Environmental Lead after Hurricane Katrina: Implications for Future Populations

Background: As a result of Hurricane Katrina, > 100,000 homes were destroyed or damaged and a significant amount of sediment was deposited throughout the city of New Orleans, Louisiana. Researchers have identified the potential for increased lead hazards from environmental lead contamination of soils. Objectives: We assessed the distribution of residential soil and dust lead 2 years poststorm and compared soil lead before and after the storm. Methods: We conducted a cross-sectional study in New Orleans in which households were selected by stratified random sampling. A standard residential questionnaire was administered, and lead testing was performed for both the interior and exterior of homes. Logistic regression was used to identify significant predictors of interior and exterior lead levels in excess of allowable levels. Results: One hundred nine households were enrolled; 61% had at least one lead measurement above federal standards. Of homes with bare soil, 47% had elevated lead and 27% had levels exceeding 1,200 ppm. Housing age was associated with soil lead, and housing age and soil lead were associated with interior lead. Race, income, and ownership status were not significantly associated with either interior or exterior lead levels. The median soil lead level of 560 ppm was significantly higher than the median level of samples collected before Hurricane Katrina. Conclusions: The high prevalence (61%) of lead above recommended levels in soil and dust samples in and around residences raises concern about potential health risks to the New Orleans population, most notably children. Steps should be taken to mitigate the risk of exposure to lead-contaminated soil and dust. Further research is needed to quantify the possible contribution of reconstruction activities to environmental lead levels.

To assess the hurricane's impact, the U.S. Environmental Protection Agency (EPA) and the Louisiana Department of Environmental Quality initiated an investigation into the floodwater sediment contamination in resi dential neighborhoods both before the flood waters receded and before cleanup. Sampling results indicated that resi dential soils con tained lead; however, the U.S. EPA found that the hurricane did not significantly affect the distribution of lead because the post hurricane geography of lead distribution resembled pre hurricane distributions (U.S. EPA 2005). Another lead assessment, con ducted in 2006, reported a 46% decrease in median soil lead from preKatrina levels (Zahran et al. 2010). Both of these studies were conducted in the immediate aftermath of Hurricane Katrina and preceded the exten sive renovation effort that would be required to rebuild the city.
A 2007 report by the Agency for Toxic Substances and Disease Registry (ATSDR) highlighted the potential risk of lead exposure to families returning to New Orleans in light of the extensive amount of renovation and demolition that would be required to rebuild the city. According to the 2000 U.S. Census, > 100,000 homes in New Orleans were built before 1950, an estimated 83% of which have lead hazards (ATSDR 2007). The report con cluded by stating that despite surveys indicat ing no increase in environmental lead levels, the actual extent of lead hazards would be determined only after soil data collected sub sequent to reconstruction activities became available (ATSDR 2007). A recently pub lished survey of schoolyard soil also suggested the need for more extensive assessment of residential lead hazards (Presley et al. 2010).
To our knowledge, there have been no environmental surveys of residential haz ards in New Orleans in the aftermath of Hurricane Katrina. The primary goal of the New Orleans Home Health Hazard project was to assess the burden of numerous envi ronmental health hazards (e.g., allergen levels, mold, lead) in the homes of returning resi dents. The present analysis was conducted to charac terize the distribution of residential soil and dust lead levels in a representative sample of New Orleans homes after the reconstruc tion effort in the city had begun, to compare the soil lead distribution pre and post storm, and to address the potential lead hazard to residents of New Orleans after the devasta tion of Hurricane Katrina.

Materials and Methods
Sampling and recruitment. We conducted a crosssectional survey of lead hazards in a representative sample of New Orleans homes. The target population was all occupied homes in New Orleans. Repopulation of the city was a dynamic process; as a result, determining population estimates was a challenge. The Louisiana Public Health Institute worked with the U.S. Census Bureau and the Centers for Disease Control and Prevention (CDC) to produce accurate and reliable estimates of the size and charac teristics of the New Orleans population during the postHurricane Katrina recovery period (Louisiana Public Health Institute 2006). Their population estimates, which were stratified by planning district, were used for the present study. Planning districts are geographical units of the city designated by federal and state requirements for economic development planning. New Orleans is composed of 13 planning districts ( Figure 1). We excluded 3 planning districts from the study because they are unique areas and do not represent the New Orleans urban core (Venetian Isles, Village de L'est, and English Turn). One other planning district, the Lower Ninth Ward, was also excluded because of the slow pace of reconstruction in that district.
Mass population shifts occurred during the period after the hurricane, making exist ing household sampling frames (e.g., residen tial rosters) inappropriate for use. Therefore, to select sample households, we used the Sewerage and Water Board address list, a ros ter of all residential addresses billed for water service (the best available list of reoccupied homes after the storm), as the sampling frame. Sewerage and Water Board addresses were geocoded to a census tract and assigned to the appropriate planning district.
Households were recruited from January 2007 through May 2008. Recruitment was staggered over this 17month period to incorporate the dynamic repopulation of the city in an effort to include the maximum number of neighborhoods. Households were eligible for inclusion if the house was deemed occupied and the house was a primary residence (i.e., the location where the head of household slept ≥ 4 nights/week. Federal Emergency Management Agency (FEMA) trailers were excluded from the study. Proportionate random sampling, stratified by city planning district, was used to select 109 study households. An initial recruitment letter was hand delivered to each selected household after field staff determined that the home was occupied using a welldefined protocol developed and pilot tested by the study team. If a selected household was deemed to be ineligible, another household was randomly chosen from the relevant planning district strata. A second recruitment letter was mailed after 2 weeks if no response was received from the initial recruitment letter. Up to five recruitment letters were sent to homes deemed occupied per the study protocol. If eligible homes did not enroll after five recruitment attempts, they were deemed to be non responsive. Active informed consent was obtained for each enrolled household. The study was approved by the Tulane University Institutional Review Board.
Data collection. Each head of household was adminis tered a survey questionnaire by a field researcher. The questionnaire was adapted from the National Survey of Lead and Allergens in Housing [Department of Housing and Urban Development (HUD) 2001]. Field staff also recorded the home's proximity to a major road and type of house (single family home, lowrise apartment, duplex, etc.). Lead sampling was conducted in accor dance with sampling procedures as described by the U.S. EPA (1995), HUD, and the American Society for Testing and Materials (ASTM 1997a(ASTM , 1997b. Surface lead dust was collected via wipe sampling from six locations in the interior of the home. Floor samples from the kitchen, bedroom, and living room floors were taken if there was bare floor, and windowsill samples were collected from the same rooms if there was a window in that room. An outdoor sample was collected from the bare soil just outside the front door of the home; if there was no bare soil in the entrance area, a sample was taken from the middle of the yard. If there was no bare soil in either of these areas, a sample was taken from the backyard. If there was no bare soil in any of these areas, no sample was taken. Results from a soil lead survey conducted between 1998 and 2000 were obtained and used to compare the distribution of lead in the city before and after Hurricane Katrina. The soil lead survey sampling protocol has been described in detail previously (Mielke et al. 2005). Briefly, soil samples were collected from 286 census tracts in New Orleans from the top 2.5 cm of bare soil following standard soil sampling proce dures. Soil lead data for the present study were geocoded to a census tract. For the 55 census tracts with soil lead data from both studies, the results were compared to assess soil lead distribution pre and post storm.
Sample analyses. Dust and soil lead samples were analyzed for metals by BTS Laboratories (Waldorf, MD)-a U.S. EPArecognized, American Industrial Hygiene Association-accredited laboratory-using U.S. EPA method 7010 with graphite furnace atomic absorption spectrophotometry (U.S. EPA 2007). The results were reported in parts per million.
Statistical analyses. We assessed variables known to be associated with elevated soil and dust lead levels, including age of housing, home owner ship status, race, income, and proximity of the home to a busy street. For interior lead, soil lead level was also considered. Descriptive statistical analyses were performed to examine the distribution of socio demographic and household charac teristics and lead levels. Twentynine missing income data points were imputed from median income level of the census tract block to which the address belonged, according to the 2000 Census (U.S. Census Bureau 2003). We used logistic regression models to estimate crude and adjusted odds ratios (ORs) with 95% confidence intervals (CIs) to identify factors related to elevated soil lead levels and interior lead levels, respectively. Race, income, home owner ship status, and housing located near a busy street were included in the multi variable model based on a priori consideration, as these variables have been shown in the litera ture to be strongly associated with lead levels. Wilcoxon signed rank test for paired data was run on median census tract soil lead level to explore changes in the distribution of soil lead pre and post hurricane. Homes that lacked sample data because there was no bare floor to sample, no windowsill in the designated room, or no bare soil around the property were excluded from the relevant analysis. All analyses were performed using SAS software (version 9.1.3; SAS Institute Inc., Cary, NC).

Results
The sample consisted of 109 homes from nine planning districts in New Orleans; thus, power estimates were calculated on the sample size of 109 households. With an exposure prevalence rate of 61.2%, an α of 0.05, and an alter native proportion of 0.45, study power was 92%. Response rates varied from 14.2% in the French Quarter (mainly a commercial area) to 75.6% in Lakeview. The overall response rate was 32.5%. Most respondents were Caucasian (61.5%) and had an annual house hold income > $30,000 (67%) ( Table 1). Most of the homes were built before 1946  (64.2%), and most participants were home owners (68.8%). Four participant house holds (3.7%) reported receiving some form of govern ment assistance. Nine homes (8.3%) were located near a busy street or intersection. Using standard HUD/U.S. EPA cut points of > 40 µg/ft 2 for floor dust and > 250 µg/ft 2 for windowsill dust (U.S. EPA 2008a), 50.5% of homes had at least one interior sample that was elevated (Table 2). Nearly half (46.7%) of the 90 homes with bare soil had levels > 400 ppm, and 26.7% had levels > 1,200 ppm. Considering both interior and exterior samples, 61.4% of homes had at least one measurement in excess of the HUD/U.S. EPA standard.
In unadjusted models, age of housing was the only factor significantly associated (p < 0.05) with elevated soil lead (OR = 82.0; 95% CI: 10.3, 651.5), whereas both age of housing (OR = 12.8; 95% CI: 4.7, 35.1) and elevated soil lead (OR = 21.2; 95% CI: 7.2, 62.7) were significantly associated with increased interior lead. In adjusted models, after controlling for housing age, soil lead remained significantly associated with interior lead levels above the HUD/U.S. EPA health based standards (Table 3).
Soil data were geocoded to the 55 census tracts for which soil data were available for both pre and postHurricane Katrina time periods. We compared the soil lead levels

Discussion
The traditional lead distribution in New Orleans preKatrina was like that of many other older cities, with a higher concentration in and around deteriorating older housing. Besides age of housing, socio demographic variables, including AfricanAmerican race, low income, poverty, and renter status, have consistently been shown to be associated with elevated lead in urban cities (Gaitens et al. 2009;Jacobs et al. 2002;Lanphear et al. 2005). Hurricane Katrina disproportionately affected lowincome, minority residents. The diaspora lacked the means to quickly return to the city, and our study sample reflects this repopulation pattern. Study households included a large proportion of non minority home owners living in areas of the city with low poverty. This demo graphic group would normally be considered to have a low risk for lead exposure; therefore, the widespread lead contamination in and around study homes-particularly the degree of soil contamination-was unexpected.
More than 60% of households had an elevated lead sample either inside or outside of the home. Nearly half the homes with bare soil had elevated soil lead, and 27% of those homes had soil lead > 1,200 ppm, three times the HUD/U.S. EPA standard. Neither soil nor interior dust lead levels were sig nificantly associated with socio demographic variables commonly found to be related to elevated lead, although low income was non significantly associated with interior lead in adjusted models. We hypothesize that lead contamination may be the result of the unprecedented amount of home renovation and demolition that was required as a result of Hurricane Katrina damage both in high and lowincome neighborhoods.
In New Orleans, approximately 135,000 structures sustained hurricane damage, with major or severe damage to 105,000 of these homes. As of May 2009, an estimated 9,000 homes had been demolished and thousands more repaired or renovated (Brookings Institute 2009;HUD 2006). Both the reno va tion and demolition of these structures, many of which were built before 1950, pose a potential new lead exposure source and a potential health hazard for children (Chou et al. 2010;U.S. EPA 2006). Numerous studies have shown that when leadsafe practices are not employed during renovation, lead is dispersed into the environment and residential soil becomes contaminated (Clark et al. 2004;Dixon et al. 2007;Lanphear and Roghmann 1997;Reissman et al. 2002). Demolition activity is also a source of environ mental lead and has been associated with an increase in children's blood lead levels (Farfel et al. 2003(Farfel et al. , 2005Rabito et al. 2007).
We found that the median soil lead level for the sampled homes was 37% higher than the median soil lead level for samples col lected for a 1998-2000 lead survey. Our find ing that soil lead levels in samples collected from the same census tracts in [2007][2008] were higher than pre storm levels contradict those from two soil lead distribution studies   conducted after Hurricane Katrina (Walsh et al. 2006;Zahran et al. 2010). The U.S. EPA Louisiana Department of Environmental Quality survey found no change in the dis tribution of soil lead levels, and Zahran et al. (2010) reported a 46% decrease in median soil lead from preKatrina levels. Both of these studies were conducted in the immedi ate aftermath of Hurricane Katrina and before most reconstruction activity. Furthermore, the reduction in soil lead observed immediately after the storm may have reflected relatively low levels in soil sediment from surround ing water bodies, which was deposited dur ing widespread flooding of the city (Zahran et al. 2010). Natural and anthropogenic activities can redistribute soil lead, and it was predicted by Potera (2010) that the cleaner soil sediment brought by Hurricane Katrina may not persist. The ATSDR (2007) specifi cally warned of the risk of lead dispersion as a result of home renovation and demolition and predicted that if leadsafe practices were not employed, the level and the extent of soil and interior contamination would increase. As a result of Hurricane Katrina's wide spread destruction, homes previously in good condition required renovation, and the sub sequent disturbance of lead in old homes makes the risk of lead exposure universal. In this population, 64.2% of the homes were built before 1946, a period when lead paint was in widespread use. Lead paint can be released directly into the soil via air resuspension during renovation activities. Power sanding, a popular method of exterior paint removal, can release a large amount of lead dust into the environment. HUD (1995) estimated that 1 ft 2 of pulverized lead paint produces a settled dust lead level of 9,300 µg/ft 2 . Although power sanding is prohibited in New Orleans by city ordinance (New Orleans, Louisiana, Code of Ordinances 2001), lack of oversight in the post disaster environment resulted in widespread sanding of homes under going renovation. Therefore, it seems plausible that subsequent soil contamination may have occurred. If renovation/demolition activities are contributing to the high lead levels, the lack of association with commonly known socio demographic factors is not surprising given the extent of housing damage.
There is abundant evidence that soil lead exposure is a significant contributor to blood lead levels in children (Clark et al. 2004;Dixon et al. 2007;Lanphear et al. 2005;Ren et al. 2006). Estimates of the contribution of exterior soil to indoor dust range from 20-30% (Culbard et al. 1988;Davies et al. 1985;Rutz et al. 1997) to as high as 85% (Fergusson and Kim 1991;Roberts et al. 1991;Trowbridge and Burmaster 1997). Urban soils can integrate lead from numerous sources, including paint on the exterior of homes, leaded gasoline emissions, and incinerator or industrial lead emissions that have accumulated in the environ ment (Mielke 1999;Mielke et al. 2005). Because regulations are in place to limit industrial emissions and because leaded gasoline and paint have been banned, of particular concern is new contamination as a result of renovation and demolition of old structures. According to a 1998-2000 national survey (Jacobs et al. 2002), 24 million housing units contain leadbased paint hazards; these housing units serve as a reservoir of lead hazards that can pose a risk to children via dust and soil for years to come. The preponderance of old homes coupled with extensive renovation/ demolition activities and high dust and soil lead suggests that children in New Orleans are at substantial risk of environ mental lead exposure in and around their homes. An alternate explanation is that soils contaminated with heavy metals may have been carried by flood water and redeposited to new locations postKatrina (Presley et al. 2010). However, regardless of the actual source of contamination and the findings from the pre and postKatrina comparative analysis, our findings indicate that lead contaminationparticularly soil contamination-is prevalent and has significant public health implications for residents, especially children. The current targeted screening and public health intervention efforts to prevent childhood lead poisoning may need to be expanded to capture a population that previously was not considered at risk of environ mental lead exposure.
Regulatory action is an effective tool for reducing lead exposure. The key, however, is the extent to which regulations are enforced and their degree of coverage. A U.S. EPA rule governing renovation on childoccupied structures built before 1978 was adopted in April 2008 (U.S. EPA 2008b). Although the rule provides a formal framework for leadsafe work practices, it has several limitations. First, it includes work performed by contractors and exempts work performed by homeowners, tenants, and day laborers. In a study of home renovation among children with lead levels ≥ 20 µg/dL in New York, 14% of elevated levels were attributed to renovation, and owners or tenants performed 66% of that work (CDC 2009). The rule also exempts homes without children. Although a child may not currently reside in the home, there can be no prediction for future occupancy. Furthermore, children residing near homes where lead dust is dispersed are put at risk from resuspension of lead, particularly in highdensity areas, such as urban cores, where the vast amount of old housing exists. Testing urban soils should be a priority, and policy officials should support efforts to halt power sanding and other work practices that result in resuspension of lead into the environment. This study has several limitations. Although households were selected by strati fied random sample, the overall response rate was low at 32.5%. Although likely an under estimation because of the conservative approach taken to define occupancy, there were no reliable population estimates because occupancy was a dynamic and fluid process postKatrina. The possibility of selection bias due to differential participation by households concerned about lead hazards should be con sidered; however, the primary goal of the proj ect was measure ment of allergens and mold in homes, and mold exposure was the princi pal health concerns for residents returning to New Orleans after the storm. Furthermore, media extensively reported the U.S. EPA find ing of no new lead exposure in the wake of the storm; therefore, we feel it is unlikely that participation was motivated by concern for lead hazards. A further limitation of the study is the lack of information on reno va tion activi ties for participating households and structures nearby to support the ad hoc hypothesis that renovation activities are contributing to the high residential lead levels. Although > 86% of households in this study reported that they had completed renovations, were in the pro cess of renovating, or still needed renovation, no data were collected as to the extent of dam age or what type of renovations had taken place; therefore, renovation status could not be included in the multi variate model, and the role that renovation and demolition may have played remains unknown. Finally, our com parison of soil lead data pre and postKatrina has some limitations. Although our approach was similar to that of other studies, it is likely that differences in sampling methodology may explain some of the observed differences in censustract-specific soil lead levels. In the present study, samples were collected exclu sively from soil around participating houses. In the 2000 soil survey (Mielke et al. 2000), although data were collected primarily from around the house, some samples were collected near streets in residential neighborhoods. Moreover, the number of soil samples taken per census tract varied between the surveys.

Conclusion
To our knowledge, this is the first study of residential lead hazards in New Orleans in the aftermath of Hurricane Katrina. Survey results indicate that 61% of homes had lead levels that exceed the U.S. EPA standards, indepen dent of race, income, and ownership status. New Orleans children are at risk for elevated blood lead levels, including children who were not considered at high risk previously and for whom lead reduction has been considered a public health success. Enhanced surveillance and lead hazard mitigation efforts are needed to safeguard the health of New Orleans residents.