CDC 1999 Estimates: Methods

Estimating foodborne illnesses due to known foodborne pathogens

For the pathogens with surveillance data available, Mead et al. gathered data on the number of illnesses reported to surveillance systems.  They corrected for underreporting² using generic multipliers based on the similarity of symptoms for several known pathogens, and then estimated the total number of illnesses (both reported and unreported) that occurred in the population. To that figure, they applied the estimated proportion of illnesses transmitted by food, rather than by some other route of transmission. Finally, they added the estimates for each of the pathogens to arrive at a grand total estimate for foodborne illnesses from known pathogens (Figure 1).

¹ In the 2011 estimates of the burden of foodborne illnesses, Scallan et al. use the term unspecified agents instead of unknown agents.

² In the 2011 estimates of the burden of foodborne illnesses, Scallan et al. refer to underreporting and under-diagnosing. Mead et al. did not differentiate between underreporting and under- diagnosis and used the term underreporting to describe both processes.

Figure 1. Estimating illnesses due to pathogens known to cause foodborne illness, 1999

Because of a lack of surveillance information, Mead et al. did not include specific estimates for some pathogens only occasionally transmitted by food nor did they develop specific estimates for known noninfectious foodborne agents (e.g., mushrooms, marine biotoxins, metals, or inorganic toxins).

Top of Page

Estimating foodborne illnesses due to unknown agents

Unknown agents fall into four general categories:

• Agents with insufficient data to estimate burden
• Known agents not yet recognized as causing foodborne illness
• Microbes, chemicals, or other substances known to be in food whose pathogenicity is unproven
• Agents not yet described

To estimate foodborne illnesses from unknown agents, Mead et al. used symptom-based data from three surveys to estimate the total number of acute gastroenteritis illnesses (AGI) and then subtracted the number of cases accounted for by known gastroenteritis pathogens. Finally, they multiplied this number by the proportion of illnesses attributed to foodborne transmission, using the relative frequency of foodborne transmission for known gastroenteritis pathogens as a guide (Figure 2).

Figure 2. Estimating foodborne illnesses due to unknown agents, 1999

Top of Page

Estimating hospitalizations and deaths from foodborne illnesses due to known pathogens

For each known pathogen with surveillance data available, Mead et al. multiplied the number of reported illnesses by the pathogen-specific hospitalization rate from surveillance data, surveys, and outbreak data. The product was doubled to correct for underreporting. They multiplied the adjusted number by the proportion of illnesses transmitted by food (described previously). Finally, the estimates for all pathogens were added for a total number of hospitalizations. Deaths were calculated in the same way (Figure 3).

Figure 3. Estimating hospitalizations and deaths* from foodborne illnesses due to known pathogens, 1999

^*Process for estimating hospitalizations was repeated for deaths.

Top of Page

Estimating hospitalizations and deaths from foodborne illnesses due to unknown agents

Mead et al. used the average hospitalization rate for acute gastroenteritis illness, derived from three national surveys, to estimate the total number of acute gastroenteritis hospitalizations. They then subtracted the estimated number of hospitalizations due to known gastroenteritis pathogens. Finally, they multiplied the adjusted number by the proportion of illnesses transmitted by food (described previously). Deaths were estimated in a similar way, with the death rate derived from multiple-cause-of-death data (from death certificates) and in-hospital death data (Figure 4).

Figure 4. Estimating hospitalizations and deaths* from foodborne illnesses due to unknown agents, 1999

^*Process for estimating hospitalizations was repeated for deaths using multiple-cause-of-death data and in-hospital death data.

Top of Page

Data Sources for the 1999 Estimates

Mead et al. used five general types of data sources to create the 1999 estimates.

Active surveillance (public health officials actively gather data from state and local health departments, laboratories, hospitals, etc.)

• Foodborne Diseases Active Surveillance Network (FoodNet)

Passive surveillance (public health officials rely on state and local health departments, laboratories, hospitals, etc. to report data to surveillance systems)

• National Notifiable Diseases Surveillance System (NNDSS)
• Public Health Laboratory Information System
• Gulf Coast States Vibrio Surveillance System

Outbreak surveillance

• Foodborne Disease Outbreak Surveillance System

Surveys

• FoodNet Population Survey
• FoodNet Laboratory Survey
• National Ambulatory Medical Care Survey (NAMCS)
• National Hospital Ambulatory Medical Care Survey (NHAMCS)
• National Hospital Discharge Survey (NHDS)
• National Health and Nutrition Survey (NHANES)
• Tecumseh study
• Cleveland study

Vital (Government) Statistics

• Multiple-cause-of-death data (from US death certificates)
• US Census

Tables 5a and 5b provide detailed information about these data sources. Please visit the links below to view these tables.

Table 5a. Data sources used to estimate illnesses, hospitalizations, and deaths due to known foodborne pathogens in the United States, 1999 (from Mead et al.)

Table 5a. Data sources used to estimate illnesses, hospitalizations, and deaths due to known foodborne pathogens in the United States, 1999 (from Mead et al.)

Data Source	Data	Pathogen(s)	Geographic Coverage	Time Frame	Adjustments
Foodborne Diseases Active Surveillance Network (FoodNet)	Number of laboratory-confirmed illnesses, proportion hospitalized, proportion who died	Campylobacter spp.; Cryptosporidium parvum; Cyclospora cayetanensis; Shiga toxin-producing Escherichia coli O157:H7; Listeria monocytogenes; non-typhoidal Salmonella; Shigella spp.; Vibrio, other spp.; and Yersinia enterocolitica	FoodNet sites1	1996–1997	Geographical coverage
Foodborne Disease Outbreak Surveillance System (FDOSS)	Number of foodborne outbreak-associated illnesses	Bacillus cereus; Clostridium perfringens; Staphylococcus aureus; and Streptococcus spp., Group A	United States	1988–1992 (1983–1987 for Staphylococcus aureus)	Underreporting
	Proportion hospitalized and proportion who died in foodborne outbreaks	Bacillus cereus; Brucella spp.; Clostridium botulinum;  Clostridium perfringens; Staphylococcus aureus; Streptococcus spp., Group A; and Trichinella spiralis	United  States	1988–1992	Underreporting
Gulf Coast States Vibrio Surveillance System	Number of case-patient reports, proportion hospitalized, proportion who died	Vibrio cholera and Vibrio Vulnificus	4 Gulf Coast states	1989	Underreporting;
National Electronic Telecommuni-cations System for Surveillance (NETSS)	Number of case-patient reports	Trichinella spiralis	United States (not all states report to NETSS; those who don’t report through Public Health Laboratories Information System)	1995–1998	Underreporting
National Health and Nutrition Examination Survey (NHANES)	Seroprevalence	Toxoplasma gondii	United States	1994	Rate of infection over time and percentage symptomatic
National Notifiable Diseases Surveillance System (NNDSS)2	Number of case-patient reports	Clostridum botulinum; Brucella spp.; Hepatitis A, Salmonella Typhi	United States	1992–1997	Underreporting
Various acute gastroenteritis data sources (see Table 5b)	Acute gastroenteritis illnesses, hospitalizations, and deaths	Norovirus	See Table 5b	See Table 5b	Fraction of acute gastroenteritis attributable to norovirus

¹ FoodNet co'nducts population-based active surveillance for selected foodborne infections in 8 sites with a total population catchment area of 20.5 million Americans.

² Passive surveillance data reported by physicians and laboratories

Table 5b. Data sources used to estimate illnesses, hospitalizations, and deaths due to acute gastroenteritis foodborne pathogens in the United States, 1999 (from Mead et al.)

Table 5b. Data sources used to estimate illnesses, hospitalizations, and deaths due to acute gastroenteritis foodborne pathogens in the United States, 1999 (from Mead et al.)

Data source	Data	Definition	Geographic coverage	Time frame
FoodNet  Population  Survey	Rate of illnesses	Average annual rate of diarrheal illness derived by multiplying the average monthly prevalence by 12, where an episode of acute diarrheal illess was defined as diarrhea (≥3 loose stools in 24 hours) lasting >1 day or resulting in restricted daily activities or vomiting in the past month.	FoodNet  sites1	1996–1997
Monto &  Koopman, 1980	Rate of illnesses	Vomiting and respiratory symptoms	850 households in Tecumseh, Michigan	1965–1971
Dingle et al., 1964	Rate of illnesses	Vomiting and respiratory symptoms	86 families in Cleveland, Ohio	1948­–1957
National  Hospital  Discharge  Survey  (NHDS)	Hospitalization  rate	Acute gastroenteritis hospitalizations were identified from discharges with one of the first three listed diagnoses classified by ICD-9-M diagnostic codes 001–008 (infectious gastroenteritis of known cause);  009 (infectious gastroenteritis); 558.9 (other and unspecified noninfectious gastroenteritis and colitis).	Nationally representative sample of discharge records from ~ 475 US hospitals	1992–1996
National Ambulatory Medical Care Survey (NAMCS); National Hospital Ambulatory Medical Care Survey (NHAMCS)	Hospitalization rate	Acute gastroenteritis hospitalizations were identified from patient visits to clinical settings, including physician offices, hospital emergency and outpatient departments with a diagnosis of infectious enteritis (ICD-9-CM diagnostic codes 001–008 (infectious gastroenteritis of known cause); 009 (infectious gastroenteritis); and 558.9 (other and unspecified noninfectious gastroenteritis and colitis) and reason for visit classification {RVC} codes 1595, 1530, 1540, for diarrhea, vomiting and gastroenteritis, respectively.	Nationally representative sample of US clinical settings	1996
Multiple-cause-of-death data from the National Vital Statistics System	Death rate	Acute gastroenteritis deaths were identified from the underlying or contributing cause of death classified by ICD-10 diagnostic codes A00.9–A08.5 (infectious gastroenteritis of known cause) A09 (diarrhea & gastroenteritis of presumed infectious origin); and  K52.9 (noninfectious gastroenteritis &colitis, unspecified)	United States	1998

¹FoodNet conducts population-based active surveillance for selected foodborne infections in 8 sites with a total population catchment area of 20.5 million Americans.

Top of Page