A critical finding of the present study is that in homes that held flood waters for several weeks, our flood cleanup techniques were associated with a reduction in mold and endotoxin levels. In all three homes, the interventions decreased mold levels and in some cases, the decreases were several orders of magnitude. In fact, the only measure that increased from pre-intervention to postintervention was endotoxin in house 2. This house had the most extensive flood damage, and during the intervention, endotoxin was higher than in the other houses. Because endotoxin is a cell wall component of gram-negative bacteria, we speculate that the bacteria were killed but that endotoxin remained in the settled dust after the intervention. Nonetheless, the levels of mold (as determined by three different methodologies) were drastically reduced after the intervention in house 2.
A strength of our study was the multi-pronged assessment of mold exposure. From our data, it is clear that the use of any one of the analytical methodologies (culture, microscopy, or PCR) would have shown that baseline levels of mold were high compared with agricultural, industrial, or other home environments. However, each analytical method gave different insight that we hope will inform future investigations in homes in New Orleans and in other environments. In contrast to culture methods to detect and enumerate fungi, spore microscopy and qPCR do not require viable fungal elements. In addition, spore microscopy and qPCR detect not only dead fungi but those that compete poorly on the various media used to grow fungi. The culture-based analyses (on both malt extract agar and DG18 agars) can underestimate the populations of some Aspergillus
species by orders of magnitude when compared with qPCR (Meklin et al. 2004
). In contrast, Basidiomycota spores (the common mushroom is in this fungal phylum) generally do not grow well on culture plates, and a probe was not available for PCR analysis; therefore, only spore microscopy enabled detection of this potentially allergenic group of molds (Horner et al. 1995
; Lehrer et al. 1986
). The low frequency of Trichoderma
detection by PCR compared with the moderate to high frequency of recovery by direct microscopy and, to a lesser extent, culture likely reflects the airborne presence of taxa (e.g., T. harzianum
and Alternaria tenuissima
) for which PCR probes were not used or available. The spore microscopy also revealed that Curvularia
was common in the houses. Although a PCR probe for Curvularia
was not available for analysis of these samples, culture analysis should have been able to grow Curvularia.
The low prevalence of culturable Cladosporium
contrasts with the high prevalence, as determined by microscopy and PCR. Cladosporium
is one of the most common fungal taxa recovered indoors and outdoors throughout the world (Chew et al. 2001
; IOM 2004
; Levetin 1995
; Su et al. 2001
competes well with many of the other taxa that we recovered, so the reason for its decreased culturability in our samples remains elusive.
At baseline and particularly during intervention, we observed household levels of mold and endotoxin that equaled or surpassed those in wastewater treatment plants, cotton mills, and agricultural environments (Christiani et al. 1993
; Lee J et al. 2006
; Lee S-A et al. 2006
; Spaan et al. 2006
). Interestingly, the outdoor measurements were also higher than expected, ranging from 10 to 25 EU/m3
. These exceed the average outdoor endotoxin measurements from 13 Southern California communities of 0.44 EU/m3
(Mueller-Anneling et al. 2004
). In addition, the baseline levels of culturable mold in the three houses averaged 352,701 CFU/m3
, which was much higher than the average (2,190 CFU/m3
) or maximum level (48,760 CFU/m3
) in homes sampled after the 1993 Mississippi River flood (Ross et al. 2000
). The comparison with the Mississippi River flood study may somewhat overstate the disparity between the two floods because sampling in the earlier study occurred 1 year after the flood and only 40% of the residents in that study reported flood damage in their homes. Nonetheless, the levels of mold in the homes in the present study were extremely high. Given the level of visible mold observed when we first entered our demonstration homes, we were concerned about the level of respiratory protection required for entry and cleanup, and our measurements of both the mold levels and the respirator WPFs supported our concern.
By the time we planned the initial visit to house 3, we were able to characterize the particle sizes and conduct experiments to test the WPF of a disposable N-95 respirator and an elastomeric half-facepiece respirator. Our data indicate that counts of larger particles decreased drastically during the lunch break, taking almost 30 min to reach the lowest point, and reached counts of 50, 500, and 5,000/L of air (i.e., 50,000, 500,000, and 5,000,000 particles/m3) depending on the size range. These data, combined with the measurements of mold spores, culturable mold, and endotoxin, suggest that a substantial portion of the particle counts could be composed of fungal and bacterial material even during periods of inactivity.
The choice of respirator depends on the expected level of contamination. Currently, there are no threshold limit values for mold or endotoxin in the United States; however, a Dutch occupational health standard for endotoxin (50 EU/m3
) existed for a brief time (Douwes et al. 2003
). Also, excessive mold can be cited as a health concern by the Occupational Safety and Health Administration (OSHA) under their General Duty Clause (Occupational Safety and Health Act of 1970
). Still, no defined level of mold is listed that warrants a specific level of respiratory protection. An assigned protection factor, which can be used in the initial selection of respirator type, gives the level of the respiratory protection that a properly functioning respirator or class of respirators would be expected to provide to properly fitted and trained users in the workplace (OSHA 2006
). The assigned protection factor for both filtering facepiece and elastomeric half-facepiece respirators is 10 (e.g., 5,000,000 particles/m3
would be reduced to 500,000 particles/m3
) (OSHA 2006
). We found that, in the field, the WPF was lower for the N-95 respirator. In a previous study (Lee et al. 2005
) in which the WPF against fungal spores was measured in agricultural environments, the geometric mean and geometric standard deviation of 21 WPF data points was 25 and 9.9, respectively. In addition, the WPF decreased with decreasing spore size. The values obtained from the N-95 respirator in house 3 were somewhat lower than those in agricultural environments. This could be due the composition of fungal load in the moldy building where small Aspergillus/Penicillium
spores predominated. On the other hand, the WPF for the elastomeric respirator was clearly higher than the values obtained for the N-95 respirator in this study and in the previous agricultural study (Lee et al. 2005
). Although the number of WPF data points in the present study is low, the results suggest that the elastomeric half-facepiece respirators offer at least 10 times the protection against fungal spores. Because the fungal spore concentrations were extremely high during the renovation, we question whether the protection offered by the N-95 filtering facepiece or the elastomeric respirators is sufficient. Furthermore, our WPF values are based on microscopic counting of intact spores. Recent studies have shown that exposure to fungi occurs also through sub-micrometer fungal fragments (Foto et al. 2005
; Gorny et al. 2002
; Green et al. 2005
). These particles may penetrate at even higher rates as intact spores because the filter materials commonly used in N-95 respirators have maximum particle penetration of approximately 0.03–0.07 μm (Balazy et al. 2006
The postintervention findings in house 1 highlight the critical importance of fully cleaning and drying a house. The upper walls and ceilings in the house were not vacuumed as part of the initial cleanup procedures. In addition, a small water leak saturated a portion of the concrete floor after work was completed. Because possessions remained inside, the house was closed and the humidity levels could have created a climate hospitable for further mold growth. The fact that culturable mold levels in this home were not substantially lower after intervention than before is likely related to these factors. After post-intervention sampling was completed, the water leak in house 1 was fixed and cleaning and mechanical drying was conducted.
House 1 also offers a cautionary note about the risks involved with leaving some of the original drywall in a home. Some flood cleanup guidance suggests that, in homes with minimal flooding, removing drywall on walls to 1.2 m (4 ft is the width of a standard sheet of drywall) instead of to the ceiling can save thousands of dollars in restoration costs (American Red Cross and Federal Emergency Management Agency 1992
). However, many homes were submerged for weeks after Hurricane Katrina; although the water may have only wicked from the water line to the first 1.2 m, the homes were usually closed and the summer heat resulted in humidity levels similar to that of a terrarium. In our study, the house with the lowest water line (house 1) had visible mold growth in the wall cavities above 1.2 m after treatment. Although we cannot be certain that the growth would have occurred had cleaning and drying been adequate, the potential health risks of leaving the original drywall in the home must be taken into consideration.
The question of whether household bleach is an effective treatment mechanism was one of the most debated topics among the advisory group. Although it can have adverse environmental health effects, we used a dilute solution of bleach because of widespread concern of bacterial contamination and evidence that it could denature allergens (Chen and Eggleston 2001
; Matsui et al. 2003
) and possibly inactivate mycotoxins (Wilson et al. 2004
). Bleach was selected primarily on the basis of federal guidance (American Red Cross and Federal Emergency Management Agency 1992
; CDC 2005
) and because it was widely accessible. In house 2, bleach was applied to the wooden building members, whereas it was not used in house 3. Because the postintervention findings for mold were similar in the two homes, we are encouraged that intensive dry cleaning followed by the application of borates appears to control mold growth. The use of dry cleaning without wet cleaning the wood had the added benefit of reducing the time of flood cleanup because the workers did not need to allow the wood to dry before applying borates. Research to examine alternatives to bleach is under way. If effective alternatives are identified, we would encourage their use to be incorporated into the federal emergency response protocols. Ideally the products should be accessible to consumers (both available and inexpensive) to enable their adoption.
There are several noteworthy limitations to the present study, including a
) the small sample size; b
) inconsistency of the number of samples collected; and c
) possible lack of generalizability to other homes because of the home-selection process (Hung et al. 2005
). The number of homes was necessarily small so that we could quickly try different types of cleanup procedures and assess their feasibility and efficacy. To conduct the interventions in a larger set of homes would have prevented expeditious reporting of findings. We had an inconsistent number of samples because of the lack of electricity; as of 22 July 2006, this was still a problem for much of New Orleans (Nossiter 2006
), as was the lack of access to a fully functioning laboratory in New Orleans. We selected houses with a range of flood damage; however, the houses were typical New Orleans building structures, and the level of flooding was typical of many homes in the affected areas of New Orleans. The homes in Mississippi that were directly in the path of Hurricane Katrina sustained heavy wind damage, and we do not believe that our intervention results can be generalized to those homes. Nonetheless, our discussion of respiratory protection should be applicable to those homes with extensive mold growth.
The main goals for this pilot project were to synthesize, field test, and evaluate existing flood cleanup methods. For all houses, we removed ≥ 1.2 m of drywall, conducted HEPA vacuuming, used a borate salt solution to help prevent mold growth, and used bleach as a disinfectant. Using a variety of sampling and analytical methods, we observed airborne levels of mold and endotoxin, which often increased orders of magnitude during the intervention, and determined that workplace protection factors of some respirators can be suboptimal in such conditions. Although the generally accepted mold remediation protocols reduced bioaerosols in the demonstration houses, myriad issues including the qualifications of those performing the work (including homeowners), depth and duration of flooding, and the availability of electricity and supplies can affect the feasibility and ultimately the success of flood cleanup efforts. Our pilot project was not designed for determining whether the demonstration homes were safe for reoccupancy. Rather, we examined the extent to which homes that experienced significant and prolonged exposure to flood waters could be satisfactorily cleaned to enable reconstruction. Future research may include revisiting these homes after reconstruction to determine whether the low bioaerosol levels persisted or even continued to decline.