Results

Soybean allergen levels (table 1) in air filters collected during 1995–6 on soybean unloading days were consistently low, being below the detection limit on 10 days with the highest level being 268 U/m³, and were not significantly different from levels on non-unloading days (p = 0.47). The levels on both unloading and non-unloading days were similar to those recorded during the period 1988–9 after the installation of bag filters at the top of the responsible silo.

Of a total of 316 asthmatic patients surveyed in 1989, 31 had died by 1995 (18 (10.6%) with epidemic asthma and 13 (8.8%) with non-epidemic asthma) and 29 had moved to another geographical area (12 (7.1%) with epidemic asthma and 17 (11.5%) with non-epidemic asthma). One subject with epidemic asthma and two controls died from asthma. A total of 203 (79%) of the 256 eligible individuals were studied; there were no statistical differences in participation rates between epidemic and non-epidemic asthmatic subjects (p = 0.28). The characteristics of both groups in 1995 are shown in table 2. The subjects with epidemic asthma were older and there were fewer women (p<0.05). No differences were observed in current smoking status, the percentage predicted basal FEV₁, bronchial responsiveness, or use of inhaled steroids. Since a close similarity between both groups was also observed in 1989,11 no further analysis on lung function and bronchial responsiveness was carried out. Epidemic asthma patients showed a trend to be more atopic with a higher geometric mean total IgE (p<0.05), although skin reactivity to common aeroallergens was similar (sensitisation to at least one allergen 60.9% versus 53.7%, p = 0.43).

A similar proportion of subjects with soybean asthma who reported symptoms in 1989 and controls reported the absence of symptoms in 1995 so, for most symptoms, the relative risks (RRs) of improvement were near to 1 (table 3). The only statistically significant differences were for being woken up by attacks of coughing (RR improvement 0.47; 95% CI 0.22 to 0.99) and for visits to the emergency room (RR improvement 0.63; 95% CI 0.41 to 0.96). Accordingly, when subjects not reporting a given symptom in 1989 were considered, soybean asthma subjects were slightly more likely to report the symptom in 1995 but none of the relative risks for worsening achieved statistical significance (table 3).

Because of limited availability of serum samples a paired comparison of changes in sensitisation to soybean between 1995 (kU/l; CAP technique) and 1989 (U/ml; RAST technique) was only possible in 52 subjects with soybean asthma and 52 controls. Of the patients with soybean asthma 23 had measurable levels of soybean IgE (⩾0.35, both U/ml and kU/l) on both occasions while 27 did not. Only one reverted and one converted. Of the 52 controls 48 were negative on both occasions, one was positive, two converted, and one reverted.

Discussion

In a previous study we found that, two years after an intervention which eliminated the occurrence of outbreaks of soybean asthma, about 50% of subjects affected by the outbreaks showed measurable levels of specific IgE to soybean compared with 74% before the intervention,10 subjects with epidemic asthma having fewer asthma symptoms than controls.11 In the present study, eight years after the intervention, symptoms of asthma were more likely to improve than to worsen in both study groups but subjects with epidemic asthma were less likely to improve than controls, the latter being statistically significant only for being woken up by coughing and for visits to an emergency room. Of those with epidemic asthma, 55% were still sensitised to soybean whereas lung function and bronchial responsiveness to methacholine were similar in both groups in 1989 and in 1995. These results strongly suggest an initial improvement in epidemic asthma in the two years following the intervention with no further improvement in subsequent years. It should be emphasised that the findings refer to individual outcomes and are in contrast to previously reported evidence of a dramatic improvement in the occurrence of asthma outbreaks.10

A clear cut interpretation of these results is limited since no information was available on the status of asthma prior to the occurrence of the outbreaks in the period 1980–7. However, an interpretation by analogy to both allergen avoidance trials and occupational asthma is possible. The results obtained in the present study appear to contrast with the view that avoidance of mite allergen is effective in improving asthma morbidity,7 but are in agreement with occupational studies in which asthma has been reported to persist in a substantial proportion of cases for years—and possibly indefinitely—after avoidance of exposure to causative agents.8

Evidence for the efficacy of allergen avoidance in improving asthma symptoms has been mainly provided by randomised controlled trials of mite allergen avoidance in both children and adults with asthma and sensitisation to house dust mite. Studies in adults have shown statistically significant improvements in symptoms and/or bronchial responsiveness.1-3 One study showed no effect, probably because there was an insufficient reduction in the level of mite allergen.4 Randomised controlled trials of mite allergen avoidance in asthmatic children have shown either positive effects5 ,6 or no effects at all.17 ,18 The different results from these studies may be due to the different type of interventions since bed covers have been found useful in the short term but the efficacy of acaricides remains unproven.7 The evidence for the persistence of occupational asthma after avoidance of the causative factors includes persistence of specific IgE to tetrachlorophthalic anhydride (TCPA),19 persistence of symptoms in patients with Western red cedar asthma,20 ,21and in those with asthma due to isocyanates.22 Persistent asthma after avoidance of the responsible factors has also been reported for allergy to trimetillic acid (TMA),23 wasp venom,24 and platinum salts.25

In contrast to both mite allergen avoidance and occupational asthma, our study involved a population exposed to an airborne allergen in the community and we cannot exclude the presence of high levels of allergen on a few sporadic days as a relevant factor. During the study period we are aware of at least one such day on 29 October 1994.26The finding in the present study that, among those who reported a visit to an emergency room in 1989, controls were more likely not to report it in 1995 than were epidemic subjects may reflect the occasional exposure to unmeasured high levels of soybean allergen on sporadic days. These uncontrolled exposures to high levels of allergen may also be important in clinical trials of allergen avoidance in which the levels of allergen are measured on a few occasions during the study period and do not include measurement of exposures at work, school, or public places. The persistence of sensitisation and asthma symptoms in subjects with soybean asthma may also have resulted from continuous exposure to low levels of allergen over a long period of time. One possible explanation for the parallelism between our results and those obtained in occupational studies may be the longer follow up compared with trials on mite allergen. In contrast, the persistent exposure to low levels of allergen and the possible occurrence of sporadic exposures to high levels of allergen in our population is unlikely in some types of occupational asthma in which the causative molecules are absent in the non-occupational environment.

Other potential explanations for the persistence of asthma after avoidance or reduction to allergen exposure, such as the retention of antigen in lung tissue as mentioned by Venables et al 19 or the presence of IgE autoantibodies to lung tissues,27 cannot be excluded. Saettaet al studied bronchial biopsy specimens of 10 subjects with occupational asthma to isocyanates at diagnosis and six and 21 months after cessation of exposure and found a significant decrease in both bronchial responsiveness to isocyanates and thickness of subepithelial fibrosis, but no change in non-specific bronchial responsiveness or in the number of macrophages and eosinophils.28 Unfortunately, no data were available in our study to consider any of these possibilities.

In interpreting our study results some methodological limitations need to be considered. Because controls were subjects who visited the emergency room for asthma but who had never been treated on an epidemic day, misclassification was unlikely. Misclassification may, however, have occurred with cases. Taking into account both the distribution of the expected number of visits on an epidemic day29 and the sampling scheme of soybean asthma cases in this study, we estimate that about 20 of the 213 cases may have been misclassified. The consequence of this misclassification pattern, if any, would have been a slight attenuation of the differences observed in our study. Assessment of sampling variation in relative risks for symptoms involved the calculation of a large number of confidence intervals, and the levels of significance achieved for being woken up by coughing and treatment in the emergency room should be taken with caution. On the other hand, the size of the study population was small and, although most of the relative risks for worsening indicated that worsening was more likely in subjects with soybean asthma than in controls, their confidence intervals were wide.

Another complicating factor is that in Barcelona several hundred asthmatics are still at risk of having severe attacks of asthma if sporadic releases of dust occur during soybean unloading operations at the city harbour. After the study period, on 12 June 1996, a peak of soybean allergen of 5032 U/m³ was recorded and considered responsible for 27 asthma visits to emergency departments, and possibly one death. Soybean epidemic asthma has been well documented in New Orleans, USA,30 Naples, Italy,31 Cartagena, Spain,32 and Tarragona, Spain.33 In all these areas subjects who have had soybean epidemic asthma may be at risk of severe asthma attacks which may eventually be fatal in the case of re-exposure. In order to keep the levels of airborne soybean allergen in these areas as low as possible, the use of highly efficient control measures of emissions may be necessary. Since the outbreak which occurred in 1996, the soybean industries and the harbour authorities in Barcelona have established improved filtering measures to achieve additional reduction of soybean dust emissions and a monitoring system of airborne soybean allergen is being operated by the City Council.

The finding that, in subjects previously affected by soybean epidemic asthma, this condition persisted during a long period despite a considerable reduction in exposure to soybean dust, highlights the importance of primary prevention. Because soybeans are traded worldwide the use of effective technology to control the emission of soybean allergenic particles from soybean industries may be necessary in other places with populations living in their vicinity.