Multistate Surveillance for Food-Handling, Preparation, and Consumption Behaviors Associated with Foodborne Diseases: 1995 and 1996 BRFSS Food-Safety Questions

Samantha Yang, MPH (1); Marilyn G. Leff, MSPH (3); Doris McTague, MS (4); Kathryn A. Horvath (5); Jeanette Jackson-Thompson, PhD (6); Theophile Murayi, PhD (6); Georgette K. Boeselager, MS (7); Thomas A. Melnik, DrPH (8); Mark C. Gildemaster (9); David L. Ridings (10); Sean F. Altekruse, DVM, MPH (2); Frederick J. Angulo, DVM, PhD (1)

(1) Division of Bacterial and Mycotic Diseases

National Center for Infectious Diseases (2) Center for Food Safety and Applied Nutrition

U.S. Food and Drug Administration (3) Colorado Department of Public Health and Environment (4) Florida Department of Health (5) Indiana State Department of Health (6) Missouri Department of Health (7) Center for Health Statistics

New Jersey Department of Health and Senior Services (8) New York State Health Department (9) South Dakota State Department of Health (10) Tennessee Department of Health

Abstract

Problem/Condition: In 1995, CDC, the Food and Drug Administration (FDA), and several state health departments collaboratively developed questions regarding food safety. This set of questions was used to collect data about food-handling, preparation, and consumption behaviors that have been associated with foodborne diseases in adults. These data will help characterize persons at high risk for foodborne illness and assist in developing food-safety education strategies for consumers and foodhandlers that are intended to reduce foodborne illness.

Reporting Period Covered: January 1995-December 1996.

Description of System: Data were collected by using the 12 food-safety questions, which were administered with the 1995 Behavioral Risk Factor Surveillance Systems (BRFSS) in Colorado, Florida, Missouri, New York, and Tennessee, and the 1996 BRFSS in Indiana and New Jersey. In addition, data were collected in South Dakota from two of the standardized questions that deal with consumption of undercooked eggs and pink hamburgers. The BRFSS is a state-based system that surveys noninstitutionalized adults by telephone about their health behaviors and practices.

Results: This study included 19,356 completed questionnaires (2,461 in Colorado; 3,335 in Florida; 2,212 in Indiana; 1,572 in Missouri; 3,149 in New Jersey; 2,477 in New York; 2,110 in South Dakota; and 2,040 in Tennessee). During the previous 12 months, 50.2% of respondents reported eating undercooked eggs (95% confidence interval {CI} = 49.2-51.2); 23.8% reported eating home-canned vegetables (95% CI = 22.5-24.5); 19.7% reported eating pink hamburgers (95% CI = 18.9-20.5); 8.0% reported eating raw oysters (95% CI = 7.5-8.5); and 1.4% reported drinking raw milk (95% CI = 1.2-1.6). The prevalence of not washing hands with soap after handling raw meat or chicken and not washing a cutting board with soap or bleach after using it for cutting raw meat or chicken were 18.6% (95% CI = 17.8-19.4) and 19.5% (95% CI = 18.6-20.4), respectively. Less than half of respondents (45.4%, 95% CI = 44.2-46.6) reported seeing safe food-handling label information on raw meat products. In addition, among those persons who reported they remembered seeing the label information, 77.2% (95% CI = 76.0-78.4) remembered reading the label information, and 36.7% reported changing their meat and poultry preparation habits because of the labels (95% CI = 35.2-38.2). When population characteristics were considered in the analysis, all high-risk food-handling, preparation, and consumption behaviors were more prevalent in men than in women. Eating pink hamburgers during the previous 12 months was more commonly reported by whites (22.3%) than by blacks (6.5%). The prevalence of reported consumption of pink hamburgers during the previous 12 months decreased with age (18-29 years: 21.8%, 30-59 years: 21.9%, and 60-99 years: 13.2%); increased with education (less than grade 12: 12.0%, high school graduate: 16.5%, and any college education: 24.0%); and increased with income (less than $15,000: 11.8%, $15,000-$34,999: 17.6%, $35,000-$49,999: 22.0%, and greater than or equal to $50,000: 28.6%).

Interpretation: During 1995-1996, several high-risk food-handling, preparation, and consumption behaviors were common, and some were particular to specific population groups. Based on this analysis, interventions are needed to reduce the prevalence of these risky behaviors. All consumers and foodhandlers could benefit from food-safety education.

Actions Taken: Behavioral surveillance systems can provide data that identify persons or groups in which behaviors associated with foodborne diseases are more common and who are at higher risk for foodborne illness. State-specific data can assist in developing food-safety education programs and, if collected periodically, can be used to evaluate program effectiveness.

INTRODUCTION

Foodborne illness is a substantial problem in the United States. Each year, an estimated 6.5-33 million persons become ill from foodborne diseases, and up to 9,000 die (1). One strategy to reduce foodborne illness involves implementing food-safety education programs for consumers and foodhandlers. These education programs should include approaches that focus on reducing the prevalence of food-handling, preparation, and consumption behaviors associated with foodborne diseaseses. Safe food-handling, preparation, and consumption behaviors are important for persons who are particularly susceptible to foodborne illness, including pregnant women, young children, older adults, immunocompromised persons, and persons with reduced access to medical care (e.g., persons with low socioeconomic status). To aid in designing these programs, data are required that identify the population groups in which these risky behaviors are more common. Limited data have been collected to monitor these behaviors and to assess risk reduction secondary to educational campaigns. The Behavioral Risk Factor Surveillance System (BRFSS) can be used to provide such data. The BRFSS has been widely used to determine the prevalence of personal health behaviors -- including those among specific population groups -- related to morbidity and mortality from both chronic and acute disease (2).

METHODS

Sources of Data for Food-Handling, Preparation, and Consumption Behaviors

Data were collected through a standard set of 12 food-safety questions that were added to the 1995 BRFSS in Colorado, Florida, Missouri, New York, and Tennessee, and to the 1996 BRFSS in Indiana and New Jersey. South Dakota added two of the standardized questions (one regarding undercooked egg consumption and one regarding pink-hamburger consumption) to its 1996 BRFSS (Appendix). The BRFSS is a state-based system that surveys noninstitutionalized adults (greater than or equal to 18 years of age) by telephone about their health behaviors and practices, using random-digit-dialing techniques. The BRFSS uses either a three-stage cluster sampling design based on the Waksburg Method or the disproportionate stratified random sampling method (3,4).

The set of food-safety questions included a) two questions about actions taken after handling raw meat or chicken; b) six questions about consumption of specific high-risk food items (i.e., home-canned vegetables, pink hamburgers, undercooked eggs, raw oysters, and raw milk); c) three questions about respondents' awareness of safe food-handling labels on raw meat products and any changes in their raw meat or poultry preparation methods after reading these labels; and d) one question about the occurrence of diarrhea. (Data collected for the question regarding diarrhea will not be discussed in this summary.)

The BRFSS coordinator for each participating state sent data to CDC for review and analysis. Descriptive analyses were performed using SAS * (5) and SUDAAN ** (6), and a weighting factor was assigned to each survey respondent. This weighting factor adjusted for the respondent's probability of selection and age-, race-, and sex-specific population from the 1990 census data and was used to estimate the prevalence of high-risk food-handling, preparation, and consumption behaviors for each state's population (7,8).

Definitions

For analysis of the two questions about actions taken after handling raw meat or chicken, responses were categorized as follows: a) persons who usually "continue cooking" or "rinse and/or wipe hands then continuing cooking" after handling raw meat or chicken were classified as persons who usually do not wash hands with soap after handling raw meat or chicken; b) persons who usually "continue using a cutting surface as is" or "rinse and/or wipe cutting surface, then continue cooking" after using the cutting board, counter top, or other surface for cutting raw meat or chicken were classified as persons who usually do not wash cutting boards with soap or bleach after using it to cut raw meat or chicken. For analysis of the question about pink-hamburger consumption during the previous 12 months, persons who responded that they had "never" eaten a hamburger during the previous 12 months also were classified as not having eaten a pink hamburger during the previous 12 months.

The variable for residential area was created by matching the county of the respondent's residence with Economic Research Service (ERS) rural-urban continuum codes (i.e., Beale codes). These ERS rural-urban continuum codes divide counties in the United States into nine groups. For our survey data, central counties of a metropolitan area (counties with at least 50% of the population of a central city), were classified as urban areas. Other counties in metropolitan areas with populations of 1 million persons also were classified as urban areas (1993 Beale codes = 0, 1, 2, and 3). Fringe counties of a metropolitan area (counties where at least 50% of the employed workers residing in the county commute to the central county/counties) were classified as suburban or small town areas. Counties with populations of greater than or equal to 20,000 persons that were not adjacent to metropolitan areas also were classified as suburban or small town areas (1993 Beale codes = 4, 5, and 6). The remaining counties with populations of 2,500-19,999 not adjacent to metropolitan areas or completely rural counties (total population less than 2,500 persons) were classified as rural areas (1993 Beale codes = 7, 8, and 9) (9,10).

In general, participant responses were excluded from analysis if a participant responded "don't know," "not sure," or refused to answer the question. For analysis of hand-washing, responses were excluded from analysis if participants reported they had "other" hand-washing techniques not provided on the questionnaire or they did not handle raw meat or chicken. Similarly, for analysis of the cutting board question, responses were excluded from analysis if participants had "other" cutting board washing techniques not provided on the questionnaire or they did not cut raw meat or chicken.

RESULTS

In this study, 19,356 interviews were completed (2,461 in Colorado; 3,335 in Florida; 2,212 in Indiana; 1,572 in Missouri; 3,149 in New Jersey; 2,477 in New York; 2,110 in South Dakota; and 2,040 in Tennessee) (Table_1). The results of this surveillance system are presented for each of the food-safety questions by state (Table_2) and by state and population characteristics (Table_3, Table_4, Table_5, Table_6, Table_7, Table_8, Table_9, Table_10, Table_11, Table_12, Table_13). During the previous 12 months, 50.2% of respondents reported eating undercooked eggs (95% confidence interval {CI} = 49.2-51.2); 23.8% reported eating home-canned vegetables (95% CI = 23.0-24.6); 19.7% reported eating pink hamburgers (95% CI = 18.9-20.5); 8.0% reported eating raw oysters (95% CI = 7.5-8.5); and 1.4% reported drinking raw milk (95% CI = 1.2-1.6). The prevalence of not washing hands with soap after handling raw meat or chicken and not washing a cutting board with soap or bleach after using it for raw meat or chicken were 18.6% (95% CI = 17.8-19.4) and 19.5% (95% CI = 18.6-20.4), respectively. In addition, 45.4% (95% CI = 44.2-46.6) of respondents remembered seeing safe food-handling label information on packages of meat and poultry, and 36.7% (95% CI = 33.8-39.6) of those who remembered seeing the label reported that the label information changed their meat and poultry preparation methods.

The prevalence of several behaviors associated with foodborne diseases varied by state. For example, the prevalence of reported consumption of pink hamburgers and undercooked eggs during the previous 12 months and the prevalence of not washing hands with soap or not washing the cutting board after contact with raw meat or chicken were higher in Colorado respondents than in respondents from the other six states in which this question was asked on their BRFSS. During the previous 12 months, a higher percentage of participants from Tennessee than from any other state reported eating home-canned vegetables, and consumption of raw oysters was reported more commonly in Florida than in any other state. The proportion of respondents who remembered seeing the safe food-handling label information was lowest in New York (36.4%, 95% CI = 34.2-38.6) and highest in Missouri (54.6%, 95% CI = 51.6-57.6). Of the respondents who remembered seeing the label information, the proportion who remembered reading the label was lower in Indiana (72.9%) than in any other state (Colorado: 73.0%; Missouri: 74.5%; Florida: 76.0%; New Jersey: 77.3%; New York: 80.4%; and Tennessee: 83.0%).

Some high-risk food-handling, preparation, and consumption behaviors were more common in specific population groups (Table_3, Table_4, Table_5, Table_6, Table_7, Table_8, Table_9, Table_10, Table_11, Table_12, Table_13). For example, the prevalence of reported consumption of pink hamburgers was higher among men (24.3%) than among women (15.6%) and among whites (22.3%) than among any other race (Asians/Pacific Islanders: 13.7%; Hispanics: 13.5%; and blacks: 6.5%). The prevalence of several food-consumption behaviors associated with foodborne diseases decreased with age, increased with education, and increased with yearly salary. For example, the prevalence of consumption of pink hamburgers decreased with age (18-29 years: 21.8%, 30-59 years: 21.9%, and 60-99 years: 13.2%); increased with education (less than grade 12: 12.0%, high school graduate: 16.5, and any college education: 24.0%); and increased with yearly salary (less than $15,000: 11.8%, $15,000-$34,999: 17.6%, $35,000-$49,999: 22.0%, and greater than or equal to $50,000: 28.6%). Similar patterns with age, education, and income were found for food-handling and preparation behaviors associated with foodborne diseases. In addition, awareness of safe food-handling label information was more common in certain population groups. Of respondents who remembered seeing label information, the proportion who remembered reading label information was higher in women (82.4%) than in men (68.7%) and higher in whites (78.7%) than in other races (Asians/Pacific Islanders: 74.3%; blacks: 74.1%; and Hispanics: 66.7%).

DISCUSSION

General Interpretation of Surveillance Data for Food-Handling, Preparation, and Consumption Behaviors Associated with Foodborne Diseases

The survey data described in this report indicate that several behaviors associated with foodborne diseases were common in 1995 and 1996. For example, approximately 50% of respondents reported eating undercooked eggs during the previous 12 months, and 20% reported not washing the cutting board with soap or bleach after using it for cutting raw meat or chicken. Prevalence estimates in previous studies differ from those in this study. In a nationwide survey conducted in 1993, the estimated prevalence of not washing hands after handling raw meat or chicken was higher than that from our survey (37% versus 19%, respectively) (11). In this nationwide survey conducted in 1993, 23% of survey respondents reported serving pink hamburgers in their homes (12). In 1986, a study in Oregon indicated that 23% of home food preparers reported serving rare hamburger (13), and in 1991, a study in Nebraska indicated that 42% of survey respondents did not prepare hamburgers to a well-done stage (14). In the survey described in this report, the prevalence of reported consumption of pink hamburgers was 19.6% during the previous 12 months. A study based on the 1992 California BRFSS indicated that 23% of respondents reported eating raw shellfish (15); in comparison, the survey described in this report indicated that in 1995 and in 1996, 8% of respondents reported eating raw oysters during the previous 12 months. Differences in survey design and methodology might explain some of the differences between prevalence estimates in previous studies and those in this survey. However, our survey estimates demonstrate that although some high-risk food-handling and consumption behaviors were still common in 1995 and 1996, they might have been improving.

Prevalence of high-risk behaviors varied among the states. Regional differences in high-risk food-handling, preparation, and consumption behaviors might result from socioeconomic or cultural differences and variations between state laws enacted to discourage risky behaviors.

The findings of this survey indicate that high-risk food-handling, preparation, and consumption behaviors were more common in certain population groups. All behaviors associated with foodborne diseases were more prevalent in men than in women. Other studies support this finding (11,14). In a 1991 study in Nebraska in which safer behaviors were assigned higher scores, men demonstrated lower food-handling scores than women (14). The 1993 FDA Health and Diet Survey indicated that men were less likely than women to wash their hands after handling raw meat or poultry (53% versus 75%) (11). In our survey, prevalence of high-risk food-handling, preparation, and consumption behaviors also varied by age group, race/ethnicity, socioeconomic status, and residential area. Results from previous studies agree with our survey in that the prevalence of high-risk food-handling, preparation, and consumption practices (except eating undercooked eggs) increased as age decreased (11,12,15). Similar to our survey, a previous study found that consumption of pink hamburgers is more common in whites than in any other racial/ethnic group (12).

In our survey, a direct relation was observed between education level and the frequencies of some high-risk food-handling, preparation, and consumption behaviors (e.g., consumption of pink hamburgers or raw oysters and failure to wash hands or cutting boards after contact with raw meat or chicken). These findings contrast with findings that persons with an education beyond high school are more likely than other persons to pursue other health-promoting behaviors (e.g., using seat belts, abstaining from cigarette smoking, and engaging in regular exercise) (12). This difference in results suggests that some highly educated persons might not know or choose to ignore the hazards associated with behaviors that have been related to foodborne diseases. Despite knowing the hazards associated with high-risk behaviors, highly educated persons might continue to perform such behaviors because of cultural influences or social norms. Decisions about behavior frequently are guided by risk perception rather than risk awareness (16). Factors that can influence risk perception include media coverage, opinions of scientific experts and peer groups, perceived control over risk, and knowledge about a potential hazard (16).

Persons can be aware of risks but choose to continue such behaviors if they believe they or others can control the risk. A food-safety survey of 2,197 homemakers concluded that homemakers rely on government inspection for the prevention of bacterial contamination of raw meat and poultry. Perceiving that the hazards in raw meat and poultry were controlled, many homemakers in the study underrated their responsibility for safe food-handling and preparation practices and were not aware of the sources of pathogens in the environment and in the human body (14). Furthermore, persons might believe that, although negative events occur, such events are relatively unlikely to harm them personally (16). In a 1991 national consumer survey, participants thought food-safety problems were most likely to occur at food manufacturing facilities (34%), followed by restaurants (32%), and homes (16%) (17).

Limitations

The findings in this report are subject to at least two limitations. First, because the analysis is based on self-reported data, the findings might be subject to reporting bias: respondents might have answered questions according to what they perceive as being the correct answer rather than what they actually practice. Second, this analysis did not address possible confounders, such as socioeconomic status (e.g., education and annual income), of the relation between other population characteristics and high-risk food-handling, preparation, and consumption behaviors. Further analysis that adjusts for socioeconomic status will be conducted.

CONCLUSIONS

This survey found that the prevalence of behaviors associated with foodborne diseases vary by sex, age, race/ethnicity, education and income. In addition, this survey presents data indicating that persons who might be more susceptible to foodborne illness were more likely to have safer food-handling, preparation, or consumption practices than those who usually are perceived to be less susceptible to foodborne illness.

The results of this survey should be used in conjunction with results from studies that evaluate risk perception and knowledge of food-safety issues to develop foodborne disease intervention and prevention strategies. These strategies should be aimed at persons who are more susceptible to foodborne illness, more likely to perform behaviors associated with foodborne diseases, more likely to perceive personal invulnerability to foodborne illness, and more likely to have little or no knowledge of food safety. Future surveys should include questions that assess consumers' and foodhandlers' perceptions of risk, food-preparation experience, and knowledge of food safety.

All consumers could benefit from food-safety education. To effectively decrease foodborne illness, strategies should reduce the prevalence of behaviors associated with foodborne diseases, increase consumers' awareness of risks from foodborne illness, and motivate them to change their high-risk behaviors.

Behavioral surveillance systems can provide data to assist in identifying persons in which behaviors associated with foodborne diseases are more common. Since 1996, some states have voluntarily added all or some of the standard 12 food-safety questions to their BRFSS. For example, for the first time, Arizona added the full set of 12 food-safety questions; Idaho added the question about hand washing; and Vermont added the question about consumption of raw milk to their 1997 BRFSS. In 1997, New York added to its BRFSS the same questions about hamburger and pink-hamburger consumption that were asked during administration of its 1995 BRFSS to monitor the prevalence of hamburger and pink-hamburger consumption and evaluate the effectiveness of its slogan, "It's clear, a safer hamburger is cooked brown in the middle." That these states and others will add food-safety questions to their BRFSS in future years to monitor trends in high-risk consumer behaviors and assess the effectiveness of food-safety education strategies is anticipated.

SAS, a computer software for data access, management, analysis, and presentation; for additional information, contact SAS Institute, Inc., SAS Campus Drive, Cary, NC 27513.

** SUrvey DAta ANalysis, a computer software for the statistical analysis of correlated data; for additional information, contact Research Triangle Institute, 3040 Cornwallis Road, Research Triangle Park, NC 27709 (Telephone: 919-541-6000).