During 1987 a coincidence between the unloading of soybeans in the harbor and outbreaks of asthma was observed. In this study we investigated the association between the unloading of products in the harbor, particularly soybeans, and the occurrence of asthma epidemics during the years 1985 and 1986.

All daily emergency room admissions of Barcelona residents with asthma were recorded during 1985 and 1986. Data were collected from the clinical records of the four largest urban hospitals, which account for approximately 90 percent of the emergencies in the city. An emergency room admission for asthma was defined as a visit during which any asthma-related diagnosis was recorded.¹⁴ “Asthma” refers to a disease in which wide variations in resistance to flow in the airways of the lungs occur over short periods.¹⁷

The present study included only patients over 14 years of age, because no abnormal increases in the number of cases of acute severe asthma in childhood had ever been observed in the pediatric emergency services in Barcelona.^14,15

An “unusual asthma day” was defined as a day on which the number of emergency room visits was so high that the probability that such a number or a higher one was the result of chance was 0.025 or less. This probability was calculated by assuming a Poisson distribution, with the 15-day moving average representing the number of cases expected.¹⁸

An “asthma-epidemic day” was defined as an unusual asthma day on which the cases were clustered on an hourly basis. An “hourly cluster” was defined as the occurrence in one four-hour period of so many emergency visits for asthma that the probability that such a high number of visits was the result of chance was 0.05 or less. The four-hour periods were selected on the basis of the distribution of cases during asthma outbreaks. This probability was calculated by the Knox and Lancashire approximation¹⁹ to the scan method.²⁰

All products identified as having been loaded or unloaded during at least one asthma outbreak were studied. We recorded the days on which each product was loaded or unloaded during 1985 and 1986. In the case of soybeans, additional variables were examined, including the use of two harbor silos (A and B) for soybean storage,

the form of the soybeans (in bulk, or as derivatives in the form of meal or pellets), and the country of the port of origin.

During three nonconsecutive outbreaks—on November 11, 1986, February 8, 1987, and September 7, 1987—particles were collected in small samplers with cellulose ester filters²¹ placed in the district bordering the harbor, where most cases had occurred. Morphologic analysis was performed by optical microscopy on mineral oil-mounted slides. Lugol’s solution was used for starch staining, and Coomassie blue solution for the detection of protein.^22,23 Soybean samples, collected from the hold of the ship that had unloaded soybeans during the two most recent documented outbreaks (September 4 and 7, 1987), were examined by optical microscopy and transmission and scanning electron microscopy.^22,23 Meteorologic information was collected on epidemic days from four stations in different parts of the city and a fifth at the airport, 5 miles to the southwest.

For each product unloaded, a risk ratio was calculated between the probability of an asthma epidemic on the days when the product was being unloaded and the probability of an epidemic on the days when it was not being unloaded. The unit of observation and analysis was the 24-hour day.

Confidence intervals for the risk ratios were estimated by the Miettinen method.²⁴ When there was a zero in one cell of the two-by-two table, the lower limit of the confidence interval was calculated with use of the exact confidence-limit test²⁵; in such a case, the null hypothesis was also tested with Fisher’s exact test.²⁶ Evaluations of the independent effect of each product on the occurrence of asthma epidemics, with control for the other products, were carried out with use of the Mantel-Haenszel method.²⁷

Thirteen of 23 “unusual asthma days” identified during the study period had hourly cluster patterns