Marilyn L. Browne, Carole L. Ju, Gregg M. Recer, Lee R. Kallenbach,
James M. Melius and Edward G. Horn
although the counts of A. fumigatus spores measured in air in the community near the
composting facility were somewhat higher than those for the reference community
(Syzdek and Haines 1995; Recer et al., 2001), increased rates of health symptoms in re-
sponse to elevations in A. fumigatus counts were not demonstrated in this study.
An important limitation of the study is that data providing reliable estimates of
average daily A. fumigatus spore counts were only available for approximately one-
quarter of the original study period. The sharp short-term fluctuations observed in
spore levels were not anticipated. Modifying the spore-count method to average over
these short-term fluctuations reduced the sensitivity of the original approach for re-
lating daily spore exposure to symptom incidence. Also, fungal spore counts at the
sampling locations are probably only an approximate measure of individual exposure
to fungal spores, which may have further limited our ability to detect associations. In-
door spore counts may differ significantly from those measured outdoors, and many
people spend most of their time indoors. In addition, other indoor and outdoor envi-
ronments away from home may have an effect on symptom occurrence.
It was our intent to have each subject act as his or her own control by using a
method of data analysis which produces a summary estimate based on separate logis-
tic regression of data for each subject in the study (Korn 1979). Losses due to dropouts
from the study and gaps in the data due to incomplete or missing diaries did not per-
mit the use of that method of data analysis. For Poisson regression analysis of incidence
rates, participants do not serve as their own controls. Therefore, there is the potential
for differences in individual susceptibility to confound the relationship being evaluat-
ed. While participants do act as their own controls in the paired t-test of differences in
symptom prevalence, data sparseness limited that analysis to 12 days and 48 partici-
pants. The t-test analysis had a 63% to 91% probability of correctly identifying in-
creases of 10% to 15%, respectively, in the frequency of allergy or asthma symptom re-
porting on high A. fumigatus days versus low A. fumigatus days. Thus, while it is likely
that a true increase in symptom frequency of 15% or more would have been detected,
a 10% or smaller increase could have been missed.
Despite the limitations described above, it is interesting to note that symptom in-
cidence was associated with ragweed, ozone, temperature and time since start of the
study. This was true for participants living near the composting facility and, to a less-
er degree, participants from the reference community. The presence of a dose-response
trend suggests that changes in one or more of these variables are having an effect on
symptom incidence. If ragweed, ozone and temperature are independently related to
symptom incidence (and not just in association with time since start of the study), our
ability to assess the effect of environmental influences on allergy and asthma symp-
tom occurrence would be supported. However, time since start of the study was
strongly correlated with ragweed, ozone and temperature and could provide an alter-
native explanation for the decrease in symptom incidence observed over the course of
the study. A general tendency to report fewer symptoms as a study progresses, termed
a “fatigue effect,” has been noted in other studies using diary data (Abramson 1990).
Separating the relative importance of the environmental factors and time since start of
the study was not possible due to their strong intercorrelations.
The results of this study suggest that if elevated concentrations of A. fumigatus
spores due to operations at the composting facility are causing increases in allergy and
asthma symptom prevalence, the increase was too small for us to detect given the study
limitations. We propose that future studies employ more objective measures of im-
mune response and respiratory function. Also, less labor-intensive exposure measures
that can reliably assess individual daily bioaerosol exposures would be desirable.
248 Compost Science & Utilization
Summer 2001