A continuous telephone survey conducted by state health departments in collaboration with CDC, the BRFSS provides state-specific estimates of behaviors that relate to the leading causes of death among adults in the United States. In each state, random-digit dialing is used to select an independent probability sample of residents aged 18 years or older, and trained interviewers administer identical core questionnaires. Data are weighted by race/ethnicity, age, and sex to reflect the respondents' probability of selection and the state's population. A detailed description of BRFSS methods is available elsewhere (9
We used data from all 50 U.S. states and the District of Columbia that participated in the BRFSS for the years from 1994 through 2005 in which the fruit and vegetable module was part of the core questionnaire: 1994, 1996, 1998, 2000, 2002, 2003, and 2005. We aggregated state estimates to analyze overall and sex-specific consumption over time of fruit juices, whole fruits, green salad, carrots, nonfried potatoes, and "all other" vegetables. We also used BRFSS data to calculate body mass index (BMI), and to evaluate consumption by participation in leisure-time activity and by smoking status.
The BRFSS module on fruits and vegetables has six questions that have remained the same over time. Interviewers begin the module with the following statement: These next questions are about the foods you usually eat or drink. Please tell me how often you eat or drink each one: for example, twice a week, three times a month, and so forth. Respondents are then asked the following questions: 1) How often do you drink fruit juices such as orange, grapefruit, or tomato? 2) Not counting juice, how often do you eat fruit? 3) How often do you eat green salad? 4) How often do you eat potatoes, not including french fries, fried potatoes, or potato chips? 5) How often do you eat carrots? and 6) Not counting carrots, potatoes, or salad, how many servings of vegetables do you usually eat? Participants are not given a definition of serving size. At the end of the fruits and vegetables interview, respondents are asked to report their weight and their height without shoes.
To create an index of fruit and vegetable consumption, we summed the daily frequency of consumption of food items mentioned in the module (11
). We calculated total daily fruit consumption from responses to questions 1 and 2 and total daily vegetable consumption from responses to questions 3–6. To calculate consumption in times per day, we divided weekly frequencies by 7, monthly frequencies by 30, and yearly frequencies by 365. For consistency with past analyses, the answer to question 6 was treated as times per day. We calculated BMI as self-reported weight in kilograms divided by height in meters squared and grouped respondents into three categories: normal weight (BMI <25), overweight (BMI 25 to <30), and obese (BMI ≥30) (12
About leisure-time activity, BRFSS respondents are asked: During the past month, other than during your regular job, did you participate in any physical activities or exercises such as running, calisthenics, golf, gardening, or walking for exercise? We classified smoking status as current smoker, former smoker, and nonsmoker (never smoked). We used the U.S. Census Bureau definition for regions of the United States: Northeast, South, Midwest, and West (13
A total of 1,394,471 people completed the interviews. We excluded people not reporting race/ethnicity (n = 11,217), education level (n = 2898), smoking status (n = 4138), leisure-time physical activity (n = 5989), and weight or height (n = 58,781) and people who did not answer more than one question in the fruit and vegetable module (n = 103,808) or who reported eating 25 or more fruits and vegetables per day (n = 229). Because some individuals were missing more than one covariate, we had 1,227,969 people in our final sample: 95,571 from 1994; 107,522 from 1996; 130,086 from 1998; 157,179 from 2000; 211,507 from 2002; 224,807 from 2003; and 301,297 from 2005.
To allow comparison with earlier dietary recommendations (14
) and previous reports of BRFSS data (7
), we calculated the percentage of respondents who ate five or more fruits or vegetables or both per day and the mean daily consumption. Because the frequency of consumption was skewed, we calculated geometric mean daily consumption to provide a valid measure of center. Participants who reported not eating any fruits or vegetables were assigned 0.1 times per day to allow for their inclusion in the analysis.
We used state census population estimates for the survey years to weight data for age, race/ethnicity, and sex. Because of the large sample size, we set statistical significance at P
< .01. To account for the complex sampling design and to report weighted findings, we used SAS 9.1 (SAS Institute, Inc, Cary, North Carolina) and SAS-Callable SUDAAN 9.0 (RTI International, Research Triangle Park, North Carolina). Changes in the geometric mean frequency of daily consumption and in the percentage of respondents eating fruits and vegetables five or more times daily from 1994 through 2005 were standardized to the sex, age, and racial/ethnic distribution of the 2000 BRFSS population. To analyze temporal changes in the geometric mean frequency, we used multivariable regression that controlled for year, age, sex, and race/ethnicity for both linear and quadratic time effects (16
). Quadratic trends can be statistically significant but nonlinear over time because of a leveling off or significant change in direction, whereas trends with significant linear and quadratic components demonstrate nonlinear variation along with an overall increase or decrease over time.