Quantitative Risk Assessment for Viral Contamination of Shellfish and Coastal Waters

Transcription

1 1043 Journal of Food Protection, Vol. 56, No. 1, Pages (December 1993) Copyright, International Association of Milk, Food and Environmental Sanitarians Quantitative Risk Assessment for Viral Contamination of Shellfish and Coastal Waters JOAN B. ROSE 1 * and MARK D. SOBSEY 'Environmental and Occupational Health, 1301 Bruce B. Downs, College of Public Health, University of South Florida, Tampa, Florida 3361 and ^Environmental Science and Engineering, School of Public Health, University of North Carolina, Chapel Hill, North Carolina 7599 (Received for publication January 13, 1993) ABSTRACT Human pathogenic viruses have been detected from approved shellfish harvesting waters based on the fecal coliform indicator. Until recently it was difficult to assess viral contamination and the potential impact on public health. Risk assessment is a valuable tool which can be used to estimate adverse effects associated with microbial hazards. This report describes the use of quantitative risk assessment for evaluating potential human health impacts associated with exposure to viral contamination of shellfish. The four fundamental steps used in a formal risk assessment are described within and include i) Hazard identification, ii) Doseresponse determination, iii) Exposure assessment, and iv) Risk characterization. Dose-response models developed from human feeding studies were used to evaluate the risk of infection from contaminated shellfish. Of 58 pooled samples, 19% were found to be positive for viruses. Using an echovirus-1 probability model, the individual risk was determined for consumption of 60 g of raw shellfish. Individual risks ranged from. x 10 4 to 3.5 x 10". These data suggest that individuals consuming raw shellfish from approved waters in the United States may have on the average a 1 in 100 chance of becoming infected with an enteric virus. Using the rotavirus model which represents a more infectious virus, the risk rose to 5 in 10. The potential for use of a risk assessment approach for developing priorities and strategies for control of disease is immense. Epidemiological data have demonstrated the significance of shellfish-associated viral disease and, although limited, appropriate virus occurrence data are available. Additional information on virus occurrence and exposure is needed, and then scientific risk assessment can be used to better assure the safety of seafood. Assessing human health risks associated with the pollution of coastal waters has been done empirically since infectious disease transmission via exposure to contaminated shellfish was first documented. Traditionally, microbial hazards were identified through epidemiological investigations and outbreak documentation. Bacteriological standards or indicators were then used to define marine water quality to evaluate potential contamination and to protect public health. These standards have been shown to be adequate for prevention of enteric bacterial outbreaks such as typhoid (37). However, there is now significant evidence that the bacterial indicators do not adequately predict all microbial health risks. This includes bacterial pathogens such as Vibrio vulnificus and natural seafood toxins. The relationships between indicator concentrations and risk of human illness caused by viral pathogens are limited. No correlation between the indicator bacteria and the presence (or absence) of enteric viruses has been demonstrated for bivalve molluskan shellfish or their waters (13,39). However, viral outbreaks in the United States associated with shellfish harvested from approved water continue to occur (9). A recent National Academy of Science report on "Seafood Safety" concluded that the fecal coliform indicator is inadequate for determining the microbial quality and safety of marine waters and shellfish (7). According to the "National Water Quality Inventory, 1988, Report to Congress", an average of 3% of the estuaries in 17 coastal states (9,43 square miles) were impaired, and 46% of this was due to the microbial quality of the water (34). Yet very little information is available regarding the types of microbial contaminants, sources, levels, persistence, and accumulation in sediments and shellfish. Therefore, the potential impact on public health and coastal resources cannot be adequately assessed. Risk assessment is a valuable tool which has been used to estimate adverse effects associated with particular hazards and is an essential component of risk management. In particular, quantitative risk assessment (QRA) is a statistical estimate of the probability of an event taking place based on exposure and dose-response models. This type of assessment has provided managers with valuable information on the identity and characterization of the risks which can then be used to develop appropriate control strategies. Risk assessment approaches have been used for estimating human health impacts for chemical contaminants in fish and shellfish (7). However, quantitative evaluation of risks associated with exposure to specific microbial contaminants is a relatively new approach and has only been used thus far for drinking water (9,31,3). This report describes the use of quantitative risk assessment or evaluating potential human health impacts associated with exposure to viral contamination of coastal waters and shellfish. The four fundamental steps used in a formal risk assessment include i) Hazard identification, ii) Dose-response determination, iii) Exposure assessment, and iv) Risk characterization. JOURNAL OF FOOD PROTECTION, VOL. 56. DECEMBER 1993

2 1044 ROSE AND SOBSEY MATERIALS AND METHODS Identification of the microbial hazard Hazard identification is accomplished by observing and defining the types of adverse health effects in humans associated with exposure to enteric viral agents. This includes morbidity rates, severity and duration of illness, and mortality rates. Epidemiological evidence which links the various diseases with the particular exposure route is an important component of this identification. Through an evaluation of the current literature, a summary of the viral hazards associated with contaminated shellfish was assessed. Dose-response determination The risk of infection is defined as a mathematical probability of infection from a given unit dose or exposure (18). The two components of the model which aid in characterizing the risk are i) the level of exposure and ii) the interaction of the particular pathogen and host (defined by the dose-response curve). Dose-response experiments for microorganisms of concern have been conducted in human volunteers for a variety of viruses. In these experiments, several sets of volunteers were exposed to doses of viruses at various average concentrations measured as plaque-forming units (PFU) or as multiples of the 50% tissue culture infectious dose. The resulting percentages of infected individuals were then determined by excretion of the virus in the feces and sero-conversion. Three dose-response models have been previously compared using these experimental data sets for determining low dose risks (18,9). The log-normal model assumes a threshold effect where each microorganism has an inherent minimal infectious dose that is the lowest dose capable of initiating an infection. If the exposures are greater than this threshold of concentrations of microorganisms, a response will be observed in the host. Haas (18) found that this model did not provide an adequate fit to the data set. The single-hit exponential model is a two-step process: the host exposure, followed by the ability of the microorganism to cause the result. It includes factors such as host resistance, nonspecific pathogen decay, and the presence of noninfective particles. Laboratory dose-response studies generally have been conducted under conditions where the counts of microorganisms in the administered dose approximate the Poisson distribution. Under these conditions, if one microorganism is sufficient to cause an infection, and if host-microorganism interactions are constant, then the probability of an infection (P,) resulting from ingestion containing an average number of organisms from a single exposure may be defined by the exponential model (Table 1). In this equation, r is the fraction (proportion) of microorganisms that are ingested which survive to initiate infections ("hostmicroorganism interaction probability"). When the dose-response curves were plotted (P, versus log [N], the exposure), the slopes were generally less steep than predicted by the exponential model. It was suggested that this difference was due to the heterogeneity of either the infectivity of the individual microorganisms or the sensitivity of the individual hosts, or both. Thus the model can be generalized by assuming that r is actually not a constant value but is itself described by a probability distribution (in particular, a beta distribution). Haas (18) generalized this model, and with the previously mentioned assumptions, the beta-poisson model was developed as an alternative (Table 1). The parameters a and B characterize the doseresponse curve, and as a increases, the model becomes closer to the exponential model. Table 1 summarizes the best point estimates for a and 3 and r for a number of studies for the enteric viruses. Except in one case, the beta-poisson model was found to provide an improved fit to the experimental data set relative to the exponential model TABLE 1. Best fit dose-response parameters for various human feeding studies (18,9). Organism Echovirus 1 Rotavirus Poliovirus I Poliovirus I Poliovirus III Best model beta-poisson beta-poisson exponential beta-poisson beta-poisson Model parameters a = = a = = 0.4 r = a = B = 154 a = B = Models: P. = 1 - (1 + N/3)" (beta-poisson model). P. = 1 - exp(-rn) (exponential). P = Probability of infection. a, 3, r = Parameters defining the dose-response curve. N = Exposure (PFU). Original reference (33) (40) (3) (1) (19) (18,9). The echovirus-1 model and the rotavirus model were chosen and used consistently throughout this study to represent moderately and highly infectious viruses, respectively. Exposure assessment Exposure is determined by calculating the numbers of viral PFU per gram of shellfish multiplied by the grams ingested. Estimates of per capita consumption have been used to evaluate exposures to inorganic contaminants for risk assessment (7). These data were generated from landing information contained in "Fisheries of the United States, 1987", and reported weight of the meat, excluding the shells, divided by the approximate population in the United States (4 million). For clams, oysters, and other shellfish, the U.S. average annual per capita consumption was 50, 74, and 53 g, respectively. This approach assumes an equal exposure for each individual throughout a year's time. In contrast to chemicals, a single exposure from microorganisms can result in an adverse effect; therefore, the risk from a single serving is a better estimate of risk and this approach was used in this risk assessment. Studies have determined that an average serving may consist of 6 to 1 shellfish, each ranging in weight between 10 and 0 g (unpublished data). These values of 60 and 40 g were used to represent low and high exposures in a single serving. Table gives a summary of some of the studies on the occurrence of viruses in shellfish and their overlying waters in areas open and closed for harvesting, based on the bacteriological indicator standards and sanitary survey criteria (10,1,17,35,36,39). Greater prevalence and concentrations of viruses were found in shellfish compared to the water column and in closed areas, as compared to open areas. Between 8 and 63% of the samples of shellfish were positive for enteroviruses in closed areas (0. to 4 PFU/100 g) and 9 to 40% of the samples were positive in open areas (0.3 to 00 PFU/100 g). Approximately 15 to 0 shellfish represented a sample, ranging in weight between 76 to 150 g per sample. For the purposes of this risk evaluation, only the levels of viruses reported in shellfish harvested from approved waters were used in this study. Average PFU/100 g were determined, and arithmetic averages were calculated for each site based on the total number of positive and negative samples. Levels of PFU per 60 and 40 g were used as the basis for virus exposure. Risk characterization Risk estimates were determined using the following: i) betadistributed probability models for echovirus-1 and rotavirus, ii)

3 TABLE. Enteric virus isolations from shellfish harvesting areas. RISK ASSESSMENT FOR VIRAL CONTAMINATION OF SHELLFISH 1045 Area State or location Open Closed Shellfish beds 8 Shellfish beds' Water (%V Shellfish Water Shellfish PFU7100 L PFU/100 g PFU/100 L PFU/100 g (Type) Gulf MS (10) Mississippi ND d (9%) ND" (34%) Oysters Sound c 0.-. c TX (17) Galveston (50%) (0%) (63%) (40%) Oysters Bay c Atlantic North Carolina Central ND (5%) ND (37%) Clams (39) Region Southern ND (1%) ND (37%) Region New Jersey/ Great South (4.3%) (40%) (43%) (8%) Oysters New York Bay (36) Oyster Bay (1%). (5%) (0%) (37%) Clams " Based on Bacteriological Indicator Standards and Sanitary Survey b Percent of samples positive for viruses. c Ranges of plaque forming units (in the positive samples). 11 Not determined. (1,35). the levels of viruses found in shellfish, and iii) the two different levels of consumption (Table ). Morbidity (disease) and mortality (death) estimates were compared for various exposures for rotavirus, representing gastrointestinal viruses and HAV, assuming infectivity rates of hepatitis A virus (HAV) similar to these of echovirus-1. The morbidity and mortality rates (see Table 3, hazard identification, results section) were multiplied by the probability of infection determined by the models, and the risk of disease and death were estimated (i.e., R x morbidity rate x mortality rate = risk of mortality). Outbreak data were compared to the output data of models for an assessment of the uncertainty and appropriateness for estimating adverse health outcomes. Desenclos et al. (9) reported attack rates (AR) for hepatitis A per 1,000 dozen oysters served in an outbreak in Florida. These data were used, assuming a 50% morbidity rate x P. to estimate the probability of illness. Then, using the models for echovirus and rotavirus, the levels of viruses (N) were calculated. RESULTS Identification of the microbial hazard Poliovirus, echoviruses, and other unidentified enteroviruses have all been isolated from shellfish (10,17,36,39). These enteric viruses are obligate human pathogens and do not replicate outside the human host. They are small and have a simple structure consisting of a protein coat surrounding a core of genetic material (DNA or RNA). When human enteric viruses are ingested, they enter the alimentary tract. Once past the stomach they replicate within the intestinal tract, are then excreted in the feces, and able to survive in the marine environment due to the relatively stable nature of their structure (). There are over 10 enteric viruses which may be found in human wastes and domestic wastewaters. Table 3 lists some of the better described viruses, including the enteroviruses (polio-, echo-, and coxsackieviruses), hepatitis A virus, rotavirus, and Norwalk virus. As shown in this table, these viruses may produce an immense range of diseases in humans, such as diarrhea, aseptic meningitis, paralysis, conjunctivitis, myocarditis, and hepatitis. Cases of poliomyelitis are low in the United States due to almost universal immunization with attenuated virus; however, 98,000 cases of hepatitis and more than 30 million cases for all the other enteric viruses have been documented annually (). Aseptic meningitis continues to be a problem associated with echoviruses and coxsackie B viruses and some states reported a - to 10-fold increase in the number of cases in 1991 (5). Morbidity and mortality rates for various enteric viruses are also shown in Table 3. Not everyone who may become infected with enteric viruses will become clinically ill. Asymptomatic infections are particularly common among some of the enteroviruses. The development of clinical illness depends on numerous factors including the immune status of the host, age of the host, virulence of the microorganisms according to their type and strain, and route of infection. For HAV the percentage of individuals with clinically observed illness is low for children (usually <5%) but increases greatly with age (77). The frequency of clinical hepatitis A virus in infected adults is estimated at 75%. However, during waterborne outbreaks it has been observed as high as 97% (0). In contrast, the frequency of clinical symptoms for rotavirus is greatest in childhood (14) and lowest in adulthood, and is likely due to development of immunity as a consequence of repeated infections. The observed frequencies of symptomatic infections for various enteroviruses may range from 1% for poliovirus to more than 75% for some of the coxsackie B viruses (6) (Table 3). Severity of the viral diseases may be defined by the types of symptoms and duration. Norwalk virus, for example, certainly may be ranked low because it produces

4 1046 ROSE AND SOBSEY TABLE 3. Characteristics of enteric viruses (,16). Virus group 1985 Incidences Mortality rates (%) Morbidity rates (%) Diseases interovirus: Polioviruses Coxsackieviruses A B Echoviruses: Hepatitis A virus 6,000, , Paralysis Aseptic meningitis Herpangina Aseptic meningitis Respiratory illness Paralysis fever Pleurodynia Aseptic meningitis Pericarditis Myocarditis Congenital heart anomalies Nephritis Respiratory infection Aseptic meningitis Diarrhea Pericarditis Myocarditis, fever, rash Infectious hepatitis Adenoviruses: 10,000, Acute conjunctivitis Diarrhea, Respiratory illness Eye infection Rotaviruses: 8,000, Infantile gastroenteritis Norwalk virus: (a calicivirus) 6,000, Gastroenteritis generally mild diarrhea lasting an average d. Aseptic meningitis may be ranked as moderate, while hepatitis and poliovirus may be ranked as relatively severe. In addition, there is now evidence that coxsackievirus type B5 may be associated with insulin-dependent diabetes mellitus with severe consequences (38). Other types of sequelae associated with coxsackievirus type B5 are encephalitis, neonatal disease, and carditis. Mortality rates are also affected by many of the same factors that determine the likelihood of the development of clinical illness. The risk of mortality for hepatitis A virus has been estimated at 0.6% () and appears to be particularly significant for adults. Mortalities from other enterovirus infections in North America and Europe have been reported to range from <0.1 to 1.8% and are predominantly seen in children under 15 years of age (7). There is a significant body of information linking hepatitis and viral gastroenteritis with consumption of contaminated shellfish (8,15,30). Over 100 outbreaks have been documented in the United States, increasing from less than 10 in the years to more than 50 in the years (30). However, there are many enteric viruses which have not been linked epidemiologically with contaminated shellfish. From 1983 to 1987, an average of nine outbreaks from all etiological agents per year were associated with shellfish. However, for more than 50% of these outbreaks the etiology was unknown (4). These could be due to viruses. Overall, this number of reported outbreaks most certainly represents a great underestimation due to the difficulty in identifying and documenting them. While epidemic levels of shellfish-associated disease are underestimated, it becomes even more difficult to determine what level of endemic infectious disease may be due to shellfish. Both hepatitis A and non-a non-b hepatitis have been linked epidemiologically to shellfish-associated illness (8,19a). The incidence of hepatitis A has decreased since 1970, but it began increasing in 1983 with 1.6 cases per 100,000 in the United States for An estimated 19 to 5% of this may be due to contaminated shellfish (3,8,19a). The other type of illness associated epidemiologically with contaminated shellfish is gastroenteritis caused by Norwalk virus, Snow Mountain agent, astroviruses and calciviruses, and other small round viruses. Of these viral agents, the most extensive data base is for the Norwalk virus (8,19a). Other end points of disease potentially associated with enteric viruses that contaminate shellfish, such as aseptic meningitis, may be too rare for detection through the current disease surveillance system. Dose-response determination The risk of infection from exposure to one viral PFU may range from.8 x 10 1 to 7. x 10" 5 depending on the

5 RISK ASSESSMENT FOR VIRAL CONTAMINATION OF SHELLFISH 1Q47 virus model. Rotavirus and poliovirus 3 were found to be more infective then the echovirus 1 and poliovirus 1 (Fig. 1). The beta-distributed model when used to extrapolate to low doses predicted an increased risk, as compared to the log-normal model. It has been suggested as a better estimator for conservative protection of public health (18). The use of the rotavirus model was chosen not only for this reason, but also because it provided the best-fit to the data with development of 95% confidence limits. This model was used to represent gastrointestinal viruses (18,9). The echovirus model was chosen to represent other moderately infectious enteroviruses which have been isolated from shellfish. Probability of Infection Virus Exposure (PFU) ^Echovirus 1 ^-Rotavirus -*-Poliovirus Poliovirus 3 Figure 1. Dose-response a Based on the probability models. relationships for various enteric viruses". Exposure assessment Four different studies were used to evaluate the potential for virus exposure from shellfish consumption (10,17,36,39). Viruses were isolated using cell culture techniques, and PFU levels were then assessed from these data. The types of viruses identified included poliovirus types 1 and, echovirus types 1, 7, 0, 3, 4, and 7, as well as unidentified enteroviruses. The results are summarized in Table 4. Of 58 total samples, 19% were found to be positive for viruses. Average virus levels ranged from 0. to 31 PFU/100 g of shellfish. Exposures for a single serving ranged from 0.11 to 18.6 PFU for a 60 g serving and from 0.43 to 74.4 PFU for a 40 g serving. The estimated average virus exposure per 60 g serving is 6 PFU. Risk characterization Using the echovirus-1 and rotavirus probability models, the individual risk was determined for consumption of raw shellfish (Table 4). The quantities of viruses ingested per serving ranged from 0.1 to 18.6 PFU and averaged 6 PFU. For a moderately infectious virus like echovirus 1, individual risks ranged from. x 10 4 to 3.5 x 10 for a small serving of shellfish (60 g), and with an increased exposure (40 g) the risk was four times greater. Based on the average virus level from all studies, individuals consuming a single serving of raw shellfish from approved waters in the United States may have a 1/100 probability of becoming infected with a moderately infectious enteric virus. However, the risk of infection becomes a 5/10 probability (with a lower 95% confidence limit of a 4/10 risk) if exposed to a highly infectious virus, such as the rotavirus at an average virus exposure of 6 PFU/60 g. Risks associated with virus levels in shellfish from New York, Texas, and North Carolina were all similar to the average; however, risk associated with oysters in Mississippi was 10 times lower (Fig. ). Morbidity and mortality risks were estimated from probabilities of infection and varied depending on the virus. Figure 3 compares risks of infection, illness, and death for rotavirus and hepatitis A virus (HAV). In this comparison, the echovirus model was used for the probability of HAV infectivity. Exposure was set at a small serving of shellfish (60 g) with an average of 6 PFU per serving. Risks of disease were 3.1 x 10' 1 and 9 x 10 3 for rotavirus and HAV, respectively. Although the risk of disease may be potentially less for HAV (about 30-fold less), depending on its infectivity, the severity of the disease and the risk of death are much more significant than for a gastrointestinal virus, particularly for a healthy adult, which is the primary population consuming shellfish. This demonstrates that even with a model for a moderately infectious virus, when the more severe end-points of disease and death are considered, the consequences of infection overshadow the lower infectivity of the virus. Using data from an outbreak of hepatitis A (9), the reported attack rate estimates (AR) per 1,000 dozen oysters served were compared to probability of infection models prediction of exposure. The AR ranged from 1.1 to 18, with an average of per 1,000 dozen oysters served (Table 5), and the rotavirus model predicted virus levels per 1,000 dozen oysters of 3.6 to 57 PFU. This suggests that very few oysters contained viable viruses, but those that did were able to initiate infection and disease. In contrast, for a moderately infectious virus such as the echovirus-1, the levels predicted were 80 to 330 times greater per 1,000 dozen oysters, suggesting a much greater contamination. During the environmental investigation of oysters from the approved areas traced to this outbreak, the presence of HAV was detected in 50 and 100% of the samples by immunological and molecular techniques, respectively (9). Each sample consisted of two specimens. This suggests that at least one out of four specimens contained virus protein antigen and one out of two contained viral nucleic acid. DISCUSSION This paper presents and applies an approach for assessing risks associated with viral contamination of bivalve molluskan shellfish. Two pieces of information are essential in preparing a QRA. First, models are needed which estimate the probability of infection after exposure. Second, monitoring data are needed to estimate the exposure levels. Various dose-response (exposure level-infection) models for bacteria, protozoa, and viruses have been developed (31). These models could be useful for evaluation of public health risks associated with exposure to coastal waters and

6 1048 ROSE AND SOBSEY TABLE 4. Virus levels in shellfish from waters open to harvesting and the related risk of infection due to consumption of shellfish. Study site Shellfish Total No. of samples collected Total samples Positive (levels 1 ) Average level of viruses PFU/100 g Individual risk b 60 g c 40 g c Mississippi Oysters (0.3) (3.6) x x 10-4 New York Clams 5 (10) (30) x x 10- New York Oysters 8 (48) (00) x x 10-1 North Carolina Clams 13 3 (0.8) (0.8) (48.0) x x 10- Texas Oysters 10 (17) (59) x x 10- Total/Averages x x 10- a Levels are PFU/100 g. b Echovirus-1 model. c Exposures of 60 and 40 g Probability of Infection 11 TJ\ YA // //' f \ State : MS NY Rotavirus" Echovirus" Exposure b /7 // '!> >/, '/i \ NY Virus Model 7 '// 77 '// 'ii NC H Rotavirus O Echovirus 1 11 X //, /> TX fij // '/ ' //, '/, II Average Figure. Risk of viral infection associated with consumption of shellfish from areas open to harvesting. " Risk from a single serving assuming 60 g consumption of shellfish. b Virus levels PFU/60 g, see Table 4. shellfish if better monitoring data were available for the variety of microbial contaminants of concern. When using microbial QRA, several assumptions are made regarding both the models and the exposure. In this example, the following assumptions were used: i) The Poisson distribution was appropriate for the model and the distribution of the viruses in shellfish; ii) The shellfish were consumed raw and 60 g represented an average exposure at a single serving; iii) The enterovirus levels detected in shellfish could Estimated Risk* - I i P 7/7 Infection,, l/i), 1 ////J '////, //// llll Morbidity 0 Rotavirus 0HAV Mortality Figure 3. Risks of infection, morbidity, and mortality for consumption of contaminated shellfish. a Risk from a single serving for six small shellfish, 60 g at average virus levels of 10 PFU/100 g. be averaged to represent an average exposure for a variety of enteric viruses; and iv) The echovirus and the rotavirus dose-response models represented moderately and highly infectious viruses, respectively. These assumptions may overestimate or underestimate the true risks. One of the greatest uncertainties is the levels of specific viruses of concern (i.e., HAV) in contaminated shellfish. Williams and Fout (41) have reviewed methods for the recovery and detection of viruses from shellfish. Both extraction and concentration, as well as adsorption- JOURNAL OF FOOD PROTFCTION. VOL. 56, DECEMBER 1993

7 RISK ASSESSMENT FOR VIRAL CONTAMINATION OF SHELLFISH 1049 TABLE 5. Estimation of virus levels per 1,000 dozen oysters based upon reported attack rates for an outbreak situation. Reported Attack rates during an outbreak of hepatitis A virus (9) Estimated levels of viruses using the rotavirus model Using the echovirus model" 1,100,000 19,000 Estimated using the risk assessment models and the corresponding attack rates, 0.50 morbidity rates, to estimate P. and solving for N (see Table 1). elution-concentration procedures, have been developed to recover viruses. The efficiency of the methods for recovery may range from to 47%. The virus losses due to the methodology were not taken into account in this study, hence resulting in an underestimation of human exposure and the subsequent risk. While cell culture systems have been widely used for the detection of viable viruses, they detect and quantify only a fraction of the hundreds of viruses present in wastewater or contaminated coastal water and shellfish. The inability to detect some important shellfish-borne viruses, such as HAV and Norwalk, by cell culture infectivity also contributes to the underestimation of true risks. In addition, the levels of virus detected were assumed to be the levels of exposure, without regard to viral inactivation; this assumption contributes to the overestimation of risk. The monitoring data used in this study are minimal and are not adequate to reliably predict public health risks associated with consumption of raw shellfish. Better virus monitoring programs are needed to provide information on the prevalence of viruses and the quantitative distribution of low, average, and maximum levels of contamination. New procedures, such as nucleic acid (gene) probes and polymerase chain reaction amplification of target viral genomes, provide more rapid, specific, and sensitive approaches for detection of viruses especially the fastidious ones (i.e., HAV and Norwalk virus). During the investigation of the outbreak of HAV associated with the consumption of raw oysters, polymerase chain reaction was used to detect the presence of HAV in oysters and scallops from both unapproved and approved waters (9). Although new molecular biological approaches have great promise, quantitative information is needed in order to develop a QRA. Molecular biological methods should be developed to provide quantitative information and to distinguish between inactivated and viable viruses. Comprehensive studies are needed that incorporate the tools of epidemiology, QRA, and improved methodologies for detecting and quantifying viruses, and perhaps candidate viral indicators. In this way, models can be assessed and validated, and appropriate methodologies for undertaking a comprehensive microbial risk assessment for bivalve molluskan shellfish and harvesting waters can be developed. Such an approach would lead to better information from sanitary surveys for assessing microbial water quality and shellfish quality. The sources of human enteric viruses and many other enteric pathogens in coastal waters and shellfish are human wastes primarily in the form of sewage treatment plant effluents and septic tank discharges. Approximately, 10 trillion liters of sewage per year with various levels of treatment are discharged to estuaries from coastal states (5). Stormwaters with and without direct sewer inputs also have been reported to contain enteric viruses in levels ranging from 69 to,800 PFU/100 L (8). The Gulf of Mexico continues to be the fastest growing coastal region in the nation, with a population increase of 30% from 1970 to A large percentage of this population has settled in urban complexes that have developed around port cities (i.e., Tampa Bay, New Orleans, and Houston) (5). The Gulf of Mexico estuaries produced over 11,818 kg of oysters in 1985 and continue to produce most of the oysters harvested in the United States. This area also receives discharges from a greater number of municipal sewage treatment facilities than any other coastal region. Furthermore, nonpoint sources such as stormwaters and septic tank discharges, which do not originate from the end of a pipe, are more difficult to control and may contribute to greater than 50% of the water quality problems. Shellfish harvesting is prohibited from 34% of the estuaries in the Gulf of Mexico and from 13 to 31% of the other regions in the United States (6). These closures are based on identifying sources of fecal contamination and on the levels of total or fecal coliform bacteria in the water. The sources and amounts of fecal contamination and the potential levels of viruses impacting coastal waters need to be better defined. The inability to identify sources of fecal contamination and the potential for misclassification of waters on the basis of coliform bacteria levels are on-going problems in shellfish sanitation. The public's view on the safety of our seafood may be changing. Consumer Reports (7) reviewed the contamination level of fish and shellfish in the United States in the report "Is Our Fish Fit to Eat?" They reported that 38% of the shellfish exceeded the fecal coliform standard and 44% of the fish and shellfish, overall, were contaminated based on the same indicator. This type of information is insufficient to determine the public health risks, especially from enteric viruses, and it leads to tremendous uncertainty due to the inadequacy of the bacterial indicator system. Information on the pathogens of concern is needed. The National Advisory Committee on Microbiological Criteria for Foods (4) has recommended "proper growing water classification and restriction of harvesting" for protection of public health from shellfish-associated disease. It is recognized that more suitable criteria are needed for growing water classification. In conjunction with the fecal coliform/'escherichia coli guideline, QRA could be used to characterize microbial water quality, not only for pathogenic viruses, but for bacterial pathogens of concern as well. This would lead to a risk-based classification of waters, development of appropriate monitoring strategies, and better approaches to control microbial contamination of coastal waters. As we address the complex issues of pollution control and public health during the next decade, the use of scientific risk assessment will lead

8 1050 ROSE AND SOBSEY toward the greater assurance of seafood safety and protection of public health. REFERENCES 1. Assaad, F., and 1. Borecka Nine-year study of WHO virus reports on fatal virus infections, Bull. W.H.O. 55: Bennett, J. V., S. D. Homberg, M. F. Rogers, and S. L. Solomon Infectious and parasitic diseases. Am. J. Preventive Med. 55: Centers for Disease Control Summary of notifiable diseases, United States. Morbid. Mortal. Weekly Rep. 39: Centers for Disease Control Waterbome and foodborne disease outbreaks. Morbid. Mortal. Weekly Rep. 39: Centers for Disease Control Aseptic meningitis - New York State and United States, weeks 1-36, Morbid. Mortal. Weekly Rep. 40: Cherry, J. D Non-polio enteroviruses: Coxsackieviruses, echoviruses, and enteroviruses, pp In R. D. Feigin and J. D. Cherry, (ed.), Textbook of pediatric infectious diseases. W. B. Saunders, Philadelphia. 7. Consumer Reports Is our fish fit to eat? Consumer Rep. 57: DeLeon, R. and C. P. Gerba Viral disease and transmission by seafood. In J. O. Nriagu and M. S. Simmons (ed.), Food contamination from environmental sources. John Wiley and Sons, Inc.. New York. 9. Desenclos, J. C. A., K. C. Klontz, M. H. Wilder, O. V. Nainan, H. S. Margolis, and R. A. Gunn A multi-state outbreak of hepatitis A caused by the consumption of raw oysters. Am. J. Public Health 81(10): Ellender, R. D., J. B. Map, B. L. Middlebrooks, D. W. Cook, and E. W. Cake Natural enterovirus and fecal coliform contamination of Gulf Coast oysters. J. Food Prot. 4: Evans, A. S Epidemiological concept and methods, pp In A. S. Evans, (ed.), Viral infection of humans. 3rd ed. Plenum Publishing Co., New York. 1. Food and Drug Administration. 199 revision. Sanitation of shellfish growing areas. National Shellfish Sanitation Program Manual of Operations. Part 1. Center for Food Safety and Applied Nutrition, Division of Cooperative Programs, Shellfish Sanitation Branch, Washington, DC. 13. Gerba, C. P., S. M. Goyal, R. L. LaBelle, I. Cech, and G. F. Bogdan Failure of indicator bacteria to reflect the occurrence of enteroviruses in marine waters. Am. J. Public Health 69: Gerba, C. P., S. N. Singh, and J. B. Rose Waterbome viral gastroenteritis and hepatitis. Crit. Rev. Environ. Control. 15: Gerba, C. P Viral disease transmission by seafoods. Food Technol. 4: Gerba, C. P., and J. B. Rose Estimating viral risk from drinking water. In C. R. Cofhem, (ed.), Comparative environmental risk assessment. Lewis Publishers, Boca Raton, FL. 17. Goyal, S. M., C. P. Gerba, and J. L. Melnick Human enteroviruses in oysters and their overlying waters. Appl. Environ. Microbiol. 37: Haas, C. N Estimation of risk due to low doses of microorganisms: A comparison of alternative methodologies. Am. J. Epidemiol. 118: a. Jaykus, L. A., M. T. Hemard, and M. D. Sobsey Human enteric pathogenic viruses. In M. D. Pierson and C. R. Hackney (ed.), Environmental indicators of shellfish safety, van Nostrand Reinhold, New York. 19. Katz, M., and S. A. Plotkin Minimal infective dose of attenuated poliovirus for man. Am. J. Public Health. 57: Lednar, W. M., S. M. Lemon, J. W. Kirkpatrick, R. R. Redfield, M. L. Fields, and P. W. Kelley Frequency of illness associated with epidemic hepatitis A virus infections in adults. Am. J. Epidemiol. 1: Lepow, M. L., R. J. Warren, V. G. Ingram, S. C. Daughtery, and F. C. Robbins Sabin Type 1 oral poliomyelitis vaccine effect of dose upon drinking water. Am. J. Dis. Child. 104: Melnick, J. L., and C. P. Gerba The ecology of enteroviruses in natural waters. Crit. Rev. Environ. Control 10: Minor, T. E., C. I. Allen, A. A. Tsiatis, D. B. Nelson, and D. J. D'Alessio Human infective dose determination for oral Poliovirus Type 1 vaccine in infants. J. Clin. Microbiol. 13: National Advisory Committee Microbiological criteria for raw molluscan shellfish. J. Food Prot. 55: National Oceanic and Atmospheric Administration Coastal environmental quality in the United States, Chemical contamination in sediment and tissues. A Special NOAA 0th Anniversary Report, Rockville, MD. 6. National Oceanic and Atmospheric Administration The 1990 National Shellfish Register of Classified Estuarine Waters, Rockville, MD. 7. National Research Council Sea food safety. The National Academy Press, Washington, DC. 8. O'Shea, M. L., and R. Field Detection and disinfection of pathogens in storm generated flows. Can. J. Microbiol. 38: Regli, S., J. B. Rose, C. N. Haas, and C. P. Gerba Modeling risk for Giardia and viruses in drinking water. J. Am. Water Works Assoc. 83: Richards, G. P Outbreaks of shellfish associated enteric virus illness in the United States: Requisite for development of viral guidelines. J. Food Prot. 48: Rose, J. B., and C. P. Gerba Use of risk assessment for development of microbial standards. Water Sci. Technol. 4: Rose, J. B., C. N. Haas, and S. Regli Risk assessment and control of waterbome giardiasis. Am. J. Public Health 81: Schiff, G. M., G. M. Stefanovic, E. C. Young, D. S. Sander, J. K. Pennekamo, and R. L. Ward Studies of Echovirus 1 in volunteers: Determination of minimal infectious dose and the effect of previous infection on infectious dose. J. Infect. Dis. 150: U.S. Environmental Protection Agency National water quality inventory Report to Congress. EPA/ Office of Water, Washington, DC. 35. U.S. Public Health Service Publ. No. 33. Manual of operations national shellfish sanitation program. Part 1. Sanitation of shellfish growing areas. Washington, DC. 36. Vaughn, J. M., E. F. Landry, T. J. Vicale, and M. C. Dahl Isolation of naturally occurring enteroviruses from a variety of shellfish species residing in Long Island and New Jersey marine embayments. J. Food Prot. 43: Verber, J. L Shellfish borne outbreaks. U.S. Public Health Service, Food and Drug Administration, Davisville, RI. pp Wagenknecht, L. E., J. M. Roseman, and W. H. Herman Increased incidence of insulin-dependent diabetes mellitus following an epidemic of coxsackievirus B5. Am. J. Epidemiol. 133: Wait, D. A., C. R. Hackney, R. J. Carrick, G. Lovelace, and M. D. Sobsey Enteric bacterial and viral pathogens and indicator bacteria in hard shell clams. J. Food Prot. 46: Ward, R. L., D. I. Berstein, and E. C. Young Human rotavirus studies in volunteers of infectious dose and serological response to infection. J. Infect. Dis. 154: Williams, F. P., and G. S. Fout Contamination of shellfish by stool-shed viruses: Methods of detection. Environ. Sci. Technol. 6: