Chapter 11 Risk characterization “integrates the results of [dose response and exposure assessment] into a risk statement that includes one or more quantitative estimates of risk” [1], along with both qualitative and quantitative uncertainties. One decision to be made in framing a risk estimate is what outcome(s) are most appropriate to the needs of decision makers and other stakeholders. For example, the following outcome measurements might be relevant for microbial risk assessment (each of the measurements would be evaluated for a spectrum of policy choices to ascertain the relative consequences of a given decision): Even given these potential endpoints, we realize that there may be a span of outcomes in term of illness—ranging from mild to grave (and death). It may be desirable to integrate these outcomes into a metric. Each of these measurements (the list is by no means exhaustive) consists of a single numerical value (for a given policy alternative) and is termed a “point estimate.” The first two quantities attempt to depict a central measure of a consequence. The next two quantities attempt to depict some upper value or “conservative” measure of consequence. The final quantity required information of the dynamics (incubation, duration) of disease to assess the case burden as a function of exposure. What none of these (and other “point estimates”) depict is a range of values that denote the uncertainty and variability of the input quantities and assumptions of the characterized risk. In contrast, an interval estimate of risk is presented as either a confidence region or a full probability distribution of the resulting risk. The assessment of uncertainty and variability in a risk characterization is important since it has been argued that “a decision made without taking uncertainty into account is barely worth calling a decision” [2]. Indirectly, the third and fourth outcomes enumerated in the preceding list attempt to account for underlying uncertainty and variability, but as will be noted do so in an unclear manner. There are a number of approaches to valuing residual outcomes. One approach would be to seek to monetize the outcomes (or reduction in adverse outcomes) and to balance this against the costs to achieve that benefit. Conceptually, the idea can be shown in Figure 11.1. If the costs of residual risk (benefits) can be estimated in commensurate terms (e.g., money) with the costs of compliance 1 for a particular level of decision (stringency), then an optimal decision is given by minimizing the total costs. However (as indicated by the width of these curves), the uncertainty in estimating both sets of costs may be quite high, resulting in a broad optimum (that may be decided by other factors such as equity.). Figure 11.1 Concept of minimizing social costs. Once the extent of outcomes, in terms of total infections (or cases), or in terms of community impact (via epidemic models), has been assessed, the risk may need to be quantified in terms of economic impacts to a community. One method (although not the only method) of performing risk management is by the use of a risk-benefit calculus. If a number of different interventions (say levels of water treatment) are evaluated for their relative economic costs, and if the microbial (and other) risks resulting from these alternatives are also evaluated for their economic costs, then a balancing approach in commensurable units may be made. It should be noted that there are a number of technical, as well as ethical, issues in determining the economic value of disease impacts. Some of the technical issues include how to place an economic value on levels of illness and on lost time other than in employment (e.g., leisure time) [4]. In addition, Harrington et al. [4] noted that valuation of “anxiety” or permanent changes in behavior (or attitude) as a result of illness needs to be better considered. In massive outbreaks, there may be a permanent loss of value to a city, or to a company or employer (the latter as in the case of a foodborne outbreak). Finally, the impact cost estimates that have been reported typically are on a direct cost-incurred basis, rather than on a “willingness to pay” to avoid impact basis (with the latter frequently being greater than the former) [4]. Estimates based on “willingness to pay” approaches are discussed below. Most of the available literature focuses on estimating impacts associated with outbreak situations. There is an unstated presumption that the direct costs to individuals (lost wages, medical costs) from endemic illnesses are similar to those incurred in outbreaks of the same infectious agent. However, in the endemic situation, it is reasonable to expect that a number of “avoided” costs would be less or nonexistent. For example, in endemic waterborne illness, where the affected individuals did not become aware that their source of illness was from the water, one would not expect behavior-modifying activities such as switching to alternative (e.g., bottled) water supplies or use of home water purifiers to occur. The issue of economic valuation of waterborne and foodborne disease is therefore one of the most open technical areas. We shall review a number of prior studies, with the aim of providing prototypical values that might be used as a point of departure in subsequent efforts. A 1981-waterborne outbreak of gastroenteritis in Eagle-Vail, Colorado, affected 48.2% of a community of 3540 with an estimated total case burden of 1706 [5]. From a telephonic survey, the estimated costs incurred by affected individuals were determined. The aggregate and per case (of illness) costs incurred are summarized in Table 11.1. The costs incurred by individuals, per case of illness, were US$97.20 (1981 dollars). There were apparently no severe illnesses (and no fatalities) incurred in this event. The long-term costs for plant upgrading were substantial. Table 11.1 Costs (1981 dollars) of Eagle-Vail Waterborne Gastroenteritis Outbreak Source: Modified from Ref. [5]. Harrington et al. [4] assessed the economic impact of a 1983 outbreak of giardiasis in Luzerne County, Pennsylvania. Of the community of 25,000, there were 6,000 total cases and 370 clinically confirmed cases of illness. Surveys were sent to these confirmed cases to estimate incurred costs. The authors, as part of the survey, addressed direct costs incurred by the ill persons, as well as costs (such as lost wages) incurred by caregivers. A separate survey of the community was conducted to estimate costs of behavior alterations, particularly switching costs of water supply (during the outbreak). Costs of long-term behavior alterations were not assessed. Alternative supplies were obtained by hauled water or boiling water, with virtually no switching to bottled water. The costs of aversive action were determined by applying an hourly wage rate to the time required. A summary of the results of these authors is given in Table 11.2, on a total community basis and per case (total) basis. Note that the per case direct costs (medical plus loss of work) are somewhat more substantial than in the Eagle Vail outbreak (Table 11.1) due to the greater severity of giardiasis (relative to gastroenteritis). What is also very noteworthy about Table 11.2 is the substantially greater cost from averting illness during an outbreak (obtaining water from alternative supplies) than the direct medical and lost work and productivity/leisure costs. As stated before, the costs of aversive behavior would likely not be incurred in a nonoutbreak situation 2 ; hence, there is a much greater (economic) value to avoiding illnesses in an outbreak situation than in an endemic (nonoutbreak) situation. Table 11.2 Economic Costs of Luzerne County, PA, Giardiasis Outbreak (1984 dollars) Source: Modified from Ref. [4]. The three scenarios use a different assumed after-tax wage rate for unemployed, homemakers and retirees of US$6.39 (high), US$6.08 (medium), and US$2.65 (low) per hour. The direct medical and productivity costs associated with several foodborne illnesses were estimated by Mauskopf and French [6]. These do not include costs of aversive actions; the authors were particularly interested in endemic situations. For mild, moderate, and severe salmonellosis, the methods yielded cost estimates of US$197 and US$222, US$622 and US$890, and US$86,895 and US$743,000 per case, respectively. The much larger impacts from severe salmonellosis result from the relative large (13%) mortality ratio assumed. However, note that the cost impact of mild salmonellosis is similar to that for gastroenteritis (Table 11.1), while the impact of moderate salmonellosis is similar to that for giardiasis (Table 11.2) (excluding aversion costs). For the 1993-Milwaukee Cryptosporidium outbreak, the total cost per person for mild, moderate, and severe illness was estimated as US$116, US$475, and US$7808 per case (1993 US dollars), including medical costs and costs of lost productivity (for both patients and caregivers) [7]. However, these authors noted the substantial additional costs not accounted for in this figure from consequences to business, modifications of the local treatment costs, and provision of alternative water sources. Based on foodborne disease data, and detailed consideration of the courses of illness for different pathogens, Hoffmann et al. [8] developed estimates for costs associated with mild, moderate, and severe manifestations of different infectious agents. Costs were based on medical costs and direct lost wages. The information is given in Table 11.3. Mild cases are those involving no doctor’s visit. Moderate cases seek a doctor’s attention, but do not require hospitalization, and severe cases (nonfatal) require hospitalization. Examination of this table reveals a high degree of similarity of costs among mild and moderate cases (interestingly, the two protozoan illnesses—cryptosporidiosis and cyclosporiasis—have somewhat higher costs for the mild cases than the other mild cases). There is a higher degree of divergence in the cost of severe cases, ranging from a low of approximately US$12,075 per case (Campylobacter and Salmonella) to a high of US$71,200 for Listeria. Table 11.3 Cost of Illness per Case for Various Infectious Agents (in 2009 dollars) Source: Adapted from Ref. [8]. The costing methods discussed above rely on a bottom-up examination of direct (and possibly indirect) costs associated with microbial risk (or reduction in microbial risk). Some have advocated using a “top down” approach based on consumer surveys of willingness to pay (WTP) for various interventions that could reduce risks. Two examples of this approach in the field of food safety are studies of Teisl and Roe [9] and Haninger and Hammitt [10]. While some have argued that WTP could lead to higher estimates of risk cost, the empirical tabulation presented by Roberts [11] and summarized in Table 11.4 seem to contrast this. Comparing Table 11.4 with Table 11.3, the WTP for mild and moderate cases is far greater than the direct + lost wages costs, while for severe cases, the WTP is (for the most part) less than the direct + lost wages costs. Further research is clearly needed as to the relatively appropriateness of the direct and WTP methods for microbial risk reduction benefit assessment. Table 11.4 Willingness to Pay for Different Severities Source: Compiled from Ref. [11]. a Roberts terms this as Moderate, with no medical care. b The Moderate category here are cases who sought a physician’s care, but require no testing (lower value) or require diagnostic tests (higher value). An alternative approach to the monetization of different degrees of severity of an illness is to use another approach to use a nonmonetary weighted index to express the diminution of health and then to compare this (for decision making purposes) with the costs (regulatory, technological) needed to achieve a reduction in this adverse effect. Two such are disability adjusted life years (DALY) and quality adjusted life years (QALY). The DALY approach will be described first in detail and then the differences of the QALY approach. DALYs were used by WHO to determine the global burden of disease in a way to integrate morbidity, duration, mortality, and time of mortality in a systematic way [12, 13]. In the terminology of Havelaar et al. [14], DALYs lost is the summation of two quantities: In mathematical terms, this can be written as:
Risk Characterization and Decision Making
Introduction
Valuing Residual Outcomes
Classical Economics
Per Case (US$)
Direct costs
Medical (physician visits, medications)
3.25
Time lost from work
81.06
Bottled water
12.90
Total individual costs
97.20
Short-term emergency response to treatment plant
54.16
Total direct costs
151.37
Indirect costs
Long-term infrastructure repairs
545.84
Cost of epidemiologic investigation
5.10
Total indirect costs
550.94
Total costs
702.30
Loss Category
Costs per Case
High (US$)
Medium (US$)
Low (US$)
Direct medical costs
178
175
172
Direct lost wages
358
272
208
Value in loss of productivity and leisure time
630
485
385
Estimated loss due to averting actions
6418
2157
2,020
Total
7585
3088
2785
Agent
Mild (US$)
Moderate (US$)
Severe (US$)
Campylobacter
52.09
558.83
12,076.10
Cryptosporidium parvum
104.49
598.53
19,047.62
Cyclospora cayetanensis
99.28
604.69
18,181.82
E. coli O157:H7
30.44
553.80
13,470.53
Listeria monocytogenes
71,202.75
Norovirus
52.05
524.68
19,645.43
Salmonella, nontyphoid
52.02
552.82
12,075.92
Shigella
52.35
560.85
20,810.44
Toxoplasma gondii, adult
51.53
524.36
58,310.75
Vibrio vulnificus
68,817.20
Yersinia enterocolitica
51.76
561.57
21,575.98
Severity
US$ Per Case (Adults)
Milda
11,100
Moderateb
11,700/14,400
Severe
16,100
DALYs and QALYs
Definition
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
