relationships between indicator organisms and specific pathogens

D. K. MISKIMIN,
K. A. BERKOWITZ,
l4 C FRANKE,
M. SOLBERG,
Food Science
Rutgers-The
State
University
W. E. RIHA
R. L. BUCHANANand
Dept.,
JR.
V. O’LEARY
Cook College
of New Jersey, New Brunswick,
NJ 08903
RELATIONSHIPS BETWEEN INDICATOR ORGANISMS AND
SPECIFIC PATHOGENS IN POTENTIALLY HAZARDOUS FOODS
ABSTRACT
Quality audit data collected as part of a mass feeding quality assurance
program were analyzed to determine the relationships between the indicator tests (total aerobic plate count, coliform count and Escherichiu
colz) and the common food-borne pathogens (Staphylococcus aureus,
Clostridium perfringens and Salmonella). 132 raw foods and 593 readyto-eat foods were evaluated. The indicators were grouped into ranges
and compared to the pathogens and to each other in terms of detectability. There were correlations between the pathogens and the indicator ranges and between the indicators and the indicator ranges. The
value of the indicators in the e’valuation of food safety was tested by
setting standards and determining the numbers of correct and incorrect
decisions which would be made relative to the pathogens detected in
the foods. None of the indicators was suitable as a screening agent for
food safety.
INTRODUCTION
FOOD POISONING
is a universal illness which affects people
without
regard for age, sex, geography
or intelligence.
Approximately
75% of the food poisoning
outbreaks
in the
United States are caused by C. perfringens, S. aureus or Salmo-
nella species. The majority
microorganisms
have their
of outbreaks caused by these three
origins
in foods
served in food
ser-
vice establishments (Brachman et al., 1973). There has been a
search for methods which would provide a basis for controlling
and eliminating food poisoning outbreaks. Attempts have been
made to adapt microbiological methods which were of proven
value in nonfood systems, such as water. Since food systems
support growth of microorganisms, while water does not, the
adaptations have been subject to considerable question. The
methods have attempted to utilize nonpathogenic indicator
organisms. Among the most popular indicator tests are the
total aerobic plate count (TAPC), the coliform count and the
E. coli count.
It is generally believed that high TAPC in a food is indicative of a greater risk of pathogens being present, and the
presence of the coliform group of microorganisms signals unsanitary
conditions
and therefore
potential
pathogenicity.
ever, no correlation was found between coliforms and Salmonella species. Duitschaever et al. (1973) found staphylococci in
98% of the samples and coliforms in 95% of the samples of
raw ground beef they examined. Tomkin and Keuper (1973)
observed that the rate of SalmoneZZu isolations rose from 12%
to 70% as TAPC increased from <l O4 to lo7 organisms/g in
rendered animal by-products. Drion and Mossel (1972) observed that the ratio of Enterobacteriaceae to Salmonella in
dried foods was lOOO:l, and therefore, suggested testing l-g
aliquots for Enterobacteriaceae instead of testing sixty 25-g
samples for SuZmonelZa. Seligmann and Frank-Blum (1974)
studied barbecued chickens and discovered that when the
TAPC was less than lo4 organisms/g, C. perfringens, S. aureus
and enterococci were not present. As the TAPC rose to 104,
10’ and lo6 organisms/g or greater, the pathogenic organisms
were observed in 5%, 9% and 39% of the chickens, respectively.
The purpose of this study was to extract from quality audit
data, collected as part of a mass feeding food safety assurance
program, information
which established the relationships and
lack of relationships between the indicator organisms and the
pathogenic organisms most often associated with food poisoning outbreaks in both raw and ready-to-eat foods.
The
absence of coliforms is assumed to be an indication of safe
sanitary conditions (Buttiaux and Mossel, 1961; Thatcher and
Clark, 1968). E. coli has been the traditional indicator of fecal
contamination
and therefore of Sulmonella organisms (Buttiaux and Mossel, 1961). In a somewhat new approach, Read
and Baer (1974) clearly differentiate between the microbiological safety of foods, whether the foods contain pathogens
or their toxins; and the microbiological
quality of foods,
whether the numbers of indicator organisms exceed certain
limits. They also state that standards for safety factors and
quality factors are separate, as are the tests for them.
Correlations between indicators and pathogens have been
found by several authors in specific foods. Hagberg et al.
(1973) reported a correlation between coliforms and both S.
aureus
and C. perfringens in turkey processing plants. How-
MATERIALS & METHODS
Collection of samples
Samples of potentially hazardous foods were collected aseptically
from the kitchen and serving lines of 17 university dining halls and
snack bars. A sterile plastic bag (Whirl-pak) was used for each food. The
samples were transported to the laboratory in insulated chests containing ice packs. The samples were subsequently kept under refrigeration for 2-4 hr while awaiting further processing.
Salmonello enumeration
50g of each food sample were weighed out aseptically and combined
with 450 ml of lactose broth (Difco) in a sterile steel Waring Blendor
and blended at high speed for 1 min. The blended liquid was transferred
aseptically to a sterile flask The pH was checked and adjusted when
necessary to values between 6.5 and 7.0. The flask was incubated at
37°C for 4 hr to allow for repair of cells which may have been damaged
by mild treatments such as mixing or chilling and to prevent cell multi-
plication so that most probable number (MPN) counts of Salmonella
could be made. Tetrathionate broth (Difco) tubes of 10 ml each were
prepared Serial dilutions of lo-‘, 10e2 and lOA of the lactose broth
were implanted in triplicate into the tetrathionate broth and incubated
at 37°C for 18-24 hr. After incubation, a loopful from each tube was
streaked onto brilliant green agar (Difco) plates which were incubated
for 18-24 hr at 37°C and checked for suspect colonies with surrounding red agar. All negative plates were incubated at 37°C for another
24hr period One suspect colony from each plate was suspended in
tubes containing 10 ml of brain-heart infusion (Difco) and incubated at
37°C for 4-6 hr. A loopful from.each tube was transferred to a well of
the slide provided in the Fluoro-Kit for Salmonella Screening (Clinical
Science Inc.) and manufacturer’s directions for fluorescent antibody
staining were followed. The slides were examined with a Zeiss immunofluorescence microscope.
Volume 41 (1976)-JOURNAL
OF FOOD SCIENCE-1001
1002-JOURNAL
OF FOOD SCIENCE-Volume
41 (1976)
For the last 13-month period of this study, which lasted a total of
34 months, positive reactions with the fluorescent antibody method
were checked with the improved Enterotube (Roche Diagnostics). A
colony identified as Salmonella was considered representative of a
positive tetrathionate tube. An M P N count was determined.
Table
containing
Sample preparation
50g of each food sample were combined with 450 ml of 0.1% peptone and blended at high speed for 1 min in a sterile stainless steel
Waring Blendor. Serial dilutions were prepared for use in enumerating
C perfringens, S, aureus, total aerobes, coliforms and E. coli.
Total
C peifringens enumeration
Tryptose-sulfitccycloserine
agar was prepared according to the
method of Harmon et aL (1971) with the exception that the base
ingredients were dissolved in 1000 ml instead of 900 ml. Spread plates
were prepared in duplicate depositing 0.5 ml on the 10-l plates and 0.1
ml on each of the 10d2 and 10m3 plates. The plates were overlaid and
incubated under nitrogen in Anaerojars (Case Laboratories) at 37°C for
18-24 hr. Large black colonies with opaque zones were checked microscopically for correct morphology and counted as positive. The colonies
on the 10-l plates were added together to obtain a count of organisms
per g. The counts on the other dilutions were obtained by averaging the
number of positive colonies on the duplicate plates.
S. uureus enumeration
Spread plates of Baird-Parker medium (Difco) were prepared in
duplicate, depositing 0.5 ml on the 10-l plates and 0.1 ml on each of
the lo-’ and 10m3 plates. The plates were incubated at 37°C for 20-22
hr. Round black convex colonies with a clear zone surrounding the
colony were inoculated into tubes containing 10 ml of tryptic soy
broth (Difco) and incubated at 37°C for 18-24 hr. The coagulase tube
test was performed with rabbit coagulase plasma-EDTA (Difco). The
coagulase positive colonies were enumerated as previously described for
C perfringens.
Total aerobic plate count
Serial dilutions of 10-l to 10d6 were dispensed into petri dishes in
duplicate and pour plates using total plate count agar (Difco) were
prepared. They were incubated at 37°C for 18-24 hr. Plates containing
30-300 colonies were counted. If difficulty was encountered in differentiating food particles and colonies, the plate was flooded with a
0.3% solution of 2, 3, 5 triphenyltetrazolium
chloride and allowed to
stand at room temperature for 20 min to 2 hr. Red colonies that
developed were counted (Solberg and Proctor, 1960).
Coliform enumeration
Serial dilutions of IO-’ to lob6 were dispensed in triplicate into test
tubes, each containing 10 ml of lauryl tryptose broth (Difco) and an
inverted Durham tube. Each tube was mixed using a Vortex stirrer
(Scientific Industries, Inc.) and incubated for 42-48 hr at 37°C. The
tubes were examined for gas production. A loopful from each positive
tube was transferred to a test tube which contained 10 ml of 2% brilliant green bile broth (Difco) and an inverted Durham tube. They were
incubated at 37°C for 42-48 hr. Positive brilliant green tubes were
used to determine an M P N count for coliforms.
E. coli enumeration
A loopful from each positive brilliant green tube from the coliform
test was streaked onto a plate of Levine’s eosin methylene blue agar
(Difco). The plates were incubated at 37°C for 18-24 hr. The plates
were examined for green sheeny colonies. One such colony from each
plate was selected and an Enterotube (Roche Diagnostics) was inoculated. The Enterotube was incubated at 37°C for 18-24 hr and read
using the flow chart provided. Plates which contained colonies identitied as E. coli were considered as representative of positive brilliant
green tubes. An M P N count was determined.
Data analysis
The data were subjected to regression analysis. Ranges of counts for
each indicator test were selected. Chi square analysis was performed
comparing each indicator range to the presence or absence of pathogens. Ranges of TAPC and coliform counts were also compared to the
presence or absence of E. coli. From the Chi square analyses, the percent of samples positive for each pathogen and E. coli within each range
of indicator was calculated. For all computations which involved
ranges, a geometric mean of samples within each range was used so that
regression analysis could be performed. Using specific levels of indicators as standards, tables were constructed to show the percentage of
correct and incorrect decisions, relative to accepting or rejecting foods,
that would have been made.
Number
l-The
distribution
pathogens
of food
samples
tested
C. perfringens
S. aureus
Salmonella
Total pathogens
% with pathogens
RESULTS
of raw and ready-to-eat
Raw
132
27
66
8
76
58
food
samples
Ready-to-eat
593
18
21
8
40
7
& DISCUSSION
TABLE
1 shows the number of food samples in which pathcgenie organisms were found. A total of 725 potentially
hazardous food samples were tested:
132 samples were raw foods,
and of these, 119 were raw hamburger.
Other raw meats,
liquid eggs and raw pancake batter were also included as raw
foods. The other 593 food samples were classified as ready-toeat foods. This grouping included:
protein salads such as tuna
salad, egg salad, and turkey
salad; side dish salads such as
potato salad, macaroni salad, and cole slaw; cooked meats and
fish; processed luncheon meats; combination
main dishes; and
desserts. The total number of samples with pathogens is not a
sum of the number of samples containing
C perfringens, S.
aureus and Salmonella because some food samples were contaminated with more than one pathogen. Within the raw foods
58% contained
pathogenic
organisms, while only 7% of the
ready-to-eat
foods fell into this potentially
unsafe category.
This reduction
in the percent of foods contaminated
demonstrates the significant
contribution
of food preparation
procedures to reduction
of possible food borne illness transmission.
Table 2 represents the arbitrarily
selected ranges for the
indicator
tests and the ratios and percentages of both raw and
ready-to-eat
foods which contained
each of the three pathogens within
each of the selected ranges. In raw food samples
with TAPC greater than lo4 organisms/g,
approximately
50%
of the samples were positive for 5’. aureus. This level was fairly
constant regardless of the TAPC. There is no apparent explanation for this independence
of S. aureus from the TAPC.
Samples contaminated
with C. perfringens increased as the
TAPC increased.
Lillard
er al. (1973) reported
a similar relationship
between C. perfringens and TAPC in the lung tissue
of broilers. A possible explanation
of this relationship
between
an anaerobic organisim and an aerobic indicator test is that th’e
aerobic organisms used a major proportion
of the oxygen,
making the food a good medium for the anaerobic
C. perfringens. Both S. aureus and C. perfringens also showed a
reasonably
steady increase in the percentage of positive samples as the coliform count increased except within the 5 X lo3
to 1 O4 organisms/g range of the coliform count. By examining
the number of samples which make up the percentages of the
positive samples for these pathogens in Table 2, it is evident
that there were considerably
fewer samples within that range
than in most other ranges of coliform counts selected for consideration.
This would seem to diminish the importance
of the
higher percentages of C. perfringens and S. aureus within the
coliform
count range of 5 X lo3 to lo4 organisms/g.
When
coliforms
were not detected
(at the <3 level), 29% of the
samples contained S. aureus. All of these samples were samples
of raw hamburger
patties. These results are in general agreement with Duitschaever
et al. (1973) who found high coliform
and high staphylococci
counts in raw ground beef and with
Hagberg et al. (1973) who found positive coliform
counts as-
INDICATORS
Table 2-Ratios
coli counts within
and percentages
various ranges
of raw and ready-to-eat
food
samples which
contained
detectable
pathogens
and TAPC’s,
Indicator
Total
aerobic
test
plate count
I
I
Coliform
count
from
10
lo4
5 x lo4
105
5x lo5
lo6
3
50
100
500
5,000
lo4
lo5
5x
Escherichia
coli count
3
50
100
lo3
IO4
Salmonella
ratio
%
to
< 10
10'
5x104
105
5x105
106
>106
<3
lo5
<3
50
100
1,000
IO4
>104
o/12
2126
o/a
3/32
l/l 1
2122
O/6
J/24
o/24
l/21
3126
50
100
500
5,000
lo4
105
5x
>105
O/l
lo5
l/6
2/l 1
Of3
l/25
3i49
219
l/12
l/9
012
C. perfringens
ratio
%
0
0
a
0
9
9
9
O/l
l/16
1 I30
2/l 1
9/39
3/l 1
11124
0
4
o/7
1128
3114
3127
6130
516
4112
0
5
12
17
ia
0
4
6
22
a
11
0
sociated with higher counts of S. aureus and C. perfringens in
turkey processing plants.
The relationship between the raw food samples containing
C. perfringens or S. aureus and the various ranges of E. coli
contamination
selected was reasonably similar to that previously described when the data were evaluated with respect to
coliform organisms and TAPC. Although 29% of the samples
were positive for S. aureus when the E. coli count was <3
organisms/g, there was a steady increase in the percentage of
samples containing S. aureus as the E. coli increased. It is
interesting to,note that the S. aureus containing food samples
increased to the 50-70% levels as the E. coli counts increased
to the levels of the often cited standard of <lOO E. coZi/g and
remained at this level. The rise to 100% of the samples containing S. aureus in those samples contaminated with >104 E.
coli organisms/g is the result of only two samples, and therefore, may be of little significance.
In raw foods, when presumptive positive tests for Salmonella species were encountered, 8-9% of the samples were
contaminated regardless of TAPC. Although the Salmonella
data are derived from a small number of positive samples,
there is a consistency, which may be seen in Table 2, showing
that the TAPC ranges which have no Salmonella present
always contained a total number of samples lower than the
other ranges, and therefore, a single positive response might
still be unexpected. When the percentage of raw food samples
presumptively positive for Salmonella was evaluated in terms
of the coliform count ranges, a fairly steady increase in contaminated samples was evident as the coliform range increased.
The linear correlation coefficient was 0.704. This relatively
high degree of correlation may not be of true significance
because there were only eight presumptively positive Salmo-
1/3
O/27
9157
l/10
6115
4110
212
S. aureus
ratio
%
Salmonella
ratio
%
counts
or E.
foods
C. perfringens
ratio
%
S. aureus
ratio
%
oiia
41361
0151
1114
2125
o/9
1 I40
0
1
0
7
a
0
3
0180
91368
3154
l/14
O/25
0110
5140
0
2
6
7
0
0
13
if80
81369
3154
II15
l/25
l/l0
6l40
I
2
6
7
4
10
15
217
29
13129
45
3/14
21
13128
46
la/28
64
100
616
7112
58
314
75
-
21367
31105
O/l 7
O/25
2127
o/4
II13
o/4
O/7
1
3
0
0
7
0
a
0
0
61374
31106
i/la
2126
4129
014
ii13
o/4
l/7
2
3
6
8
14
0
a
0
14
61376
l/l06
Ifra
5126
6129
o/4
O/13
214
o/7
2
1
6
19
21
0
0
50
0
al28
29
29159
49
6110
60
10115
67
7110
70
212
100
71494
1145
o/2
o/10
o/3
o/1
1
2
0
0
0
0
al502
2i48
213
4/10
l/3
2
4
67
40
33
100
0
6
3
O/l
2115
17130
0
13
57
ia
23
27
46
5111
21137
6111
15126
45
58
55
58
0
4
21
11
20
a3
33
33
0
16
10
40
40
100
coliform
Ready-to-eat
Raw foods
Indicator
range
(organisms/g)
AND PATHOGENS-1003
111
101504
3148
113
4/10
II3
O/l
2
6
33
40
33
0
nella samples. The percentage of raw food samples which
showed a presumptive Salmonella count was erratic with
respect to the increasing E. coli count ranges. This was surprising since both organisms are considered to be of similar
origin and both demonstrate similar cultural characteristics.
The small number of samples involved may explain the apparent anomaly.
Since E. coli may also be considered a potential pathogen,
its relationship to the TAPC and the coliform count was evaluated in both raw and ready-to-eat foods. The results are presented in Table 3.
When the presence of E. coli was compared to the TAPC in
raw foods, it showed a rapid rise to the 80-100% contaminated sample level by the time the TAPC was 5 X lo4 to 10’
organisms/g. It maintained this level at higher TAPC. This
shows a correlation between higher TAPC and presence of E.
coli in raw foods.
When E. coli was compared to the coliform count in raw
foods, the percentage of samples positive for E. coli rose to
60-80% when the coliform count was 50-100 organisms/g
and showed an upward trend at higher coliform counts. In raw
foods, therefore, some of the coliform count usually consisted
of E. coli. Thus, the results show that more samples of raw
foods contained E. coli as the TAPC and the coliform count
became higher. This is in agreement with the study of Hans et
al. (1973), who found interrelationships
in raw beef steaks
between TAPC, coliforms and E. coli.
A very significant relationship was found between TAPC
and E. coli in the ready-to-eat foods. When all of the ready-toeat foods were grouped together as shown in Table 3, there
was a fairly steady rise in the percentage of samples which
contained E. coli as the TAPC increased.
1004-JOURNAL
OF FOOD SCIENCE-Volume
41 (7976)
Table 3-Ratios
and percentages
of raw and ready-to-eat
food
samples which contained detectable
E. coli organisms and TAPC’s or
coliform
counts within various ranges
Escherichia
Indicator
test
TAPC
Indicator
range
(organisms/g)
from
to
Raw food
Ratio
%
IO
O/l
8/l 5
lo4
5x104
105
5x105
lo6
Coliform
count
-
3
50
100
500
5,000
1 o4
IO5
5x105
<IO
IO4
5 X lo4 20130
lo5
IO/IO
5 X IO5 26132
coli
Ready-to-eat
foods
Ratio
%
0 oil30
53 261362
67 9153
100
3114
81 4127
11112
28131
92 4111
90 19135
<3
50
O/J
23129
0
79
62
85
88
83
8/l 3
500
5,000
22126
30134
516
IO4
lo5
5x
>5x
IO5
lo5
12112
3/4
-
17
21
15
IO6
>106
100
0
7
3’3
54
O/386
211103
6/l 7
11122
14128
O/4
J/l3
75 O/3
- 6/J
100
In ready-to-eat food samples, there were only 7% which
contained pathogenic organisms in comparison to the 58% for
the raw foods (see Table 1). The vast majority of the ready-toeat food samples also contained low levels of the indicators.
When the various ranges of TAPC were compared to the
percentage of samples containing pathogens in ready-to-eat
foods as shown in Table 2, the only constant relationship that
could be found was a slow rise in the percent of samples which
contained S. aureus. The microflora contained in ready-to-eat
foods are the result of its handling in post heating preparation
since the original contamination is often eliminated by some
form of heat treatment.
The percentage of samples containing C. perfringens also
increased but none was found in 10’ -lo6 organisms/g range.
Table 4-Chi
square analyses
Thirty-five of the samples fall into this range of TAPC. Twelve
of these 35 samples are protein salads which are not a common
vehicle of C. perfringens. Only 7 of the 35 are luncheon meats
or cooked hamburger which are foods in which C. perfringens
is more likely to be found.
When the percentage of ready-toeat
food samples containing pathogens was compared with coliform count ranges,
as shown in Table 2, the percentage of samples which contained C. perfringens and S. aureus increased until the coliform
count reached 5000 organisms/g. At counts of >5000 organisms/g, the relationship broke down, probably because of the
low number of samples in that range. The relationship of the
percentage of C. perfringens contaminated samples and the E.
coli count was erratic in the ready-to-eat foods, and due to the
small number of samples, inconclusive. The percentage of samples positive for S. aureus within the various ranges of E. coli
counts rose to 33-40% when there were more than 50 E. coli
organisimslg in the foods.
There were only eight ready-to-eat food samples out of 578
(1.4%) tested for Salmonella which were presumptively positive. Therefore, no results with any real meaning could be
obtained. The apparent lack of correlation between E. coli and
Salmonella is evident just as it was in the raw foods. Of 16
samples which contained more than 50 E. coli organisms/g,
none contained Salmonella.
When E. coli was compared to the ranges of coliform count
in ready-to-eat foods, the percentage of samples positive for E.
coli increased from 0% at <3 coliform organisms/g to a 50%
level at about 500 coliform organisms/g. This 50% level was
generally maintained at the higher coliform counts (Table 3).
All of the data from Tables 2 and 3 were subjected to Chi
square analyses to evaluate statistically the various relationships. The statistical analyses are presented in Table 4. The low
correlations in both raw and ready-to-eat foods between all
indicators and the presence of Salmonella stand out. When
considering the presence of C. perfringens, there is high correlation with coliforms in ready-to-eat foods and E. coli in
both raw and ready-to-eat foods. There is high correlation
demonstrated for the presence of S. aureus with both coliform
counts and E. coli counts in ready-to-eat foods. When considering the correlations between the indicator tests among
themselves, there is a high level of agreement between all
combinations with the exception of TAPC and E. coli in raw
foods. Thus, it appears that all three indicator tests are closely
related to one another but their relationship to the presence of
pathogens is inconsistent.
for raw and ready-to-eat
food
samples.
Raw foods
Total
Chi square
TAPC vs C. perfringens
TAPC vs S. aureus
TAPC vs Salmonella
TAPC vs E. coli
TAPC vs Coliforms
Coliforms vs C. perfringens
Coliforms
vs S. aureus
Coliforms vs Salmonella
Col iforms vs E. co/i
E. coli vs C. perfringens
E. coli vs S. aureus
E. coli vs Salmonella
17.661
11.066
2.056
18.529
90.058
27.486
16.209
6.794
34.723
29.422
11.983
9.521
Degrees of
freedom
6
6
6
6
42
9
9
9
9
7
7
7
Ready-to-eat
Probability
> Chi square
0.0073
0.0856
0.9142
0.0052
0.0001
0.0012
0.0623
0.6599
0.0001
0.0001
0.1003
0.2165
Total
Chi square
18.117
20.949
13.922
92.133
403.318
46.237
93.848
16.101
210.243
127.075
62.526
0.429
foods
Degrees of
freedom
Probability
> Chi square
6
6
6
6
48
10
10
9
10
5
5
5
0.0061
0.0020
0.0304
0.0001
0.0001
0.0001
0.0001
0.0644
0.0001
0.0001
0.0001
0.9927
INDICATORS
Table B-Linear
correlation
coefficients
obtained
when
sion analysis was performed
on the number of pathogens/g
the number of indicator organisms/g.
Ready-to-eat
foods
Raw
foods
TAPC vs C. perfringens
TAPC vs S. aureus
TAPC vs Salmonella
TAPC vs E. coli
TAPC vs Coliforms
Coliforms vs C. perfringens
Coliforms vs S. aureus
Coiiform vs Salmonella
E. coli vs C. perfringens
E. coli vs S. aureus
E. coli vs Salmonella
regresversus
0.231
0.221
0.371
0.352
0.607
0.344
0.342
0.385
0.450
0.139
0.245
0.012
0.344
0.480
0.207
0.487
0.173
0.428
0.704
0.479
0.352
0.708
When the actual numbers of organisms/g of food were compared to one another by regression analyses, the linear correlation coefficients presented in Table 5 were obtained. From
these data, it is obvious that the number of indicator organisms present has essentially no relationship to the number of
pathogens which may be present in a food sample.
By comparing Tables 4 and 5, it can be concluded that a
high TAPC is indicative of the presence of E. coli or coliforms
but it is not indicative of a high number of E. coli or coliforms
being present. This same trend was present throughout all of
the comparisons with the exception, possibly, of the Salmonella counts in raw foods. The small sample sizes involved
make it difficult to evaluate fully these data.
The real relationships between the indicator tests and the
pathogens can probably best be seen by evaluating the effect
of utilizing various indicator standards upon the number of
Table B-Comparison
of correct and incorrect
decisions
subsequently
applied in the evaluation of food safety.
which
would
correct or incorrect decisions, which would subsequently be
made with respect to food samples. Table 6 shows the responses if the indicator tests had been used as the only criteria
for accepting or rejecting the foods tested. For an indicator to
successfully fulfill the function of indicating the safety of
foods, it cannot permit foods containing pathogens to be accepted. The incorrect decision of rejecting a food which does
not contain pathogens is an economic error which one cannot
afford to make. The incorrect decisions of accepting a food
containing pathogens is potentially unsafe and illegal, since
such a food is legally unwholesome. It may be seen in Table 6
that a standard of acceptance set at the level of <lo4 organisms/g of TAPC would give more correct decisions than any
of the other suggested standards for raw foods. Even at this
level, there were 32% incorrect decisions (the sum of those
rejected without pathogens and those accepted with pathogen), most of which would have rejected wholesome food.
For the ready-toeat foods, if an unrealistic standard of <lO
organisms/g of TAPC were used, no ready-to-eat food samples
with pathogens would be accepted. However, 80% of the total
samples would be rejected when safe. At a standard of <lo6
organisms/g of TAPC, 90% of the decisions would have been
correct, and 5% of the total samples would be accepted containing pathogens when only 7% of the total samples contained pathogens.
Considering the coliform count as an index of safety, Table
6 shows that if the standard for raw food was set at <lOO
organisms/g, which was the optimum level, 34% of the decisions would be incorrect. To eliminate the rejection of foods
which were free of pathogens, the standard would have to be
set at <lo’ coliform organisms/g, and this would allow 58% of
the foods accepted to contain pathogens. The assurance of
safety could not be guaranteed in either raw or ready-to-eat
foods because even when the standard was set at <3 organisms/g, there were 2% of the food samples which would have
been accepted containing pathogens. In the ready-to-eat foods,
if the standard was set at <lo’ coliform organisms/g, 92% of
the decisions would have been correct but all of the samples
which contained pathogens would have been accepted. The
only reason that the majority of the decisions would have been
be made if various
standards
were established
Raw foods
Correct
Indicator
Total
aerobic
Coliform
Escherichia
test
plate count
count
coli count
Rejected
containing
pathogens
1%)
decisions
Accepted
without
pathogens
(%I
AND PATHOGENS-1005
for the indicator
Ready-toeat
Incorrect
Rejected
without
pathogens
W)
decisions
Accepted
containing
pathogens
(%I
Standard
Total
no. of
samples
<lo’
<IO4
<1os
<IO6
124
124
124
124
61
59
40
14
1
9
19
34
38
30
19
5
22
47
<3
<IO2
<lo3
<104
<IO5
119
119
119
119
119
59
43
30
9
3
4
23
31
37
39
35
17
9
3
0
<3
<lo*
<IO”
<IO4
114
114
114
114
53
18
8
2
17
37
38
39
23
3
2
0
0
2
Correct
Rejected
Total
containing
no. of pathogens
samples
(%I
decisions
Accepted
without
pathogens
(%I
tests and
foods
Incorrect
decisions
Rejected
without
pathogens
(%)
Accepted
containing
pathogens
(%I
7
3
2
2
13
72
83
88
80
21
10
5
2
17
30
51
58
581
581
581
581
571
571
571
571
571
5
3
2
1
0
62
82
85
90
92
31
11
8
3
1
7
42
52
59
557
557
557
557
3
1
0
0
84
92
93
93
9
1
0
0
4
6
7
7
1006-JOURNAL
OF FOOD SCIENCE-Volume
41 (1976)
correct is that most of the ready-to-eat foods did not contain
pathogens.
Using E. coli organisms as an indicator, the most correct
decisions were made in raw foods when the standard was set at
<3 E. coli organisms/g. In the ready-to-eat foods, the most
correct decisions were observed when the standard was set at
<lOO organisms/g. It is also evident that if all raw food samples without pathogens were to be accepted, the standard
would have to be set at <IO4 E. coli organisms/g, which would
require 59% of the samples containing pathogens to be accepted. Thus, to solve the economic problem, using E. coli as
an indicator test would create a potentially dangerous raw
food supply. In ready-to-eat food samples, if the standard was
set at <3 organisms/g, 87% of the decisions were correct but
4% of the samples would have been accepted containing pathogens. If the standard was set at <IO3 E. coli organisms/g, all
ready-toeat food samples would have been accepted, although
7% of them contained pathogens.
CONCLUSIONS
NONE of the three indicators evaluated could serve as a food
safety assurance test for raw or ready-to-eat foods.
TAPC, coliform count and E. coli count are related to one
another in both raw and ready-to-eat foods. Any of the three
indicator tests is, therefore, suitable to insure the procedural
integrity of food preparation activities.
TAPC is less time consuming and is a direct count method,
whereas both the coliform count and the E. coli count are
indirect most probable number techniques.
The TAPC is, therefore, the most suitable method for the
evaluation of microbiological quality of foods and the search
for specific pathogens is necessary to insure the safety of
foods.
REFERENCES
Brachman,
P.S., Taylor, A.. Gangarosa, E.J., Merson, M.H. and Barker,
W.H. 1973. Food poisoning in the, U.S.A. In “The Microbiological
Safety of Foods.” Proceedings of the Eighth International
Symposium on Food Microbiology,
Reading, England. September,
1972.
Academic Press.
Buttiaux.
R. and Mossel. D.A.A. 1961. The sinnificance
of various oraanisms’ of faecal origin in foods and dringing
water. J. APP~. Bacteriol. 24: 353.
Drion, E.F. and Mossel, D.A.A. 1972. Mathematical-ecological
aspects
of the examination
for Enterobacteriaceae
of food processed for
safety. J. Appl. Bacterial. 35: 233.
Duitschaever,
C.L.. Arnott,
D.R. and Bullock, D.H. 1973. Bacteriological auality
of refrigerated
around beef. J. Milk Food Technol. 36:
375:
E.A. and Arnold,
E.A. 1973.
Hagberg,
M.M.. Busta. F.F., Zottola.
Incidence
of potentially
pathogenic
microorganisms
in furtherprocessed turkey products.
J. Milk Food Technol. 36: 625.
Hans, J.C., Fievez, L. and Granville,
A. 1973. Qualite bacteriologique
des steaks de boeuf crus prets a la cuisson (Bacteriology
of raw.
kitchen-ready
beef steaks). Ann. Med. Vet. 117: 157. [Cited in Biol.
Abs. 57: 229.1
Harmon, S.M., Kautter,
D.A. and Peeler, J.T. 1971. Improved
medium
for enumeration
of Clostridium
perfringens.
Appl. Microbial.
22:
688.
Lillard, H.S., Klose, A.A.. Hegge, R.I. and Chew, V. 1973. Microbiological comparison
of steam (at sub-atmospheric
pressure)
and immersion-scalded
broilers. J. Food Sci. 38: 903.
Read, R.B. Jr. and Baer, E.F. 1974. Role of the regulatory
in setting
microbiological
quality standards. Food Technol. 28(10): 42.
Seligmann,
R. and Frank-Blum,
H. 1974. Microbial
quality
of barbecued chickens
from commercial
rotisseries.
J. Milk Food Technol.
37: 473.
Solberg. M. and Proctor,
B.E. 1960. A technique
utilizing
2,3,5-triohenyltetrazolium
chloride
for recognition
of bacterial colonies in
the presence of large numbers of food particles. Food Technol. 14:
343.
in Foods: their
Thatcher,
F.S. and Clark, D.S. 1968. “Microorganisms
significance
and methods of enumeration.”
University
of Toronto
Press.
Tompkin,
R.B. and Kueper. T.V. 1973. Factors influencing
detection
of salmonellae in rendered animal by-products.
Awl. Microbial.
25:
485.
M S received 9127175; revised 2126176; accepted 2128176.
A paper of the Journal Series New Jersey Agricultural
Experiment
Station,
Cook College, Rutgers-The
State University
of New Jersey,
Dept. of Food Science, New Brunswick,
New Jersey.