INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A number of investigators have determined links between asthma symptoms/admissions and particular atmospheric situations. Weather has a dual role, affecting the asthmatic through direct and indirect means. Among direct influences are rapid meteorological fluctuation, such as the onset of cold weather in the fall, which appears to be associated with decreased lung function in asthmatics (1). Increased visits to New York City emergency rooms were found to be associated with decreases in temperature, especially during and after the first and second onset of cold periods in the fall; another study incorporating a subjective identification of weather events found nearly all asthma epidemics in both New Orleans and New York City were preceded by the passage of a cold front followed by a high pressure system (4). Asthmatic behavioral responses are an important indirect effect of weather. Increases in emergency clinic admittances during the first onset of cool or cold weather in fall often occurred in United States cities with the first seasonal use of indoor heating (4, 7). This seasonal increase could be related to the transport of dust particles accumulating in heating systems used for the first time since the previous spring (7).

These studies use simple regression and/or comparative techniques, and there is a notable lack of sophisticated climatological modeling. The goal of this research is to utilize recently-developed procedures which evaluate this simultaneous impact of the entire suite of meteorological elements on the asthmatic. This study incorporates a "synoptic climatological approach" to categorize daily weather into air mass types, which are homogeneous bodies of air with distinct thermal and moisture characteristics (8). While the synoptic methodology employed here is site-specific, similar categories can be identified at different locations. This procedure has been used primarily by climatologists in a number of weather/health evaluations to identify unhealthful weather situations and the individual parameters within them that contribute to health problems (9). Considering the findings of previous researchers, this study makes an a priori assumption that certain air masses, such as those possessing cold temperatures and high barometric pressure in the fall, will be highly correlated with increased asthma admissions. However, it is also expected that other air mass types will be responsible for increasing admissions, and the use of a synoptic climatological approach will contribute considerable insight to an array of asthma/weather interactions.

Certain air pollutants have been shown to be associated with increased asthma admissions to hospitals (12), and weather can have an indirect effect on these admissions by affecting the atmospheric concentration of such pollutants. It is plausible that stagnant weather conditions permit the buildup of atmospheric pollutants which exacerbate the asthmatic condition (7). The synoptic climatological approach used here permits an evaluation of the influence of atmospheric pollutants on admissions within each weather situation. Another potential indirect effect of weather on admissions involves air-borne allergen concentrations; again, stagnant weather conditions may contribute to high spore concentrations. Several studies have reported on the coincidence of asthma epidemics with thunderstorms (13), and the unhealthful allergen conditions that result.

The synoptic climatological procedure appears well-suited to evaluate asthma hospital admission variation as impacted by the various factors discussed above.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Individual asthma admissions data for all New York City SMSA hospitals from 1982 to 1992 are provided by the New York State Health Department. Only those cases severe enough to require an overnight stay are included in the data base, and daily counts of total asthma admissions are determined. Due to the increasing number of admissions through the period of record for New York City, a standardization based on the trend of daily admission means for each year is performed. Next, a Fourier transform filter is applied to the daily 11-yr admission means to assist in removing daily and seasonal noise (Figure 1). This transformation removes a number of sample irregularities, such as the reported increase in admissions on Sundays and Mondays (18). The daily values from the resulting baseline (the solid line in Figure 1) are then compared to the corresponding values within each year, producing a daily deviation value. This deviation value is free of any confounding by season. "High admissions days" are defined as those with admissions values exceeding 1.96 standard deviations from the daily baseline.

View larger version (37K): [in this window] [in a new window]   Figure 1.   New York City 1991 asthma admissions versus period of record daily averages.

First order weather station data are obtained from the National Climatic Data Center, and are used in a synoptic climatological procedure known as the Temporal Synoptic Index (TSI). The TSI is run on a seasonal basis due to the relatively greater climatic homogeneity present within each season. Thus, air masses are identified separately for winter (December, January, February), spring (March, April, May), summer (June, July, August), and fall (September, October, November). The procedure uses observations of six separate meteorological variablesair temperature, dew point, cloud cover, atmospheric pressure, wind speed, and wind directionmeasured four times daily, to determine air mass characteristics for each day at a particular location. To ensure linkage between weather and the hospitalization data, a "day" is defined as the 24-h period from midnight to midnight of the next day. Principal components analysis and an average linkage clustering procedure are used to identify days with similar meteorological character, and to sort them into homogenous groups, such as "continental polar" and "maritime tropical" air masses, which represent the "umbrella of air" that the asthmatic experiences (refer to Kalkstein, Tan, and Skindlov [19] for a detailed discussion of the development of TSI). Mean meteorological values for the derived air mass categories can then be calculated (Table 1 shows fall air masses). While the air masses are site-specific, similar categories at different locations can often be identified by comparing these mean values.

Two air mass/asthma admissions analyses are performed. First, the daily mean admissions deviation value from the baseline is determined for each air mass for a lag of 0 to 3 d, and those air masses with high deviation values are identified (because of the exploratory nature of this work, a single optimal lag day was not selected). Second, we calculate the percentage of time each air mass is present during high admissions days. A ratio value of air mass presence on high admissions days versus daily frequency of the air mass is calculated; the typical ratio is 1.0. For example, if air mass 1 occurs on 50% of high admissions days, but typically occurs 25% of the time during the season, the ratio value equals 2. "High risk" air masses are defined as those with ratios statistically significant at the 0.05 level using a chi-square test.

Once the high risk air mass is identified, the approach allows for determination of those particular factors within the air mass most responsible for the increased admissions. These could be meteorological (e.g., cold temperatures) or non-meteorological (e.g., air mass duration, accounting for several consecutive days of the same air mass). This is achieved using within-air mass category regressions to determine which characteristics of the air mass, if any, are correlated with increased admissions. Independent meteorological variables used in the regression model are maximum and minimum dew point and temperature, 3:00 P.M. wind speed and cloud cover, and average daily pressure. Nonmeterological variables include day in season (did the day occur early or late in the season?), number of consecutive days in the high risk category, and daily means of various air pollutants.

Air pollution data for the New York City metropolitan area are taken from the Environmental Protection Agency's Aerometric Information Retrieval System (AIRS). We focus on four pollutants reported to be associated with asthma: nitrogen dioxide (NO₂), sulfur dioxide (SO₂), ozone (O₃), and particulate matter with a diameter of less than 10 µm (PM₁₀). PM₁₀ is measured once every 6 d, and the other pollutants are measured hourly, Hourly measurements are averaged over the 24-h period of each day to obtain daily values. Based on suggestions from the EPA (20), a single monitoring site within the city center with the most complete record is used. To fill in any missing data at these sites, correlation analysis is performed on a site-by-site basis to obtain the most accurate substitutes. Pollutant counts are divided into quintiles for each pollutant, and asthma admission deviation means are calculated for each.

The synoptic approach has been used to identify those air masses associated with the highest air pollution concentrations at a particular location (21). Here, concentration means are calculated for each air mass, permitting determination of which air masses possess high atmospheric pollutant levels. To investigate whether high levels of pollutants within offensive air masses act as a trigger for daily asthma admissions, within-high risk category pollution quintiles are developed for all seasons.

Six pollen variablestotal pollen, tree pollen, total weeds, ragweed, fungi, and grassesare evaluated from spring through fall during the period 1991-1992 and are obtained from the Asthma and Allergy Research Center, Newark, NJ. Due to the short period of record, only a cursory data analysis is performed. Calculations of allergen averages by air mass category and total non-standardized asthma admissions by allergen quintile are performed.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The expected seasonal variation is apparent in the admissions data, with a summer minimum followed by a sharp increase and maximum in fall for New York City (Figure 1). Admissions are moderately high through the winter period, and are quite variable in spring before decreasing into summer. This abrupt rise occurs around the same time each year, in late September and early October. The average weekly admission count more than doubles from about 400 in early September to nearly 850 in early October for New York City. However, the magnitude of this increase varies significantly from year to year.

Synoptic Climatology Results

Certain air masses are clearly associated with days possessing high numbers of asthmatic visits to hospital emergency rooms (Table 2). The highest mean daily deviations in admissions are found among air masses in the fall, and the coldest and driest fall air mass, f8, exhibits a very strong positive admissions relationship. One day after this air mass was present, a mean of 7.1 admissions above the standardized daily average occurs, and the ratio value is statistically significant at lags of 1 and 3 d. Although f8 occurred on 19.0% of all fall days, it accounts for 44.1% of all high admissions days with a one day lag, yielding a ratio value of 2.3. In addition, when f8 is present over a consecutive day period, mean excess admissions increase further. This continental polar weather situation is characterized by little cloud cover, light northwesterly winds, and high sea-level pressure (Table 1). F2, representing a post-cold front situation during fall, also has a statistically significant ratio value.

In winer, two temperate air masses (w2, a continental polar air mass, and w5, a cold modified continental air mass with zonal flow aloft) and a strong cold front passage (w10) are significantly associated with increased asthma admissions. These are among the coldest winter air masses identified; however, the coldest air mass, one of pure arctic origin, is not among the high risk groups. Although exhibiting a lower admission deviation than offensive winter and fall air masses, a spring maritime tropical air mass (sp4) possesses a statistically significant ratio value. Sp4 represents the warmest spring air mass, exhibited high dew points, and frequently occurs when the region is in the warm sector of an approaching low pressure system. No summer air masses have statistically significant ratio values.

The synoptic procedure is a better method to comprehend asthma/weather relationships than nondiscriminatory multiple regression procedures, which use temperature and other meteorological factors for all days within the season as independent variables. For example, by isolating days which occur during fall air mass f8, it is possible to determine much more about potential meteorological/asthma admissions links than by a standard multiple regression procedure for all days during the season using meteorological factors (Table 3). Results from the standard regression procedure are very weak, while the within-air mass results for f8 are considerably stronger. Thus, the identification of a high risk air mass, such as f8, may be considered a means to isolate meteorological conditions which frequently exceed the tolerance of many asthma sufferers, and this could assist in the development of a procedure to predict asthma admissions.

Table 3 suggests that not all days within the fall high risk air mass are associated with elevated admissions, and this variation within the air mass is related to several factors which significantly impact asthma admissions. Therefore within-category multiple regressions similar to the f8 example in Table 3 are developed for each high risk air mass for each season. For example, for fall air mass f8, the inverse relationship with day of season indicates that this air mass has more of a direct impact on asthma admissions if it occurs earlier in the fall season the first or second time the asthmatic experiences cold air. In addition, minimum temperature is inversely related to asthma hospital admissions within air mass f8. Thus, colder overnight temperatures are associated with increased hospital admissions. This model explains over 20% of the variance in asthma admissions when f8 is present, and those factors within the high risk air mass which can contribute to higher hospital admissions are isolated.

A variety of statistically significant factors are found using these regressions for high risk air masses during the other seasons, including direct relationships for cloud cover and maximum temperature within winter category w10, and consecutive day duration of category sp4 during spring (for a more detailed discussion on these regressions, refer to Jamason, 1996 (22). For all seasons, the within-air mass category regressions produce better model R² values than the non-discriminatory seasonal regressions.

Air Pollution and Pollen Results

The impact of pollutant concentration within high risk air mass categories varies considerably with season. It appears that, during spring and summer, the high risk categories are associated with high concentrations of various pollutants (Table 4). For example, of all spring air masses, high risk air mass sp4 possesses high concentrations for many of the pollutants evaluated. Although high risk air mass su4 is not the most polluted for the season, it demonstrates among the highest summer concentrations for PM₁₀ and O₃. The importance of pollution on asthma admissions within these air masses is more clearly expressed when pollutant concentrations are subdivided into quintiles (Figure 2). For example, for su4, it is apparent that mean daily admission deviations increase dramatically for four important pollutants as concentration quintiles increase. Possibly the most dramatic rise is associated with NO₂, where asthma admissions are close to the daily mean for the lowest concentration quintile, and rise to about 12 above the mean for the highest quintile. It should be noted that in general, pollution concentrations are highest in summer.

View larger version (31K): [in this window] [in a new window]   Figure 2.   New York City summer air mass su4 air pollutant quintile admission means.

These relationships contrast sharply with those obtained through a pollution evaluation of winter and fall air masses. The high risk air masses for these seasons are associated with comparatively low concentrations of pollutants relative to season, and it appears that pollution concentration has little impact on asthma admissions during winter and fall (Table 5, Figure 3). The three winter high risk air masses (w2, w5, w10) are comparatively "clean," and possess concentrations below the overall mean of all air masses, with the exception of ozone concentration for w10. Even this "high" ozone concentration is less than one-third that of the high risk spring and summer air masses. In fact, particulate concentrations for w10 are the lowest for any winter air mass, and among the lowest for air masses of all seasons. The highest risk air mass of all, fall air mass f8, demonstrates a similar pattern. It contains concentrations below the seasonal mean for every pollutant with the exception of SO₂. Ozone concentrations are among the lowest for all fall air masses. A pollution quintile evaluation of f8 is even more instructive (Figure 3). In contrast with the spring and summer high risk air masses, there is no systematic rise in asthma hospital admissions as pollutant quintiles increase. This important inter-seasonal differential in pollution impact within the high risk air masses suggests that the environmental factors which contribute to high admissions can be isolated from season to season using a synoptic climatological approach.

View larger version (40K): [in this window] [in a new window]   Figure 3.   New York City fall air mass f8 air pollutant quintile admission means.

While noting the brief nature of the data set, none of the fall high risk air masses are associated with unusual pollen concentrations. In addition, the highest ragweed concentrations occur well in advance of the sharp asthma hospital admissions increases for the years 1991 and 1992 (Figure 4). It is unlikely that high ragweed concentration several weeks before the fall rise can be considered an important causal mechanism. In spring, a slight increase in asthma admissions occurs near the time of elevated total pollen concentrations during 1991 and 1992, but these admissions increases are much less dramatic than those in fall. It should be noted that high risk air mass sp4 possesses the highest fungi concentrations. Thus, it appears that pollen concentration plays little role in the observed asthma increases, but it must be noted that the limited pollen dataset renders this finding far from conclusive.

View larger version (32K): [in this window] [in a new window]   Figure 4.   1991 pollen concentrations versus daily raw asthma admissions.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In New York City, relationships between asthma admissions, weather, and air pollution are obviously distinctive among the seasons. Fall and winter high risk air masses clearly have a stronger relationship with asthma admissions than their spring and summer counterparts. In addition, pollution concentration has little influence on admissions within the fall and winter high risk air masses, but is considerably more important in spring and summer. Specifically, the most important high risk air mass during the entire year (based on deviation in admissions from the mean) is fall air mass f8. A cold, clear air mass such as f8 often represents the first or second influx of cold weather into a region during the fall season. This may have both direct and indirect effects on the asthmatic. First, the cold air itself has been shown to induce asthma (1), and this may directly increase admissions to hospitals when this air mass occurs. Second, the typical behavioral response to such an event is for individuals to turn on an indoor heating system which has been dormant and collecting allergens since the previous spring, sending large quantities of dust, mites, mold spores, and other materials into a sealed environment (7). The cold air might force the asthmatic indoors, where these various indoor allergens exist, which no doubt helps explain the high admission rates for windy and cold winter air masses. In addition, the most robust air mass/asthma relationships for fall are found with a lag time of 1 to 3 d, suggesting that prolonged exposure to allergens, rather than an immediate response to weather conditions, may be responsible.

Warm and moist air masses are weakly associated with increased admissions in spring and summer, although the air pollution concentration within these air masses seems more important than the weather itself. This suggests that asthmatics respond to increased pollution levels to a greater extent during these months of the year than in fall and winter. One possible explanation is that asthmatics are outside more often in these seasons, resulting in greater exposure to harmful levels of pollutants. In addition, the pollution analysis findings indicate most air pollutants exhibit their highest yearly concentrations in summer, and this may exceed some response threshold for asthmatics. The synoptic approach allows weather to be controlled for, such that the relationship between various air pollutants and asthma admissions can be analyzed in greater detail in future studies.

A significant indirect role of weather on asthmatics is the ability of certain weather situations to permit the buildup of outdoor allergens and pollutants. Although moist conditions are usually associated with cleansing the atmosphere through rain, they also foster the growth of fungal and mold spores. Therefore, it is not surprising that in New York, high risk air mass sp4, a wet, tropical air mass, possesses the highest fungi concentrations. In other seasons, outdoor allergen/air mass relationships are less clear, possibly due to the short period of record used for the former.

Air pollution and weather appear to have a limited synergistic effect on asthma admissions. However, summer shows some weak synergism, as higher admission values occur when both high levels of NO₂ and SO₂ are coupled with the presence of hot, stagnant air (air mass su4). It appears that fall and winter asthma admissions are not related to pollutant concentrations, as the high risk air masses during these seasons are not among the most polluted.

There is a noteworthy interseasonal differential response to weather, air pollution and outdoor allergens by asthmatics. A large number of indoor and outdoor environmental parameters discussed above are known to be causative factors in precipitating asthma attacks, and do so through different physiological responses. For example, the physiological mechanisms involved in allergen-induced asthma differ markedly from those of cold-air induced asthma (1, 18, 23, 24). It seems plausible that some threshold is exceeded within each of these seasons (i.e., changes to, and the presence of, cold air in fall and winter; excessive air pollution and outdoor allergen concentrations in spring and summer), and as a result, a seasonally distinct physiological mechanism within the asthmatic triggers an asthma attack. Therefore, different environmental parameters initiate asthma attacks through their own respective biological triggers, and this may help explain the variety of offensive weather, air pollution, and outdoor allergen variables that are unhealthful to asthmatics in different seasons.

Conclusions

The use of a synoptic climatological approach in a study of this type offers a concise, holistic method to determine the role of climate, and enables an evaluation of the potential synergistic role of air pollutants and outdoor allergens on asthma admissions to New York City hospitals. The a priori hypothesis that cold fall air masses possessing high pressure are associated with increased admissions to hospitals is confirmed. These results are specific to the location analyzed, and temporal or spatial cross-validation is required to conclude whether the above findings are valid for other cities. However, it is clear that the determination of realistic combinations of atmospheric components, through the identification of air masses, allows for the identification of high risk atmospheric conditions. These possess numerous meteorological and non-meteorological elements that act on the asthmatic to produce stressful conditions.

While the findings of this study need to be verified at other locations, this approach lends itself to the possibility of developing a watch/warning system, which may alert asthmatics and public health officials to the impending arrival of an air mass which has been historically associated with increased asthma hospital admissions. With present-day weather forecasting technology, it is possible to predict the arrival of a high risk air mass up to 48 h in advance. Thus, susceptible individuals and interested health officials could be provided with sufficient notice to mitigate the possible impact of the high risk situation. A similar system for heat-related mortality has already been developed, using a synoptic approach, for the city of Philadelphia. The Philadelphia hot weather/health watch-warning system was instituted in June, 1995, and has been used by the Philadelphia Department of Public Health to warn people of conditions which have been associated with increased heat- related mortality (10). This system, through greater public awareness of excessive heat conditions, may have played an important role in reducing Philadelphia's total heat-related deaths during the summer of 1995. There are plans to expand the system to other cities in subsequent summers, and to develop a national system for major cities vulnerable to heat- related morbidity and mortality. Considering the results described here, such a system could be of considerable value to asthmatics who are trying to avoid situations where hospitalization might be necessary. Of course, further research is needed to determine whether the results of this study can be replicated at other locations, and although the analyses here are encouraging, much more must be learned before a feasible asthma/weather watch-warning system can be implemented.