• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services


  • Print
  • Share
  • E-mail

M-I-04-5: Staphylococcus Enterotoxin Production in Foam and Current Dairy Industry Balance Tank Practices

<< Back to Index of Memoranda of Information (M-I) Main Page  

June 4, 2004 

TO: All Regional Food and Drug Directors
Attn: Regional Milk Specialists

FROM: Milk Safety Branch (HFS-626)

SUBJECT: Staphylococcus Enterotoxin Production in Foam and Current Dairy Industry Balance Tank Practices

Milk Safety Branch (MSB) has for some time been concerned about the possibility of Staphylococcus enterotoxin formation in foam within milk and/or cream balance tanks that contain transitory skim milk or cream. It has been the practice to debit Item 17p-Cooling of Milk on Form FDA 2359-Milk Plant Inspection Report during routine inspections, state ratings and check ratings if it is determined that a significant violation is observed involving a foam layer, which is out of temperature (>45°F) over an extended period of time (generally 4 hours or more).

Recently, industry and a few State Regulatory Agencies have asked MSB to request the Center for Food Safety and Applied Nutrition's (CFSAN) Office of Plant and Dairy Foods' Division of Dairy and Egg Safety and Division of Microbiological Studies to conduct a scientific evaluation of this concern and to provide a scientific review for MSB to make a final determination. Included in this M-I is the report that MSB received from the Office of Plant and Dairy Foods' microbiologists.


Industry has reported that the vast majority of milk separation across the country is being conducted between 122°F to 150°F, with the product being held at those temperatures within the balance (surge) tank. The foam generated in these balance (surge) tanks typically can accumulate for up to 20 hours, with industry reporting a range of either 10-15 hours or 18-20 hours. Typical transit times for products passing through these balance (surge) tanks were reported by industry to be 3-4 minutes. Based on this industry information and the analysis of CFSAN's microbiologists the following conclusion has been arrived at:

CFSAN finds that there is no indication of a threat to human health by current industry practices. If a plant is processing within the above cited industry practices; milk separation practices between 122°F to 150°F, with the product being held within that temperature range within the balance (surge) tank, and foam accumulation up to 20 hours, this issue should not be a health concern and; therefore, should not be debited on routine inspections, state ratings or federal check ratings.

Following is the actual report that MSB received from CFSAN's Office of Plant and Dairy Foods:

"Staphylococcus enterotoxin (SET) is of concern to the dairy industry because of the toxin's ability to withstand pasteurization time/temperature combinations commonly used for a range of dairy products. Recently, the Milk Safety Team (MST) has had some concern over the possibility of SET formation in foam within milk and/or cream balance tanks that contain transitory cream or skim milk at temperatures ranging from 122° to 150°F. The foam generated in these tanks typically can accumulate for up to 20 hours, with industry reporting a range of either 10-15 hours or 18-20 hours. The MST concern is centered on whether the foam that accumulates upon the surface of the liquids within these tanks could create an environment that would permit toxin formation and contamination of the bulk product. To assess the likelihood of this scenario, we considered several key factors regarding Staphylococcus aureus (Sa) in milk and the generation of SET.


Sa is an occasional contaminate of raw milk and when present is usually due to the presence of mastitis within the herd which generated the milk (2). When Sa is the causative organism of a mastitic condition in cattle, up to 15% of Sa strains prove to be toxigenic.(3, 10). Studies in Germany reported that the levels of Sa in milk ranged from 39.6% having levels of less than 500/ml to 7.2% having more than 2,000/ml (2). One study reported that 5-22% of milk samples were found to be positive for Sa (8). A Wisconsin study found 9.7% of bulk tank samples positive for Sa, while another study from Virginia reported 30% of herds tested were estimated to have 10% of cows infected with Sa (9, 12).

Therefore we can assume for the sake of this discussion that there is a reasonable likelihood of Sa contamination in approximately 10% of raw milk. It can also be assumed that when present in milk, numbers could occasionally be as high as 103 CFU/ml. Determining the potential distributions of Sa densities in milk and the incidence of toxin producing Sa strains would require that a quantitative risk assessment be undertaken. We do not think, at this time, that a risk assessment is necessary.

Heat sensitivity: 

At 122, 131, 140 and 149°F, Sa has a reported D-value of 10, 3.1, 0.9 and 0.2 minutes respectively in whole milk (5). The same study also reported injury to be more rapid than death up to 140°F. Above 140°F death accelerated faster than injury (5). Therefore, within the balance tanks in question, containing product heated from 122 to 150°F there is sufficient heat to cause death and /or injury to Sa cells in raw dairy products. At 122°F, the milk would need to be heated for 30 minutes to eliminate Sa at the highest levels likely found (103 CFU/ml). At product holding temperature of 140°F, only 3 minutes are needed to inactivate the same level of Sa. Typical transit times for products passing through these storage tanks were reported by industry to be 3-4 minutes.

An increased fat content dairy product (e.g. cream) decreases heating efficiency and thus increases the time (D-value) needed for the same level of inactivation. On the other hand, skim milk has a reduced level of fat which increases heating efficiency and thus lowers the time to inactivate Sa cells. The heating of the dairy product before going into the balance tank will result in both inactivation and injury of Sa. These effects will greatly increase the lag phase needed for the surviving cells in the foam to begin growing and will lower the number of cells with the capacity to produce toxin.


It is commonly understood and accepted that Sa is a poor competitor against the normal microflora in unpasteurized dairy products and typically Sa is found to cause problems in foods with lower water activity, where it out competes other bacteria (13). The thermal resistance of Sa is not exceptional and competitive microorganisms are likely to survive heat treatments that would permit Sa survival. For cream, the increased fat (36%) does not sufficiently lower the water activity to provide Sa a competitive advantage over other microorganisms. In the scenario in question, i.e., cream or skim milk heated and held at temperatures ranging from 122 to 150°F, competing microflora are likely to be present in product foam where Sa has survived. Industry data presented to the FDA appears to support this fact. Where the industry found Sa in product inside these tanks, they also found higher numbers of other microorganisms. The presence of a competitive microflora (14) further adds to the time needed for repair of injury, sufficient growth, and detectable toxin production by Sa.

Growth and toxin production:

Sa growth in pasteurized milk has been reported (7) and the data are in relatively good agreement with the USDA Pathogen Modeling Program (PMP) for growth in the absence of competitive microflora. We looked at the possibility that a temperature gradient exists in the foam in the headspace of the balance tank which could provide a favorable environment for Sa. Given the lack of data regarding the temperature in the foam, we assumed that there was an area of the foam that could equilibrate at an optimum growth temperature for Sa. The PMP predicted growth of Sa in the absence of competitors and injury from 103 to105 CFU/g at 104°F (optimal for toxin production) in 3.5 hours. The rate of growth in skim milk was reported to be close to that of whole milk at 98.6°F, whereas with cream, the predicted growth rate was about half as fast (7).

Because of the design of the PMP, these estimated growth rates do not accurately depict the growth rate of Sa in balance tank foam. The PMP growth rate predictions are based on sterile broth growth studies using pure cultures. The growth prediction program may accurately predict the risk of Sa growth in pasteurized dairy products but does not predict the risk of Sa growth under conditions of injury and competition. The growth model is not yet sophisticated enough to predict the typical and expected increased lag phase duration due to injury, nor to predict the effect of competition of spoilage and other bacteria in a product that was not pasteurized.

It is important to note that Sa not only has to grow, but that it also has to produce toxin. Donnelly et al. (4) reported 10 ng of toxin production within 12 hours in pasteurized milk at 35°C (95°F). In that study the initial population of Sa was reported as 104 to 106 CFU/ml. In the same study, toxin was reported produced in raw milk at 37°C (98.6°F) within 12 hours; however, an initial inoculation of 106 cells of Sa was needed before toxin production was observed. Another study examined the production of toxin in UHT milk in the presence of competitors and found no toxin production after 24 hours at 98.6°F with 105 CFU/ml Sa cells added (10). This information suggests that if growth occurs in the foam of cream or skim milk balance tanks at all, the time needed for measurable toxin production in a partially heated raw product with a starting Sa population of 103 cfu/ml would take appreciably longer than 24 hours.


The smallest dose of SET to cause illness in susceptible individuals has been reported to be around 100 ng of toxin per serving (13). If the typical serving was estimated to be 250ml (8.2 oz) then 0.4 ng/ml would be needed in the milk within the balance tank to cause illness. Multiplied over a 30 gallon tank, 45.4 µg of SET would have to be produced to cause illness. We feel that because of the fact that the product is well mixed and non-viscous, it is reasonable to use the whole volume of the tank in the calculation. To generate this amount of toxin would require a much longer time than industry practices would allow even with a margin of error. Many other factors militate against the formation of toxin to detectable levels, much less a level that causes illness. The volume of fluid is relatively small which limits the amount of toxin that can be generated. Furthermore, the probable transfer coefficient for the toxin from the foam to the bulk fluid is likely to be fairly low. Estimating a conservative transfer coefficient of 1/10 and 1/100 would require the generation of 28 µg and 280 µg of SET produced in the foam of one 30 gallon balance tank respectively to reach the level of detection of 0.25 ng/ml. Either amount of toxin production is many times higher than what Donnely et al. (4) were able to obtain for Sa in pasteurized milk. Because of the injury of the Sa due to heating, the presence of competitors, the relatively small volume of fluid in the foam, and the need for any generated toxin to diffuse through the foam to the bulk liquid, generation of detectable SET in the final product does not appear to be possible for Sa in balance tank foam.


Sa food poisoning usually takes only 2-4 hours to produce symptoms. The most common symptoms of intoxication include nausea and vomiting. Because of the short incubation time before the onset of symptoms, contaminated foods are more readily identified than many other foodborne illnesses with longer onset times; therefore, it is unlikely that unrecognized outbreaks of SET involving contaminated milk products would commonly occur. Given the almost impossible levels of toxin generation needed to achieve detectable levels of SET in finished product from Sa contamination of foam, it is understandable that contamination of pasteurized milk with SET to the level of causing an outbreak is even more remote a possibility.

Over the last 30 years, there have been only a handful of SET poisoning cases involving milk products reported, and none implicating milk balance tanks. The primary cause of contamination in the few cases reported were post pasteurization contamination because of faulty equipment and poor product handling practices (1). Considering that there is an estimated 87 billion servings of milk a year in the US, conditions that are commonly occurring now with skim milk and cream holding tanks would likely have been implicated in some foodborne illness cases if conditions were permitting Sa growth and the necessary level of SET formation. The fact that no incidents have been linked to toxin formation prior to pasteurization or in balance tanks is further evidence of the low risk of current dairy practices.


Sa is occasionally found in bulk milk and the level of contamination can reach as high as 103 CFU/ml. There are very few reported outbreaks attributed to Sa in milk and no cases attributed to current pre-pasteurization balancing practices.

Sa grows well in pasteurized milk, however, available information indicates that competition, injury, and cell death likely limit significant growth of Sa in the balance tanks in question here. At the lowest product heating temperature (122°F), competition with thermotrophic bacteria will be the greatest. A fortiori, at the highest product heating temperature (150°F), most Sa are either killed or injured within seconds. The combination of both thermotrophic competition and thermal killing of Sa probably allows for these products to be manufactured safely using current industry flow control practices.

The lack of outbreaks implicating illness and the multiple barriers against Sa survival, growth, and toxin production all argue against a public health concern over current industry practices.

A more definitive answer to the question of the risk of current industry practices would require a quantitative microbiological risk assessment. However, based on the analysis presented here, we feel there is no indication of a threat to human health by current industry practices. At present, our resources for risk analysis are better deployed on food safety issues where there is a definite risk.

B. Shawn Eblen, Microbiologist, Division of Dairy and Egg Safety

Mark Walderhaug, Ph.D., Director (Acting), Division of Microbiological Studies

Donald Zink, Ph.D., Sr. Food Scientist, Office of Plant and Dairy Foods

Reginald Bennett, Chief, Microbiology Methods Branch, Division of Microbiological Studies


Literature Cited:

  1. Asao, T., Kumeda, Y., Kawai, T., Shibata, T., Oda, H., Haruki, K., Nakazawa, H. and Kozaki, S. (2003) 'An extensive outbreak of staphylococcal food poisoning due to low- fat milk in Japan: estimation of enterotoxin A in incriminated milk and powdered skim milk', Epidemiology And Infection 130:33-40.
  2. Asperger, H. 'The Significance of Pathogenic Microorganisms in Raw Milk', International Dairy Federation 24-39.
  3. Cenci-Goga, B.T., Karama, M., Rossitto, P.V., Morgante, R.A. and Cullor, J.S. (2003) 'Enterotoxin production by Staphylococcus aureus isolated from mastitic cows', Journal of Food Protection 66: 1693-1696.
  4. Donelly, C.B., Leslie, J.E. and Black, L.A. (1968) 'Production of Enterotoxin A in milk', Applied Microbiology 16:917-24.
  5. Firstenberg-Eden, R., Rosen, B. and Mannheim, C.H. (1977) 'Death and injury of Staphylococcus aureus during thermal treatment of milk', Canadian Journal of Microbiology 23: 1034-1037.
  6. Hahn, G. 'Pathogenic Bacteria in Raw Milk-Situation and Significance', Kiel (Germany): Federal Dairy Research Centre: 67-71.
  7. Halprin-Dohnalek, M.I. and Marth, E.H. (1989) 'Growth of Staphyloccocus aureus in Milks and Creams with Various Amounts of Milk Fat', Journal of Food Protection 52:540-543.
  8. IDF (1980) "Factors influencing the bacteriological quality of raw milk" Bulletin International Dairy Federation 120.
  9. Jones, G.M. (1991) 'Analysis of DHI cow and herd bulk tank milk samples for Staphylococcus aureus antibody levels', Journal of Dairy Science 74:161.
  10. Joshi, R. and Purohoit, S.K. (1997) 'Studies on important pathogens of public health significance from marketed raw milk and cream in Bikaner City',Indian Veterinary Medical Journal 21:197-200.
  11. Khalid, A.S. and Harrigan, W.F. (1984) 'A study of the effects of bacterial competitors, sodium chloride and medium on growth of toxin producing Staphylococcus aureus strain', Lebensmittel-Wissenschaft und Technologie 17:137-41
  12. Makovec, J.A. and Ruegg, P.L. (2003) 'Results of Milk Samples Submitted for Microbiological Examination in Wisconsin from 1994 to 2001', Journal of Dairy Science 86:3466-3472.
  13. Roberts, T.A. (1996) Micro-Organisms in Foods 5: Characteristics of Microbial Pathogens. Blackie Academic & Professional, London.
  14. Sayler, Allen. Personal Communication"

An electronic version of this memorandum is available for distribution to Regional Milk Specialists, State Milk Regulatory Agencies and State Milk Sanitation Rating Officers in your region. The electronic version should be widely distributed to representatives of the dairy industry and other interested parties and also will be available on the FDA Web site at http://www.cfsan.fda.gov (Updated Web Address) at a later date.

If you would like an electronic version of this document prior to it being available on the CFSAN Web Site, please e-mail your request to Robert.Hennes@fda.hhs.gov


CAPT Robert F. Hennes, RS, MPH, Chief
Milk Safety Branch

Grade "A" Pasteurized Milk Ordinance - 2001 Revision (document temporarily unavailable online)