Methods

Pregnant women who lived in the City of St. Louis and who provided informed consent were eligible to participate. Recruitment occurred between August 2007 and December 2009. Because prenatal home inspection and remediation had become standard practice of the City of St. Louis Health Department and because the project was considered an evaluation of an existing city program, institutional review board approval was deemed exempt (Washington University X-07-079). Because we believed that this program would lower children’s blood-lead levels, we did not include a prospective control population.

Health Department workers, building inspectors, and lead contractors were funded through the City of St. Louis federal Housing and Urban Development (HUD) grant and city tax revenue. Recruitment primarily occurred in the outpatient clinics of the Barnes-Jewish Hospital, with a few patients coming from other urban clinics that also served populations at high risk by virtue of geography, race, and income. Basic information was recorded that included patient name, address, expected delivery date, names of other children, and 3 contact phone numbers.

Home inspections were conducted by certified inspectors from the City of St. Louis Building Division in accordance with HUD regulations, which included visual paint inspection, x-ray fluoroscopy, and dust swipes. Hazards were defined in one of the following ways: visibly deteriorated paint that registered >1.0 mg/cm ² on x-ray fluoroscopy; dust wipe >40 μg/sq ft on the floor or 250 μg/sq ft on window sills; and soil >400 μg/g for play areas or >1200 μg/g for other residential yards. These levels were also used to define clearance levels.

Remediation efforts included paint stabilization, window replacement, and cleaning as needed. The work was performed either by the city’s certified lead workers or by a contractor who was paid through the city’s HUD grant. Information on inspections was maintained by the Building Division in the Lead Housing Tracking database. All homes that underwent remediation passed clearance dust wipes.

Contact with each family was done at 3- to 6-month intervals through phone calls with the use of a standardized script. At these times health workers updated information regarding the names of the children and demographics and encouraged the mothers to have their children tested for lead. Most of the children were tested by their pediatricians, although a few parents used free testing programs that are offered by the Health Department. Lead levels were checked by capillary or venous blood samples. Children with multiple blood-lead levels had the highest number of samples that were used for analysis. All data were kept on a password-protected database at the Health Department or the Lead Housing Tracking database. Missouri mandates the reporting of blood-lead level data, and these data are maintained in Missouri Health Strategic Architecture and Information Cooperative Database.

To identify control subjects, the Missouri Health Strategic Architecture and Information Cooperative Database was searched for all children who lived in the same census tracts as the program participants, who were born ±0.2 years of the date of the birth of the participants, and who had blood-lead levels drawn within the same calendar year as the participants. We chose census tract as the geographic unit of comparison because this has been found to be a good correlate for blood-lead levels, as compared with zip code data. The list was parsed for data duplication, and children whose homes had received an intervention were removed from the control subjects. Each study participant was matched with 2 control subjects who were selected randomly from the pool of control subjects within the matching census tract with the correct age and date of blood draw.

Mean blood-lead levels were compared with the use of a t test. The number of children with levels of ≥5 μg/dL was analyzed by χ ² test and levels ≥10 μg/dL by 1-tailed Fisher’s exact test.