Using COI and WTP Methods to Estimate the Societal Costs of Foodborne Illness

1
Using COI and WTP Methods to Estimate the
Societal Costs of Foodborne Illness

• Tanya Roberts, Economic Research Service (The
  views in this presentation are my own and do not
  reflect any
• official position of ERS or the U.S. Department
  of Agriculture)
• IRAC Public Conference Risk Assessment,
• Economic Analysis, and Foodborne Illness
  Regulations
• November 16, 2007

2
Overview

• Why a public role in foodborne illness data?
• Epidemiological information is missing
• Missing information?low economic incentives
• COI/WTP estimates of human foodborne illness
• Illness (morbidity) cost estimates
• Value of a statistical life (unidentified person
  dies)
• Discussion of valuation

3
CDC Estimates of Human Illnesses Caused by
Foodborne Pathogens

• 200 foodborne pathogens linked to human illness
• CDC estimates for U.S. each year
• 76 million acute FBI
• 325,000 acute hospitalizations
• 5,200 acute illnesses result in death
• Unknown number of sequellae
• Mead et al. (1999) Emerging Infectious Diseases,
  5(5)607-625www.cdc.gov/ncidod/eid/vol5no5/mead.h
  tm

4
Weak Incentives for Pathogen Control

• Identifying illness/food/pathogen connection is
  difficult
• Illness often occurs days after pathogen
  consumption
• Known outbreaks are best chance for ID food
• 2005 FoodNet data on outbreaks
• Only 410 outbreaks (10,781 illnesses) have
  pathogen ID
• Illnesses in an outbreak with pathogen ID are
  0.014 of all 76 million foodborne
  illnesses
• Weak private sector incentives for pathogen
  control
• Even smaller probability that ill person will be
  compensated
• Market failure because of missing
  information/compensation
• Signal for government intervention to correct
  economic incentives

5
ERS Foodborne Cost EstimatesMorbidity-COI
Mortality-WTP
Pathogen 2000 (Billions) Salmonella 2.4 Li
steria monocytogenes 2.3 Campylobacter
spp. 1.2 E. coli O157H7 0.7 E. coli,
non-O157 STEC 0.3 Total
6.9 billion www.ers.usda.gov/briefi
ng/FoodborneDisease/features.htm
6
Societal costs market/regulatory failure?

• 6.9 billion/yr estimate for human illnesses is
  low
• Only 5 pathogens underestimates public health
  problem
• COI method for morbidity (medical costs,
  productivity losses)
• Research results Benefit/Cost Analysis of
  HACCP
• Found public health protection benefits gt
  industry costs
• Too little food safety provided by private sector
• www.ers.usda.gov/publications/aer755/
• Role for regulators to improve food safety
• Economic incentives are embedded in
• Regulations
• Enforcement

7
ERS needs better estimate of societys value for
safe food

• COI morbidity valuation is based on medical costs
  and productivity losses
• WTP valuation for value of a statistical life is
  a comprehensive valuation measure
• Includes value for lost leisure time
• Includes value for pain and suffering
• More consistent with economic theory than COI
• ERS needs to use WTP for morbidity valuation
• Vast majority of foodborne cases do NOT die
• 5,200 deaths/year out of 76 million cases/year
• WTP would value these illnesses more accurately

8
ERS Future Willingness to Pay Method

• Few consumer surveys on WTP for food safety
• Accuracy a question when so few
• ERS commissioned two studies
• Hammitt and Haninger is one of these studies
• How use the data in Hammitt and Haninger?
• Assess what data are most accurate best
  estimates of societal values
• Combine WTP estimates with CDC/FoodNet estimates
  of foodborne illness, by severity category

9
Issues for new valuation estimates

• Premature Death value of a statistical life
  (VSL)
• Include all foodborne deaths in estimates
• Compare WTP in consumer surveys to WTP in risky
  jobs
• Morbidity for acute foodborne illness
• Use what illness severity categories?
• Use WTP to reduce food risk from HH survey

10
Severity of foodborne illnesses 76 million cases
of acute foodborne illness each yeara
Deathsa 0.007
Hospitalizationsa 0.425
Medical visitsb 20 of cases
Ill person with no medical visitb 80 of cases
aMead 1999 Emerg. Inf. Dis. bJones 2007
Epidemiol. Infect.
11
FoodNet Data by Age, 2001-2005

• Severity/Age 0-14 15-69 70
• 
  
  
• Pathogen test 41 55 4
• Hospitalization 31 55 14
• Deaths 10 43 47
• Illness severity categories are mutually
  exclusive.
• .

12
HH Estimates of WTP to Avoid the Risk of a
Foodborne Illness

• Adult moderate case of foodborne illness
• 1 day 11,100
• 3 days 11,700
• 7 days 14,400
• Child moderate case of foodborne illness
• 1 day 28,000
• 3 days 30,400
• 7 days 26,500
• Smaller sample size for children
• Big problem, small variation with days of illness

13
Impose Daily Values on WTP per Foodborne Illness

• Disease severity Duration Value per
  case
  days child adult
• No medical care 1 3,800 2,100
• (1 day moderate illness)
• See physician/no test 3 11,400
  6,300(3 days moderate illness)
• See physician/ test 7
  26,500 14,400(7 days moderate illness)
• Hospitalized cases 7
  26,700 16,100(7 days severe illness)
• linear transformation of moderate 7 day value,
  HammittHaninger

14
WTP/VSL to Avoid Foodborne Deaths

• Age 1 Death Cases Total cost
• (million ) ()
  (billion )
• Child (0-14) 10 520
  5.2
• Adult (15-69) 6 2,236 13.6
• Elderly (70) 4.2 2,444 10.3
• TOTAL 5,200 29 billion

15
WTP Values Acute Foodborne Illnesses

• Severity of cases Cases Total
  costs
  billion US
• No medical care 60,800,000 170
• (1 day moderate)
• See physician/no test 12,878,489
  108
• (3 days moderate)
• See physician/test 1,991,311
  39
• (7 days moderate)
• Hospitalized cases 325,000
  6
• (7 days severe)
• Premature deaths 5,200
  29
• TOTAL (preliminary) 76 million 350
  billion
• No sequellae included in this estimate, based on
  HH

16
Discussion of Valuation

• Preliminary WTP estimates morbidity
• Higher values for morbidity than Cost of Illness
  method
• Will other WTP studies replicate these high
  values?
• WTP estimates mortality
• Death by food doesnt appear to be valued
  differently than other deaths
• How will new CDC/FoodNet estimates change?
• Changes in total number of cases or severity?
• Change the distribution of illnesses by
  pathogens?
• Will sequellae be included this time?