Cases studies

|

-

Centers for Disease Control  and Prevention National  Center for Environmental  Health Case Studies in Applied Epidemiology No. 002-800

TOS (Toxic Oil Syndrome) An Epidemic of Mass Proportions in Spain, 1981

After completing this case study, the participant should be able to:

-
-
-

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Public Health Service

Part I

In 1981, Spain had just joined the ranks of the other more developed European countries, after a brutal civil war in the 1930s and a long dictatorship under General Francisco Franco. Spain was now a constitutional monarchy with Ministers heading the Departments of Health and Consumer Affairs, Agriculture, State, and other major government functions.

In the late 1970s Spain had started a national healthcare system, providing essentially free care to all of its 35 million citizens. The educational level of physicians in Spain had always been high, and with the new emphasis on health care, they started to bring Spanish medicine into line with other developed countries.

The Ministry of Health and Consumer Affairs was in charge of the national health care system, and also ran the national medical schools and the schools of public health. While Spain had many fine research epidemiologists at that time, few were knowledgeable or experienced in field epidemiology.

Spain’s major export products at that time were agricultural, including fruits, vegetables, meat, wines, and olive oil. Although it was a major export product, domestic demand for olive oil always exceeded supply. Spaniards typically went home for a hot lunch every day and consumed a Mediterranean diet of salads, fruits, vegetables, fish, olive oil and wine.

While the standard of living was improving overall, many Spaniards lived in rural or semi-rural areas and shopped at little open-air markets ("mercadillos"), where most household goods and food items were available through itinerant vendors.

Madrid is the capital city of Spain, and is centrally located. The population of the metropolitan Madrid area then was approximately 6 million, and many people lived in newly developing areas on the outskirts of the city.

On May 1, 1981, an eight-year-old boy was pronounced dead on arrival at a hospital in Madrid, Spain. Within days, five other members of his family were examined at the same hospital with similar signs and symptoms which included interstitial pulmonary infiltrate on chest X-ray, headache, asthenia, scalp itchiness, and slight fever. This syndrome was initially diagnosed as an atypical pneumonia. Soon, the number of patients admitted to Madrid hospitals with this clinical presentation increased dramatically. The clinical features of these early patients are listed in Table 1.

Table 1. Some clinical features of early patients, Madrid, 1981

Clinical feature Percent
 

myalgia 80.0%

eosinophilia 78.0%

pulmonary findings 70.0%

fever 59.0%

rash 39.0%

scalp itch 22.0%

acute encephalopathy 1.2%

Table 2 shows the number of new cases reported daily during the first 4 weeks.

Table 2. New cases of epidemic illness and time of identification, Madrid, 1981

Day # # Cases identified
 

1 1

2 0

3 2

4 1

5 0

6 1

7 1

8 2

9 2

10 0
 

11 6

12 5

13 4

14 9

15 32

16 55

17 96

18 130

19 155

20 199

Day # # Cases identified
 

21 233

22 129

23 210

24 202

25 161

26 200

27 211

28 187

Early in the epidemic the clinical picture was thought to be consistent with an infectious agent as the cause of the illness. Later, two findings lent support to this hypothesis: one was the isolation of Legionella gormany from a patient's sputum. The second, made by a group of pathologists, was of a morphological structure compatible with Mycoplasma pneumoniae in the lung of a deceased patient.

With an infectious hypothesis in mind, clinicians at a Madrid pediatric hospital compared children treated with erythromycin to groups who received other treatments, including antihistamines, other antibiotics, and placebo, but they found that the clinical course of the disease was similar in all groups. A dietary study done early in the epidemic included many non-specific questions about foods. However, the results did not suggest any as a possible cause.

Concurrently, Dr. Manuel Tabuenca, also a pediatrician, noted that children with the new disease developed a clinical syndrome different from adults, with a rash suggestive of a reaction to an external agent. Although cases of illness had been frequent in children, babies under six months of age appeared to have been spared. Thus, when an infant less than six months old was identified with the disease, Tabuenca focused on exposures the baby might have had in common with older children and adults.
 
 

Part II

Dr. Tabuenca learned that the infant's grandmother had added a particular type of cooking oil, sold in an unlabeled 5 liter plastic container, to the baby's formula as a dietary supplement.
 
 

Part III

Mothers of children admitted to the hospital with the new disease were interviewed. Their responses were compared to those from mothers of children admitted for trauma or surgery.

The crude data from Dr. Tabuenca’s study showed that all of the 62 children with the new disease had consumed oil from unlabeled 5 liter plastic containers. Of the 62 children admitted for trauma or surgery, only 4 of 62 had consumed that type of oil.

The data from an epidemiologic study can be summarized by cross-tabulating exposure status and disease status in a two-by-two table, as shown below.
 

A measure of association quantifies the strength or magnitude of the statistical association between the exposure and the health outcome of interest. Two commonly used measures of association are the relative risk and the odds ratio.

The relative risk reflects the risk of disease in the exposed group versus the risk in the unexposed group. In most case-control studies, since the true size of the exposed and unexposed groups is not known, you do not have a denominator with which to calculate an attack rate or risk. When case-control data are used, the odds ratio is calculation as an approximation of the relative risk.

What level are the intended students? Do we really need to explain a 2-by-2 table?
 
 

Part IV

Dr. Tabuenca’s study established that consumption of oil sold in unlabeled 5 liter plastic containers was a risk factor for developing the new disease.

Following Dr. Tabuenca’s discovery, oils were collected from some households and open air markets and analyzed by the official customs laboratory. They were found to have come from rapeseed oil (the generic name for Canola® oil) originally denatured with 2% aniline, but which now contained other aniline-derived compounds.

Officials learned later that the customs laboratory had known of the importation of aniline-denatured rapeseed oil for several months. On June 10, 1981, 40 days after the epidemic started, in an attempt to prevent further cases, the Ministry of Health and Consumer Affairs (MHCA) alerted the population about the relationship of the epidemic disease with the consumption of unlabeled adulterated oil.
 
 

Part V

Once the oil-disease relationship was announced, the number of new cases of what came to be called toxic oil syndrome (TOS) decreased rapidly. However, some patients who had been discharged in good condition were readmitted, leading the

MHCA to suspect that these patients may have been re-exposed. Thus an official oil recall program began on June 30, 1981.

People concerned about their oil exchanged it for pure olive oil at government expense.
 

Figure 2. Cases of Toxic Oil Syndrome by week, Madrid, 1981

-
-
-

Part VI

While the fall in new cases prior to the announcement of the implicated oil could have resulted from several factors, most likely the etiologic oil had all been sold and further exposure was limited to households who already had the oil. Another possibility would be that all the susceptibles had already been affected, and the remaining non-ill people were not susceptible.

Around the time of Dr. Tabuenca’s study, a Dr. Rigau performed four population-based case-control studies in the town of Navas del Marqués, 47 miles northeast of Madrid. The first of these studies examined various hypotheses including respiratory transmission, but did not ask about oil consumption, and found no potential risk factors for illness. After the oil-disease relationship was announced, a second study was done in the same town, which asked about the use of oil from unlabeled 5 liter containers and which confirmed Tabuenca’s results. This study not only showed an association between the oil and illness, but also showed an association between a specific vendor of unlabeled oil and TOS.

Another of the four studies suggested a dose response effect, but this was never replicated. The use of oil in family meals meant that persons were usually unaware of the quantity or type of oil they consumed. The study also assumed that all family members consumed similar amounts of the same oil, and this was estimated from weekly oil consumption, typically reported by the housewife. The difficulty inherent in determining individual oil consumption impeded later attempts to study the individual dose-response relationship of the oil to TOS.

Eight additional case-control studies were conducted after those of Tabuenca and Rigau. Although the studies’ methods were similar and all used the diagnosis of an atypical pneumonia as a case definition, they were performed in different geographic areas by different research teams. All of these studies showed a strong association between development of TOS and the ingestion of oils sold by itinerant salesmen in unlabeled 5 liter plastic containers that had not been subject to sanitary control.
 
 

Part VII

A potential problem with many of the studies was that participants were aware of the suggested relationship between consumption of oil from unlabeled 5 liter plastic containers and TOS. This knowledge could have introduced a recall bias into the studies performed. However, the short time between the possible exposure and most of the studies, the use of only one case per affected household, and the frequent need to interview someone other than the ill person to determine the type of oil used by the household, likely kept recall bias to a minimum.

Although outbreaks of disease in close-knit groups were not commonly reported, two convents reported several cases of TOS each. Since convents are closed communities and residents there share virtually all their meals, a study was done to examine in detail the use of food oil in the convents. Both convents housed laywomen in addition to nuns. Meals for the laywomen and nuns in each convent were prepared in the same kitchen, and with the exception of the oil used for cooking and salads, all the other food that was prepared and served was the same for all the residents in each convent. In each convent a number of the nuns who were served the oil from unlabeled 5 liter containers developed TOS, while none of the laywomen, who were served soybean oil, became ill.

A study done in Orcasur [is this the neighborhood survey?] analyzed TOS by the places people purchased unlabeled 5 liter bottles of oil, including grocery stores, supermarkets, oil pressing facilities at olive farms, warehouses, itinerant salesmen, and open air markets. Only oil sold by itinerant salesmen or in open air markets was associated with TOS. These results, plus information obtained through the Ministry of Health (which regulates oil sales) pointed to a particular oil distributor, RAELCA, which used itinerant salesmen rather than the usual outlets for food products.

Investigators went to the RAELCA plant.
 
 

Part VIII

The RAELCA plant contained large holding tanks of oil, a plastic injection mold for making plastic bottles from raw materials, and identical bottles containing oils of different chemical compositions. Investigators collected samples from the holding tanks and individual bottles.

***

When the government oil recall took place, attempts were made to place identifying information on each oil specimen. Not all specimens were clearly labeled, thus many of the over 100,000 containers of oil collected during the recall were unsuitable for use in epidemiologic or chemical studies. Nevertheless, in two case-control studies TOS investigators systematically looked for oil containers stored in the official warehouses by the MHCA, and identified those containers potentially useful for research from the thousands in storage.

The first toxico-epidemiologic study was conducted in two locations in the Madrid province, Alcorcón and Leganés. Among the thousands of bottles of oil available, investigators selected for testing those they judged most likely to be associated with TOS -- those that appeared to be the original 5 liter plastic container, having a shape typical of those found at the RAELCA plant), and not full (a full bottle means no oil had been consumed, and therefore it was not associated with disease). Identifying information was used to attempt to contact the family who had turned in the oil. Sixteen other criteria were then applied to the containers of oil initially selected to arrive at the case and control oils. These criteria included sufficient identifying information to locate the family, fulfillment of the case definition for TOS by at least one family member, and a red top on the oil container. Specimens of the oil were chemically analyzed for 21 oil constituents and two measures of oil rancidity, and these results were statistically analyzed.

The second toxico-epidemiologic study used a similar design in the choice of case and non-case oils, but covered all the geographic areas affected by the epidemic.

Three fatty acid anilides – oleyl, linoleyl and palmityl anilide – were highly statistically associated with the oil having come from a family with a member who was ill with TOS. Of these three fatty acid anilides, oleyl was present in the highest concentrations. The relationship between amount of oleyl anilide found in case and control oils for both toxico-epidemiologic studies are shown in Table 4.

Table 4 Amount of an aniline-derived chemical compound (oleyl anilide) in case and control oils, Toxico-epidemiologic studies I and II, Toxic Oil Syndrome, Spain, 1981

Toxico-epidemiologic study IToxico-epidemiologic study II

Oleyl Odds Oleyl Odds

anilide (g/g) Case oilsControl oils Ratio anilide (g/g) Case oilsControl oilsRatio

1,201+ 7 0 Infinite 901+ 9 0 infinite

601-1,200 6 3 8.73 601-900 8 1 16.55

101-600 3 7 1.87 301-600 8 4 4.14

1-100 2 6 1.45 1-300 5 5 2.07

0 11 48 1.0 (Ref) 0 29 60 1.0 (Ref)

Total 29 64 -- Total 59 70 --
 
 

Part IX – Conclusion

Research Issues

While the exact identity of the etiologic agent in TOS remains unknown, work on TOS continues. Follow up clinical studies and long term mortality studies are underway. Investigation of the mechanisms involved in TOS continues. The identification of suspect chemical compounds, their characterization, and effects will hopefully one day contribute to the prevention of other similar diseases.

In the fall of 1989, the United States experienced an epidemic of a new syndrome characterized by severe myalgias and eosinophilia. This syndrome, dubbed eosinpohila-myalgia syndrome (EMS), was quickly linked to ingestion of contaminated L-tryptophan. Because EMS was clinically similar to TOS, American researchers who had participated in the investigation of TOS were able to direct epidemiologic and clinical research appropriately. The pathogenesis of EMS appears to have much in common with TOS, although its complete evolution remains unknown.

Long term follow-up in TOS is of great importance. In addition to documenting the clinical outcomes of TOS patients, follow-up will assist investigators in the United States who continue to study EMS. The clinical resemblance of these two disorders is remarkable, and they could be similar etiologically. Further, features of TOS, especially in its chronic phase, are similar to those of other immunologic diseases.

Recent work in drug-induced lupus suggests that different environmental exposures can provoke connective tissue diseases and that an immunologic similarity may exist among these diseases. Continuing investigation of chemical compounds in TOS oils or in EMS-associated L-tryptophan may one day establish the identity of the etiologic agent in these diseases and help explain the etiology of connective tissue diseases such as systemic lupus erythematosus, scleroderma and rheumatoid arthritis, whose causes remain unknown.

Social Issues

A variety of factors played a role in the social and political repercussions of the TOS epidemic. The beginning of the epidemic was acute and massive – within the first month and a half after the epidemic began 80 people had died and over 12,000 had become ill. TOS was a new systemic disease without any known treatment that primarily affected people of middle to low socio-economic status. The immediate cause of TOS was unknown for at least a month and a half after the epidemic began and rumors surrounding possible causes were rampant. TOS was also a toxic illness associated with illegal trade in a fraudulent food product. [This was never brought up in the case study] As the epidemic progressed serious deficiencies in the control of the food oil market by regulatory authorities were documented.

Taken together, these factors caused a societal upheaval, first based on fear and anxiety about an unknown and potentially deadly illness, and later based on compassion for affected citizens. In the end, the government provided economic and social assistance to mitigate needs that TOS had created in families, and the judicial process sentenced those who were held legally responsible.

These social attitudes made the work of the investigators and physicians difficult. On one hand the media exerted pressure regarding what should be investigated at any particular time, increasing public anxiety and unease. On the other hand the association of possible economic benefits with a person's health caused confusion for physicians regarding the authenticity of the complex symptomatology of this disease, an attitude which persists even today.

References

1. Philen RM, Posada de la Paz M. Toxic oil syndrome and eosinophilia-myalgia syndrome: May 8-10, 1991, World Health Organization meeting report. Semin Arthritis Rheum 1993;23(2):104-24.
2. Tabuenca JM. Toxic-allergic syndrome caused by ingestion of rapeseed oil denatured with aniline. Lancet 1981;2(8246):567-68.
3. Bernert JT, Jr., Kilbourne EM, Akins JR, et al. Compositional analysis of oil samples implicated in the Spanish Toxic Oil Syndrome. J Food Sci 1987;52(6):1562-69
4. Hill RH, Schurz HH, Posada de la Paz M, et al. Possible etiologic agents for toxic oil syndrome: fatty acid esters of 3-(N-phenylamino)-1,2-propanediol. Arch Environ Contam Toxicol 1995;28(2):259-64.
5.  Posada de la Paz M. Diet and food contaminants. In: Steenland K, Savitz DA, eds. Topics in environmental epidemiology. Oxford: Oxford University Press, 1997:64-88.
6.  Tabuenca JM. Discovery of toxic oil as the cause of the epidemic. In: World Health Organization, ed. Toxic Oil Syndrome: mass food poisoning in Spain. Copenhagen: World Health Organization, Regional Office for Europe, 1984: 39-46.
9. Casado J, Ortíz JM, Otero J, et al. Síndrome Tóxico en niños por consumo de aceites vegetales: modelo clínico de la enfermedad, en la fase aguda. Pediatrika 1982;2(Supl.1):22-6.
10.  Posada de la Paz M, Philen RM, Abaitua Borda I, et al. Toxic oil syndrome: traceback of the toxic oil and evidence for a point source epidemic. Food Chem Toxicol 1996;34(3):251-57.
11. Posada de la Paz M, Abaitua Borda I, Kilbourne EM, et al. Late cases of toxic oil syndrome: evidence that the aetiological agent persisted in oil stored for up to one year. Food Chem Toxicol 1989;27(8):517-21.
12. Díaz de Rojas F, Castro Garcia M, Abaitua Borda I, et al. The association of oil ingestion with toxic oil syndrome in two convents. Am J Epidemiol 1987;125(5):907-11.
13. Grandjean P, Tarkowsky S. Subgroup discussion. In: World Health Organization , eds. Toxic Oil Syndrome: mass food poisoning in Spain. Copenhagen: World Health Organization, Regional Office for Europe, 1984: 17-27.
14. Grandjean P, Tarkowsky S. Review of investigations and findings. In: World Health Organization eds. Toxic Oil Syndrome: Mass food poisoning in Spain. Copenhagen: World Health Organization, Regional Office for Europe, 1984:3-16.
15. Posada de la Paz M, Castro Garcia M, Kilbourne EM, et al. Toxic-oil syndrome: case reports associated with the ITH oil refinery in Sevilla. Food Chem Toxicol 1987;25(1):87-90.

Case Definition of Toxic Oil Syndrome Proposed by the Spanish Clinical Commission, August 3, 1981.*

Major criteria

Appendix B

Epidemiologic studies of toxic oil syndrome, Spain, 1981.