METHODS

There were different phases during this process. The plants operated without restrictions until the June 1996 outbreak, when unloading operations were suspended as a cautionary measure, a review was carried out of the structural aspects of the plants, their maintenance, and the routine of unloading operations. Subsequently, unloading was authorised on a boat to boat basis and under special restrictions covering the unloading flux and its timetable; simultaneous unloading in the two docks was not permitted, and there were restrictions linked to weather conditions. These restrictions were reviewed as data on allergen concentrations suggested variations linked to their operation. The identification of dry filtration technology as a promising option for reducing emissions led to its adoption in the unloading installations in November 1997, replacing the previous standard polyester sleeves by new micro pore size (MPS) filter sleeves, or by tetratex filters in series with polytetrafluroethylene (PTFE) membrane filter cartridges. These modifications were implemented while under the special regimen of boat to boat unloading authorisation, as the firms tested it and subsequently requested new licences for plant operation. These licences were approved in the summer of 1998 by the city government, which maintained as a cautionary measure some operation restrictions under adverse weather conditions, and required the monitorisation of soybean allergen emission and inmision levels by the firms in a pilot phase.

On the basis of this evolution, the study period may be subdivided into four phases: phase 1 covers eight days around the time of the June 1996 episode; phase 2 involves crude control methods based on the restriction of activity until the installation of new filters in November 1997; phase 3 covers the tentative use of the new filters until the pilot phase in summer 1998; and finally phase 4 involves the pilot phase evaluation until December 1998. During this period four indicators were analysed: allergen levels in the city, allergen emissions by the three firms involved, aggregated numbers of asthma cases in city emergency services, and the health status of a panel of patients. Not all the indicators were active during the entire period: emissions data could not be collected until 1997, and the panel of patients was set up towards the end of 1997. Table 2 shows the four phases and the indicators available in each phase, along with other relevant variables.

Aggregation of asthma patients attending emergency services

The emergency departments of the four university hospitals in the city monitored the daily volume of patients. As defined in the initial study of the epidemics, any day on which more than 12 asthma patients were seen was considered an abnormal or potential asthma outbreak day.² The names of asthma patients on such days were checked against a list of people who had been repeatedly involved in episodes of epidemic asthma. The simultaneous detection of three such patients in emergency services was considered to be a marker of an episode of epidemic asthma.⁷

Health status of a panel of patients

After a clinical epidemiological study of soybean asthmatic patients, a sentinel surveillance system was set up based on the continuous follow up of symptoms in a small group of patients. Such a system ought to facilitate early detection of episodes of allergy. A group of high risk patients was selected for this purpose. It involved 10 asthmatic “repeaters” resident in the areas closest to the port, and four patients not sensitised to soybean dust. To follow symptoms daily the patients were trained to measure their peak flow (a measure of respiratory function) in the morning and at night, and to fill in daily a questionnaire on symptoms. This information was complemented through periodic follow up telephone interviews. In the case of any sudden increase in acuteness of symptoms, they were required to immediately call a contact telephone number set up for this purpose. Concurrently, a plan was established to follow variations in the patients’ symptoms, with successive levels of assessment and decision, including the alarm signal, investigation, notification, and verification. Follow up of this panel took place during the period from February to May 1998.

Allergen concentration

This is measured by the radioallergoabsorbence test (RAST) applied to filters installed in samplers.^9,10 The measurements from sampling filters yield a 24 hour average. In this article the data presented centre on results obtained from the municipal atmospheric contamination control station situated in the port authority customs building (city district of Poble Sec) as this station was both closest to the plants and to the neighbourhoods with the greatest number of affected patients. Initally the filters were analysed in the Mayo Clinic (Rochester, USA) but the technique used was implemented in the laboratory of the Vall d’Hebron Hospital (Barcelona) from the autumn of 1997, and after an excellent correlation between the results obtained by the two laboratories had been reached, analysis of the filters was performed only in Barcleona.

Allergen emission

The facilities for unloading, manipulation, storage, and grinding of soybean in the port of Barcelona use pneumatic transportation circuits. In 1996, all foci of channelled emissions were identified. In these foci air is expelled to maintain the low pressure in the pipes; as this air is filtered before reaching the atmosphere, the emissions thus produced are channelled emissions. The steps undertaken to measure allergen emission led to the identification of almost 40 foci of channelled emissions in the three plants. The level of emission at each point and the global value for each plant were measured.¹¹ The RAST inhibition analytical technique provided a measurement of emissions in terms of allergen units per volume of air; thus, with the known air volume flow rate at each point, emissions can be estimated in allergen units per unit of time. Sampling was carried out in accordance with the German VDI methods, which are reference methods in several European Union countries (isokinetic procedure VDI 2066 adapted to the analytical technique using PTFE coated filters).

RESULTS

Epidemic outbreaks

After the June 1996 episode, no further epidemic asthma outbreaks have been detected in the city. The epidemiological surveillance system continues to monitor emergency services and has actively checked the situation on several occasions when events, or evolution of indicators, seemed to suggest that it was necessary, but without detecting either an epidemic asthma outbreak or clustering of “repeater” patients.

Panel of patients

Symptoms referred by the panel may be characterised by their instability over the study period, and by an intensity from low to moderate. There were several false alarms, caused by worsening of symptoms in a single panel member. However, a single cluster of symptoms was detected: five patients presented a sudden worsening in their symptoms (out of eight sensitised patients contacted) on a given day. During this episode, none of the controls manifested the slightest modification in their symptoms, no changes were detected in the number of asthma cases seen in emergency departments, but an increase in symptoms was confirmed in other known asthma patients not members of the panel who were contacted by telephone. On this day there was soybean unloading under “non-optimal” weather conditions. Levels of soybean allergen measured were comparatively low (225–415 U/m³), although they were the highest recorded during the panel follow up period.

Allergen concentrations

Table 3 presents allergen levels data in RAST units per m³ for the four phases analysed. Even though the measurements were in general low, some important fluctuations are visible towards the end of November 1996, when values are higher for some days. These results led necessarily to a review of unloading conditions. After the installation of the new filters there is a fall in both the measured values and their variation. In the final pilot phase for normalisation, as unloading activities increased, allergen concentrations do not rise: they are only above 300 U/m³ on two days, and there was no unloading on either of them. In this final phase, mean allergen levels in unloading days (67 U/m³) and in days without unloading (63 U/m³) are similar; they are even lower on days with simultaneous unloading (45 U/m³).

Allergen emission

Table 4 shows emission data for the three plants. Very significant decreases were achieved in channelled emissions of allergen, between 95% and 99% lower than two years before. The emissions by plants A and B (those involved in unloading operations) are now of the same order of magnitude as those of plant C (which does no unloading).

DISCUSSION

A controlled trial is the most solid design for research in the fields of human health, but it would not be possible for this particular problem. We believe that these data show the effectiveness of the new filters in controlling allergen emissions: after November 1997 emissions are reduced, and allergen levels in the city also fall. They also show that during the period of restricted unloading (before normalisation) it was possible to obtain reductions in risk by limiting the flux and timetable of unloading operations, complemented by weather conditions. The Barcelona experience shows that industrial soybean plants may be located near urban settings if strict control criteria are applied. The identification of similar problems in port cities with soybean harbour facilities in Spain,^12–14 and in other countries,^15,16 points to the external relevance of this process. To date, no country has developed legislation to regulate this environmental risk.

Meteorologically, epidemic days were anticyclonic with atmospheric stability, a low altitude mixing layer, and light winds, with long calm spells. Soybean allergen levels in the city depend closely on weather conditions facilitating dust dispersion: wind speed and direction, altitude, and vertical turbulence of the air mixing layer. Thus, allergen concentration in the city may be seen as a problem of gas dispersion: with stable emission, they will be a function of weather conditions, in accordance with a probabilistic model. This model was used to define a preventive tool, which was important in the initial stage of restricted unloading, as it defined the weather associated with higher levels of allergen, and under certain weather conditions unloading was suspended. The solution to the problem is in the reduction of the allergenic particles emitted by the plants, but weather related controls play an additional safety part.

Allergen concentrations during the pilot phase showed a small increase on two days during which no unloading took place; on one of these days grain was unloaded from a train in a railway yard adjacent to one of the plants, while on the other day the conveyor belt linking the two plants was in operation. This shows that operations other than ship unloading may give rise to a certain amount of allergen emission. Railway and conveyor belt operations in the plants are now subject to the same meteorological restrictions described for the unloading of soybean carrying ships.

The process described indicates the feasibility of using a panel of asthmatic patients allergic to soybean as part of the surveillance system. It is worth noting, however, that the delay in symptom reporting by patients requires a very active follow up. This follow up provided useful elements for the evaluation of the effectiveness of risk control. Although the control measures drastically reduced both emissions and the concentration of soybean allergen in the city air, active surveillance through the panel showed that in given circumstances small increases in comparatively low allergen concentrations may be sufficient to produce symptoms (although not severe) in a subgroup of the population allergic to soybean.

A general principle in our public health law is that industrial activities must be compatible with the health of the population. This legislation provided the grounds for the involvement of municipal authorities in dealing with this problem and controlling it. However, the only specific legislation applicable to this type of industry refers to general dust, clearly inappropriate for soybean dust, which may cause allergy. In fact, there is no environmental regulation in any country for soybean dust or any other type of allergenic products, beyond occupational standards. Our contacts with the World Health Organisation and the European Union confirm that there is no legislation based on the applicable scientific knowledge, nor any group of experts working in this area. The international nature of the problem and its economic implications may justify the involvement of the International Chemical Safety Program promoted by the International Labor Office, the WHO, and the United Nations Environment Program, to provide a global frame of reference.

The process followed in Barcelona combines research and action. It provides an example of public health applied to risk control under incomplete knowledge conditions. In such a process many uncertainties arise, which require a pragmatic response. The key to progress was intersectoral work, in interaction with applied epidemiology, clinical practice, laboratory and environmental sciences, industrial engineering, and physics of meteorology, but also with the legal, management, and public administration fields, and a prudential use of trial and error.

There is need for further research. Key questions to be answered are the evolution of soybean sensitisation in patients after the reduction achieved in the allergen levels to which they are currently exposed. There is also a need for the development of analytical techniques for improved monitoring of emissions, where promising advances are underway.^17,18