|
|
| |||||||||||||||||||||||||||||||||||||||||||||
Background and Methods Between June 30th and August 30th, 1983, 49 patients in Massachusetts acquired listeriosis. Seven cases occurred in fetuses or infants and 42 in immunosuppressed adults; 14 patients (29 per cent) died. Of 40 Listeria monocytogenes isolates available for testing, 32 were serotype 4b. Two case–control studies, one matching for neighborhood of residence and the other for underlying disease, revealed that the illness was strongly associated with drinking a specific brand of pasteurized whole or 2 per cent milk (odds ratio = 9, P<0.01 for the neighborhood-matched study; odds ratio = 11.5, P<0.001 for the illness-matched study).
Results The association with milk was further substantiated by four additional analyses that suggested the presence of a dose–response effect, demonstrated a protective effect of skim milk, associated cases with the same product in an independent study in another state, and linked a specific phage type with the disease associated with milk. The milk associated with disease came from a group of farms on which listeriosis in dairy cows was known to have occurred at the time of the outbreak. Multiple serotypes of L. monocytogenes were isolated from raw milk obtained from these farms after the outbreak. At the plant where the milk was processed, inspections revealed no evidence of improper pasteurization.
Conclusions These results support the hypothesis that human listeriosis can be a foodborne disease and raise questions about the ability of pasteurization to eradicate a large inoculum of L. monocytogenes from contaminated raw milk.
Methods
Case Definition and Case Finding
A case was defined as a patient from whom L. monocytogenes was isolated from either blood or cerebrospinal fluid, or from placenta in conjunction with fetal death.
All microbiology laboratories in Massachusetts were surveyed for reports of L. monocytogenes isolated from clinical specimens received between January 1982 and August 1983. Records of clinical isolates of L. monocytogenes submitted to the state laboratory in Massachusetts for bacteriologic confirmation were reviewed. Infection-control practitioners in major hospitals were notified of the problem, and prospective surveillance was established.
Case–Control Studies
Two studies evaluated differences in exposures between cases and controls during the month before the onset of illness in the case. In the first study, two controls were matched with each case for age (within 10 years), sex, and neighborhood of residence. Control subjects for this study were selected from town listings made available by the Harvard School of Public Health. In the second study, one control was matched with each case for age (within 10 years) and primary underlying illness in the case. Among mother–baby pairs, maternal exposures were assessed, and pregnancy was considered the primary underlying illness. Control subjects for the second study were identified with the help of infection-control practitioners where cases were hospitalized. In both studies, cases whose isolates were of a known serotype other than the epidemic serotype were excluded from the analysis.
Connecticut Cases
Patients in Connecticut who acquired listeriosis during the time of the outbreak in Massachusetts were enrolled in a separate illness-matched case–control study using the same methods as in Massachusetts.
Environmental Sampling and Bacteriology
With the help of state health and agricultural departments and plant personnel, milk was sampled from individual farms, the cooperative through which milk was shipped, and the pasteurizing plant.
Clinical isolates were sent to the state laboratory in Massachusetts for bacteriologic confirmation and then to the Centers for Disease Control for serotyping. Milk samples were cultured in the Respiratory and Special Pathogens Laboratory Branch at the Centers for Disease Control.
All serotype 4b isolates were phage-typed in a blinded fashion by the method of Audurier et al.3
Statistical Analysis
Except where stated otherwise, probability values and odds ratios with 95 per cent confidence intervals were calculated with conditional logistic regression for matched data.4
Results
The Outbreak
Between June 30 and August 30, 1983, 49 patients were hospitalized in Massachusetts with meningitis, septicemia, or abortion caused by L. monocytogenes. The attack rate of approximately 1 per 118,000 was four to five times higher than the rate observed in the state during previous summers.
Forty-two of the cases (86 per cent) occurred in adults, and seven (14 per cent) in mother–infant pairs (Table 1). All the adults had preexisting illnesses or conditions causing immunosuppression. The symptoms at presentation were variable and depended on the age of the patient and the degree of immunocompromise. Illness was community-acquired in 48 of the 49 cases and nosocomial in 1. There was no obvious geographic clustering of cases within the state. Isolates from 40 of the 49 cases were available for serotyping. Of these, 32 were serotype 4b, subsequently defined as the epidemic strain.
|
The data obtained from the 19 patients and 38 controls enrolled in the first case–control study did not support person-to-person, airborne, or waterborne transmission of the disease. The illness was not associated with consumption of meats, fresh vegetables, coleslaw and other salads, or unpasteurized dairy products. One trend was noted, however. Cases appeared to be more likely than their neighborhood-matched controls to have purchased groceries from one particular foodstore chain (subsequently called "chain A").
Five persons who shopped for cases were taken to the chain A stores where they normally bought food. When lists of the items each could recall purchasing in the month before the illness were compared, chain A brand pasteurized whole milk or milk containing 2 per cent fat was the only item common to all five. Cases and controls were recontacted and asked about consumption of pasteurized dairy products — items not included on the initial questionnaire. Fourteen of 19 cases, as compared with 11 of 38 neighborhood-matched controls, recalled having drunk chain A whole or 2 per cent milk (odds ratio = 9.3, P<0.01).
We were concerned that the association of listeriosis with milk consumption might be attributable to the comparison of pregnant or chronically ill patients with their healthy neighbors. A second case–control study, using illness-matched controls, was therefore performed. In this study, 40 of the 49 patients who had listeriosis during the outbreak were enrolled. Of the nine remaining patients, one had died and no next of kin was available for questioning, and eight (including the patient with nosocomial illness) had infections caused by serotypes other than 4b.
The results of the second study confirmed those of the first. A strong association was found between chain A whole or 2 per cent milk and the development of listeriosis (Table 2). This association persisted when the analysis was limited to the cases and controls who shopped at chain A stores (Table 2). If the five cases who were used to generate the hypothesis are excluded, the association of chain A milk with listeriosis remains statistically significant (P<0.05 for the neighborhood-matched study, P<0.005 for the illness-matched study).
|
Because pasteurized milk has not been shown to be a common vehicle for infection, the dose–response effect, the protective effect of skim milk, the Connecticut cases, and phage typing were analyzed to assess the validity of our case–control findings.
Using the illness-matched controls, we evaluated whether a dose–response effect could be demonstrated. Only three of the six controls who had drunk any chain A whole or 2 per cent milk had drunk more than a third of a glass per day, and none had drunk more than one glass per day. In contrast, 19 of the 27 cases (71 per cent) had drunk more than a third of a glass per day, and 11 (40 per cent) had drunk more than one glass per day.
A corollary hypothesis was tested. If only whole or 2 per cent milk was contaminated, then drinking 1 per cent or skim milk, by decreasing the chance of exposure to whole milk, should have protected against the development of disease. When cases and illness-matched controls were questioned about their consumption of low-fat milk, only 6 of 40 cases (15 per cent) reported drinking 1 per cent or skim milk, as compared with 18 of 40 controls (45 per cent) (odds ratio = 0.25, 95 per cent confidence interval = 0.08 to 0.75, P<0.02).
To assess whether the association of chain A milk with listeriosis could be demonstrated in a neighboring state, we questioned 12 of the 13 patients hospitalized in Connecticut during the time of the outbreak in Massachusetts. Although only 2 of the 2600 grocery stores in Connecticut sold milk pasteurized by the facility supplying milk to all the chain A stores in Massachusetts, 3 of the 12 patients had drunk milk purchased from one of these two stores (P<0.0001, binomial distribution). Among the other nine patients, no association of illness with any food or food store was present.
Isolates were available from 32 of the 40 cases enrolled in the illness-matched case–control study. Isolates from all 19 cases who reported drinking chain A whole or 2 per cent milk were phage type 2425A, whereas only 6 of the 13 isolates (46 per cent) from cases who did not report drinking the milk were of this phage type (P<0.001, Fisher's exact test).
Investigations of the Pasteurizing Facility and the Source of Milk
The milk implicated in this outbreak was pasteurized at a single plant. All milk was processed with the same equipment and was pasteurized at times and temperatures exceeding current regulations of the Food and Drug Administration. Inspections by the FDA and the Food and Drug Division of the Massachusetts Department of Public Health, performed both before and after the managers of the plant had been notified of a possible problem, revealed that the facility was clean, modern, and well run. No defect that could have resulted in improper pasteurization and no source of contamination after pasteurization were identified. Records on temperatures and phosphatase testing during the outbreak period were consistent with proper pasteurization.
Milk was supplied to the pasteurizing plant by a milk cooperative of 450 farms. As a result of milk-truck delivery patterns one group of farms supplied chain A whole and 2 per cent milk, and another group supplied chain A 1 per cent and skim milk.
During the first three weeks of September, milk was sampled from the pasteurizing plant, the cooperative, and individual farms. In retrospect, however, it was recognized that the outbreak had ended by that time, and consumption of this milk was not associated with the development of listeriosis. L. monocytogenes was not isolated from pasteurized samples. The organism was isolated from 15 of 124 samples obtained before pasteurization (12 per cent) and 2 of 14 milk filters (14 per cent). Several different serotypes were identified, including 1a, 3b, 4a-b, and 4b, the epidemic serotype. The phage type associated with the epidemic was not identified among the isolates tested.
Veterinarians from the region of the cooperative reported that listeriosis was usually quite rare in animals. However, in the late spring and early summer of 1983, four cases of bovine listeriotic encephalitis had been diagnosed. All were in dairy cows from farms supplying milk to the cooperative, and two were in cows from farms supplying whole milk to chain A.
Discussion
Cases from a large outbreak of listeriosis occurring in Massachusetts during the summer of 1983 were associated in two case–control studies with consumption of a specific brand of whole and 2 per cent pasteurized milk. This association was substantiated by additional analyses, which suggested the presence of a dose–response effect, demonstrated a protective effect of skim milk, associated cases with the same product in an independent study in another state, and linked a specific phage type with the disease associated with milk consumption. These epidemiologic results were supported by the microbiologic finding that L. monocytogenes could be isolated from unpasteurized milk even after the outbreak, and by veterinarians' records documenting listeriosis in the dairy cattle supplying the milk associated with disease.
Because the milk implicated in this outbreak was apparently properly pasteurized, the question of when it became contaminated assumes some importance. Although postpasteurization contamination cannot be excluded, it seems unlikely for two reasons: multiple inspections of the plant did not reveal a potential source, and it is difficult to postulate a mechanism whereby postpasteurization contamination occurring over an extended period would have affected only whole milk and not skim milk, since both were processed with the same equipment each day.
On the other hand, intrinsic contamination of the milk and survival of some organisms despite adequate pasteurization is both consistent with the results of this investigation and biologically plausible. Listeria-associated encephalitis, mastitis, and abortion occur in cattle, and the organism is excreted in milk.5 Consumption of raw milk has been associated with L. monocytogenes disease in a case report,6 and preliminary experiments have indicated that, as compared with other bacteria, L. monocytogenes is quite resistant to heat.7 Although a large inoculum would be necessary before survival of any organisms during pasteurization would be expected, postpasteurization storage of contaminated milk at refrigerator temperatures (in essence, cold enrichment) would permit selective growth of the remaining organisms.8 In fact, a relatively small infectious dose may explain why all the adult patients in this outbreak were immunocompromised; studies have shown that immunosuppressed animals are more susceptible than immunocompetent animals to a small inoculum of L. monocytogenes.9
The ability of L. monocytogenes to exist as an intracellular parasite may have increased the likelihood that some organisms survived pasteurization under the circumstances encountered in Massachusetts. Before pasteurization, most homogenized milk undergoes a centrifugal filtering process called clarification, which removes leukocytes.10 In Massachusetts, the whole milk was passed through a milk filter rather than clarified, so that leukocytes were not removed. Pasteurization may have been only partially effective because the entire large load of intracellular lipid-shielded organisms could not be killed.
The association of listeriosis with consumption of milk supports the hypotheses that L. monocytogenes is a pathogen transmitted to human beings from infected animals or their byproducts and that ingestion of the organism is one mechanism of infection. Our results suggest that milk (in addition to raw vegetables11) should be considered a possible vehicle of infection in sporadic listeriosis.
Our results also suggest that although pasteurization is a highly effective method of eliminating bacterial pathogens from milk, it may not always be 100 per cent effective. This finding must be kept in perspective, however. Human listeriosis is a rare disease. Even if contaminated pasteurized milk is the vehicle for some cases of listeriosis, it is still a distinctly uncommon cause of disease, especially in relation to the frequency of exposure. It is important to remember the potential benefits of this food product and the amount of human illness caused by consumption of unpasteurized milk.
We are indebted to the members of the Communicable Disease Division of the Massachusetts Department of Public Health; to Lewis Graves, Gloria Ajello, Chris Thacker, Morris Potter, and Khalil Sharefzadeh; and especially to the infection-control practitioners and the workers at the clinical microbiology laboratory in Massachusetts.
Source Information
From the Respiratory and Special Pathogens Epidemiology Branch, the Enteric Diseases Branch, the Respiratory and Special Pathogens Laboratory Branch, and the Statistical Services Branch, Division of Bacterial Diseases, Center for Infectious Diseases, Centers for Disease Control, Atlanta; the Division of Epidemiology, Vermont Department of Health, Burlington; the State Laboratory Institute, Massachusetts Department of Public Health, Boston; and the Laboratoire de Microbiology, Faculte de Medicine, 37032 Tours, France. Address reprint requests to Dr. Fleming at the Respiratory and Special Pathogens Epidemiology Branch, Bldg. 1, Rm. 5405, Centers for Disease Control, Atlanta, GA 30333.
References
| |||||||||||||||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
