Caballero B. (ed.) Encyclopaedia of Food Science, Food Technology and Nutrition. Ten-Volume Set

Подождите немного. Документ загружается.

(smokers having intakes of many nutrients which are

different from nonsmokers, and smokers having an

increased risk of a wide range of diseases). In practice

it is found that most factors, dietary and other, which

are predictive for a disease, are interrelated to a

whole network of other factors relevant to that dis-

ease. Unscrambling these relationships presents great

difficulties, especially when the different factors are

estimated with different degrees of precision. Other

difficulties arise because of the possibility of over-

control for confounding, and because errors in the

measurement of an independent factor, such as a

dietary factor, lead to the underestimation of an asso-

ciation.

Intervention Trials

0020 There are two broad uses of intervention trials in

dietary research. First, there is the small, short-term

intervention trial that examines a mechanism or tests

the association between factors. Thus, the intake of a

food item, such as salt, saturated fat, fatty fish, and

vitamin C, or the intake of dietary fiber, fruit

and vegetables, or dairy produce can be changed

and changes observed in biochemical or hemato-

logical factors known to be relevant to disease.

Possible determinants of platelet function, oxidized

lipoproteins, and blood pressure can be thoroughly

evaluated using this strategy.

0021 The second, and ultimately far more important use

of intervention trials is to test the relevance of a factor

to the incidence of a disease. The rules for such RCTs

have been well worked out and if an RCT is to yield

reliable evidence then there must be no compromise

on certain cardinal aspects. These last include

random allocation of subjects to the intervention

(‘treatment’) or the ‘control’ group; the follow-up of

all subjects admitted to the trial, whatever their level

of compliance, is essential if bias is to be avoided

(‘intention-to-treat analysis’); and the assessment of

outcome events must be made ‘blind’ with regard to

whether the subject was in the treated or the control

group. Unfortunately, in dietary RCTs it is rarely

possible to comply with a further point which is

regarded as important in RCTs, namely ‘blindness’

of the participants as to whether they are receiving the

intervention measure or not. This last is easily

achieved in drug trials, but the use of a dummy

placebo is seldom possible in dietary trials.

0022 While RCTs give evidence with a high level of

certainty, they still have limitations. First, they usu-

ally have to be large and long-term. For example,

even with the inclusion of thousands of subjects,

and prolongation for several years, almost all RCTs

so far reported have failed to give conclusive evidence

on the relevance of dietary fat intake to survival.

Hence there needs to be overviews of all available

and relevant evidence, as described below.

0023Other limitations of dietary RCTs include the diffi-

culty in persuading subjects to make a change in their

diet which is of sufficient size, and is maintained for

a sufficient time, to be likely to affect the incidence

or the progression of a disease. In fact, setting-up

a trial with random allocation to an intervention

group often leads to ‘contamination’ of the control

group, either through contact between subjects them-

selves, or through reports in the media. At the very

least, dietary and other changes in the two groups,

intervention and control, must be carefully moni-

tored. Then it is important to avoid as far as possible

changes in dietary factors other than the one of inter-

est, and those changes that do occur should be moni-

tored and recorded. Thus an increase of, say, fish

consumption is likely to lead to a decrease in meat

intake, and an increase in wholemeal bread may lead

to an increase in the use of spreading fats.

Limitations of Approach

0024Difficulties in epidemiological approaches to research

on nutrition and disease include the following:

0025The etiology of most diseases is complex and for

very few diseases is there a single cause. Not only

are food items likely to interact together, and not

only is a change in one dietary item likely to be

accompanied by changes in other food items, but

other lifestyle factors are likely to be related to the

dietary intake of any particular nutrient. Smoking

and exercise level are two nonnutritional factors

that confound almost every possible nutritional

relationship. Social class and education are also

relevant to most nutritional factors and to the

risk of almost every disease.

0026The difficulties in adequately characterizing the

diet of any individual subject are enormous and

in most nutritional epidemiological studies this is

necessary for thousands of subjects. Methods of

estimating dietary intakes are dealt with elsewhere

and it has to be accepted that even the most careful

and exacting of these give at best only a crude

estimate of what is required: that is, the usual or

the average dietary intake of subjects over a very

long period of time. This is a most important

limitation because uncertainties in the measure-

ments of independent variables will reduce the

power of a study and true effects may escape

detection.

0027Third, the effects of food items or dietary pat-

terns are likely to be very long-term, and in the

EPIDEMIOLOGY 2147

study of relationships with disease it is possible

that the diet at the time the disease is diagnosed is

no longer of any relevance, yet the estimation of

dietary intake of an individual in the past – per-

haps 10 or 20 years previously – is clearly almost

impossible.

0028 There are often difficulties in recruiting enough

subjects to give power in a study to test a hypoth-

esis adequately. Recruitment can be especially dif-

ficult in nutritional studies in which the collection

of dietary data, or the making of a change in diet,

can be quite disruptive. In estimating the power of

a proposed study it is important that a realistic

estimate is made of the likely size of the predicted

effects. It is all too easy to overestimate this, just as

it is easy to overestimate the validity of estimates

of dietary intakes. The power of a study to detect a

real effect depends ultimately on the size of the

effect, and this is usually smaller than one would

like, and the size of the effect, or intervention,

being studied.

0029 In uncommon diseases, and in diseases with a long

‘incubation’ period, a case-control study may be

all that is possible, and this strategy has the

greatest limitations. On the other hand, while

RCTs yield the most conclusive evidence, they

are the most difficult of all the strategies and, as

has been indicated, they raise additional difficul-

ties in nutritional research.

Progression in Epidemiological Research

0030 There is a logical sequence in epidemiology, from the

observation perhaps of area differences in the preva-

lence of a disease, or a change over time, to cross-

sectional studies that determine prevalence and test

associations with dietary or other factors, to inter-

vention trials which examine a preventive strategy

focused, for example, on a nutritional factor. A pro-

gression such as this will of course be governed to

some extent by findings in clinical and in laboratory

research and the different approaches are not in any

way totally independent.

0031 Such a sequence is seldom seen, and opportunistic

studies which explore some aspect are often all that

are feasible. However, research on magnesium does

illustrate a logical but incomplete sequence. The dem-

onstration of magnesium as an essential element in

many enzymes and its involvement in many biological

mechanisms led to studies of tissue magnesium in

samples taken postmortem. These have led to long-

term prospective studies of dietary Mg intakes and

the incidence of vascular disease. These led to RCTs

of the benefit of Mg infusions given immediately after

a myocardial infarction. The sequence will not be

completed, and the relevance of dietary Mg to disease

will not be apparent until long-term RCTs of an

increased Mg intake have been completed and

examined in a metaanalysis.

Interpretation of Data

0032A clear distinction must always be made between the

testing of a hypothesis which was defined before a

study was set up, and the generation of new hypoth-

eses from data collected for another purpose. This

latter, which is often termed ‘fishing’ or ‘dredging,’

is permissible only if it is recognized for what it is and

if the further associations detected are tested in new

ad hoc studies. The dangers inherent in generating

multiple hypotheses, and in then treating those that

achieve an acceptable level of statistical significance

as if they were prior hypotheses, are very great in

dietary studies because of the number of independent

variables.

0033Firm conclusions should only be based on evi-

dence drawn from a number of studies. The play

of chance can never be ruled out, whatever the

level of statistical significance, and consistency in a

number of studies is a better guide to truth than the

results of a single study. Hence the importance of

‘overviews’ of studies, or metaanalyses of data from

a number of different studies. It is most important

that data from every relevant study is included in an

overview. Unfortunately, certainty on this last is im-

possible, as there is always a ‘publication bias,’ that

is, reports which confirm a null hypothesis are far

less likely to be published than those which show

an effect.

0034There is another aspect of this last point. Not only

does one look for consistency in the results of differ-

ent studies using the same design strategy; one

should also look for consistency in the results of

studies with different approaches. This is necessary

even when there is convincing evidence from, say, a

number of RCTs, because an increase in the con-

sumption of one nutrient or one foodstuff is usually

accompanied by a change in the consumption of

some other nutrients or food items. Confidence is

enhanced, however, if prospective studies demon-

strate an association between a food item and a

disease, if RCTs show a change in the disease inci-

dence, and if these effects are consistent with the

results of metabolic studies. Examples of such con-

sistency are rare in dietary research but the benefit of

fatty fish in heart disease has been shown in histor-

ical studies, in comparisons between communities,

in laboratory-type studies of fish oil and vascular

risk factors, and in (admittedly only one) interven-

tion RCT.

2148 EPIDEMIOLOGY

The Cochrane Collaboration

0035 There have been a number of basic advances in re-

search methodology and foremost among these must

be the introduction of the RCT. Another, of perhaps

equal importance, has been the development of the

overview or metaanalysis. The recognition of the

need for these throughout clinical practice led to es-

tablishment of the Cochrane Collaboration. This is

now an international organization that seeks to iden-

tify every acceptable RCT within medical care, makes

these available within selected diseases or disease

groups as databases, and organizes systematic reviews

within different health care interventions.

0036 The Collaboration is an exceedingly valuable re-

search resource. For example, a recent search under

the headings ‘heart disease’ and ‘diet’ produced,

within a minute or two, 328 references to RCTs.

See also: Aging – Nutritional Aspects; Cancer:

Epidemiology; Food Safety; Malnutrition: The Problem

of Malnutrition

Further Reading

Alderson M (1983) An Introduction to Epidemiology, 2nd

edn. London: Macmillan.

Armstrong BK, While E and Saracci R (1992) Principles of

Exposure Measurement in Epidemiology. Oxford:

Oxford University Press.

Bradford-Hill A (1971) Principles of Medical Statistics, 9th

edn. London: The Lancet.

Burr ML (1979) Medical memorandum: criteria for plan-

ning a research project. Community Medicine 1:

157–159.

Burr ML (1985) Progress in food and nutritional science.

Nutritional Epidemiology 9: 149–183.

Burr ML, Elwood PC, Fehily AM and Sweetnan PM (1983,

1984) Epidemiology for nutritionists: a series of five

articles on epidemiological methods. Human Nutrition:

Applied Nutrition 37A: 259–264, 265–269, 339–347,

419–425; 38A: 215–222, 329–334.

Cochrane AL (1972) Effectiveness and efficiency. Random

reflections on health services. The Nuffield Provincial

Hospital Trust 1972. London: Royal Society of Medi-

cine Press Ltd.

Last JM (1988) A Dictionary of Epidemiology, 2nd edn.

Oxford: Oxford University Press.

Margetts BM and Nelson M (1977) Design Concepts in

Nutritional Epidemiology, 2nd edn. Oxford: Oxford

University Press.

Willett W (1998) Nutrition Epidemiology. Oxford: Oxford

University Press.

ESCHERICHIA COLI

Contents

Occurrence

Detection

Food Poisoning

Occurrence and Epidemiology of Species other than

Escherichia coli

Food Poisoning by Species other than

Escherichia coli

Occurrence

K S Venkitanarayanan, University of Connecticut,

Storrs, CT, USA

M P Doyle, University of Georgia, Griffin, GA, USA

Introduction

0001 Escherichia coli was first described by Dr. Theodore

Escherich in 1885. The natural habitat of E. coli

is the large intestine of warm-blooded animals,

including humans. The bacterium is Gram negative,

rod-shaped, and can be easily grown on laboratory

culture media. However, most strains generally do not

survive very well outside of the host’s gastrointestinal

tract. Therefore, its presence in the environment is

considered as evidence of recent contamination with

mammalian or avian feces. Because many fecal-borne

microorganisms are pathogenic in animals and

humans, the presence of E. coli in water and foods

is indicative of a potential hazard.

0002Environmentally, E. coli is most often associated

with fecally contaminated waters. Sources of contam-

ination include animal feces and discharge of un-

treated sewage. This has become a cause of concern

ESCHERICHIA COLI

/Occurrence 2149

in recreational areas and beaches, as such areas are

often closed when populations of E. coli increase.

E. coli-contaminated water is also of concern to the

food industry. Shellfish, through their filter-feeding

apparatus, can concentrate in their tissue E. coli and

other bacteria present in water. Many shellfish are

eaten raw or poorly cooked whereby contaminating

bacteria are ingested. In addition, the use of fecal-

contaminated water for irrigation of crops and the

use of animal manure as a fertilizer can cause contam-

ination of vegetables and other foods of plant origin.

Not only is E. coli an indicator of the possible pres-

ence of fecal-borne pathogens, but some strains are

themselves pathogenic. Most strains of E. coli are

harmless commensals and appear to be avirulent in

healthy humans. However, a few strains, with well-

characterized traits that are known to be associated

with pathogenicity, cause infections of the urinary

tract, peritoneum, central nervous system, wounds,

and gastrointestinal tract. Those of greatest concern

in water and foods are the intestinal pathogens, which

are classified into five major groups: the entero-

toxigenic E. coli (ETEC), the enteroinvasive E. coli

(EIEC), the enteropathogenic E. coli (EPEC) the

enterohemorrhagic E. coli (EHEC), and the entero-

aggregative E. coli (EAEC).

Entry into the Food Chain

0003 Since E. coli is naturally occurring in the intestinal

tract of animals, its entry into foods most often occurs

by direct or indirect contact with fecal material. Since

humans are the principal source of ETEC, EPEC,

EIEC, and EAEC, contamination with human feces

is the principal factor in transmission of these patho-

gens. E. coli O157:H7, the predominant serotype

among the EHEC, can colonize cattle and other

ruminants such as sheep and deer. In cattle, E. coli

O157:H7 localizes primarily in the rumen and colon,

and is shed in feces. Studies have revealed that

E. coli O157:H7 can survive in bovine feces for

many months but also retains its ability to produce

toxins. Two important routes by which foods may

become fecally contaminated with E. coli include:

0004 Contamination from food handlers. This can be

prevented by education of food handlers in proper

hygienic practices, e.g., washing hands thoroughly

after defecation.

0005 Contamination of animal products with feces

from the animal. During slaughter and subsequent

processing operations, contamination of carcasses

with E. coli O157:H7 from the digesta or feces of

cattle can occur. Hygienic slaughter practices and

treatments to clean and disinfect carcasses may be

inefficient in completely removing bacteria from

carcasses. During milking of dairy cattle, milk can

become contaminated with E. coli. One study has

found that 46% of raw milk samples and 29% of

milk products contained E. coli.

0006Water can also serve as a vehicle for contamination

of foods by E. coli. Water-borne outbreaks of E. coli

O157:H7 infection have been reported in the USA,

Scotland, Canada, and Japan. E. coli O157:NM was

implicated in a water-borne outbreak in South Africa.

At least four water-borne outbreaks of E. coli

O157:H7 infection in the USA were associated with

swimming water in lakes. Since natural waters often

contain E. coli of animal origin, and may be contam-

inated with animal feces and untreated human

sewage, their use in food harvesting or processing

can cause contamination of the food products, e.g.:

0007Direct contamination of seafood. This is especially

a problem with filter feeders such as shellfish, also

but also has been observed in many other types of

seafood.

0008Contamination of plants during irrigation or fer-

tilization. In some areas, raw human sewage is

used to fertilize crops intended for human con-

sumption. Fruits and vegetables grown on soil

fertilized with cattle manure or irrigated with

water contaminated with cattle manure have the

potential of being a vehicle of E. coli O157:H7. An

outbreak of disease caused by E. coli O157:H7

occurred after handling potatoes that had been

fertilized with cow manure.

0009Contamination of food-processing equipment or

food contact surfaces.

Occurrence in Foods

0010Since raw foods of animal and plant origin are subject

to potential fecal contamination, E. coli has been

isolated from a wide variety of foods. Foods that

have been associated with outbreaks of E. coli infec-

tions include meat and meat products, fish, poultry,

milk and dairy products, vegetables, and water. The

incidence of E. coli in foods varies greatly in different

countries. Investigators in India isolated entero-

toxigenic E. coli from 33–52% of uncooked chicken

sausages and 10–43% of uncooked pork sausages.

E. coli was detected in 4–7% of 597 retail seafood

products in San Francisco, USA. A survey of retail

meats revealed that E. coli O157:H7 was isolated

from 1–2% of beef, pork, poultry, and lamb samples

in Madison, USA. A study in Mexico revealed that 74

of 75 Mexican cheeses contained E. coli over accept-

able levels, whereas seven of 25 non-Mexican cheeses

contained unacceptable levels. In Egypt, 28% of fresh

2150

ESCHERICHIA COLI

/Occurrence

soft white cheese samples contained E. coli, of which

half the isolates were enterotoxigenic. Twenty-nine

percent of dairy products tested in the former

Yugoslavia contained E. coli. E. coli, including some

pathogenic serotypes, was isolated from 82 of 105

milk samples from various sources in India. However,

a survey of commercially available cheese and milk

samples for enterotoxigenic E. coli in the United

States revealed that none of the E. coli isolates

recovered was enterotoxigenic. A 1997 survey con-

ducted on the occurrence of E. coli in ready-to-eat

food products in the UK revealed that the bacterium

was present in 2.2% of beef products, 2.8% of chicken

products, 1.9% of ham products, 1% of pork prod-

ucts, 5% of plant products, and 2.8% of turkey

products. A study of retail foods in Seattle, USA,

found Shiga toxin-producing E. coli in 18% of pork,

12% of chicken, 7% of turkey, 10% of fish, and 5%

of shellfish samples. A recent survey of ground beef

for E. coli O157:H7 conducted by the United States

Department of Agriculture found only 59 positive

samples out of 7009 specimens, indicating that the

prevalence of this pathogen in US ground beef is

about 0.8%.

Fate of

E. coli

in Foods

0011 The behaviour of E. coli in foods is similar to that of

other Gram-negative bacteria. They are generally

sensitive to heat and are killed in foods that are

adequately cooked. Studies with E. coli O157:H7

have revealed the organism to be slightly more heat-

sensitive than Salmonella spp., with D-values in

ground beef at 57.2, 60, 62.8, and 64.3



C of 270,

45, 24, and 9.6 s, respectively. Heating apple juice at



C for 1.6 min reduced the E. coli O157:H7 popu-

lation by 4log

. D-values of E. coli in shrimp paste

were 5.8 and 0–3 min at 57 and 65



C, respectively.

Hot-water sprays are useful for reducing the bacterial

load on animal carcasses. Treating sides of beef with

water at 83.5



C for 20 s can reduce E. coli popula-

tions 3 log

without any detrimental effects on meat

color. E. coli O157:H7 is as sensitive to radiation

as most other enteric pathogens. For example, in

chicken, the D-values of E. coli O157:H7 at 5 and

5



C were 0.27 and 0.42 kGy, respectively. In

ground-beef patties at 4 and 16



C, the D-values

for E. coli O157:H7 were 0.24 and 0.30 kGy,

respectively, compared with 0.62 and 0.80 kGy,

respectively, for Salmonella.

0012 E. coli O157:H7 can grow in a variety of foods,

including meat, milk, cantaloupe, and watermelon

cubes, and salad vegetables such as lettuce and cu-

cumber. A study on the survival and growth charac-

teristics of E. coli O157:H7 in unpasteurized milk

revealed that the pathogen survived at 5



C for 28

days, and increased in populations at 8 and 15



with rapid growth growth observed at the latter tem-

perature. In pasteurized milk samples held at 22



E. coli O157:H7 populations increased by approxi-

mately 5.0 log

colony-forming units (CFU) ml

1

7 days, and then decreased to undetectable levels by

day 28. When inoculated in large populations, E. coli

O157:H7 could survive in mayonnaise (pH 3.6–3.9)

for 5–7 weeks at 5



C and for 1–3 weeks at 20



C. E.

coli O157:H7 also can survive for extended periods

of time in several acidic foods such as cheese, yogurt,

and apple cider. E. coli O157:H7 survived in cottage

cheese for more than 60 days. The pathogen can

survive the fermentation process in yogurt for 5 h at



C, and for 7 days in yogurt held at 4



0013Freezing can kill approximately 1 log

of E. coli in

foods. Prolonged frozen storage will not completely

kill E. coli unless the organism goes through several

freeze/thaw cycles. However, such conditions would

normally destroy the palatability of food. E. coli

O157:H7 survived for up to 9 months with no

decrease in population in ground beef held frozen at

20



C with no freeze/thaw cycles. E. coli is more

resistant to sodium chloride and sodium nitrite than

Salmonella. E. coli in brain–heart infusion broth can

grow in 4% sodium chloride and 400 mg of sodium

nitrite per milliliter at 15 and 35



0014Several studies have determined the fate of E. coli

in a variety of dairy products. Generation times of

E. coli in skim milk were 28–35 min at 32



C and 66–

109 min at 21



C. When lactic starter cultures were

included in the milk, growth of E. coli was inhibited

at both temperatures. A study of the fate of E. coli

during the manufacture of Camembert cheese revealed

that E. coli grew slowly until the curd was cut and

hooped, then increased rapidly. During overnight

storage at 10



C, as the pH of the cheese decreased

to 5.0 or below, the viable E. coli population

decreased sharply. Depending on the strain tested,

E. coli was not detectable in the cheese after 0–9

weeks of storage at 10



C. Inoculating E. coli directly

onto the surface of ripening Camembert cheese

resulted in rapid growth and subsequent survival for

up to 7 weeks of storage.

0015Parameters that influence the growth of E. coli in

foods roughly parallel those of other Gram-negative

bacteria. The minimum water activity (a

) for growth

ranges from 0.94 to 0.97. The optimum pH for

growth is approximately 7.0, with a minimum and

maximum pH for growth of 4.5 and approximately

9.0, respectively. Growth for some strains can occur

from 7 to 46



C, with maximal growth rates occur-

ring at 35–37



C. E. coli O157:H7 grows very poorly

or not at all in many media at temperatures at and

ESCHERICHIA COLI

/Occurrence 2151

above 44



C. E. coli is capable of very rapid growth in

high-nutrient foods. Populations can double within

30–40 min under optimal conditions.

Acid Tolerance of

E. coli

O157:H7

0016 E. coli O157:H7 has an unusual acid tolerance, as

evidenced by involvement in outbreaks associated

with high-acid foods such as apple cider, yogurt, and

fermented meat sausage. E. coli O157:H7 can survive

for extended periods of time in synthetic gastric fluid.

Further, exposure of E. coli O157:H7 to mild or

moderate acidic environments can induce an acid-

tolerance response, which enables the pathogen to

survive extreme acidic conditions. For example,

acid-adapted cells of E. coli O157:H7 survived longer

in apple cider, fermented sausage, and hydrochloric

acid than nonacid-adapted cells. However, E. coli

O157:H7 is not unusually heat-resistant or salt-

tolerant unless cells are preexposed to acid to become

acid adapted. Acid-adapted E. coli O157:H7 cells

also have increased heat tolerance.

0017 The effect of diet and gastrointestinal acidity in

cattle on E. coli O157:H7 and its acid tolerance has

received recent attention from microbiologists. A

study published in 1998 evaluated the role of diet

(grass vs grain) and the resulting volatile fatty acids

(VFA) levels and pH in the rumen and colon on

induction of acid tolerance in E. coli. The study

revealed that the rumen and colon fluids of grain-

fed cattle had higher VFA concentrations and lower

pH values than those of cattle fed only grass. Further,

larger populations of E. coli and acid-resistant E. coli

were found in rumen and colonic fluids of cattle fed a

high-grain diet than that in cattle fed grass only. A

following study by a different group of investigators

evaluated the effect of diet (high grain vs hay-rich

diets) on the duration of shedding of E. coli

O157:H7 by cattle and the acid resistance of the

pathogen shed by the animals. However, these investi-

gators observed that hay-fed cattle shed E. coli

O157:H7 longer than grain-fed animals, and the

acid resistance of E. coli O157:H7 was similar, irre-

spective of the diet.

Prevention of

E. coli

Contamination

0018 The most important considerations for preventing

E. coli contamination are:

0019 preventing fecal contamination by use of clean

water supplies, avoiding fecal material during

slaughter and milking, and practicing hygienic

food handling techniques;

0020thorough cooking of all foods, especially foods of

animal origin, including seafood, and pasteuriza-

tion of milk and fruit juices;

0021avoiding cross-contamination of cooked foods

with raw foods of animal origin and unsanitized

surfaces of food-processing equipment;

0022proper cleaning and sanitizing of food-processing

equipment;

0023practicing good personal hygiene;

0024storing cooked or perishable food at 4



C or lower.

See also: Cheeses: Types of Cheese; Surface Mold-

ripened Cheese Varieties; Cleaning Procedures in the

Factory: Overall Approach; Escherichia coli: Food

Poisoning; Food Poisoning by Species other than

Escherichia coli; Food Poisoning: Classification; Food

Safety; Lactic Acid Bacteria; Microbiology:

Classification of Microorganisms; Microflora of the

Intestine: Role and Effects; Milk: Analysis;

Pasteurization: Pasteurization of Liquid Products;

Shigella; Spoilage: Bacterial Spoilage; Starter Cultures;

Vibrios: Vibrio cholerae

Further Reading

Banwart GJ (1989) Basic Food Microbiology, 2nd edn.

New York: Van Nostrand.

Diez-Gonzalez F, Callaway TR, Kizoulis MGv and Russel

JR (1998) Grain feeding and the dissemination of acid-

resistant Escherichia coli from cattle. Science 281:

1666–1668.

Doyle MP and Padhye VV (1989) In: Doyle MP (ed.) Food-

borne Bacterial Pathogens, pp. 235–281. New York:

Marcel Dekker.

Gross RJ and Rowe B (1985) Escherichia coli diarrhea.

Journal of Hygiene, Cambridge 95: 531–550.

Hovde CJ, Austin PR, Cloud KA, Williams CJ and Hunt CA

(1999) Effect of cattle diet on Escherichia coli O157:H7

acid resistance. Applied Environmental Microbiology

65: 3233–3235.

International Commission on Microbiological Specifica-

tions for Foods (1980) Microbial Ecology of Foods.

Food Commodities, vol. 2. New York: Academic Press.

Jay JM (2000) Modern Food Microbiology, 6th edn.

Gaithersburg, MD: Aspen.

Kornacki JL and Marth EH (1982) Foodborne illness

caused by Escherichia coli: a review. Journal of Food

Protection 45: 1051–1067.

Meng and Doyle MP (1998) Microbiology of Shiga toxin-

producing Escherichia coli in foods, pp. 92–107. In:

Kaper JB and O’Brien AD (eds) Escherichica coli and

Other Shiga Toxin-producing E. coli Strains. Washing-

ton DC: ASM Press.

Meng J, Doyle MP, Zhao T and Zhao S (2001) Escherichia

coli O157:H7. In: Doyle MP, Beuchat LR and Montville

TJ (ed.) Food Microbiology: Fundamentals and Fron-

tiers, pp. 193–213. Washington, DC: ASM Press.

2152

ESCHERICHIA COLI

/Occurrence

Detection

K S Venkitanarayanan, University of Connecticut,

Storrs, CT, USA

M P Doyle, University of Georgia, Griffin, GA, USA

Use of

Escherichia coli

as Indicator

Organism

0001 Escherichia coli is the predominant member of the

facultatively anaerobic portion of the human colonic

normal flora. The bacterium’s only natural habitat is

the large intestine of warm-blooded animals and,

since E. coli, with some exceptions, generally does

not survive well outside of the intestinal tract, its

presence in environmental samples, food, or water

usually indicates recent fecal contamination or poor

sanitation practices in food-processing facilities. The

population of E. coli in these samples is influenced by

the extent of fecal pollution, lack of hygienic prac-

tices, and storage conditions. The mere presence of

E. coli in food or water does not indicate directly that

pathogenic microorganisms are in the sample, but it

does indicate that there is a heightened risk of the

presence of other fecal-borne bacteria and viruses,

many of which, such as Salmonella spp. or hepatitis

A virus, are pathogenic. For this reason, E. coli is

widely used as an indicator organism to identify

food and water samples that may contain unaccept-

able levels of fecal contamination.

0002 Most testing done to isolate E. coli from food

samples is undertaken to assess the acceptability

of manufacturing practices, including sanitation

regimes, used in food-processing facilities. Several

methods have been developed to achieve this.

The traditional cultural approach is a three-step

procedure. The first step estimates the number of

coliforms in the sample, the second step estimates

the number of fecal coliforms, and the final step

isolates E. coli. The coliform group is defined as

members of the family Enterobacteriaceae that pro-

duce acid and gas from lactose at 37



C. Included in

this group are E. coli and other species of the family

such as Citrobacter, Enterobacter, and Klebsiella.

Many of these genera are not primarily of fecal origin,

and so the presence of coliforms in a sample does not

necessarily indicate fecal contamination. The fecal

coliforms group includes those coliforms that produce

acid and gas from lactose when grown at temperatures

between 44 and 45.5



C. These organisms are typically

of fecal origin. E. coli is usually the dominant bacter-

ium of the fecal coliform group. The two most com-

monly used methods for enumerating coliforms, fecal

coliforms, and E. coli in foods both utilize the most

probable number (MPN) technique of statistically

estimating the number of bacteria in the sample. In

the method recommended by the American Public

Health Association, the Association of Food and

Drug Officials of the USA, and the Association of

Official Analytical Chemists (AOAC), the food or

water sample is decimally diluted, and 1 ml of each

dilution is transferred into each of three separate

tubes containing 10 ml of lauryl sulfate tryptose

(LST) broth with inverted Durham tubes. These

tubes are incubated at 35–37



C for 24 h. From each

tube with growth and gas production, a loopful of

culture is transferred into a tube containing 10 ml of

brilliant green lactose bile (BGLB) broth and an

inverted Durham tube. These tubes are incubated at

35–37



C for 24 h. Gas production is indicative of

lactose fermentation by coliform bacteria. The

number of positive tubes in each dilution should be

recorded and an MPN table consulted to determine

the number of coliforms in the sample. Both LST and

BGLB tubes should be reincubated for an additional

24 h to detect slow lactose fermenters.

0003For the quantitation of fecal coliform bacteria, a

loopful from each gas-producing LST culture is trans-

ferred into a tube containing 10 ml of E. coli (EC)

broth and an inverted Durham tube. These tubes are

incubated at 44.5



C for 24 h. Gas production is indi-

cative of fecal coliform bacteria. To continue with the

isolation of E. coli, gas-producing cultures in EC

broth are streaked onto plates of eosin-methylene

blue (EMB) agar or Endo agar and incubated for

24–48 h at 35–37



C. Representative colonies are

chosen and biochemically tested (Table 1).

0004Another method of detecting E. coli, recommended

by the UK Ministry of Health, and more commonly

used in European countries, also uses the MPN pro-

cedure. Decimal dilutions of food or water samples

tbl0001Table 1 Biochemical properties of E. c oli

Te s t E. coli ETEC EPEC EIEC O157:H7

Lactose fermentation þþþv þ

Gas from lactose þþþþ

Sorbitol fermentation þþþv 

Motility þþþþ

Indole þþþv þ

Methyl red þþþþþ

Voges–Proskauer 

Citrate 

Lysine decarboxylation þþþv þ

Ornithine decarboxylation v v v þ

b-Glucuronidase þþþþ

ETEC, enterotoxigenic E. coli; EPEC, enteropathogenic E. coli ; EIEC,

enteroinvasive E. c oli.

þ, > 90%positive; , <10%positive; v, 10–90% positive.

ESCHERICHIA COLI

/Detection 2153

are inoculated into tubes of MacConkey broth con-

taining inverted Durham tubes, with triplicate tubes

receiving 1 ml of each dilution. These tubes are incu-

bated for 24–48 h at 35–37



C, and the MPN of coli-

forms is determined using the number of tubes in each

dilution showing gas production. To detect fecal coli-

forms, a loopful from each tube of MacConkey broth

showing gas production after 24 h is transferred into a

tube containing 10 ml of BGLB broth and an inverted

Durham tube, and into a second tube containing 10 ml

of peptone water. These tubes are incubated at 44



for 24 h. The indole test is performed on the peptone–

water culture. Samples showing gas formation in

BGLB broth and indole production are considered

positive for fecal coliforms. These cultures can then

be streaked onto EMB or Endo agar, and representa-

tive colonies are biochemically confirmed as E. coli.

0005 Several agar media can be used for isolation of

E. coli. EMB agar contains eosin Y and methylene

blue to inhibit Gram-positive bacteria, and lactose for

differentiation of lactose-fermenting colonies. Strong

lactose fermentation by E. coli will cause formation

of a typical green sheen over dark colonies. A

medium commonly used for isolation of E. coli from

clinical specimens is MacConkey agar, which con-

tains crystal violet and bile salts to inhibit Gram-

positive organisms. E. coli will form bright pink to

red colonies, surrounded by zones of precipitated bile

salts/neutral red complexes. There has been a great

interest in recent years in the use of substrates of

b-glucuronidase in the detection of E. coli. Studies

have indicated that 94–99% of E. coli strains produce

this enzyme, whereas few other species do. Two

substrates that have garnered the greatest atten-

tion are 5-bromo-4-chloro-3-indolyl-b-glucuronide

(BCIG), which forms a blue-colored compound

when hydrolyzed by the enzyme, and methylumbelli-

feryl glucuronide (MUG), which forms a fluorescent

compound upon hydrolysis. Of the two, BCIG may

be more desirable because it does not require the use

of an ultraviolet light, and the color remains more

localized. Many investigators have suggested the add-

ition of one of these compounds to standard media to

further differentiate between E. coli and other bacter-

ial species. Researchers in one study filtered water

samples through membrane filters, incubated the

filters on Endo agar, and transferred filters onto nu-

trient agar containing MUG. They demonstrated that

98% of E. coli isolates hydrolyzed MUG to produce

fluorescent colonies within 4 h at 35



C with no false

positives. It is indicated by the AOAC that incorpor-

ation of MUG in LST broth in the MPN technique

may enhance the specificity of the method for

detecting E. coli. The AOAC is currently studying the

use of a hydrophobic grid membrane/b-glucuronidase

method for detection of E. coli. However, the b-glu-

curonidase test is not 100% sensitive or specific. Shi-

gella spp. typically also produce the enzyme. Several

b-glucuronidase-positive Salmonella strains have also

been identified. Some E. coli strains do not produce

the enzyme. One study has shown that a mean of

34% and a median of 15% of human fecal isolates

of E. coli were b-glucuronidase-negative when grown

in LST containing MUG. Among the consistently

b-glucuronidase-negative E. coli strains, there are

some pathogenic isolates, including the enterohemor-

rhagic E. coli O157:H7 strains.

Isolation of Pathogenic

E. coli

0006Methods used to isolate E. coli as an indicator organ-

ism from foods have not proved to be efficient for

isolating pathogenic strains of E. coli. This is largely

because pathogenic strains often differ considerably

from nonpathogenic E. coli in growth patterns.

Pathogenic strains frequently show delayed growth

at 44 and 45.5



C, particularly when initially present

in low populations. Some pathogenic strains will not

produce acid and gas from lactose in LST, BGLB, or

EC broths within 48 h. It has also been shown that

growth in media containing sodium lauryl sulfate and

growth at 44.5



C can cause a loss of plasmids,

known to encode many virulence factors associated

with pathogenic E. coli strains. One study indicated

that up to 95% of E. coli cells lost plasmids during

selective enrichment cultures. Therefore, the methods

commonly used for detection of E. coli as an indicator

organism should not be used to attempt isolation of

pathogenic strains from foods or water.

0007The US Food and Drug Administration Bacterio-

logical Analytical Manual describes a procedure for

isolating pathogenic E. coli that overcomes many of

the problems associated with the MPN procedures.

This method is not quantitative. Its goal is the isol-

ation, but not enumeration, of pathogenic E. coli

from foods. A 25-g sample of food is transferred

into a flask containing 225 ml of brain heart infusion

broth and is incubated for 3 h at 35



C to allow for

resuscitation of injured cells. The culture is mixed

with 225 ml of double-strength tryptone phosphate

broth and incubated at 44



C for 20 h. Representative

colonies are picked and tested for biochemical

characteristics.

Isolation of Pathogenic

E. coli

from

Clinical Specimens

0008The standard methods for detection and isolation of

E. coli from foods are generally not practiced in clin-

ical laboratories. The numbers of viable E. coli cells in

2154

ESCHERICHIA COLI

/Detection

clinical specimens are generally much higher than in

foods. Clinical specimens are usually streaked directly

onto selective-differential media such as MacConkey

or EMB agar without any enrichment or selective

enrichment cultures. Typical lactose-fermenting col-

onies are picked and tested with a battery of biochem-

ical tests (Table 1).

Detection of

E. coli

Virulence Factors

0009 The major problem in isolating pathogenic strains of

E. coli is that most E. coli possess no characteristics

that have been directly associated with production of

disease, such as production of toxins or invasiveness.

Therefore, it is assumed that most E. coli strains are

avirulent in healthy humans. Hence, when attempting

to isolate pathogenic strains, isolated colonies of

E. coli must be assayed for characteristics related to

pathogenicity. This is particularly a problem in fecal

specimens where large numbers of nonpathogenic

E. coli are usually present.

0010 The process of screening E. coli isolates most often

involves determining the serogroup (O antigen) by

agglutination in O-specific serum. Polyvalent sera

containing antibodies against somatic (O) antigens

frequently associated with enterotoxigenic E. coli

(ETEC), enteroinvasive E. coli (EIEC), enteropatho-

genic E. coli (EPEC), and enterohemorrhagic (EHEC)

strains are commercially available. This test is easy

and quick to perform. However, the disadvantages

are that the procedure does not specifically test for

disease-producing potential (e.g., not all strains of an

ETEC-associated serogroup are ETEC), and sero-

logical cross-reactions can occur between serogroups.

0011 Other tests to determine the virulence potential of

E. coli isolates include assays for toxicity. Among the

best known is the ligated rabbit ileal loop test, in

which cell suspensions are injected into a ligated seg-

ment of rabbit intestine. Heat-labile enterotoxin (LT)-

producing bacteria will cause fluid accumulation

within the segment. LT can also be detected by vascu-

lar permeability in animals and by inducement of

morphological changes in tissue culture cell lines Y–

I and Chinese hamster ovary (CHO). Culture filtrates

of heat-stable enterotoxin (ST)-producing ETEC

strains cause fluid accumulation in the intestines of

infant mice. Identification of EIEC and Shigella spp.

may include the Sereny test for invasiveness. This

involves applying a suspension of bacteria to the eye

of a guinea-pig. If the bacteria are capable of cellular

invasion, ulceration and opacity will occur in the eye

within 3 days. This test has been replaced to some

extent with invasiveness assays in tissue culture cells.

0012 The advent of molecular biological techniques has

provided methods to test directly for virulence factors

known or believed to be involved in pathogenesis,

often without the need for bacterial isolation at all.

Factors for which such tests have been developed

include toxins and adhesins. Polymerase chain reac-

tions (PCR) targeting heat-labile toxin genes of ETEC

and Shiga toxin genes of EHEC have been reported.

An E. coli species-specific PCR based on the malB

operon (gene encoding maltose permease) that detects

all E. coli strains was developed and applied for

detecting E. coli in water and soft cheese. Another

E. coli-specific PCR was also developed using primers

specific for 16S ribosomal RNA genes of the organ-

ism. Recently, a multiplex PCR that can simultan-

eously detect ETEC, Shiga toxin-producing E. coli,

and attaching and effacing strains of E. coli was also

reported. These methods have been widely researched

because they allow confirmation of the presence of a

particular microorganism faster and more accurately

than cultural methods. However, they often require

special equipment or reagents that most clinical,

public health, and food-microbiology laboratories

do not possess. This is particularly true in developing

countries, where there is a greater need for E. coli

detection systems. Another molecular method be-

coming commonly used for rapid and specific detec-

tion of various types of E. coli involves the use of

genetic probes. Radiolabelled or chemiluminescent-

labelled segments of DNA complementary to regions

of genes encoding toxins or other virulence factors are

hybridized with the sample, and the presence of the

target gene is indicated by the amount of bound

radioactivity. This procedure is very sensitive; the

specificity is dependent on the structure of the probe

and the conditions used during hybridization. Note

that genetic probes detect the genes for virulence

factors, and therefore presumably the organism itself,

whereas ELISA detects the virulence factor that is

expressed by the organism. Genetic probes have

been used to detect LT- and ST-producing bacteria in

food, water and stool, adhesins of ETEC, and the

invasiveness proteins of EIEC. DNA probes that util-

ize chemiluminescent substrates such as streptavidin/

biotin system eliminate the need for radioactivity,

with little reduction in sensitivity.

0013Enzyme-linked immunosorbent assay (ELISA) is an

immunological method used for detection of E. coli

strains. There are several forms of ELISA, but all are

variations on a single theme. The advantages offered

by ELISA methods include rapidity of diagnosis with

little or no decrease in sensitivity from tissue culture

or animal methods. ELISA has been used to detect LT

and ST, adhesins of ETEC, and invasiveness-related

outer-membrane proteins of EIEC in stool specimens

and culture filtrates. While having great utility in

detection of E. coli virulence factors in clinical

ESCHERICHIA COLI

/Detection 2155

samples, ELISA methods often do not perform as well

in food samples, due to interference from food com-

ponents.

Detection of

E. coli

O157:H7

0014 Isolation of enterohemorrhagic E. coli O157:H7

must be approached differently than using the

methods for isolating other E. coli strains because

E. coli O157:H7 grows poorly or not at all at 44



in many media used to isolate E. coli. None of the

methods described above can dependably isolate

E. coli O157:H7. However, E. coli O157:H7 has

some biochemical differences from most other E.

coli strains that can be exploited in isolation and

identification methods (Table 1). E. coli O157:H7

ferments sorbitol slowly or not at all, and does not

produce functional b-glucuronidase, whereas most

of the other E. coli strains are positive in both tests.

Further, E. coli O157 strains do not ferment rham-

nose on agar plates, whereas 60% of non-sorbitol-

fermenting E. coli belonging to other serogroups

ferment rhamnose on agar plates. Clinical laborator-

ies that test for E. coli O157:H7 usually streak fecal

specimens onto plates of sorbitol MacConkey agar

(SMAC) (which contains sorbitol instead of lactose).

Sorbitol-nonfermentative colonies are chosen and

serologically tested for the O157 and H7 antigens.

Positive colonies should then be tested for Shiga

toxin production. (See Immunoassays: Principles.)

0015 Several methods for isolation of this organism from

foods have been suggested. Media that can be used to

test both sorbitol fermentation and b-glucuronidase

activity have been developed. For example, Sorbitol

MacConkey agar containing MUG is commonly used

for differentiating E. coli O157:H7 from other E. coli

strains. In this medium, E. coli O157:H7 strains form

colorless and nonfluorescent colonies due to their

inability to ferment sorbtiol and break down MUG.

However, non E.coli O157:H7 strains in Sorbitol

MacConkey agar containing MUG, form pink col-

onies that are fluorescent under UV light. The recov-

ery rate of E. coli O157:H7 on Sorbitol MacConkey

agar can be improved by previous enrichment in

trypticase soy broth supplemented with cefixime

(0.5 mg l

1

) and vancomycin (8 mg l

1

). Another

method for improving the selectivity of Sorbitol Mac-

Conkey agar for isolating E. coli O157:H7 is inclu-

sion of cefixime and potassium tellurite (2.5 mg l

1

)

that permit growth of E. coli O157:H7 but inhibit

most of the other E. coli strains. Another study

indicated that addition of novobiocin can increase

the selectivity of the medium for isolating E. coli

O157:H7. Injured or stressed cells of E. coli

O157:H7 do not grow well on selective media such

as Sorbitol MacConkey agar, especially in the pres-

ence of selective agents such as novobiocin, cefixime,

and potassium tellurite. A study published from the

UK reported that delaying the exposure of injured E.

coli O157:H7 cells to selective agents by inclusion of

a recovery step such as resuscitating the cells on a

membrane placed on the surface of tryptic soy agar

followed by transfer of the membrane onto Sorbitol

MacConkey agar increased the detection of stressed

cells of the pathogen without sacrificing specificity.

Another medium, called Rainbow agar, has been

reported to be superior to Sorbitol MacConkey agar

in recovering heat-injured cells of E. coli O157:H7.

Rainbow agar contains chromogenic substrates

specific for b-galactosidase and b-glucuronidase, and

produces a spectrum of colored colonies ranging from

black to gray to red to blue to violet, which differen-

tiates E. coli O157:H7, E. coli O26:H11, other Shiga

toxin-producing E. coli, and Shiga toxin-negative

E. coli. Enterohemolysin agar is a nonselective, differ-

ential medium used for detecting enterohemolysin

expressed by about 90% of Shiga toxin-producing

E. coli.(See Salmonella: Detection; Shigella.)

0016DNA probes and polymerase chain reaction (PCR)

provide powerful and rapid tools for specific and

sensitive detection of E. coli O157:H7. Most probes

are directed against genes encoding Shiga toxins for

detecting E. coli O157:H7 and other Shiga toxin-

producing E. coli (STEC) strains. Shiga toxin produc-

tion is the only trait shared by all STEC that is not

found in other E. coli strains. Probes against all Shiga

toxins of strains producing human illness, including

Shiga toxin 1 (Stx1) and Shiga toxin 2 (Stx2), would

be necessary to detect all STEC strains, because some

strains produce only one Shiga toxin. Specificity

would not be 100%, because probes against Stx1

genes would also be expected to react with Shiga

toxin-producing Shigella strains. Additionally, it is

not known if all STEC are human pathogens. Many

studies have shown the capability of such Shiga toxin

gene probes to detect STEC strains in colony blots

and enrichment broth cultures. The probes have

also been successful at detecting the Shiga toxin-

producing bacteria in feces and in foods, including

ground beef, raw milk, and salads. Further, a variety

of PCR assays specific for attaching and effacing

(eae), hemolysin, plasmid pO157, and 60-Mda plas-

mid genes have been developed for detecting E. coli

O157:H7. Finally, a number of immunoassays based

on somatic (O) and flagellar (H) antigens specific for

E. coli serotype O157:H7 have been reported. An

ELISA procedure utilizing a monoclonal antibody

(4E8C12) specific for an outermembrane protein of

E. coli O157:H7 was capable of detecting E. coli

O157:H7 in foods in less than 20 h.

2156

ESCHERICHIA COLI

/Detection