Health Effects of Exposure to Water-Damaged New Orleans Homes Six Months After Hurricanes Katrina and Rita

Abstract

Objectives. We investigated the relation between respiratory symptoms and exposure to water-damaged homes and the effect of respirator use in posthurricane New Orleans, Louisiana.

Methods. We randomly selected 600 residential sites and then interviewed 1 adult per site. We created an exposure variable, calculated upper respiratory symptom (URS) and lower respiratory symptom (LRS) scores, and defined exacerbation categories by the effect on symptoms of being inside water-damaged homes. We used multiple linear regression to model symptom scores (for all participants) and polytomous logistic regression to model exacerbation of symptoms when inside (for those participating in clean-up).

Results. Of 553 participants (response rate=92%), 372 (68%) had participated in clean-up; 233 (63%) of these used a respirator. Respiratory symptom scores increased linearly with exposure (P<.05 for trend). Disposable-respirator use was associated with lower odds of exacerbation of moderate or severe symptoms inside water-damaged homes for URS (odds ratio (OR)=.51; 95% confidence interval (CI)=0.24, 1.09) and LRS (OR=0.33; 95% CI=0.13, 0.83).

Conclusions. Respiratory symptoms were positively associated with exposure to water-damaged homes, including exposure limited to being inside without participating in clean-up. Respirator use had a protective effect and should be considered when inside water-damaged homes regardless of activities undertaken.

In August and September 2005, Hurricanes Katrina and Rita caused unprecedented flooding in New Orleans, Louisiana. In the aftermath, visible mold growth occurred in approximately 44% of area homes.1 Air sampling for mold spores conducted in October and November 2005 showed high levels both indoors and outdoors.2 Similar sampling in October 2005 showed elevated levels of endotoxin, a bacterial cell wall component, in water-damaged homes.3

A 2004 Institute of Medicine report concluded that sufficient evidence exists for associating the presence of mold or other dampness-related agents in damp buildings with nasal and throat symptoms, cough, wheeze, asthma exacerbations in sensitized asthmatics, and hypersensitivity pneumonitis.4 Since that report, there has been additional evidence to suggest that asthma can develop during childhood5 and in the occupational setting6 as a result of exposure to dampness or mold.

The conditions in New Orleans after the hurricanes thus posed a potential health risk to thousands of returning residents. In light of these circumstances, public health officials made recommendations on the use of personal protective equipment. Among these was the recommendation for the general public to use particulate respirators when around mold-contaminated dust, such as might be encountered during clean-up activities.7^,8 The risk of health effects for residents involved in activities less likely to disturb mold-contaminated materials, such as visually inspecting the interior of an affected home or collecting belongings, was thought to be lower. Thus, for such activities, the public was advised that respirators were not usually needed.8

Although respiratory illness and asthma exacerbations have been noted following flooding,9 the contributory role to respiratory disease of postflood exposure to water-damaged homes has not been well documented. We sought to better understand the relation between respiratory symptoms and exposure to water-damaged homes in posthurricane New Orleans. Given the recommendation about respirator use and the observation that respirator use was common among the public following the hurricanes,10 a second objective was to determine the effect of respirator use on symptoms. We conducted a population-based investigation to address these issues.

METHODS

Participants

We randomly selected sampling locations in Orleans Parish (city of New Orleans) using geographic information system software (ArcGIS version 9.1, ESRI, Redlands, California). To focus on residential areas, we eliminated 6345 of the parish’s 10181 census blocks.11 Eliminated blocks included those with very low or very high housing density and those in neighborhoods that remained uninhabited. We generated 600 random way-points (unique locations on the basis of latitude and longitude) across the remaining 3836 census blocks.

Each waypoint served as a starting point to locate participants. A survey team navigated to a waypoint using a global positioning system device and identified the nearest home. English-speaking adults (18 years and older) associated with the home as owner, current occupant, or relative or friend of the owner or current occupant were eligible. Individuals at a home as paid employees (e.g., remediators) were not eligible. However, we did not exclude remediators encountered at their own homes. If unable to conduct an interview at the first encountered home, the team proceeded in a systematic fashion to the next home. Once the team conducted one interview at the waypoint, they navigated to the next waypoint and repeated the process.

Questionnaire

From March 4 through March 11, 2006, we interviewed participants using a 10-minute anonymous questionnaire regarding respiratory symptoms; physician diagnoses of asthma, eczema, and hay fever; smoking history; race and ethnicity; experiences with water-damaged homes; and experiences with respiratory protection. Questions on respiratory protection addressed whether, during clean-up since the hurricanes, the participant had ever used noncertified dust and surgical masks (hereafter, “masks”) or National Institute for Occupational Safety and Health (NIOSH)-certified disposable and reusable respirators (hereafter, “respirators”), including disposable N-95 filtering facepiece respirators (hereafter, “N-95 FF respirators”) and reusable half-face and full-face respirators (hereafter, “reusable respirators”).12 Participants who had used more than 1 type of respiratory protection could indicate each type. We displayed photographs and examples of masks and respirators during the interviews.

Questions on respiratory symptoms addressed upper respiratory symptoms (URS; “stuffy, itchy, runny nose,” “sinus problems,” and “hoarseness or dry, sore, or burning throat”) and lower respiratory symptoms (LRS; “wheezing or whistling in chest,” “chest tightness,” “attacks of shortness of breath,” and “coughing attacks”). We asked participants to grade the extent to which they had experienced each respiratory symptom since the hurricanes as “none,” “mild,” or “moderate or severe.” Those who reported respiratory symptoms were asked how being inside a water-damaged home affected symptoms (“same,” “worse,” or “better”), aggregated as URS or LRS.

Statistical Analyses

Exposure to water-damaged homes.

We set a “clean-up score” equal to the sum of the reported number of homes cleaned that had less than 50% of walls and ceilings covered with mold and twice the reported number of homes cleaned that had 50% or more covered with mold. We identified groupings using a cluster analysis on the natural logarithm of this score.13 From that we defined an ordinal exposure variable (“exposure to water-damaged homes”) as follows: 0 = had not been inside a water-damaged home; 1= had been inside but had not participated in clean-up activities; 2 = participated in clean-up activities, clean-up score of 1; 3= participated in clean-up activities, clean-up score of 2, 3, or 4; 4 = participated in clean-up activities, clean-up score of 5 or more.

Respiratory symptoms and exposure.

We examined the associations between individual symptoms of any severity and exposure to water-damaged homes using contingency tables and the Cochran–Armitage trend test.

To further model the relation between symptoms and exposure, we developed a “symptom score” for each participant. We assigned a value to each symptom on the basis of the reported grade (0 = none; 1 = mild; 2= moderate or severe) and then summed the values. The URS score varied from 0 (no URS) to 6 (all 3 URSs, graded moderate or severe). The LRS score varied from 0 (no LRS) to 8 (all 4 LRSs, graded moderate or severe), and the overall symptom score varied from 0 (no URS or LRS) to 14 (all URSs and LRSs, graded moderate or severe). We conducted a principal components analysis and found high correlations between principal components and symptom scores.

Given these high correlations, we created multiple linear regression models using only symptom scores (URS, LRS, and overall respiratory symptoms) as the dependent variables. The independent variables were the categories of exposure to water-damaged homes, whether the respondent was still participating in clean-up, gender, age, race (White or other), smoking status (current, former, or never), and atopy (physician-diagnosed eczema or hay fever). Overall models and models stratified by asthma were developed. We used the F test from analyses of variance (ANOVA) and the Tukey HSD (honestly significant difference) test to determine significance. For independent variables with more than 2 response levels, we assessed linearity using the method of contrasts.

Symptom exacerbations.

We used polytomous logistic regression models to investigate factors associated with exacerbation of URS and LRS when inside a water-damaged home. We created a categorical response variable as follows: 0 = no symptoms; 1 = mild symptoms, not worse (“same” or “better”) inside; 2= moderate or severe symptoms, not worse inside; 3 = mild symptoms, worse inside; 4= moderate or severe symptoms, worse inside. To assess the effect of respiratory protection, we limited the analyses to participants who reported engaging in clean-up activities. The independent variables were those noted above plus the type of respiratory protection used during clean-up activities.

We categorized respiratory protection as follows: “no mask or respirator” (those who reported using no respiratory protection), “mask only” (those who reported using only a mask), “no respirator” (those who reported using either no respiratory protection or only a mask), “any respirator” (those who reported using an N-95 FF respirator, reusable respirator, or both), and “N-95 FF respirator only” (those who reported using an N-95 FF respirator but not a reusable respirator). Participants in the “any respirator” and “N-95 FF respirator only” categories also may have reported using a mask, but those in the “mask only” and “no respirator” categories did not report using a respirator. We used the χ² test to determine statistical significance.

For all analyses, we considered 2-sided P value at .05 or below to be statistically significant, except for tests of trend for which we used 1-sided tests. We conducted analyses using SAS version 9.1 (SAS Institute, Cary, North Carolina) and JMP version 5.1 (SAS Institute).

RESULTS

Participants

We invited 659 eligible individuals to participate, 553 (84%) of whom agreed. Thus, 92% of the 600 visited waypoints were represented. Most (n = 45; 96%) unrepresented waypoints were unoccupied. The participants had a median age of 50 years (Table 1 ▶). Half of the participants were men, and about half identified their race as White and half as African American or Black.

TABLE 1—

Participant Characteristics (N = 553): Orleans Parish, Louisiana, March 2006

Characteristic	Median (range) or n/N (%)a
Age, y (n = 547)	50 (18–89)
Male gender	292/553 (53)
Race/Ethnicityb
White	241/548 (44)
African American	296/548 (54)
Asian	20/548 (4)
American Indian or Alaska Native	21/548 (4)
Native Hawaiian or other Pacific Islander	5/548 (1)
Smoking status
Current	127/551 (23)
Former	123/551 (22)
Never	301/551 (55)
Physician-diagnosed asthma
Ever diagnosed	68/553 (12)
Current asthma	33/553 (6)
Atopy	80/553 (14)
Activities in water-damaged home since Hurricane Katrina
Been inside	467/551 (85)
Participated in clean-up	372/551 (68)
Number of homes cleaned (n = 368)	2 (1–50)
Number with mold extent ≥ 50% (n = 367)	1 (0–25)
Still participating in clean-up	183/358 (51)
Respiratory protection during clean-up since Hurricane Katrinac
Used mask or any respirator	315/368 (86)
Used mask only	82/368 (22)
Used any respirator	233/368 (63)
Used N-95 FF respirator only	144/368 (39)

Note. Mask = noncertified dust mask or surgical mask; any respirator = certified respirator, including disposable N-95 filtering face-piece respirator and half-face and full-face reusable respirators; N-95 FF respirator = certified disposable N-95 filtering face-piece respirator.

^aData for some characteristics were missing for some participants.

^bParticipants could select more than 1 racial category, so numbers may total more than 100%.

^cParticipants in the “mask only” category did not report using a respirator. Participants in the “any respirator” category reported using an N-95 FF respirator or a reusable respirator and also may have reported using a mask. Participants in the “N-95 FF respirator only” category reported using an N-95 FF respirator and also may have reported using a mask but did not report using a reusable respirator.

Eighty-four (15%) participants reported that they had not been inside a water-damaged home since the hurricanes, and 95 (17%) reported that they had been inside but had not participated in clean-up activities. A total of 372 (68%) respondents had been inside and participated in clean-up activities of a median of 2 homes (range = 1–50). Of those who had participated in clean-up, 315 (86%) reported that they had used a mask or a respirator and 233 (63%) reported that they had used a respirator specifically.

Respiratory Symptoms and Diagnoses

Since the hurricanes, 360 (65%) of the 553 participants had experienced at least 1 URS of any severity: 125 (23%) reported 1, 111 (20%) reported 2, and 124 (22%) reported 3 URSs. There were 245 (44%) participants who had experienced at least 1 LRS of any severity: 110 (20%) reported 1, 70 (13%) reported 2, 39 (7%) reported 3, and 26 (5%) reported 4 LRSs. Twenty-one (4%) participants reported all 7 respiratory symptoms. Among URSs, nasal symptoms were most common (n = 270; 49%), whereas among LRSs, cough was most common (n= 203; 37%).

Lifetime and current prevalence of physician-diagnosed asthma were 12% (n = 68) and 6% (n = 33), respectively. No participant reported a new diagnosis of asthma since the hurricanes. Forty-eight (10%) of those without a diagnosis of asthma, and 11 (33%) of those with current asthma, reported experiencing at least 3 LRSs since the hurricanes. Prevalence of atopy was 14% (n= 80).

Respiratory Symptoms and Exposure

The prevalence of individual respiratory symptoms varied significantly by the extent of exposure to water-damaged homes (Figure 1 ▶). For wheeze, period prevalence was 5% among participants who had not been inside a water-damaged home, 20% among participants who had been inside but had not participated in clean-up activities, and 29% among participants in the highest exposure category (P < .001 for trend). We found the same pattern for the other respiratory symptoms.

FIGURE 1—

Period prevalence of upper and lower respiratory symptoms of any severity since Hurricanes Katrina and Rita among residents (N = 544), by extent of exposure to water-damaged homes: Orleans Parish, Louisiana, March 2006.

Note. Exposure categories were defined as follows: 0 = had not been inside water-damaged home (n = 84); 1 = had been inside but had not participated in clean-up (n = 95); 2 = participated in clean-up, clean-up score of 1 (n = 91);3 = participated in clean-up, clean-up score of 2, 3, or 4 (n = 182); 4 = participated in clean-up, clean-up score of 5 or greater (n = 92). Clean-up score was the sum of the number of homes cleaned that had fewer than 50% mold and twice the number of homes cleaned that had 50% or more mold. Statistical significance was determined by the Cochran–Armitage trend test (1 sided); P < .05 for all symptoms.

We found significant positive associations between exposure to water-damaged homes and URS score (P< .05), LRS score (P= .01), and overall symptom score (P< .01) in the multiple linear models (Figure 2 ▶). In the URS analysis, the least squares mean symptom score was 1.6 for participants who had not been inside a water-damaged home and 2.7 for participants in the highest exposure category (P< .01 for trend). However, we noted a relatively larger increase in mean symptom score between the first 2 exposure categories (“had not been inside” and “had been inside but had not participated in clean-up”) than between other consecutive exposure categories. We also found the same patterns in the models of LRS score and overall respiratory symptom score.

FIGURE 2—

Adjusted mean symptom scores for upper, lower, and overall respiratory symptoms since Hurricanes Katrina and Rita among residents (N = 522), by extent of exposure to water-damaged homes: Orleans Parish, Louisiana, March 2006.

Note. Symptom scores were on the basis of the sum of reported symptoms (3 upper respiratory and 4 lower respiratory), each graded as follows: 0 = none; 1 = mild; 2 = moderate or severe. Exposure categories were defined as follows: 0 = had not been inside water-damaged home (n = 79); 1 = had been inside but had not participated in clean-up (n = 92); 2 = participated in clean-up, clean-up score of 1 (n = 91); 3 = participated in clean-up, clean-up score of 2, 3, or 4 (n = 171); participated in clean-up, clean-up score of 5 or greater (n = 89). Clean-up score was the sum of the number of homes cleaned that had fewer than 50% mold and twice the number of homes cleaned that had 50% or more mold. Means are least squares means, adjusted for gender, age, race, smoking status, atopy, and still participating in clean-up. P < .05 for test of linear effect for each symptom location.

Those who reported still participating in clean-up activities had higher symptom scores in the models of URS score (least squares mean of 2.5 vs 2.0; P< .05), LRS score (least squares mean of 1.8 vs 1.3; P= .01), and overall respiratory symptom score (least squares mean of 4.4 vs 3.4; P< .01) than did other participants (those no longer participating in clean-up activities and those who never participated in clean-up).

Models with participants stratified by asthma diagnosis showed similar patterns to those just described, generally reaching statistical significance for participants without asthma.

Exacerbation of Respiratory Symptoms

Among the 372 participants who had been inside a water-damaged home participating in clean-up, 113 (30%) described exacerbation of URS when inside such a home. Those who reported still participating in clean-up were more likely to report exacerbation of mild URSs (odds ratio [OR] = 3.28; 95% confidence interval [CI] = 1.49, 7.25) and moderate or severe URSs (OR = 2.05; 95% CI = 1.07, 3.93) when inside than those who were no longer participating in clean-up. There was no difference between those who were still participating in clean-up and those who were not in terms of URSs that did not get worse when inside. The extent of exposure to water-damaged homes was not significantly associated with exacerbation of URSs when inside.

Participants who used only a mask during clean-up were more likely to report exacerbation of mild URSs when inside a water-damaged home (OR = 5.92; 95% CI = 1.34, 26.14) than were those who used no respiratory protection (Table 2 ▶). Those who used any type of respirator and those who specifically used an N-95 FF respirator less commonly reported both mild and moderate or severe URSs that got worse when inside than those who did not use a respirator; these differences did not reach statistical significance. There was no association between respirator use and URSs that did not get worse when inside.

TABLE 2—

Effect of Using Respiratory Protection During Clean-Up of Water-Damaged Homes on Upper and Lower Respiratory Symptoms Among Residents (N = 351): Orleans Parish, Louisiana, March 2006

		Not Worse Inside		Worse Inside
	No.	Mild, AOR (95% CI)	Moderate/Severe, AOR (95% CI)	Mild, AOR (95% CI)	Moderate/Severe, AOR (95% CI)
Upper respiratory symptoms
Mask only	78	1.12 (0.37, 3.38)	3.32 (0.75, 14.72)	5.92* (1.34, 26.14)	1.92 (0.66, 5.59)
Any respirator	228	1.05 (0.53 2.09)	1.87 (0.86, 4.07)	0.62 (0.28, 1.37)	0.58 (0.29, 1.16)
N-95 FF respirator only	140	1.09 (0.53, 2.25)	1.21 (0.51, 8.87)	0.49 (0.20, 1.20)	0.51 (0.24, 1.09)
Lower respiratory symptoms
Mask only	78	3.40 (0.67, 17.14)	1.31 (0.34, 5.14)	1.51 (0.43, 5.36)	0.52 (0.17, 1.57)
Any respirator	228	1.57 (0.78, 3.18)	0.92 (0.40, 2.12)	0.42* (0.18, 0.97)	0.47* (0.22, 0.98)
N-95 FF respirator only	140	1.86 (0.88, 3.92)	0.95 (0.38, 2.36)	0.57 (0.23, 1.39)	0.33* (0.13, 0.83)

Note. AOR = adjusted odds ratio; CI = confidence interval; mask = noncertified dust mask or surgical mask; any respirator = certified respirator, including disposable N-95 filtering face-piece respirator and half-face and full-face reusable respirators; N-95 FF respirator = certified disposable N-95 filtering face-piece respirator. The total (N) is less than the 372 who reported participating in clean-up activities in a water-damaged home since Hurricane Katrina because of missing data. AORs and 95% CIs represent odds of reporting symptom grade and exacerbation type among those using noted respiratory protection compared with those not using noted respiratory protection during clean-up activities since Hurricane Katrina. AORs were adjusted for gender, age, race, smoking status, atopy, still participating in clean-up, and extent of exposure. Participants in the “mask only” category did not report using a respirator. Participants in the “any respirator” category reported using an N-95 FF respirator or a reusable respirator and also may have reported using a mask. Participants in the “N-95 FF respirator only” category reported using an N-95 FF respirator and also may have reported using a mask but did not report using a reusable respirator. For the “mask only” category, the comparison group consisted of those who reported using no respiratory protection (n = 45); for the “any respirator” and the “N-95 FF respirator only” categories, the comparison group consisted of those who reported not using a respirator (using no respiratory protection or using only a mask; n = 123).

*P ≤ .05.

Of those who participated in clean-up, 77 (21%) described exacerbation of LRSs when inside. Those who reported still participating in clean-up were more likely to report exacerbation of mild LRSs (OR = 2.42; 95% CI = 1.04, 5.60) and moderate or severe LRSs (OR = 1.99; 95% CI = 0.95, 4.16) when inside than were those who were no longer participating in clean-up. There was no difference between those who were still participating in clean-up and those who were not in terms of LRSs that did not get worse when inside. The extent of exposure to water-damaged homes was not significantly associated with exacerbation of LRSs when inside.

Participants who used only a mask during clean-up were less likely to report moderate or severe LRSs that got worse when inside a water-damaged home than were those who used no respiratory protection, but this difference was not statistically significant (Table 2 ▶). Those who used any type of respirator were less likely to report exacerbation of mild LRSs (OR = 0.42; 95% CI = 0.18, 0.97) and moderate or severe LRSs (OR = 0.47; 95% CI = 0.22, 0.98) when inside than were those who did not use a respirator. Those who specifically used an N-95 FF respirator were less likely to report moderate or severe LRSs that got worse when inside (OR = 0.33; 95% CI = 0.13, 0.83) than were those who did not use a respirator. We saw a similar pattern for N-95 FF respirator use and exacerbation of mild LRSs when inside, but this difference did not reach statistical significance. There was no association between respirator use and LRSs that did not get worse when inside.

DISCUSSION

Participation in this survey of New Orleans residents 6 months after Hurricanes Katrina and Rita was high, reflecting widespread concern about exposure to water-damaged homes.7^,10 Respiratory symptoms were common: two thirds of respondents reported experiencing at least 1 URS and close to half reported experiencing at least 1 LRS since the hurricanes. Although we know of no other population-based surveys of respiratory symptoms following hurricanes or floods, these findings are consistent with previous investigations using hospital-based surveillance, which have generally found respiratory illnesses to be among the most common diagnoses recorded postdisaster, including following Hurricanes Katrina and Rita.8^,14^–16

We found a strong association between respiratory symptoms and exposure to water-damaged homes. Whether measured by individual-period prevalence or aggregated score (an approach used successfully in a population-based study of asthma17), symptoms increased with exposure. These results corroborate a growing body of scientific evidence of an association between exposure to indoor dampness or mold and respiratory health effects.4^,6^,18^–20 Although tests of linearity were significant, the largest increases in mean symptom scores came between the first 2 exposure categories (category 0, those who had not been inside a water-damaged home, and category 1, those who had been inside but had not participated in clean-up).

Although not captured by the covariates or asthma stratification, it is possible that participants who had been inside but had not participated in clean-up were more susceptible to respiratory symptoms than were others. Such susceptible participants may have chosen not to participate in clean-up upon entering a home and experiencing symptoms, creating an effect similar to the “healthy worker” phenomenon in occupational cohort studies.21

Following the flooding in New Orleans, public health officials were obligated to make recommendations guided by little, if any, existing data on what activities would pose a health risk.8 Our results suggest that simply entering a water-damaged home postflood, even without participating in clean-up, poses a greater risk of respiratory symptoms than previously hypothesized. These findings should inform future recommendations, including those on respiratory protection, in the postflood setting.

Participants who were still involved in clean-up at the time of the interview more frequently reported exacerbation of respiratory symptoms. This finding was independent of the number of homes cleaned and may reflect recall bias. However, it is possible that still participating in clean-up served as a marker for an unmeasured factor. For instance, 6 months after the storms, the interiors of water-damaged homes may have been drier than earlier in the postflood period, promoting greater aerosolization of respirable particles and thus greater exposure.22 Furthermore, individuals who returned to New Orleans later and were thus still participating in clean-up may have had different susceptibility to respiratory symptoms than did those who returned and participated in clean-up earlier. Although we did not address such possible explanations directly, the association we found suggests that the health risk posed by participating in clean-up of water-damaged homes can persist many months after the inciting flood. This point is particularly important for public health officials and health care providers when advising the public and caring for symptomatic individuals months after a flood.

The recommendation to use respirators during clean-up activities was widely heeded. Of all persons surveyed, two thirds had participated in clean-up activities and more than two fifths had used a respirator during cleanup. On the basis of residential population estimates done 1 month before this survey,23 these findings suggest that the recommendation was relevant to approximately 105 000 adults in Orleans Parish alone, and that of those, approximately 65 000 followed the recommendation. We found that respirator use was negatively associated with exacerbation of respiratory symptoms when inside a water-damaged home, indicating a protective effect. For both URSs and LRSs, we saw this effect regardless of respirator category. OR estimates were less than 1 for both mild symptoms and moderate or severe symptoms that got worse when inside, and many reached statistical significance. For mask use, we did not find a consistent protective effect.

The ability of certified respirators to filter fungal particles has been investigated in previous experimental and field studies.24^–26 In a pilot project conducted in New Orleans from November 2005 to January 2006, researchers evaluated 2 respirators, concluding that the reusable respirator provided greater protection against fungal spores than did the disposable N-95 FF respirator.27 To our knowledge, our investigation is the first to find that respirator use (and specifically N-95 FF respirator use) during clean-up of water-damaged homes protects against respiratory symptoms. This finding is particularly notable given that we did not collect detailed information on respirator use. The participants categorized as using a respirator during clean-up were likely heterogeneous in the amount of clean-up time they spent without a respirator and in how well they followed manufacturers’ guidelines on respirator use (e.g., maintaining a good seal, replacing dirty filter material). This heterogeneity may have led us to underestimate the true protective effect of respirator use on symptom exacerbation.

Limitations

There are several limitations to our study. Assessments of respiratory health effects and exposure were limited to participants’ reports. Validations of questions on respiratory symptoms and diagnoses suggest that they have high specificity but lower sensitivity for conditions confirmed through testing, such as asthma and rhinitis.28^,29 Thus, some participants not identified by our questionnaire may have been found to have respiratory pathology with further testing. Residents’ reports of homes’ percentage of mold coverage were validated soon after the hurricanes via independent home inspections, suggesting such reports are reliable (M. A. R., PhD, MPH, unpublished data, 2005). However, we may have found a stronger correlation between exposure and symptoms by incorporating environmental sampling.

Another limitation is that we considered participants to have asthma only if diagnosed by a physician. The fact that no participant reported a new diagnosis of asthma since the hurricanes may reflect a lack of new asthma in the population. Yet at the time of our survey, the number of physicians practicing in Orleans Parish had fallen to less than half of the pre-Katrina number, and services at functioning medical facilities were limited.30 Had health care been more accessible following the hurricanes, it is possible that some participants would have been newly diagnosed with asthma, particularly among the 10% of those without an asthma diagnosis who reported experiencing at least 3 LRSs. Finally, the small numbers in some subgroups likely limited our ability to establish the statistical significance of some results.

Conclusions

Six months after Hurricanes Katrina and Rita, respiratory symptoms were common in New Orleans. Both URSs and LRSs were positively associated with exposure to water-damaged homes, even exposure limited to being inside without participating in clean-up activities. Respirators, including disposable N-95 FF respirators, had a protective effect against symptom exacerbation when inside a water-damaged home. Public health authorities and health care providers should consider advising the public to use respirators when inside water-damaged homes, regardless of activities undertaken while inside.