Documentation

Overview

This data product contains a separate Excel file for each of 15 pathogens. Together, these pathogens cause over 95 percent of the cases of foodborne illnesses acquired in the United States in a typical year. Each Excel file consists of four worksheets—three of the worksheets contain the mean, low, and high estimates of the costs of foodborne illness in the U.S. caused by the indicated pathogen. The fourth worksheet contains the assumptions used in calculating the three sets of estimates. 

The estimates are in 2013 U.S. dollars. ERS also provides an Excel file with time series data on consumer price indexes (CPI) that can be used to update values for inflation. An additional file is provided that contains estimates for the value of a statistical life (VSL) for 1997 through the most recent year available.

The cost estimates in these files are drawn primarily from two journal articles by Sandra Hoffmann, Michael B. Batz, and J. Glenn Morris, Jr. (for complete citations, see Resources):

  • The 2012 article, "Annual Cost of Illness and Quality-Adjusted Life Year Losses in the United States Due to 14 Foodborne Pathogens" provides detailed documentation on estimating the costs of health outcomes, and
  • The 2014 article, "Disease-Outcome Trees, EQ-5D Scores, and Estimated Annual Losses of Quality-Adjusted Life Years (QALYs) for 14 Foodborne Pathogens in the United States", together with its supplemental appendix, provides detailed documentation behind estimation of the incidence of more specific health outcomes, like hospitalization in an intensive care unit (ICU).

These articles build on recent Centers for Disease Control and Prevention (CDC) foodborne illness incidence estimates (see the 2011 article, "Foodborne Illness Acquired in the United States—Major Pathogens"), hospitalization data, and relevant literature to estimate the cost of illness resulting from each pathogen. 

This page provides the following information:

Estimation methods

Cost-of-foodborne-illness estimates synthesize a variety of data and peer-reviewed literature. Estimating costs of foodborne illness involves two basic steps:

  1. Disease-outcome tree diagrams are developed to follow all cases of illness caused by a specific pathogen from infection to recovery or death, showing the percentage of cases that results in different health outcomes.
  2. The economic costs of each health outcome are estimated.

The severity of illnesses and subsequent health outcomes vary by pathogen. Health outcomes may include:

  • Non-hospitalized cases who do not seek medical care;
  • Non-hospitalized cases who seek medical care;
  • Hospitalized cases of varying severity;
  • Chronic illnesses (where data adequately supports including chronic conditions);
  • Deaths.

A detailed description of the data and assumptions used in developing the disease-outcome trees is provided in the 2014 article, "Disease-Outcome Trees, EQ-5D Scores, and Estimated Annual Losses of Quality-Adjusted Life Years (QALYs) for 14 Foodborne Pathogens in the United States." Detailed descriptions of the data and assumptions behind these cost-of-illness estimates are provided in the 2012 article, "Annual Cost of Illness and Quality-Adjusted Life Year Losses in the United States Due to 14 Foodborne Pathogens."

Economic cost is not limited to direct financial costs. The ERS cost-of-foodborne-illness estimates are a conservative approximation of individuals' willingness to pay to prevent these illnesses and their outcomes. ERS's methodology combines a human capital estimate of the cost of non-hospitalized and hospitalized health outcomes, along with use of an estimate of public willingness to pay to reduce mortality risk from foodborne illness. A more detailed discussion of conceptual issues involved in measuring the cost of illness is provided in a report:

Making Sense of Recent Cost-of-Foodborne-Illness Estimates

Estimating physical outcomes

Estimates of the incidence of illness, hospitalization, and deaths for each pathogen are taken from recent Centers for Disease Control and Prevention (CDC) research ("Foodborne Illness Acquired in the United States—Major Pathogens"). For Salmonella (non-typhoidal), Escherichia coli O157 (E. coli O157), Listeria monocytogenes, and Campylobacter (all species), existing ERS disease-outcome trees were updated to reflect the new CDC disease-incidence estimates. For the remaining 11 pathogens, new disease-outcome tree diagrams were developed based on reviews of the available scientific literature and published CDC data. For a detailed discussion of the data and assumptions used to develop the health outcome trees, see "Disease-Outcome Trees, EQ-5D Scores, and Estimated Annual Losses of Quality-Adjusted Life Years (QALYs) for 14 Foodborne Pathogens in the United States." Estimates for various physical outcomes are based on U.S. studies and data when available. Such information was not available for congenital toxoplasmosis; estimates of disease outcomes for toxoplasmosis draw upon estimates of the likelihood of different health outcomes developed for the Netherlands, and then applied to the CDC disease estimates for the United States.

Where available, the average length of a hospital stay is taken from the Nationwide Inpatient Sample (NIS) database for 2001-03, using International Classification of Diseases, Ninth Revision (ICD-9) codes to identify pathogen diagnosis; 2001-03 data are the most recent available. The NIS database was developed through a Federal-State-Industry partnership, sponsored by the U.S. Department of Health and Human Services, Agency for Healthcare Research and Quality (AHRQ)

Disease outcomes include long-term disabilities, chronic conditions, and latent impacts of acute foodborne illness that are not included in CDC’s recent incidence estimates. These chronic conditions include Guillain-Barré syndrome (GBS) from Campylobacter (all species), diarrhea relapse from Cryptosporidium parvum, end-stage renal disease (ESRD) from E. coli O157 and non-O157 Shiga toxin-producing E. coli (STEC), vision loss from Toxoplasma gondii, and numerous impacts of congenital toxoplasmosis and listeriosis such as stillbirths, neonatal deaths, and lifelong physical and mental disabilities. Other long-term health outcomes—such as reactive arthritis and irritable bowel syndrome—are not included in the current cost-of-foodborne-illness estimates because review of the scientific literature by researchers and an expert advisory panel showed that the scientific literature does not yet provide adequate quantitative support to estimate these outcomes. For permanent congenital conditions, the duration of the illness is based on the average life expectancy of the U.S. population at birth. Estimates for Guillain-Barré syndrome are taken from the 2008 article, "Economic Cost of Guillain-Barré Syndrome in the United States" (see Resources).

Estimating the cost of physical outcomes

The methods used by ERS to develop estimates of the cost of nonfatal illness are designed to be generally consistent with methods used in prior ERS cost-of-illness estimates. Estimates for disease outcomes, costs of medical treatment, and productivity loss for nonfatal Salmonella (non-typhoidal), E. coli O157, Listeria monocytogenes, and Campylobacter (all species) were based on prior ERS cost estimates, adjusted to reflect 2011 CDC disease-incidence estimates and updated for inflation and real income growth.

Medical Treatment Costs
Hospitalized Cases

For Cryptosporidium parvum, norovirus, Shigella (all species), Toxoplasma gondii, and Yersinia enterocolitica, costs of hospitalization are the mean values of hospitalization costs from the NIS database for 2001-03, based on matching ICD-9 codes for primary diagnosis. The remaining pathogens did not have sufficient coverage in the NIS, so per-visit hospitalization costs were estimated based on diseases with similar symptoms and severities. Costs per hospitalization for non-O157 Shiga toxin-producing E. coli (STEC) illnesses were assumed to be the same as those for E. coli O157, and similarly, Cyclospora cayetanesis disease was assumed to be the same as for Cryptosporidium parvum; Clostridium perfringens disease, the same as for norovirus; and Vibrio parahaemolyticus and Vibrio (all other non-cholera species), the same as for Salmonella (non-typhoidal). Vibrio vulnificus hospitalization costs were based on those of Listeria monocytogenes, adjusted for differences in intensive care unit utilization rates.

For all pathogens, hospitalized cases were assumed to have additional medical costs including physician care, emergency room services, and outpatient services; these were modeled as equivalent to cases of salmonellosis (illness caused by Salmonella).

Non-hospitalized Cases

Mild and moderate cases of foodborne illness are assumed to have the same daily health care utilization and costs as mild and moderate salmonellosis. Daily costs are multiplied by pathogen-specific illness durations.

Prior ERS estimates found that the cost of prescription and over-the-counter medication used to treat non-hospitalized cases of E. coli O157 were only 0.02 percent of total costs of illness for this pathogen. For other pathogens, scant data were available on the cost of medications used to treat non-hospitalized cases of foodborne illness. Given the small portion that such costs contributed to the total cost of treating non-hospitalized E. coli O157, medication costs were not included for non-hospitalized cases.

Productivity Loss

Cost-of-illness estimates need to reflect the opportunity costs of time spent being ill. Productivity loss is a conservative measure of this opportunity cost; it is estimated for each health state as the product of work days lost per case and daily wage rate, adjusted for the employment rate. To remain consistent across pathogens, ERS uses the mean of the age-adjusted daily wage rate and the employment rate from prior ERS estimates for Salmonella (non-typhoidal), updated to 2013 U.S. dollars. Work days lost are based on estimates of the duration of the health outcome, adjusted based on an assumed 5-day work week. Again, a more detailed discussion of this issue can be found in "Annual Cost of Illness and Quality-Adjusted Life Year Losses in the United States Due to 14 Foodborne Pathogens," and "Recent Estimates of the Cost of Foodborne Illness Are in General Agreement." 

Cost of Chronic Illness and Disabilities

Medical costs and productivity losses from congenital listeriosis, physical and mental disabilities resulting from Campylobacter-associated GBS, and ESRD following infection with STECs are based on existing cost-of-illness studies in the peer-reviewed literature. Per-case costs of diarrhea relapse following illness due to Cryptosporidium parvum are assumed to be equivalent to a mild acute case of diarrhea. Although ERS could develop a disease-outcome tree for congenital toxoplasmosis, these illnesses were not included in cost estimates due to insufficient data/research on medical costs or productivity losses.

Deaths

Economic research on valuation of the impact of deaths has been evolving. The cost-of-illness estimates provided here reflect current research and conform to guidance for economic analysis of Federal regulation. Prior ERS cost-of-foodborne-illness estimates annuitized the Value of a Statistical Life (VSL) and applied this per-year value to remaining years of life based on age of death and average U.S. life expectancy. Effectively, this practice places a lower value on preventing death as individuals age. Currently, most Federal agencies do not annuitize the VSL but instead value each death at the full VSL, regardless of age at death. This data set uses an estimate for the VSL of $8.7 million (in 2013 U.S. dollars) based on a U.S. Environmental Protection Agency (EPA) meta-analysis of existing estimates. For a review of the economic literature that led to this change in practice, see the report:

Making Sense of Recent Cost-of-Foodborne-Illness Estimates

Prior ERS estimates also assumed that 60 percent of neonatal deaths, stillbirths, and miscarriages '‘will be replaced by siblings who fully compensate for the earlier loss" and therefore include only 40 percent of such losses in cost-of-illness estimates. See the report:

Bacterial Foodborne Disease: Medical Costs and Productivity Losses

The White House Office of Management and Budget now directs Federal agencies to use values for children that are at least as high as values for adults unless there is "specific and compelling evidence to suggest otherwise" (see Circular A-4). As there are no known willingness-to-pay (WTP) studies for such losses, ERS values all miscarriages, stillbirths, and neonatal deaths by the same VSL as used for all other deaths.

Estimates of the value of deaths that occur in the future as a result of chronic disease are discounted back to present value.

Adjusting cost-of-illness estimates over time

The current cost estimates for foodborne illnesses can be adjusted for price inflation and real income growth over time. This data set contains an Excel file with consumer price indexes (CPIs) and time-adjusted VSL values that allow users to make these adjustments. Instructions on making these adjustments are provided below.

In each pathogen Excel file, the spreadsheets for low, mean, and high costs of foodborne illness are linked to the spreadsheet with assumptions used in estimating cost-of-illness estimates for that pathogen. Users who are interested in adjusting cost-of-illness estimates to a different year can substitute per-unit costs for that year for the 2013 per-unit costs in the assumptions spreadsheet. 

Adjusting Non-mortality Cost Estimates

This data set presents cost-of-illness estimates in 2013 dollars. These estimates can be adjusted over time to update values for inflation (or deflation). New CPIs will be provided as they become available; this will allow users to express the cost estimates in this data set in future dollar values. The cost estimates can also be adjusted to a previous year.

With the exception of the VSL, adjusting per-unit cost over time is intended to reflect only price inflation (or deflation). This data set includes an Excel file that contains time series data on CPIs that should be used to adjust for price inflation (or deflation). The prices of different goods and services purchased do not all change at the same rate over time. For this reason, ERS recommends using item-specific CPIs. The spreadsheet with CPI time series data also includes instructions on which price index to use with which elements of the cost-of-illness estimates.

When adjusting costs in this data set over time, 2013 should be viewed as the baseline year. One way to remember how to adjust a cost over time is to compare two ratios. The ratio of costs in year Y to costs in 2013 is equal to the ratio of the CPI in year Y to the CPI in 2013:

cost (year Y)/cost (2013) = CPI (year Y)/CPI (2013)

The relationship between these two ratios can be rearranged to solve for the cost-of-illness estimates in year Y:

cost (year Y) = cost (2013) x CPI (year Y)/CPI (2013)

As an example, look at how the cost per physician’s office visit in 2013 dollars for Shigella—$135.96—would be adjusted to 2006 dollars. First, open the Excel file on CPI indexes and find that the correct CPI component to use for "physician services" is "physician office visits" in this data set. The value of this CPI component in 2013 is $354.20; its value for 2006 is $291.90. Using the formula above, the cost of a physician’s office visit for Shigella in 2006 dollars is equal to the per-case hospitalization cost for Shigella in 2013 dollars—$135.96—multiplied by the ratio of the CPI value for 2006 to the CPI value for 2013:

$135.95 x 291.90/354.20 = $112.05

In the assumption worksheet for Shigella, replace the 2013 value of the cost per office visit ($135.95) with the 2006 value of $112.05.

Adjusting the VSL over time

The VSL is calculated from a point on the demand curve for reductions in risk of death. The demand curve for reductions in risk of death reflects the amount that people are willing to pay for different degrees of change in the risk of death. While it may seem odd to talk about choosing whether to pay to reduce risk of death, people do this all the time in very mundane ways. For example, people regularly think about the trade-off between the purchase price and safety performance when buying an automobile. 

Over time, people experience not only price inflation or deflation, but also changes in their real income. Demand for a good or service changes with its price; and usually demand changes with real income as well. For example, the demand for Rolls Royces is higher among very high income groups than it is among low-income groups. Demand for safety generally increases as real incomes rise. As a result, over time, VSL values need to be adjusted for both price inflation (or deflation) and changes in real income.

ERS has calculated adjustments to the VSL over time and provides a spreadsheet of VSL values from 2000 through the most recent year for which data on price indices and real income are available. ERS uses EPA’s estimate of the income elasticity of the VSL of 0.40 and the all-goods CPI in adjusting the VSL. This spreadsheet will be updated as new data become available.

Users should replace the 2013 VSL in the worksheet on assumptions for the specific pathogen of interest with the VSL for the year which they want to estimate cost-of-foodborne illness.

Planned updates

ERS plans to update cost-of-illness estimates on a 5-year cycle. The updated estimates will incorporate new data and scientific literature which form the basis of the cost-of-illness estimates, including changes in hospitalization and medical treatment costs, duration of hospitalizations, wage rates, epidemiological research, and the incorporation of new foodborne-illness incidence estimates as they become available.

Resources

Batz, Michael B., Sandra Hoffmann, and J. Glenn Morris, Jr., "Disease-Outcome Trees, EQ-5D Scores, and Estimated Annual Losses of Quality-Adjusted Life Years (QALYs) for 14 Foodborne Pathogens in the United States," Foodborne Pathogens and Disease 11(5): 395-402 (2014).

Anekwe, Tobenna D., and Sandra Hoffmann, "Recent Estimates of the Cost of Foodborne Illness Are in General Agreement," Amber Waves (November 2013).

Hoffmann, Sandra, and Tobenna D. Anekwe. Making Sense of Recent Cost-of-Foodborne-Illness Estimates. USDA, Economic Research Service, EIB-118 (September 2013).

Hoffmann, Sandra, Michael B. Batz, and J. Glenn Morris, Jr.,"Annual Cost of Illness and Quality-Adjusted Life Year Losses in the United States Due to 14 Foodborne Pathogens," Journal of Food Protection 75(7): 1291-1302 (January 2012).

Batz, Michael B., Sandra Hoffmann, and J. Glenn Morris, Jr., "Ranking the Disease Burden of 14 Pathogens in Food Sources in the United States Using Attribution Data from Outbreak Investigations and Expert Elicitation," Journal of Food Protection 75(7): 1270-77 (January 2012).

Scallan, Elaine, Robert M. Hoekstra, Frederick J. Angulo, Robert V. Tauxe, Marc-Alain Widdowson, Sharon L. Roy, Jeffery L. Jones, and Patricia M. Griffin, "Foodborne Illness Acquired in the United States—Major Pathogens," Emerging Infectious Diseases 17 (1), Centers for Disease Control and Prevention (January 2011). http://dx.doi.org/10.3201/eid1701.P11101

Frenzen, Paul, "Economic Cost of Guillain-Barré Syndrome in the United States," Neurology 71(1):21-27 (July 2008).

Buzby, Jean, Tanya Roberts, C.T. Jordan Lin, and James MacDonald. 

Bacterial Foodborne Disease: Medical Costs and Productivity Losses

USDA, Economic Research Service, AER-741 (August 1996).

U.S. Office of Management & Budget. Circular A-4, "Regulatory Analysis": 31 (September 17, 2003).

Recommended citation 

Economic Research Service (ERS), U.S. Department of Agriculture (USDA). Cost Estimates of Foodborne Illnesses. http://ers.usda.gov/data-products/cost-estimates-of-foodborne-illnesses.aspx.

Acknowledgments

ERS wants to acknowledge the significant contribution that Bryan Maculloch of USDA's Food Safety and Inspection Service (FSIS) made in preparing this data product for publication. ERS thanks FSIS for supporting this collaboration. ERS also thanks Michael Batz of the University of Florida for his advice in developing this data product based on articles that he co-authored with ERS economist, Sandra Hoffmann.