The addition of child’s sex, tobacco exposure during pregnancy, alcohol use during pregnancy, maternal age at delivery, marital status, and birth order did not alter the estimate, and these were not included in the model. The estimates for the covariates in the concurrent blood lead model were HOME score (, ), birth (, ), maternal IQ (, ) and maternal education (, ).