1 CFSAN values productivity losses and pain and suffering using quality adjusted life years (QALYs). This method uses a two-step procedure for valuing health losses. In the first step, the effect of a condition on health is estimated to be between zero (well-being in the full health state) and one (well-being in death). For example, a QALY loss of 0.14 for arthritis means that for every day of suffering with arthritis the affected individual has a level of well-being 14% lower than he/she would have had in the absence of arthritis. In the next step the value of a QALY is estimated and multiplied by expected QALY losses to calculate the value of health losses. QALYs are designed to measure the loss of well-being both from symptoms and from activity limitation. Consequently, both pain and suffering and productivity values are captured by this measure. Data taken from Table 2 in Raybourne, Roberts, Williams and Arthritis Working Group, draft 2001.
In addition to the common acute symptoms of foodborne illness, an estimated 1-3% of all foodborne-illness cases develop secondary illnesses or complications that can occur in other parts of the body (CAST, 1994, Table 2.2). These complications, called chronic sequelae, can occur in any part of the body including the joints, nervous system, kidneys, or heart, and may afflict the patients for the remainder of their lives and/or result in premature death. For example, Campylobacter infections are estimated to cause 20-40% of all Guillain-Barre syndrome (GBS) cases (a major cause of paralysis unrelated to trauma) in the United States. About 1.5% of E. coli 0157:H7 disease patients develop hemolytic uremic syndrome (HUS), which usually involves red blood cell destruction, kidney failure, and neurological complications such as seizures and strokes.
The medical literature indicates that the impact of the infection and its complications may vary depending on the age and health status of the individual. For example, in an ERS COI study of campylobacteriosis GBS cases, the analysis was complicated by demographic differences and the broad array of possible GBS symptoms, subsequent medical costs, and final outcomes (Buzby et al., 1997a,b). Using data from Sunderrajan and Davenport (1985), ERS divided patients with GBS into two age and treatment categories:
• Mechanically ventilated patients with an average age of 47
• Those not on mechanical ventilation with an average age of 30
Those who are mechanically ventilated face more serious complications and prognoses than those who are not, including a reduced likelihood of returning to work (see Appendix to this chapter and Fig. 7A.1 for a more detailed example of this COI analysis).
SCIENTIFIC BASIS AND IMPLICATIONS: MICROBIAL HAZARDS
Traditionally, the social costs of human illness associated with microbial food-borne pathogens have been estimated with the COI method. COI analyses have typically estimated only the individual's (or household's) medical costs, lost productivity, and value of premature death from a particular illness or injury. Other costs are usually omitted because of the lack of suitable measures, often resulting in underestimation of the true societal costs. The important advantage of a COI measure is that it employs readily available and reliable economic data (such as wages and hospitalization costs). Also, these relevant data are precise enough to allow for sensitivity analyses of the response of the measure to changes in medical costs or demographic profiles of affected individuals. Because they are so tractable, COI measures have been widely used by economists and policymakers for several decades.
Malzberg developed the COI method in 1950, and Rice codified its empirical application in 1966. The application of the method to foodborne illness is limited but increasing. The method has been applied for selected foodborne pathogens in the United States (Buzby and Roberts, 1997a; Buzby et al., 1996; Cohen et al., 1978; Roberts et al., 1998; Roberts and Marks, 1995; Todd, 1989a; Sockett and Stanwell-Smith, 1986), Croatia (Razem and Katusin-Razem, 1994), the United Kingdom (Roberts and Upton, 1997), and Canada (Todd, 1989b) and for acute infectious intestinal disease in England (Djuretic et al., 1996).
The first step in any COI analysis is to determine the incidence of a specific illness. Incidence is often expressed as the number of new cases of a disease per
100,000 individuals in a 1-year period. The quantification of foodborne disease incidence is a matter of great controversy because of uncertainties about the true incidence (CAST, 1994). Because the nature and reporting of foodborne diseases result in vast undercounting of the actual incidence of illnesses, incidence rates are often estimated by expert opinion. The ERS estimates incorporate the best available estimates from CDC (Mead et al., 1999).
For each foodborne illness, cases are classified by severity. In the ERS COI presented here, four acute illness severity groups were used: those who did not visit a physician, those who visited a physician, those who were hospitalized, and those who died prematurely because of their illnesses. For some foodborne illnesses, a fifth severity group was used for patients who developed select chronic sequelae from the acute illness.
For each severity group, medical costs were estimated for physician and hospital services, supplies, medications, and special procedures unique to treating each particular foodborne illness. Such costs reflect the number of days/treatments of a medical service, the average cost per service/treatment, and the number of patients receiving such service/treatment. Hospitalization accounts for a large proportion of these costs. Data to estimate medical costs come from nationwide data bases, such as the published Medicare reimbursement rates and per capita expenditures on physician services from the Health Care Financing Administration (HCFA), the American Hospital Association's Annual Survey of Hospitals, and the National Center for Health Statistics' National Hospital Discharge Survey (NHDS) and National Mortality Follow-Back Survey.
The incidence data combined with information on severity were also used to estimate the costs of lost productivity. Most people with foodborne illnesses restrict their usual activity for just one or two days. However, some patients die and others develop chronic complications so serious that they never return to work, regain only a portion of their pre-illness productivity, or switch to less demanding and lower-paying jobs. The total cost of lost productivity is the sum of costs for all individuals affected, primarily the patient or, in the case of ill children, their parents or paid caretakers. For those cases in which work is temporarily interrupted, we estimate the productivity loss as the product of time lost from work multiplied by the corresponding wage rate published by the Bureau of Labor Statistics. The daily wage of an individual is frequently used in economic studies as a proxy for the value of output produced in a day's work. When data are not available on time lost from work due to illness, this lost time is estimated by assuming a typical ratio of the average time spent in the hospital to time lost from work.
ERS has historically used two different methods for calculating proxy values for the forgone earnings of someone who dies prematurely or is unable ever to return to work because of their foodborne illness (Table 7.4).
TABLE 7.4. Estimating the Value of a Statistical Life
(1) Market approach:
• Estimated from studies of market behavior. Most estimates come from hedonic-wage studies, which use labor market data on how much employers must offer workers, in terms of higher wages, to induce them to take a job with some injury risks, as opposed to a similar job with no such risks. Other estimates come from studies of seat belt use and other automobile safety features as well as costs incurred to avoid contaminated drinking water and air pollution. Still others come from contingent valuation studies that study stated behavior, that is, responses to hypothetical choice situations.
• Estimated value of saving a life is $5 million for each life (1990 dollars), regardless of age, updated to current dollars. More recent valuations updated to adjust for age.
• Estimates of willingness to pay to avoid temporary or chronic illnesses vary according to individuals' risk preferences and according to the characteristics of the illness.
The market approach excludes government and industry costs as well as other costs that individuals may not consider when making choices among specific risky alternatives.
(2) Landefeld and Seskin's human capital/willingness to pay (WTP) approach:
• Generates the present value of expected lifetime after-tax income and housekeeping services at a 3% real rate of return, adjusted for an annual 1% increase in labor productivity and a risk aversion factor of 1.6. The risk aversion factor is based on the ratio of life insurance premium payments to life-insurance loss payments. In most cases, life insurance premiums represent "household WTP for potential losses associated with the death of an income-earning household member."
• Estimates the value of a statistical life, depending on age, to range from roughly $15,000 to $2,037,000 in 1996 dollars.
This more conservative approach underestimates the true costs of foodborne illnesses to society because it excludes costs, such as:
• Pain, suffering, and lost leisure time of the patient and family;
• Lost business and costs and liabilities of lawsuits affecting agriculture and the food industry;
• The value of self-protective behaviors undertaken by industry and consumers;
• Resources spent by federal, state, and local governments to investigate the source and epidemiology of the outbreak; and
• The value of reducing risks for people who do not become ill.
The first approach, the human capital approach, was used in earlier research at ERS. The human capital approach incorporated estimates of forgone earnings, adjusted by a "risk premium" from life insurance markets. The cost of a premature death was estimated, depending on age, to range from roughly $15,000 to $2,037,000 in 1996 dollars. These estimates were calculated with a combination of human capital and willingness to pay estimates developed by Landefeld and Seskin (LS) (1982). In essence, these estimates represent the value in today's dollars of an individual's lifetime stream of income if the illness had not occurred. The LS method generates the present value of expected lifetime after-tax income and housekeeping services at a 3% real rate of return, adjusted for an annual 1% increase in labor productivity and a risk aversion factor of 1.6. The risk aversion factor is based on the ratio of life insurance premium payments to life insurance loss payments. In most cases, life insurance premiums represent "household WTP for potential losses associated with the death of an income-earning household member" (Landefeld and Seskin, 1982, p. 562). The LS value of a statistical life lost is:
where T = remaining lifetime, t = a particular age, Y, = after-tax income including labor and nonlabor income, r = household's opportunity cost of investing in risk-reducing activities, and a = risk aversion factor. The major limitation of this approach is that it does not fully consider the value that individuals may place on (and pay for) feeling healthy, avoiding pain and suffering, or using their free time. Because the approach does not cover all of these valuable aspects of health, the approach is generally thought to understate the true societal costs of illness.
The second approach for calculating the value of a statistical life, the market approach, infers the value of a statistical life from behavior observed in market settings and is the foundation for ERS VOSL estimates for foodborne illness deaths. The fundamental assumption of this method is that people make tradeoffs between safety and other consumption goods in their daily lives. For example, Volvos are sold at a price premium in part because some consumers are willing to pay a higher price for goods that are safer to use, in this case, cars that provide greater protection in accidents. The increment in price attributable to the safety features reveals consumers' willingness to pay for (implicit price of) safety at the margin. The controversy over the sweetener saccharin is another example of consumers' willingness to trade safety for other goods and services. By using saccharin, weight-conscious consumers reveal a willingness to use products associated with a potentially greater risk of cancer in return for being able to eat sweet foods containing fewer calories. The majority of studies estimating the value of life in this manner use data from labor markets. Typically, employers must offer workers higher wages to induce them to take jobs with a higher risk of occupational fatalities than jobs with a lower risk.
Viscusi (1993) compared wage differentials in 24 wage-risk studies and found that the extra wages associated with the increased overall hazard of one death from risky jobs are between $3 million and $7 million (in 1990 dollars). Other studies have obtained very similar implicit values of a statistical life with information on safety features of automobiles (Atkinson and Halvorsen, 1990, Dreyfus and Viscusi, 1995). Several regulatory agencies use either Viscusi's range of estimates or the $5 million midpoint when analyzing the benefits of proposed public-safety rules. The market approach has also been used to estimate the value of lost work time due to illness or injury [see, for example, Hersch and Viscusi (1990)]. For the current ERS estimates, this $5 million midpoint was modified by taking the age distribution of deaths from each pathogen into account, in effect treating the value of life as an annuity paid over the average U.S. life span at an interest rate of 3.0%. After age adjustment, the assumed cost of each death ranges from $8.9 million for individuals who died before their first birthday to $1.7 million for individuals who died at age 85 or older in 2000 dollars (see http:\\www.ers.usda.gov\briefing\ FoodborneDiseaseloverview.htm).
Other studies have used costs incurred by consumers in avoiding or mitigating health risks to infer values of avoiding premature death, lost work time, or episodes of specific illness (for a survey see Cropper and Freeman, 1991). These costs include both explicit cash expenditures and implicit costs such as the value of lost time. Studies of this kind include Blomquist's (1979) investigation of seat belt use, Harrington et al.'s (1989) investigation of avoiding giardiasis from drinking water, and Abdalla et al.'s (1992) investigation of avoiding risks from industrial solvents in drinking water.
The market approach has been extended to encompass cases in which markets do not exist through use of the contingent valuation method (CVM). CVM constructs estimates of willingness to pay for nonmarket goods with data from surveys in which participants are asked to make choices in hypothetical situations. The most commonly used format is the discrete choice format in which respondents are presented with a choice between two goods (e.g., foods) differing in only two ways, a quality attribute such as risk of illness or death and the price. Alternatively, respondents may be asked to report the maximum additional amount they would be willing to pay for the less risky good. Several studies have used CVM to estimate consumers' willingness to pay for reductions in symptoms of illness such as shortness of breath, nausea, and headaches. The correspondence between CVM estimates and the costs of illness or averting behaviors has not been close, in part because the samples used in the CVM studies may not be representative and in part because CVM study participants did not bear the full cost of illness [see Cropper and Freeman (1991) for a review].
The latest ERS estimates of medical costs, productivity losses, and value of premature death for diseases caused by five foodborne pathogens is $6.9 billion per year (Table 7.2). The five bacterial pathogens are Campylobacter (all serotypes), Salmonella (nontyphoidal serotypes only), E. coli 0157:H7, E. coli non-0157:H7 STEC, and Listeria monocytogenes. ERS uses CDC estimates of the annual number of foodborne illnesses, hospitalizations, and deaths for these pathogens (Mead et al., 1999). ERS has also revised its methodology to take account of age in valuing premature deaths. Under the age-adjusted approach, the assumed cost of each death ranges from $8.9 million for children who die before their first birthday to $1.7 million for individuals who die at age 85 or older. Because of changes in case estimates and the economic valuation of deaths, the ERS estimates are not strictly comparable with earlier ERS estimates of foodborne disease costs.
These COI estimates undervalue the true costs of foodborne illnesses to society, however, because the analyses covers only five foodborne pathogens believed to cause human illnesses. Over 250 organisms are known to cause foodborne illnesses. Because many different organisms cause similar symptoms (especially diarrhea, abdominal cramps, and nausea), it is rarely possible to say which microbe is causing a given illness unless laboratory tests are performed to identify the microbe or the illness is part of a recognized outbreak (see http:llwww.cdc.govlncidodldiseaseslfoodlillness.htm). Estimated costs would also increase if the costs for all chronic complications linked to foodborne illnesses, such as arthritis and meningitis, were included. These estimates primarily include medical costs, lost productivity, and the value of premature deaths. Total costs would also increase with the inclusion of other societal costs, such as pain and suffering, travel to medical care, and lost leisure time as shown in Table 7.1.
REGULATORY, INDUSTRIAL, AND INTERNATIONAL IMPLICATIONS: MICROBIAL HAZARDS
In general, the COI estimates for illness due to foodborne pathogens can be used in three main ways:
• To evaluate the economic impact of foodborne diseases on the U.S. economy,
• To target pathogen reduction efforts towards the most costly diseases, and
• To compare benefits and costs of control efforts to determine the most cost-beneficial interventions.
Societal benefits of a food safety regulation arise from prevention of foodborne illness among individuals. From an economic perspective, these benefits include, at a minimum, savings in disease prevention and mitigation expenditures, increases in worker productivity, reductions in pain and suffering, and reductions in anxiety about foodborne health risk.
The costs of food safety regulations include expenditures associated with their design, implementation, and enforcement. In 1994, the Federal government budgeted $1.2 billion on food safety regulatory activities such as inspection and laboratory testing (GAO, March 1996). The food industry also incurs costs to comply with food safety rules and regulations.
One example of COI estimates for foodborne illness used in policymaking is the Food Safety and Inspection Service's (FSIS) 1996 Pathogen Reduction/
Hazard Analysis and Critical Control Point (HACCP) regulation to improve the current meat and poultry inspection. Earlier COI estimates by the USDA's Economic Research Service (ERS) provided the foundation for the estimated benefits of this HACCP regulation (USDA, 1995 and 1996).
To compare the impact of different assumptions on the calculated benefits of this HACCP rule, ERS constructed the four scenarios shown in Table 7.5. The net benefits were estimated with the FSIS estimates of costs of industry compliance with the HACCP regulations over a 20-year time horizon. ERS assumed that benefits begin five years after the HACCP regulations. Industry compliance costs are assumed to start in the first year. The results indicated that the benefits of implementing HACCP outweighed the costs, as long as four pathogens were reduced by 17% or more (Crutchfield et al., 1997). (Note that this ERS analysis did not include E. coli STEC in the publication, AER-755.) COI estimates were also used in the Food and Drug Administration's (FDA) regulation for seafood and proposed regulations for eggs (U.S. DHHS, 1995 and 1999), and the ERS COI methodology for Listeria was incorporated in the analysis supporting the USDA's 2001 proposed regulation for ready-to-eat meat and poultry products (e.g., hot dogs and luncheon meats). This regulation has provisions for mandatory in-plant testing for Listeria and higher performance standards for some pathogens as measures of process control.
The USDA's Office of Risk Assessment and Cost-Benefit Analysis (OR-ACBA) reviews regulations proposed by the USDA that concern human health and safety or the environment and have an estimated annual economic impact of at least $100 million dollars (see Table 7.5). For these regulations, the USDA conducts a thorough analysis that makes clear the nature of the risk, alternative ways of reducing it, the reasoning that justifies the proposed rule, and a comparison of the likely costs and benefits of reducing the risk (web site: www.usda.govjoceloracba). The FDA has a similar review process, see the FDA website for FDA regulation of food and their review process (http://www. cfsan.fda.govI) as well as the U.S. government-wide web site (http:I/www. FoodScifety.gov).
BACKGROUND AND HISTORICAL SIGNIFICANCE: CHEMICAL HAZARDS
Chemicals in foods fall into two main classes: (1) substances added to prevent spoilage, improve product quality, or change color and (2) residues of pesticides used to grow crops and to prevent spoilage or damage during post-harvest processing and storage. The prevalence of illness and premature death due to such chemicals in foods is difficult to ascertain. To the best of our knowledge, their incidence is extremely low, at least in developed countries with strong regulatory systems like the United States. It appears to be so low, in fact, as to be virtually undetectable from surveillance data and epidemiological studies.
TABLE 7.5. Scenarios Used to Evaluate HACCP Rule and Benefits Assumptions
Was this article helpful?