summary writing

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 1980, p. 346-351 0099-2240/80/00-0346/06$02.00/0

Vol. 40, No. 2

Estimation of Escherichia coli in Raw Ground Beef MICHAEL E. STILES* AND LAI-KING NG

Faculty ofHome Economics, The University ofAlberta, Edmonton, Alberta, Canada T6G 2M8

This study was undertaken to establish and evaluate more rapid methods of estimating Escherichia coli in ground beef than the standard most probable number (MPN) technique. Direct inoculation of and modifications to EC medium gave unreliable estimates of the presumptive E. coli count. Solid media incubated at an elevated temperature were compared to the MPN technique. Anderson and Baird-Parker's tryptone bile agar (TBA) method and prepoured plates of Endo, Levine eosin methylene blue (EMB), and violet red bile (VRBA) agars incubated at 44°C gave equivalent counts to the standard MPN method. Anderson and Baird-Parker TBA was the most selective solid medium for E. coli estimation, but all selective media incubated at elevated temperature reduced apparent E. coli counts by as much as 50%. Indole-producing and lactose-fermenting Entero- bacteriaceae, capable of growth at elevated temperature, were tested for their growth on TBA, EMB, and VRBA at elevated temperature. TBA was selective for E. coli biotype I compared to other Enterobacteriaceae that predominate in meats. VRBA and EMB incubated at elevated temperature were not as selective as TBA, but differences in colonies could be observed between typical E. coli colonies and other Enterobacteriaceae on these media. Therefore, VRBA incu- bated at elevated temperature is proposed as a quality assurance screening test for presumptive E. coli in ground meat. Resuscitation techniques and prepoured plates with VRBA increased recovery levels of presumptive E. coli, but, under the conditions of this study, not to levels that represented a significant practical difference.

The increasing concern and interest for Esch- erichia coli biotype I in food as agents of food- borne enteritis (6, 19) or as indicators of fecal contamination have resulted in a need for more rapid and direct methods of enumeration of E. coli. The standard most probable number (MPN) technique (6) is laborious, requiring up to 5 days for confirmed coliform determination and an additional 5 days to confirm E. coli biotype I (5). In a collaborative study by the International Commission of Microbiological Specifications for Foods (14), the MPN tech- nique was shown to be unreliable because of variability in the results between laboratories. Incubation at an elevated temperature between 44 and 46°C has been widely recommended for screening "fecal" coliform bacteria and/or E. coli in foods (3, 4, 15). Studies have also been conducted using selective media at elevated tem- peratures (7, 12). Anderson and Baird-Parker (1) developed a rapid and direct method for enu- meration of E. coli I, based on the observation that E. coli produce indole when grown at 44WC on cellulose acetate membrane on a tryptone bile agar medium (2). This method has been subjected to international comparative studies (13), and it was found to be preferable to the MPN technique for enumerating E. coli in raw

meats, because of lower variability, better recov- ery of E. coli from frozen meats, rapidity, and decreased cost of analysis.

Despite the marked improvement in time sav- ing with the membrane technique, its inclusion as an acceptable method should not necessitate its use for general quality assurance. In our laboratory, the Anderson and Baird-Parker membrane technique has proven relatively ex- pensive, and routine screening of food samples could possibly be achieved more economically. In a previous study by Ng and Stiles (10), it was shown that violet red bile agar (VRBA) incu- bated at 35°C for 18 h, but not exceeding 24 h, could be used to obtain a reliable estimate of total coliform and Enterobacteriaceae counts, eliminating the need for a modified VRBA me- dium, as proposed by Mossel and co-workers (8, 9). Comparisons of media for determining coli- forms and fecal coliforms in ground beef (11) did not use elevated temperature incubation for the solid selective media. The purpose of this study was to evaluate alternate elevated-temperature methods for the estimation of E. coli biotype I in ground meat. In particular, direct inoculation of meat homogenate dilutions into EC medium at elevated temperature and the use of standard selective coliform media at elevated temperature

346

ESTIMATION OF E. COLI IN RAW BEEF 347

in comparison to the Anderson and Baird-Par- ker membrane technique were considered.

MATERIALS AND METHODS Source of cultures. Enterobacteriaceae isolated

from meats and meat-handling equipment on VREA (Difco) and VRBA with 1% glucose added (after Mos- sel et al. [9]), were identified using the Minitek Disk technique (BBL Microbiology Systems) as used in an earlier study by Ng and Stiles (10). Organisms were identified using the Minitek Coding System (BBL Microbiology Systems). Organisms that produced acid from lactose (Lac') and/or indole from tryptone (Ind') and that grew at 440C in tryptic soy broth (TSB) were selected for this study. American Type Culture Collec- tion strains were used as reference organisms where appropriate. MPN technique. The standard MPN technique

(5), or its three-tube modification (6), was used as the reference indirect method (ECin) for this study. The MPN technique was used for comparison with other modifications and counts on solid media at elevated temperature. Modifications to the ECin technique in- cluded the following: (i) direct inoculation of meat homogenate dilutions into EC medium and incubation at 450C (ECdir); (ii) ECdir allowing 1-, 2-, 3-, and 4-h resuscitation time at 210C, before incubation at 450C; and (iii) addition of 0.1 and 0.01% sodium lauryl sulfate (Fisher Chemicals) or 0.1% Tween 80 (Difco) to EC medium in the ECdir technique. Three separate trials were conducted, two using commercially prepared ground-beef samples, the other using laboratory pre- pared and artificially contaminated ground beef. Ground-beef inoculation study. A 2.5-kg sample

of fresh stew beef was divided into five portions, ground aseptically in the laboratory, and handled as follows: three portions were inoculated separately with three strains of E. coli I that had been isolated from meats and meat-handling equipment, one portion was inoculated with E. coli ATCC 11775, and one portion was used as an uninoculated control. E. coli cultures for ground-beef inoculation were grown in stationary culture in TSB at 350C for 24 h for three successive days before being used to inoculate the ground beef with 104 to 105 E. coli per g. The inoculum was thoroughly mixed into each portion, using a sterile spatula, before being further divided into 10 subpor- tions and treated as follows: five subportions were frozen and stored at -20°C; the other five subportions were stored at 40C, and a sample was analyzed on each of 4 days, i.e., after 0, 2, 4, and 6 days of storage. The frozen subportions were removed from the deep freeze after 28 days of storage, thawed, and held at 4C, as described for the nonfrozen subportions. All samples were prepared for microbiological analysis by weighing 11 g and blending with 99 ml of sterile 0.1% peptone water for 5 min in a stomacher bag (10). Appropriate dilutions were prepared with 0.1% pep- tone water blanks at 210C, inoculated onto bacterio- logical media (Difco), and incubated as follows: (i) standard plate count (SPC) agar at 350C for 24 h; (ii) VRBA pour plates (VRBA-1) and prepoured plates (VRBA-2), overlaid with 5 ml ofVRBA and incubated at 44.50C for 18 h (7); the dilutions were also spread on prepoured (5 to 8 ml) nutrient agar plates, incu-

bated at 44.5°C for 2 h, similar to the method of Speck et al. (18), except for resuscitation temperature, and overlaid with 20 ml of VRBA; incubation was contin- ued at 44.50C (VRBA-R); (iii) prepoured Levine eosin methylene blue (EMB) and Endo agars incubated at 44.50C; (iv) Anderson and Baird-Parker (1) tryptone bile agar (TBA) using "Nuflow" cellulose acetate membranes (Oxoid), incubated at 44.5°C; and (v) the three-tube MPN techniques (ECdir and ECin) at 450C. Results were read at 24 h, and at 48 h in the MPN technique, where appropriate. Pure culture studies. A total of 651 Enterobac-

teriaceae isolated from meats and meat-handling equipment were screened for growth in TSB at 44°C. A total of 386 of these cultures that grew at 440C were subcultured in TSB at 350C on two successive days before preparing serial dilutions of an 18- to 24-h culture for inoculation onto growth media. All cultures were plated onto tryptic soy agar (TSA) and incubated at 350C, as a reference; Ind- cultures were plated onto prepoured EMB and pour plates of VRBA; Ind' cul- tures were also plated onto TBA. All selective media were incubated at 440C for 24 h.

Selection ofincubation temperatures. All MPN tests were incubated at 45 ± 0.020C based on the accepted Canadian Health Protection Branch method (5). For the ground-beef study, solid media were in- cubated at 44.50C, based on the recommended tem- perature of the International Commission of Microbi- ological Specifications for Foods (6), including the Anderson and Baird-Parker TBA test which has a recommended incubation temperature of 440C (1). In contrast, for the pure culture study, in which other media were being compared to the TBA technique, the incubation temperature was 440C.

RESULTS Comparison of MPN modifications. Pre-

sumptive E. coli counts on 54 commercial sam- ples of nonfrozen ground beef, determined by direct (ECdir) and indirect (ECin) inoculation of EC medium, were compared by t-test, which indicated a significantly lower count by ECdir (P = 0.001). The geometric mean by ECin was 140 per g, compared to 74 per g by ECdir. However, ECin counts could not be estimated from ECdir counts, because the test results were poorly correlated, and differences between ECin and ECdir methods ranged from 0.25- to 790- fold. In a subsequent trial of 12 samples of non- frozen ground beef using the five-tube MPN technique, a similar result was observed (P = 0.001), with the geometric mean for ECin being 260 per g and for ECdir being 38 per g. In a further study, using the three-tube MPN tech- nique on ground beef inoculated with 104 to 105 E. coli per g, the geometric mean by ECin was 4.9 x 104 per g and by ECdir it was 9.1 x 103 per g (P = 0.001, n = 32). The possible role of cell injury accounting for

the differences between ECin and ECdir was tested by allowing ECdir tubes to "resuscitate"

VOL. 40, 1980

APPL. ENVIRON. MICROBIOL.

at 21°C for periods of 1, 2, 3, and 4 h before incubation at 450C. Geometric mean MPN val- ues were 46, 49, 73, and 62 per g, respectively (P > 0.05, n = 12). As a result of this, other modi- fications to the elevated-temperature MPN technique were tested in an effort to improve gas release in the ECdir technique. The results are shown in Table 1. An analysis of variance for differences between methods was significant (P = 0.001), and the Duncan's multiple range test (95% confidence level) indicated that ECin, lauryl sulfate tryptose broth with direct inocula- tion and incubated at 450C, and ECdir + 0.1% Tween 80 were similar, but differed significantly from the other three media. Comparisons with- out ECin resulted in only ECdir + 0.01% lauryl sulfate being significantly different from the other tests. Ground beef inoculation study. Total

counts at 35 and 210C were monitored through- out this study during storage at 40C. The data (Table 2) for the uninoculated control samples were similar to the total counts on the samples inoculated with E. coli. The level of VRBA

TABLE 1. Duncan's multiple range test at the 95% confidence level for difference between MPN counts

in modified EC medium

Method Log meancount per g

1. EC medium, accepted method (ECin) 3.147 2. LST medium 45°C ........ .... 2.853 3. EC + 0.1% Tween 80 2.669 4. EC + 0.1% lauryl sulfate .... 2.431 5. EC medium direct (ECdir) .... 2.383 6. EC + 0.01% lauryl sulfate .... 2.023

Duncan's multiple range test Mean 6 5 4 3 2 1

counts for the uninoculated samples is also shown in Table 2. Although total and bile-pre- cipitating counts on VRBA at 35°C were high initially and increased during storage at 4°C, the counts on VRBA at 44.5°C were less than 0.1 per g throughout the study, indicating that there would not be any interference from adventitious E. coli in the samples. The study was designed to measure the effect of storage conditions (non- frozen versus frozen), length of storage at 40C (0, 2, 4, and 6 days), and selective growth me- dium on recovery of E. coli from ground beef. The log transformed data were subjected to analysis of variance, and only growth medium had a significant effect (P = 0.001). The geo- metric means of the counts on the selective media are shown in Table 3. There were no

significant interaction effects; hence differences between media could be compared by the Dun-

TABLE 3. Duncan's multiple range test at the 99%o confidence level for difference between selective media for enumeration of E. coli in ground beef

Selective medium Log meancount per g

1. VRBA prepoured (35°C) ............

2. VRBA pour plate (35°C) ............

3. VRBA-R (resuscitated) 44.5°C 4. EMB 44.50C ......................

5. Endo 44.50C ......................

6. VRBA-2 (prepoured) 44.5°C ....

7. ECin 450C ..................

8. TBA 44.50C (total count) ............

9. TBA 44.50C (Ind') ..................

10. VRBA-1 (pour plate) 44.50C .........

11. ECdir 450C .......................

5.047 5.027 4.775 4.721 4.714 4.694 4.694 4.662 4.650 4.471 3.960

Duncan's multiple range test Mean 11 10 9 8 7 6 5 4 3 2 1

TABLE 2. Standard plate counts (SPC) and VRBA counts of uninoculated ground- beef control sample during storage at 4°C and after freezing for 28 days

SPCO (x10) VRBAb counts

Days of Nonfrozen Frozen, then Nonfrozen Frozen, then thawedDaysof ~~~~~thawed Storage

VRBA35(xlO4) VRBA VRBA35(x104) VRBA SPC21 SPC35 SPC21 SPC35 44.5 44.5

Bile ppt' Total (xlO') Bile ppt' Total (x1O')

0 1.0 0.12 0.91 0.16 5.7 7.4 <1.0 0.84 1.1 <1.0 2 3.0 0.27 8.1 0.45 16.0 17.0 <1.0 3.4 3.8 <1.0 4 11.0 0.22 8.3 0.68 13.0 16.0 <1.0 4.2 4.8 <1.0 6 11.0 0.33 10.0 0.62 29.0 29.0 <1.0 4.4 4.6 <1.0

a SPC21, SPC incubated at 21°C, SPC35, SPC incubated at 35°C. b VRBA35, VRBA incubated at 35°C; VRBA44.5, VRBA incubated at 44.5°C. c Bile ppt, Typical bile-precipitating colonies on VRBA.

348 STILES AND NG

ESTIMATION OF E. COLI IN RAW BEEF 349

can's multiple range test (99% confidence level). These results are also shown in Table 3. VRBA incubated at 350C supported significantly higher counts than VRBA and other media at elevated temperature. VRBA-R, involving 2 h of resusci- tation on nutrient agar at 44.50C followed by overlayering with 20 ml ofVRBA, supported the highest geometric mean count at elevated tem- perature. Counts on VRBA-R were similar to counts on EMB and Endo, but significantly higher than counts on other media, including the three-tube MPN technique (ECin). The only counts significantly lower than ECin were those with VRBA-1 (i.e., VRBA using the pour plate technique) and ECdir. Pure culture studies. From 651 Enterobac-

teriaceae isolated from meats and meat-han- dling equipment, identified using the Minitek identification technique, 386 cultures were found that produced acid from lactose (Lac') and/or indole from tryptone (Ind'), and grew in TSB at 44WC (Table 4). All E. coli I cultures grew in TSB at 440C and were Ind+. Other isolates that were Lac' and/or Ind' and grew in TSB at 440C included the following: Citrobacter freundii (87.8% of 88 isolates), Enterobacter cloacae (72.5% of 48), and Klebsiellapneumoniae (94.0% of 136). Only limited numbers of Enterobacter hafniae, Klebsiella ozaenae, Serratia liquefa- ciens, and others were Lac' and capable of growth in TSB at 440C; most of these cultures were Lac-, Ind- (78.6%). The Lac' and/or Ind' isolates that grew in

TSB at 440C were plated on TSA and incubated at 350C, and appropriate dilutions were plated on VRBA, EMB, and TBA media and incubated at 440C. E. coli I grew well on all media tested (Table 5). The highest recovery of cells was on the nonselective control medium (TSA) at 350C; however, the selective media at elevated tem- perature (440C) gave highly comparable mean counts. The most inhibitory of the three selec- tive media was TBA. Highly significant corre- lations were observed for counts on the different selective media. Only one E. coli I produced gas in EC medium

at 450C and failed to grow at elevated tempera- ture on TBA, EMB, or VRBA (false negative); three out of the 109 isolates failed to produce gas in EC medium at 45°C, but grew at elevated temperature on TBA (false positive) compared to EC medium. Of the Ind+ cultures that grew on TBA at 44°C, only four were not E. coli I. They included one isolate identified and con- firmed as Citrobacter diversus and three K. pneumoniae. All other Ind+ strains (126 strains) failed to grow on TBA at 440C by a factor of >99.99%. Of the selective media, EMB at 440C supported the highest recovery of inoculated E.

TABLE 4. Lactose and indole reactions and ability to grow at 44°C in TSB of 651 Enterobacteriaceae isolates from meats and meat-handling equipment

Lac', Lac, Ind' Lac', Ind'Organism Ind Lac-, Ind-

+a + - + -

E. coli I 0 0 5b 0 104 0 0 C. freundii 64 12 6 0 2 0 4 E. aerogenes 5 7 0 0 0 0 3 E. agglomerans 6 23 2 0 2 2 38 E. cloacae 28 11 1 0 0 0 8 E. hafniae 1 0 0 0 0 0 12 K. ozaenae 2 8 1 0 0 0 2 K. pneumoniae 41 4 0 1 85 3 2 S. liquefaciens 7 8 0 0 0 0 86 Others 9 4 11 6 4 0 21

a Growth at 44°C: +, positive; -, negative. bNumber of isolates.

TABLE 5. Comparison ofmedia means and correlation coefficients for E. coli I grown on

selective media at 440C

Selective Mean Correlation coefficient for: medium count per

g (X107) VRBA EMB TBA

TSA, 35°C 142 0.95 0.96 0.89 VRBA, 44°C 122 0.92 0.86 EMB, 44°C 135 0.91 TBA, 44°C 100

coli I cells, but this medium did not select against Arizona hinshawii, C. freundii, E. cloa- cae, K. pneumoniae, and S. liquefaciens. How- ever, these organisms generally failed to form typical, large colonies with a metallic sheen, and recovery rates were reduced by a 10-fold factor compared to growth on TSA. Similarly, VRBA at 440C supported the growth of 99 out of 285 non-E. coli cultures. However, they generally formed distinguishable, small, non-bile-precipi- tating colonies, and their count was also reduced by a 10-fold factor compared to growth on S3A.

DISCUSSION Attempts to reduce the time for the standard

MPN technique for coliform bacteria and pre- sumptive E. coli determinations using direct inoculation of EC medium from meat homoge- nates (ECdir) were unsuccessful. The variably reduced counts by ECdir could not be attributed to cell injury in EC medium at 45°C, but might be resolved by adding a surface active agent, such as Tween 80. It was concluded, however, that a solid medium at elevated temperature (1, 7) might be preferable to, and more appropriate for, screening presumptive E. coli in meats than MPN techniques.

In the study of the reliability of various selec- tive media at elevated temperature to determine

VOL. 40, 1980

APPL. ENVIRON. MICROBIOL.

presumptive E. coli counts, neither storage at 40C for 6 days nor frozen storage (-20°C for 28 days) followed by storage at 40C significantly influenced the E. coli count. During the 6-day storage at 40C, maximum population of the ad- ventitious flora was achieved. Hence, increasing microbial competition did not influence the re- liability of the E. coli count. Similarly, frozen storage did not influence the count, contrary to other reports (16, 17) that coliform bacteria and E. coli are injured and reduced during frozen storage. This observation may be attributed to the use of "old" cells in the stationary growth phase as opposed to young cells that would be more susceptible to freeze injury. Stationary cul- tures were chosen for this study because it was felt that their physiological condition would more closely represent the condition of E. coli in refrigerated or frozen ground beef. Survival of frozen cells might also be attributed to the stor- age of E. coli in meat as opposed to synthetic or broth media. In this study, therefore, E. coli in appropriately stored frozen or nonfrozen ground beef may reflect the initial level of contamina- tion.

Selective growth medium was the only factor shown to influence the presumptive E. coli count in ground beef. Elevated incubation temperature reduced the apparent E. coli count by approxi- mately 50% as compared to 350C incubation on VRBA (Table 3). However, all elevated temper- ature tests, except ECdir, had geometric mean counts ranging from 3.0 x 104 to 6.0 x I10 per g, which would not nornally be considered a prac- tical difference in microbiological enumeration methods. As a result, any of these elevated tem- perature tests, except ECdir, could be used to estimate the presumptive E. coli count. The media selected for further study were the TBA membrane technique, EMB agar surface plating, and VRBA pour plates. Endo did not give suffi- cient differentiation between colony types. VRBA surface plating and resuscitation did not support sufficiently increased E. coli counts to justify the additional work under these condi- tions. The major Enterobacteriaceae with the po-

tential to interfere in the counts on selective media at elevated temperature were those that grew in TSB at 440C and produced acid from lactose and/or indole from tryptone. In this study, these organisms included C. freundii, E. cloacae, and K. pneumoniae, and, by virtue of their high incidence in ground beef (10), also Enterobacter agglomerans. E. coli I grew equally well on EMB, TBA, and VRBA (Table 5). TBA was highly selective for E. coli, only supporting the growth of 4 out of 114 non-E.

coli, Ind' isolates as false positives. In contrast, EMB and VRBA were not selective for E. coli, and differentiation between E. coli and other Enterobacteriaceae relied on morphological ap- pearance of the colonies. Nevertheless, VRBA incubated at elevated temperature represents an alternative "screening" test for E. coli which could be used in routine quality assurance pro- grams.

ACKNOWLEDGMENTS We thank Cecilia Shiu for technical work on this study. This work was supported by research contract OSU78-

00093, Agriculture Canada. M.E.S. is an Honorary Professor, Department of Microbi-

ology, The University of Alberta.

LITERATURE CITED

1. Anderson, J. M., and A. C. Baird-Parker. 1975. A rapid and direct plate method for enumerating Esche- richia coli biotype I in food. J. Appl. Bacteriol. 39:111- 117.

2. Delaney, J. E., J. A. McCarthy, and R. J. Grasso. 1962. Measurement of E. coli type I by the membrane filter. Water Sewage Works J. 109:289-294.

3. Fishbein, M., and B. F. Surkiewicz. 1964. Comparison of the recovery of Escherichia coli from frozen foods and nutmeats by confirmatory incubation in EC me- dium at 44.5'C and 45.5°C. Appl. Microbiol. 12:127- 131.

4. Hajna, A. A. 1937. Decomposition of carbohydrates and alcohols with production of gas at 46°C by members of the genus Escherichia. J. Bacteriol. 33:339-347.

5. Health Protection Branch. 1974. Determination of col- iforms and faecal coliforms in foods. Acceptable method MFA-19, May, 1974. Health Protection Branch, Bureau of Microbial Hazards, Health and Welfare Canada, Ot- tawa.

6. International Commission of Microbiological Speci- fications for Foods. 1978. Micro-organisms in foods. 1. Their significance and methods of enumeration. 2nd ed. International Commission of Microbiological Spec- ifications for Foods of the International Association of Microbiological Societies. University of Toronto Press, Toronto.

7. Klein, K., and D. Y. C. Fung. 1976. Identification and quantification of fecal coliforms using violet red bile agar at elevated temperature. J. Milk Food Technol. 39:768-770.

8. Mossel, D. A. A. 1957. The presumptive enumeration of lactose-negative as well as lactose-positive Enterobac- teriaceae in foods. Appl. Microbiol. 5:379-381.

9. Mossel, D. A. A., W. H. J. Mengerink, and H. H. Scholts. 1962. Use of a modified MacConkey agar medium for the selective growth and enumeration of Enterobacteriaceae. J. Bacteriol. 84:381.

10. Ng, L.-K., and M. E. Stiles. 1978. Enterobacteriaceae in ground meats. Can. J. Microbiol. 24:1574-1582.

11. Pierson, C. J., B. S. Emswiler, and A. W. Kotula. 1978. Comparison of methods for estimation of coli- forms, fecal coliforms and enterococci in retail ground beef. J. Food Prot. 41:263-266.

12. Powers, E. M., and T. G. Latt. 1979. Rapid enumeration and identification of stressed fecal coliforms. J. Food Prot. 42:342-345.

13. Rayman, M. K., G. A. Jarvis, C. M. Davidson, S. Long, J. M. Allen, T. Tong, P. Dodsworth, S. McLaughlin, S. Greenberg, B. G. Shaw, H. J. Beck- ers, S. Qvist, P. M. Nottingham, and B. J. Stewart.

350 STILES AND NG

ESTIMATION OF E. COLI IN RAW BEEF 351

1979. ICMSF methods studies. XIII. An international comparative study of the MPN procedure and the An- derson-Baird-Parker direct plating method for the enu-

meration of Escherichia coli biotype I in raw meats. Can. J. Microbiol. 25:1321-1327.

14. Silliker, J. H., D. A. Gabis, and A. May. 1979. ICMSF methods studies. XI. Collaborative comparative studies on determination of coliforms using the most probable number procedure. J. Food Prot. 42:638-644.

15. Skinner, C. E., and A. Njoku-Obi. 1957. Possible expla- nations for the inhibitory action of boric acid lactose broth at 43°C on fecal coliform bacteria. Appl. Micro-

biol. 5:289-292. 16. Speck, M. L, and R. A. Cowman. 1971. II. Injury and

recovery of frozen microorganisms. J. Milk Food Tech- nol. 34:548-549.

17. Speck, M. L., and B. Ray. 1977. Effects of freezing and storage on microorganisms in frozen foods: a review. J. Food Prot. 40:333-336.

18. Speck, M. L., B. Ray, and R. B. Read, Jr. 1975. Repair and enumeration of injured coliforms by a plating pro-

cedure. Appl. Microbiol. 29:549-550. 19. Todd, E. C. D. 1978. Foodborne disease in Canada-1975

annual summary. J. Food Prot. 41:910-918.

VOL. 40, 1980