|Home | About | Journals | Submit | Contact Us | Français|
Objectives To assess the health effects of exposure to smoke fromthe fifth largest US wildfire of 1999 and to evaluate whether participation ininterventions to reduce smoke exposure prevented adverse lower respiratorytract health effects among residents of the Hoopa Valley National IndianReservation in northwestern California. Design Observational study:epidemiologists from the Centers for Disease Control and Preventionretrospectively reviewed medical records at the local medical center andconducted survey interviews of reservation residents. Setting HumboldtCounty, California. Participants Interviews were completed with 289 of385 residents, representing 26% of the households on the reservation. Of the289 participants, 92 (31.8%) had preexisting cardiopulmonary conditions.Results During the weeks of the forest fire, medical visits forrespiratory illnesses increased by 217 visits (from 417 to 634 visits, or by52%) over the previous year. Survey results indicated that although 181(62.6%) of 289 participants reported worsening lower respiratory tractsymptoms, those with preexisting cardiopulmonary conditions reported moresymptoms before, during, and after the smoke episode. An increased duration ofthe use of high-efficiency particulate air cleaners and the recollection ofpublic service announcements were associated with a reduced odds of reportingadverse health effects of the lower respiratory tract. No protective effectswere observed for duration of mask use or evacuation. ConclusionsTimely actions undertaken by the clinical staff of the local medical centerappeared beneficial to the respiratory health of the community. Futureprograms that reduce economic barriers to evacuation during smoke episodes mayalso improve intervention participation rates and decrease smoke exposures.Although promising, the effectiveness of these and other interventions need tobe confirmed in a prospective community intervention trial.
Community smoke exposures resulting from wildland forest fires have beenassociated with increased emergency department and hospital admissions forchronic obstructive pulmonary disease, bronchitis, asthma, and chestpain.1,2,3Although population expansion into wildland environments continues,interventions to prevent these smoke-related adverse health effects have notbeen validated under conditions of typicaluse.4
In 1999, the fifth largest wildfire in the United States burned from August23 to November 3 near the Hoopa Valley National Indian Reservation in northernCalifornia. On 15 days, smoke from the fire produced ambient concentrations ofparticulate matter (PM10) on the reservation that exceeded the USEnvironmental Protection Agency's 24-hour air quality standard of 150μg/m3 of air. On October 21 and 22, particulate concentrationsexceeded the agency's 24-hour hazardous level of 500 μg/m3.
Concern over the health effects of the smoke prompted local officials andmedical officers to implement several interventions to reduce smoke exposurein the community. However, medical personnel were frustrated over the lack ofa scientific basis that could have been used to set public action levels or torecommend appropriate precautionary measures during this emergency. As aresult, on November 5, 1999, the Hoopa Valley Tribal Council requested thatthe Centers for Disease Control and Prevention (CDC) assist them inretrospectively assessing local adverse health effects and evaluating theinterventions that were implemented during the smoke episode.
A community survey was completed by 289 of 385 (75.1%) selected residents.We oversampled persons with preexisting cardiopulmonary conditions byattempting to interview all who were treated at the reservation medical centerin the past year for coronary artery disease, asthma, chronic obstructivepulmonary disease, or other lung diseases (n = 92). We also interviewed onerandomly sampled person per household who did not have any preexistingconditions, from 197 randomly sampled households. These persons represented26% of all of the tribal households on thereservation.5
The survey instrument included questions about family demographics,intervention participation, and lower respiratory tract symptoms linked withforest fire smoke exposures elsewhere(table1).4,6,7The respondents self-reported the frequency of chest pain, breathingdifficulty, and cough on a Likert scale (1 = never to 5 = always) for threetime periods: before the smoke episode began (which serves as a baseline),during the smoke episode (August 23-October 26, 1999), and after the smokeepisode ended (October 27-November 15, 1999). For each of these periods,individual symptom frequencies were combined to form overall respiratorysymptom scales. Two dichotomous outcomes reflect whether respiratory symptomsincreased in frequency (hereafter referred to as “became worse”)from before to during the smoke episode and from before to after the smokeepisode.
We compared weekly counts of medical visits for any respiratory problem(International Classification of Diseases, Ninth Revision[ICD-9],8 codes460-519) during August 14 through November 4, 1999, with weekly averagePM10 concentrations for the same period. We then compared thesedata with similar data from 1998 when no fires were burning.
During the fires, the staff of the local medical center and other tribalorganizations implemented several interventions:
Because of resource constraints, the distribution of free hotel vouchersand HEPA cleaners was prioritized to persons who had adverse health effectsduring the smoke or who had been treated within the past year for any of thepreexisting conditions.
Logistic regression analyses were used to examine the relationship betweenthe duration of intervention participation and the odds of reporting worseningrespiratory symptoms from baseline to the postfireperiod.9 Thepostfire period was the only time when we could be certain that reportedsymptoms occurred after participation in any interventions. Because reportedrespiratory problems were, by definition, positively correlated withintervention participation, it was necessary to evaluate interventions bylooking for dose-response relations within groups of persons who received eachintervention.
During the smoke episode in 1999, medical visits for respiratory problemsincreased by 217 (from 417 to 634 visits, or by 52%) over the previous year.The proportion of all visits for respiratory problems increased from 8.9% (95%confidence interval, 7.5%-10.3%) in September 1998 to 11.9% (10.4%-13.4%) inSeptember 1999, from 10.7% (9.1%-12.3%) in October 1998 to 19.2% (17.2%-21.3%)in October 1999, and from 13.8% (9.4%-18.2%) during the first week of November1998 to 19.5% (15.2%-23.8%) during the first week of November 1999. WeeklyPM10 concentrations were considerably higher in 1999 than in 1998and were positively correlated with the weekly number of patients presentingto the facility with respiratory illnesses (Pearson's correlations werer = 0.74 in 1999, and r = -0.63 in 1998) (Figure).
More than 60% of respondents (178/289) reported increased respiratorysymptoms during the smoke (table1). Two weeks after the smoke cleared, more than 20% (65/289)continued to report an increased frequency of respiratory symptoms overbaseline levels. Respondents with preexisting conditions reportedsignificantly more symptoms before, during, and after the smoke than others inthe community (table 1).
Of the 289 respondents, 140 (of 287 who answered the questions, or 48.8%)evacuated to a location off the reservation, 100 (of 286, or 35.0%) wore amask or face covering, and 98 (of 287, or 34.1%) ran a HEPA air cleaner intheir home at some time during the smoke. Persons with preexisting conditionswere significantly more likely to use a HEPA cleaner and marginally morelikely to evacuate the reservation, reflecting the selective targeting ofthese interventions to those with cardiopulmonary problems(table 1).
Among the evacuees, the mean duration of evacuation was 7.6 days, and 17.1%(22/129) were away from the reservation during each of the 3 days with thehighest PM10 concentrations (October 18, 21, and 22). Amongrespondents who ran a HEPA air cleaner in their home, the mean duration of usewas 14.9 days, and 48.8% (42/86) ran them during each of the 3 days withhighest PM10 concentrations(table 2). More than 80% ofrespondents (238/289) were able to correctly recall a PSA without being showna list of known PSAs. Of these, 66.0% (157/238) reported taking action toreduce smoke exposure as a result of hearing the PSA. “Staying insidemore often” was the most common action undertaken(table 2).
The duration of evacuation and mask use were not significantly associatedwith the odds of reporting worsening lower respiratory tract symptoms (table3). In contrast, odds ratios (ORs) associated with three measures of HEPAcleaner use indicated significant negative associations. Among those who ranHEPA cleaners in their home, increased duration of use was significantlyassociated with decreased odds of reporting worsening respiratory symptoms (OR= 0.54) (see table 3). This association followed a dose-response relation.Those in the highest quartile of duration of use were significantly lesslikely than those in the lowest quartile of duration of use (the referencegroup) to report worsening symptoms.
Respondents recalling a PSA were less likely than those who could notrecall a PSA to report worsening respiratory symptoms (OR = 0.25) (see table3). The number of PSAs recalled was also protective in a dose-response manner.Of the actions taken in response to hearing PSAs, only “staying insidemore often” displayed any trend toward protection.
One of the challenges health professionals face when forest fires threatentheir communities is to implement effective preventive measures when noguidelines exist for protecting the public in thesesituations.4,10,11As a result, a goal of this field study was to assess the effectiveness ofinterventions that were rapidly deployed by medical staff of the Hoopa ValleyNational Indian Reservation during an air-quality emergency.
We found mask use to be ineffective and positively associated with outdoorexposure. This finding may be explained by respondent tendencies to use masksinconsistently, without appropriate fit-testing, or by the variable filtrationeffectiveness of the masks used in thissituation.12,13,14We were also surprised that increased duration of evacuation did not appear tobe protective. However, smoke exposures from wildland fires are oftenunpredictable,15and among those who evacuated, only 17.1% were away from the reservationduring each of the 3 days with the highest smoke concentrations. In contrast,half of those who used HEPA cleaners ran them during these days, and the meanduration of HEPA cleaner use was twice as long as the mean duration ofevacuation.
The clean air delivery rate (CADR), measured in cubic feet per minute(cfm), is a function of a HEPA cleaner's efficiency of pollutant removal andrate of airexchange.4 TheAmerican National Standards Institute has approved a standard for air cleanersto have a CADR of 100 cfm for a 12- by 12-ft room and a CADR of 250 cfm for a20- by 20-ftroom.16 The HEPAcleaners provided to the population by the Hoopa Valley Tribal Council had aCADR of 150 cfm, suggesting that they would have been suitable for most roomsin the small single-family dwellings on the reservation. Although the use ofportable HEPA air cleaners has previously been reported to reduce theconcentration of fine particles indoors to an acceptable level during smokeepisodes,4 thesefindings provide additional support for their effectiveness because increasedduration of use significantly reduced the odds of reporting worseningrespiratory symptoms.
Economic and occupational barriers to leaving home may have dissuadedreservation residents from evacuating. When asked why they chose not toevacuate to a hotel, 45% of the responses of those who did not evacuateindicated an inability to take time away from work. An additional 12% citedeconomic constraints. In a locale with a 32% unemployment rate, the forestfires brought economic opportunities. This may be best illustrated by thefinding that among working-age adults, residents with preexistingcardiopulmonary conditions were as likely to work for salary in the fire campsduring the fires as others in the community. Future interventions that involvethe temporary relocation of residents may need to consider the positiveeffects of employment (and associated disincentives to evacuation) in order tobe implemented effectively.
Existing research has generally indicated that, to be most effective, therelease of PSAs should be coupled with other behavioralinterventions.17,18,19A growing body of literature, however, has suggested that PSA campaigns alonecan have significant effects on public healthbehaviors.20,21,22,23In this situation, the timely dissemination of PSAs through radio broadcastsand telephone messages was associated with a reduction in reported respiratorysymptoms in large segments of the general population. Whereas some evidencesuggests that the PSAs may have produced this effect by influencing recipientsto remain indoors, their mechanism of effect appeared mostly unmeasured by oursurvey instrument and remains an area requiring additional investigation.
In a situation where resources were limited, the interventions wereappropriately prioritized to persons with preexisting cardiopulmonaryconditions, who reported more severe respiratory problems at all time pointsin the study. The effect of each of the four interventions was of similarmagnitude in those who had and those who did not have preexisting conditionsand was also independent of the respondent's participation in any of the otherinterventions. However, a limitation to this study was the potential forrecall bias. To reduce this, respondents were asked to report symptoms foreach of three separate time periods rather than their own change in symptomsover time. Intervention dissemination was also not randomized and wasconfounded by the severity of lower respiratory tract condition. To reduce theeffects of this confounding, several interventions needed to be evaluated byexamining dose-response relations among those who received them.
As population expansion into wildland environments continues, local healthprofessionals will be repeatedly faced with the challenges of making rapiddecisions to protect their communities from forestfires.10,11,24,25,26Until randomized trials can be undertaken, these findings provide some initialguidance to the effectiveness of several possible community interventions.
Data collection efforts was made possible by Michael Lipsett, SvetlanaSmorodinsky, Kirsten Knutson, Aman Kaur, and Jane Riggan of the CaliforniaState Department of Health Services; Mark Fluchel and Chad Schaban, who werein training programs at the CDC, Atlanta; and Darla McCullagh, RobertUlibarri, Lucky Guyer, and the Humboldt County, California, health officerswho worked on the Hoopa Valley National Indian Reservation.
Competing interests: None declared